Chemical

Reaction
Engineering
Third Edition

Octave Levenspiel
Department of Chemical Engineering
Oregon State University

John Wiley & Sons

New York Chichester Weinheim Brisbane Singapore Toronto
ACQUISITIONS EDITOR Wayne Anderson
MARKETING MANAGER Katherine Hepburn
PRODUCTION EDITOR Ken Santor
SENIOR DESIGNER Kevin Murphy
ILLUSTRATION COORDINATOR Jaime Perea
ILLUSTRATION Wellington Studios
COVER DESIGN Bekki Levien

This book was set in Times Roman by Bi-Comp Inc. and printed and bound by the
Hamilton Printing Company. The cover was printed by Phoenix Color Corporation.

This book is printed on acid-free paper.

The paper in this book was manufactured by a mill whose forest management programs
include sustained yield harvesting of its timberlands. Sustained yield harvesting principles
ensure that the numbers of trees cut each year does not exceed the amount of new growth.

Copyright O 1999 John Wiley & Sons, Inc. All rights reserved.
No part of this publication may be reproduced, stored in a retrieval system or transmitted
in any form or by any means, electronic, mechanical, photocopying, recording, scanning
or otherwise, except as permitted under Sections 107 or 108 of the 1976 United States
Copyright Act, without either the prior written permission of the Publisher, or
authorization through payment of the appropriate per-copy fee to the Copyright
Clearance Center, 222 Rosewood Drive, Danvers, MA 01923, (508) 750-8400, fax
(508) 750-4470. Requests to the Publisher for permission should be addressed to the
Permissions Department, John Wiley & Sons, Inc., 605 Third Avenue, New York, NY
10158-0012,(212) 850-6011, fax (212) 850-6008, E-Mail: PERMREQ@[Link].

Library of Congress Cataloging-in-Publication Data:

Levenspiel, Octave.
Chemical reaction engineering 1 Octave Levenspiel. - 3rd ed.
p. cm.
Includes index.
ISBN 0-471-25424-X(cloth : alk. paper)
1. Chemical reactors. I. Title.
TP157.L4 1999
6601.281-dc21
97-46872
CIP
Printed in the United States of America
Preface

Chemical reaction engineering is that engineering activity concerned with the

exploitation of chemical reactions on a commercial scale. Its goal is the successful
design and operation of chemical reactors, and probably more than any other
activity it sets chemical engineering apart as a distinct branch of the engi-
neering profession.
In a typical situation the engineer is faced with a host of questions: what
information is needed to attack a problem, how best to obtain it, and then how
to select a reasonable design from the many available alternatives? The purpose
of this book is to teach how to answer these questions reliably and wisely. To
do this I emphasize qualitative arguments, simple design methods, graphical
procedures, and frequent comparison of capabilities of the major reactor types.
This approach should help develop a strong intuitive sense for good design which
can then guide and reinforce the formal methods.
This is a teaching book; thus, simple ideas are treated first, and are then
extended to the more complex. Also, emphasis is placed throughout on the
development of a common design strategy for all systems, homogeneous and
heterogeneous.
This is an introductory book. The pace is leisurely, and where needed, time is
taken to consider why certain assumptions are made, to discuss why an alternative
approach is not used, and to indicate the limitations of the treatment when
applied to real situations. Although the mathematical level is not particularly
difficult (elementary calculus and the linear first-order differential equation is
all that is needed), this does not mean that the ideas and concepts being taught
are particularly simple. To develop new ways of thinking and new intuitions is
not easy.
Regarding this new edition: first of all I should say that in spirit it follows the
earlier ones, and I try to keep things simple. In fact, I have removed material
from here and there that I felt more properly belonged in advanced books.
But I have added a number of new topics-biochemical systems, reactors with
fluidized solids, gadliquid reactors, and more on nonideal flow. The reason for
this is my feeling that students should at least be introduced to these subjects so
that they will have an idea of how to approach problems in these important areas.
iii
i~ Preface

I feel that problem-solving-the process of applying concepts to new situa-

tions-is essential to learning. Consequently this edition includes over 80 illustra-
tive examples and over 400 problems (75% new) to help the student learn and
understand the concepts being taught.
This new edition is divided into five parts. For the first undergraduate course,
I would suggest covering Part 1 (go through Chapters 1 and 2 quickly-don't
dawdle there), and if extra time is available, go on to whatever chapters in Parts
2 to 5 that are of interest. For me, these would be catalytic systems (just Chapter
18) and a bit on nonideal flow (Chapters 11 and 12).
For the graduate or second course the material in Parts 2 to 5 should be suitable.
Finally, I'd like to acknowledge Professors Keith Levien, Julio Ottino, and
Richard Turton, and Dr. Amos Avidan, who have made useful and helpful
comments. Also, my grateful thanks go to Pam Wegner and Peggy Blair, who
typed and retyped-probably what seemed like ad infiniturn-to get this manu-
script ready for the publisher.
And to you, the reader, if you find errors-no, when you find errors-or
sections of this book that are unclear, please let me know.

Octave Levenspiel
Chemical Engineering Department
Oregon State University
Corvallis, OR, 97331
Fax: (541) 737-4600
Contents

Notation /xi
Chapter 1
Overview of Chemical Reaction Engineering I1

Part I
Homogeneous Reactions in Ideal
Reactors I11

Chapter 2
Kinetics of Homogeneous Reactions I13
2.1 Concentration-Dependent Term of a Rate Equation I14
2.2 Temperature-Dependent Term of a Rate Equation I27
2.3 Searching for a Mechanism 129
2.4 Predictability of Reaction Rate from Theory 132

Chapter 3
Interpretation of Batch Reactor Data I38
3.1 Constant-volume Batch Reactor 139
3.2 Varying-volume Batch Reactor 167
3.3 Temperature and Reaction Rate 172
3.4 The Search for a Rate Equation I75

Chapter 4
Introduction to Reactor Design 183
vi Contents

Chapter 5
Ideal Reactors for a Single Reaction 190
5.1 Ideal Batch Reactors I91
52. Steady-State Mixed Flow Reactors 194
5.3 Steady-State Plug Flow Reactors 1101

Chapter 6
Design for Single Reactions I120
6.1 Size Comparison of Single Reactors 1121
6.2 Multiple-Reactor Systems 1124
6.3 Recycle Reactor 1136
6.4 Autocatalytic Reactions 1140

Chapter 7
Design for Parallel Reactions 1152
Chapter 8
Potpourri of Multiple Reactions 1170
8.1 Irreversible First-Order Reactions in Series 1170
8.2 First-Order Followed by Zero-Order Reaction 1178
8.3 Zero-Order Followed by First-Order Reaction 1179
8.4 Successive Irreversible Reactions of Different Orders 1180
8.5 Reversible Reactions 1181
8.6 Irreversible Series-Parallel Reactions 1181
8.7 The Denbigh Reaction and its Special Cases 1194

Chapter 9
Temperature and Pressure Effects 1207
9.1 Single Reactions 1207
9.2 Multiple Reactions 1235

Chapter 10
Choosing the Right Kind of Reactor 1240

Part I1
Flow Patterns, Contacting, and Non-Ideal
Flow I255

Chapter 11
Basics of Non-Ideal Flow 1257
11.1 E, the Age Distribution of Fluid, the RTD 1260
11.2 Conversion in Non-Ideal Flow Reactors 1273
 Contents Yii

Chapter 12
Compartment Models 1283

Chapter 13
The Dispersion Model 1293
13.1 Axial Dispersion 1293
13.2 Correlations for Axial Dispersion 1309
13.3 Chemical Reaction and Dispersion 1312

Chapter 14
The Tanks-in-Series Model 1321
14.1 Pulse Response Experiments and the RTD 1321
14.2 Chemical Conversion 1328

Chapter 15
The Convection Model for Laminar Flow 1339
15.1 The Convection Model and its RTD 1339
15.2 Chemical Conversion in Laminar Flow Reactors 1345

Chapter 16
Earliness of Mixing, Segregation and RTD 1350
16.1 Self-mixing of a Single Fluid 1350
16.2 Mixing of Two Miscible Fluids 1361

Part 111
Reactions Catalyzed by Solids 1367

Chapter 17
Heterogeneous Reactions - Introduction 1369

Chapter 18
Solid Catalyzed Reactions 1376
18.1 The Rate Equation for Surface Kinetics 1379
18.2 Pore Diffusion Resistance Combined with Surface Kinetics 1381
18.3 Porous Catalyst Particles I385
18.4 Heat Effects During Reaction 1391
18.5 Performance Equations for Reactors Containing Porous Catalyst
Particles 1393
18.6 Experimental Methods for Finding Rates 1396
18.7 Product Distribution in Multiple Reactions 1402
viii Contents

Chapter 19
The Packed Bed Catalytic Reactor 1427
Chapter 20
Reactors with Suspended Solid Catalyst,
Fluidized Reactors of Various Types 1447
20.1 Background Information About Suspended Solids Reactors 1447
20.2 The Bubbling Fluidized Bed-BFB 1451
20.3 The K-L Model for BFB 1445
20.4 The Circulating Fluidized Bed-CFB 1465
20.5 The Jet Impact Reactor 1470
Chapter 21
Deactivating Catalysts 1473
21.1 Mechanisms of Catalyst Deactivation 1474
21.2 The Rate and Performance Equations 1475
21.3 Design 1489
Chapter 22
GIL Reactions on Solid Catalyst: Trickle Beds, Slurry
Reactors, Three-Phase Fluidized Beds 1500
22.1 The General Rate Equation 1500
22.2 Performanc Equations for an Excess of B 1503
22.3 Performance Equations for an Excess of A 1509
22.4 Which Kind of Contactor to Use 1509
22.5 Applications 1510

Part IV
Non-Catalytic Systems I521

Chapter 23
Fluid-Fluid Reactions: Kinetics I523
23.1 The Rate Equation 1524
Chapter 24
Fluid-Fluid Reactors: Design 1.540
24.1 Straight Mass Transfer 1543
24.2 Mass Transfer Plus Not Very Slow Reaction 1546
Chapter 25
Fluid-Particle Reactions: Kinetics 1566
25.1 Selection of a Model 1568
25.2 Shrinking Core Model for Spherical Particles of Unchanging
Size 1570
 Contents ix

25.3 Rate of Reaction for Shrinking Spherical Particles 1577

25.4 Extensions 1579
25.5 Determination of the Rate-Controlling Step 1582

Chapter 26
Fluid-Particle Reactors: Design 1589

Part V
Biochemical Reaction Systems I609

Chapter 27
Enzyme Fermentation 1611
27.1 Michaelis-Menten Kinetics (M-M kinetics) 1612
27.2 Inhibition by a Foreign Substance-Competitive and
Noncompetitive Inhibition 1616

Chapter 28
Microbial Fermentation-Introduction and Overall
Picture 1623
Chapter 29
Substrate-Limiting Microbial Fermentation 1630
29.1 Batch (or Plug Flow) Fermentors 1630
29.2 Mixed Flow Fermentors 1633
29.3 Optimum Operations of Fermentors 1636

Chapter 30
Product-Limiting Microbial Fermentation 1645
30.1 Batch or Plus Flow Fermentors for n = 1 I646
30.2 Mixed Flow Fermentors for n = 1 1647

Appendix 1655
Name Index 1662
Subject Index 1665
Notation

Symbols and constants which are defined and used locally are not included here.
SI units are given to show the dimensions of the symbols.

interfacial area per unit volume of tower (m2/m3),see

Chapter 23
activity of a catalyst, see Eq. 21.4
a , b ,..., 7,s,... stoichiometric coefficients for reacting substances A,
B, ..., R, s, .,.
A cross sectional area of a reactor (m2), see Chapter 20
A, B, ... reactants
A, B, C, D, Geldart classification of particles, see Chapter 20
C concentration (mol/m3)
CM Monod constant (mol/m3),see Chapters 28-30; or Michae-
lis constant (mol/m3), see Chapter 27
c~ heat capacity (J/mol.K)
CLA,C ~ A mean specific heat of feed, and of completely converted
product stream, per mole of key entering reactant (J/
mol A + all else with it)
d diameter (m)
d order of deactivation, see Chapter 22
dimensionless particle diameter, see Eq. 20.1
axial dispersion coefficient for flowing fluid (m2/s), see
Chapter 13
molecular diffusion coefficient (m2/s)
ge effective diffusion coefficient in porous structures (m3/m
solids)
ei(x) an exponential integral, see Table 16.1
xi
~ i iNotation
enhancement factor for mass transfer with reaction, see
Eq. 23.6
concentration of enzyme (mol or gm/m3),see Chapter 27
dimensionless output to a pulse input, the exit age distribu-
tion function (s-l), see Chapter 11
E, E*, E** RTD for convective flow, see Chapter 15
Eoo, Eoc? ECO, Ecc RTD for the dispersion model, see Chapter 13
Ei(x) an exponential integral, see Table 16.1
8 effectiveness factor (-), see Chapter 18
f fraction of solids (m3 solid/m3vessel), see Chapter 20
A volume fraction of phase i (-), see Chapter 22
F feed rate (molls or kgls)
F dimensionless output to a step input (-), see Fig. 11.12
G* free energy (Jlmol A)
h heat transfer coefficient (W/m2.K),see Chapter 18
h height of absorption column (m), see Chapter 24
H height of fluidized reactor (m), see Chapter 20
H phase distribution coefficient or Henry's law constant; for
gas phase systems H = plC (Pa.m3/mol),see Chapter 23
mean enthalpy of the flowing stream per mole of A flowing
(Jlmol A + all else with it), see Chapter 9
enthalpy of unreacted feed stream, and of completely con-
verted product stream, per mole of A (Jlmol A + all
else), see Chapter 19
heat of reaction at temperature T for the stoichiometry
as written (J)
heat or enthalpy change of reaction, of formation, and of
combustion (J or Jlmol)
k reaction rate constant (mol/m3)'-" s-l, see Eq. 2.2
k, kt, II', k , k"" reaction rate constants based on r, r', J', J", J"', see Eqs.
18.14 to 18.18
rate constant for the deactivation of catalyst, see Chap-
ter 21
effective thermal conductivity (Wlrn-K), see Chapter 18
mass transfer coefficient of the gas film (mol/[Link].s),see
Eq. 23.2
mass transfer coefficient of the liquid film (m3 liquid/m2
surface.^), see Eq. 23.3
equilibrium constant of a reaction for the stoichiometry
as written (-), see Chapter 9
 Notation xiii
bubble-cloud interchange coefficient in fluidized beds
(s-l), see Eq. 20.13
cloud-emulsion interchange coefficient in fluidized beds
(s-I), see Eq. 20.14
characteristic size of a porous catalyst particle (m), see
Eq. 18.13
half thickness of a flat plate particle (m), see Table 25.1
mass flow rate (kgls), see Eq. 11.6
mass (kg), see Chapter 11
order of reaction, see Eq. 2.2
number of equal-size mixed flow reactors in series, see
Chapter 6
moles of component A
partial pressure of component A (Pa)
partial pressure of A in gas which would be in equilibrium
with CAin the liquid; hence p z = HACA(Pa)
Q heat duty (J/s = W)
r, r', J', J", J"' rate of reaction, an intensive measure, see Eqs. 1.2 to 1.6
rc radius of unreacted core (m), see Chapter 25
R radius of particle (m), see Chapter 25
R, S, ... products of reaction
R ideal gas law constant,
= 8.314 J1mol.K
= 1.987 cal1mol.K
= 0.08206 [Link]/mol.K
recycle ratio, see Eq. 6.15
space velocity (s-l); see Eqs. 5.7 and 5.8
surface (m2)
time (s)
= Vlv, reactor holding time or mean residence time of
fluid in a flow reactor (s), see Eq. 5.24
temperature (K or "C)
dimensionless velocity, see Eq. 20.2
carrier or inert component in a phase, see Chapter 24
volumetric flow rate (m3/s)
volume (m3)
mass of solids in the reactor (kg)
fraction of A converted, the conversion (-)
X ~ V Notation

xA moles Almoles inert in the liquid (-), see Chapter 24

yA moles Aimoles inert in the gas (-), see Chapter 24

Greek symbols
a m3 wake/m3 bubble, see Eq. 20.9
S volume fraction of bubbles in a BFB
6 Dirac delta function, an ideal pulse occurring at time t =
0 (s-I), see Eq. 11.14
a(t - to) Dirac delta function occurring at time to (s-l)
&A expansion factor, fractional volume change on complete
conversion of A, see Eq. 3.64
E void fraction in a gas-solid system, see Chapter 20
8 effectiveness factor, see Eq. 18.11
8 = tl? dimensionless time units (-), see Eq. 11.5
K"' overall reaction rate constant in BFB (m3 solid/m3gases),
see Chapter 20
viscosity of fluid (kg1m.s)
mean of a tracer output curve, (s), see Chapter 15
total pressure (Pa)
density or molar density (kg/m3 or mol/m3)
variance of a tracer curve or distribution function (s2),see
Eq. 13.2
V/v = CAoV/FAo, space-time (s), see Eqs. 5.6 and 5.8
time for complete conversion of a reactant particle to
product (s)
= CAoW/FAo, weight-time, (kg.s/m3), see Eq. 15.23
TI, ?",P, T'"' various measures of reactor performance, see Eqs.
18.42, 18.43
@ overall fractional yield, see Eq. 7.8
4 sphericity, see Eq. 20.6
P instantaneous fractional yield, see Eq. 7.7
p(MIN) = @ instantaneous fractional yield of M with respect to N, or
moles M formedlmol N formed or reacted away, see
Chapter 7

Symbols and abbreviations

BFB bubbling fluidized bed, see Chapter 20
BR batch reactor, see Chapters 3 and 5
CFB circulating fluidized bed, see Chapter 20
FF fast fluidized bed, see Chapter 20
 Notation XV

LFR laminar flow reactor, see Chapter 15

MFR mixed flow reactor, see Chapter 5
M-M Michaelis Menten, see Chapter 27
@ = (p(M1N) see Eqs. 28.1 to 28.4
mw molecular weight (kglmol)
PC pneumatic conveying, see Chapter 20
PCM progressive conversion model, see Chapter 25
PFR plug flow reactor, see Chapter 5
RTD residence time distribution, see Chapter 11
SCM shrinking-core model, see Chapter 25
TB turbulent fluidized bed, see Chapter 20

Subscripts
b batch
b bubble phase of a fluidized bed
C of combustion
cloud phase of a fluidized bed
at unreacted core
deactivation
deadwater, or stagnant fluid
emulsion phase of a fluidized bed
equilibrium conditions
leaving or final
of formation
of gas
entering
of liquid
mixed flow
at minimum fluidizing conditions
plug flow
reactor or of reaction
solid or catalyst or surface conditions
entering or reference
using dimensionless time units, see Chapter 11

Superscripts
a, b, ... order of reaction, see Eq. 2.2
n order of reaction
0 refers to the standard state
X V ~ Notation

Dimensionless
groups
-
D vessel dispersion number, see Chapter 13
uL
intensity of dispersion number, see Chapter 13

Hatta modulus, see Eq. 23.8 andlor Figure 23.4

Thiele modulus, see Eq. 18.23 or 18.26
Wagner-Weisz-Wheeler modulus, see Eq. 18.24 or 18.34
dup
Re = - Reynolds number
P
P
Sc = - Schmidt number
~g
Chapter 1
Overview of Chemical Reaction
Engineering

Every industrial chemical process is designed to produce economically a desired

product from a variety of starting materials through a succession of treatment
steps. Figure 1.1shows a typical situation. The raw materials undergo a number
of physical treatment steps to put them in the form in which they can be reacted
chemically. Then they pass through the reactor. The products of the reaction
must then undergo further physical treatment-separations, purifications, etc.-
for the final desired product to be obtained.
Design of equipment for the physical treatment steps is studied in the unit
operations. In this book we are concerned with the chemical treatment step of
a process. Economically this may be an inconsequential unit, perhaps a simple
mixing tank. Frequently, however, the chemical treatment step is the heart of
the process, the thing that makes or breaks the process economically.
Design of the reactor is no routine matter, and many alternatives can be
proposed for a process. In searching for the optimum it is not just the cost of
the reactor that must be minimized. One design may have low reactor cost, but
the materials leaving the unit may be such that their treatment requires a much
higher cost than alternative designs. Hence, the economics of the overall process
must be considered.
Reactor design uses information, knowledge, and experience from a variety
of areas-thermodynamics, chemical kinetics, fluid mechanics, heat transfer,
mass transfer, and economics. Chemical reaction engineering is the synthesis of
all these factors with the aim of properly designing a chemical reactor.
To find what a reactor is able to do we need to know the kinetics, the contacting
pattern and the performance equation. We show this schematically in Fig. 1.2.

I I I I

t Recycle
Figure 1.1 Typical chemical process.
2 Chapter 1 Overview of Chemical Reaction Engineering

Peformance equation
relates input to output

contacting pattern or how Kinetics or how fast things happen.

materials flow through and If very fast, then equilibrium tells
contact each other in the reactor, what will leave the reactor. If not
how early or late they mix, their so fast, then the rate of chemical
clumpiness or state of aggregation. reaction, and maybe heat and mass
By their very nature some materials transfer too, will determine what will
are very clumpy-for instance, solids happen.
and noncoalescing liquid droplets.
Figure 1.2 Information needed to predict what a reactor can do.

Much of this book deals with finding the expression to relate input to output
for various kinetics and various contacting patterns, or

output = f [input, kinetics, contacting] (1)

This is called the performance equation. Why is this important? Because with
this expression we can compare different designs and conditions, find which is
best, and then scale up to larger units.

Classification of Reactions
There are many ways of classifying chemical reactions. In chemical reaction
engineering probably the most useful scheme is the breakdown according to
the number and types of phases involved, the big division being between the
homogeneous and heterogeneous systems. A reaction is homogeneous if it takes
place in one phase alone. A reaction is heterogeneous if it requires the presence
of at least two phases to proceed at the rate that it does. It is immaterial whether
the reaction takes place in one, two, or more phases; at an interface; or whether
the reactants and products are distributed among the phases or are all contained
within a single phase. All that counts is that at least two phases are necessary
for the reaction to proceed as it does.
Sometimes this classification is not clear-cut as with the large class of biological
reactions, the enzyme-substrate reactions. Here the enzyme acts as a catalyst in
the manufacture of proteins and other products. Since enzymes themselves are
highly complicated large-molecular-weight proteins of colloidal size, 10-100 nm,
enzyme-containing solutions represent a gray region between homogeneous and
heterogeneous systems. Other examples for which the distinction between homo-
geneous and heterogeneous systems is not sharp are the very rapid chemical
reactions, such as the burning gas flame. Here large nonhomogeneity in composi-
tion and temperature exist. Strictly speaking, then, we do not have a single phase,
for a phase implies uniform temperature, pressure, and composition throughout.
The answer to the question of how to classify these borderline cases is simple.
It depends on how we choose to treat them, and this in turn depends on which
 Chapter 1 Overview of Chemical Reaction Engineering 3

Table 1.1 Classification of Chemical Reactions Useful in Reactor Design

Noncatalytic Catalytic
Homogeneous Most gas-phase reactions ..........................
..................... Most liquid-phase reactions
------------ Fast reactions such as Reactions in colloidal systems
burning of a flame
..................... Enzyme and microbial reactions
..........................
Burning of coal Ammonia synthesis
Roasting of ores Oxidation of ammonia to pro-
Attack of solids by acids duce nitric acid
Heterogeneous Gas-liquid absorption Cracking of crude oil
with reaction Oxidation of SO2to SO3
Reduction of iron ore to
iron and steel

description we think is more useful. Thus, only in the context of a given situation
can we decide how best to treat these borderline cases.
Cutting across this classification is the catalytic reaction whose rate is altered
by materials that are neither reactants nor products. These foreign materials,
called catalysts, need not be present in large amounts. Catalysts act somehow as
go-betweens, either hindering or accelerating the reaction process while being
modified relatively slowly if at all.
Table 1.1shows the classification of chemical reactions according to our scheme
with a few examples of typical reactions for each type.

Variables Affecting the Rate of Reaction

Many variables may affect the rate of a chemical reaction. In homogeneous
systems the temperature, pressure, and composition are obvious variables. In
heterogeneous systems more than one phase is involved; hence, the problem
becomes more complex. Material may have to move from phase to phase during
reaction; hence, the rate of mass transfer can become important. For example,
in the burning of a coal briquette the diffusion of oxygen through the gas film
surrounding the particle, and through the ash layer at the surface of the particle,
can play an important role in limiting the rate of reaction. In addition, the rate
of heat transfer may also become a factor. Consider, for example, an exothermic
reaction taking place at the interior surfaces of a porous catalyst pellet. If the
heat released by reaction is not removed fast enough, a severe nonuniform
temperature distribution can occur within the pellet, which in turn will result in
differing point rates of reaction. These heat and mass transfer effects become
increasingly important the faster the rate of reaction, and in very fast reactions,
such as burning flames, they become rate controlling. Thus, heat and mass transfer
may play important roles in determining the rates of heterogeneous reactions.

Definition of Reaction Rate

We next ask how to define the rate of reaction in meaningful and useful ways.
To answer this, let us adopt a number of definitions of rate of reaction, all
4 Chapter I Overview of Chemical Reaction Engineering

interrelated and all intensive rather than extensive measures. But first we must
select one reaction component for consideration and define the rate in terms of
this component i. If the rate of change in number of moles of this component
due to reaction is dN,ldt, then the rate of reaction in its various forms is defined
as follows. Based on unit volume of reacting fluid,

y . = -1 dNi= moles i formed

V dt (volume of fluid) (time)

Based on unit mass of solid in fluid-solid systems,

l , , , , , i """" mass of solid) (time)

Based on unit interfacial surface in two-fluid systems or based on unit surface

of solid in gas-solid systems,

y ; = --
I dNi = moles i formed

Based on unit volume of solid in gas-solid systems

1 dN, =
y!'t = -- moles i formed
V, dt (volume of solid) (time)

Based on unit volume of reactor, if different from the rate based on unit volume
of fluid,

1 dNi =
,.!"' = -- moles i formed
V, dt (volume of reactor) (time)

In homogeneous systems the volume of fluid in the reactor is often identical to

the volume of reactor. In such a case V and Vr are identical and Eqs. 2 and 6
are used interchangeably. In heterogeneous systems all the above definitions of
reaction rate are encountered, the definition used in any particular situation
often being a matter of convenience.
From Eqs. 2 to 6 these intensive definitions of reaction rate are related by

volume mass of surface volume vol~me ry

(
(of fluid) ri = solid ) " = (of solid) r' = (
of solid ) (
"= reactor
of )
 Chapter 1 Overview of Chemical Reaction Engineering 5

Speed of Chemical Reactions

Some reactions occur very rapidly; others very, very slowly. For example, in the
production of polyethylene, one of our most important plastics, or in the produc-
tion of gasoline from crude petroleum, we want the reaction step to be complete
in less than one second, while in waste water treatment, reaction may take days
and days to do the job.
Figure 1.3 indicates the relative rates at which reactions occur. To give you
an appreciation of the relative rates or relative values between what goes on in
sewage treatment plants and in rocket engines, this is equivalent to

1 sec to 3 yr

With such a large ratio, of course the design of reactors will be quite different
in these cases.

* t
Cellular rxs.,
1 wor'king ... ...
hard Gases in porous
industrial water Human catalyst particles *
treatment plants at rest Coal furnaces

Jet engines Rocket engines Bimolecular rxs. in which

every collision counts, at
about -1 atm and 400°C
moles of A disappearing
Figure 1.3 Rate of reactions -Ji =
m3of thing. s

Overall Plan
Reactors come in all colors, shapes, and sizes and are used for all sorts of
reactions. As a brief sampling we have the giant cat crackers for oil refining; the
monster blast furnaces for iron making; the crafty activated sludge ponds for
sewage treatment; the amazing polymerization tanks for plastics, paints, and
fibers; the critically important pharmaceutical vats for producing aspirin, penicil-
lin, and birth control drugs; the happy-go-lucky fermentation jugs for moonshine;
and, of course, the beastly cigarette.
Such reactions are so different in rates and types that it would be awkward
to try to treat them all in one way. So we treat them by type in this book because
each type requires developing the appropriate set of performance equations.
 6 Chapter 1 Overview of Chemical Reaction Engineering

/ EX4MPLB 1.1 THE ROCKET ENGINE

A rocket engine, Fig. El.l, burns a stoichiometric mixture of fuel (liquid hydro-
gen) in oxidant (liquid oxygen). The combustion chamber is cylindrical, 75 cm
long and 60 cm in diameter, and the combustion process produces 108 kgls of
exhaust gases. If combustion is complete, find the rate of reaction of hydrogen
and of oxygen.
1 C o m ~ l e t ecombustion

~ Figure E l . l

We want to evaluate

-rH2-
--1- 1 dN0,
d N ~ 2
and -yo, = --
V dt V dt

Let us evaluate terms. The reactor volume and the volume in which reaction
takes place are identical. Thus,

Next, let us look at the reaction occurring.

molecular weight: 2gm 16 gm 18 gm

Therefore,

H,O producedls = 108 kgls - = 6 kmolls

(IlKt)

So from Eq. (i)

H, used = 6 kmolls
0, used = 3 kmolls
 Chapter 1 Overview of Chemical Reaction Engineering 7

l and the rate of reaction is

- - - 1 .--6 kmol - 2.829 X lo4 mol used

3 - 0.2121 m3 s (m3of rocket) . s
1 kmol mol
-To = - 0.2121 m3 3 -= 1.415 X lo4
-
2 s

I Note: Compare these rates with the values given in Figure 1.3.

/ EXAMPLE 1.2 THE LIVING PERSON

A human being (75 kg) consumes about 6000 kJ of food per day. Assume that
I the food is all glucose and that the overall reaction is

from air '

C,H,,O,+60,-6C02+6H,0,
'breathe, out
-AHr=2816kJ

Find man's metabolic rate (the rate of living, loving, and laughing) in terms of
moles of oxygen used per m3 of person per second.

We want to find

Let us evaluate the two terms in this equation. First of all, from our life experience
we estimate the density of man to be

Therefore, for the person in question

Next, noting that each mole of glucose consumed uses 6 moles of oxygen and
releases 2816 kJ of energy, we see that we need

6000 kJIday )( 6 mol 0,

) mol 0,
= 12.8 day
2816 kJ1mol glucose 1mol glucose
 8 Chapter 1 Overview of Chemical Reaction Engineering

I Inserting into Eq. (i)

1 12.8 mol 0, used 1day mol 0, used

= 0.002
=-
0.075 m3 day 24 X 3600 s m3 . s

Note: Compare this value with those listed in Figure 1.3.

PROBLEMS

1.1. Municipal waste water treatment plant. Consider a municipal water treat-
ment plant for a small community (Fig. P1.1). Waste water, 32 000 m3/day,
flows through the treatment plant with a mean residence time of 8 hr, air
is bubbled through the tanks, and microbes in the tank attack and break
down the organic material

(organic waste) + 0, - microbes

C 0 2 + H,O

A typical entering feed has a BOD (biological oxygen demand) of 200 mg

O,/liter, while the effluent has a negligible BOD. Find the rate of reaction,
or decrease in BOD in the treatment tanks.

Waste water,
32,000 m3/day --I Waste water
treatment plant
Clean water,
32,000 rn3/day

t
2 0 0 mg O2
t
Mean residence
t
Zero O2 needed
neededlliter time t =8 hr
Figure P1.l

1.2. Coal burning electrical power station. Large central power stations (about
1000 MW electrical) using fluidized bed combustors may be built some day
(see Fig. P1.2). These giants would be fed 240 tons of coallhr (90% C, 10%

Fluidized bed

\
50% of the feed
burns in these 1 0 units

Figure P1.2
 Chapter 1 Overview of Chemical Reaction Engineering 9

H,), 50% of which would burn within the battery of primary fluidized beds,
the other 50% elsewhere in the system. One suggested design would use a
battery of 10 fluidized beds, each 20 m long, 4 m wide, and containing solids
to a depth of 1 m. Find the rate of reaction within the beds, based on the
oxygen used.

1.3. Fluid cracking crackers (FCC). FCC reactors are among the largest pro-
cessing units used in the petroleum industry. Figure P1.3 shows an example
of such units. A typical unit is 4-10 m ID and 10-20 m high and contains
about 50 tons of p = 800 kg/m3porous catalyst. It is fed about 38 000 barrels
of crude oil per day (6000 m3/day at a density p = 900 kg/m3), and it cracks
these long chain hydrocarbons into shorter molecules.
To get an idea of the rate of reaction in these giant units, let us simplify
and suppose that the feed consists of just C,, hydrocarbon, or

If 60% of the vaporized feed is cracked in the unit, what is the rate of
reaction, expressed as - r r (mols reactedlkg cat. s) and as r"' (mols reacted1
m3 cat. s)?

Figure P1.3 The Exxon Model IV FCC unit.

Homogeneous Reactions
in Ideal Reactors

Chapter 2 Kinetics of Homogeneous Reactions 113

Chapter 3 Interpretation of Batch Reactor Data I38
Chapter 4 Introduction to Reactor Design I83
Chapter 5 Ideal Reactors for a Single Reaction I90
Chapter 6 Design for Single Reactions 1120
Chapter 7 Design for Parallel Reactions 1152
Chapter 8 Potpourri of Multiple Reactions 1170
Chapter 9 Temperature and Pressure Effects 1207
Chapter 10 Choosing the Right Kind of Reactor 1240
Chapter 2
Kinetics of Homogeneous
Reactions

Simple Reactor Types

Ideal reactors have three ideal flow or contacting patterns. We show these in
Fig. 2.1, and we very often try to make real reactors approach these ideals as
closely as possible.
We particularly like these three flow or reacting patterns because they are
easy to treat (it is simple to find their performance equations) and because one
of them often is the best pattern possible (it will give the most of whatever it is
we want). Later we will consider recycle reactors, staged reactors, and other flow
pattern combinations, as well as deviations of real reactors from these ideals.

The Rate Equation

Suppose a single-phase reaction aA + bB + rR + sS. The most useful measure
of reaction rate for reactant A is then

rate of disappearance of A
F(
\ h
= -\p
1 dNA
" - (amount of
-
A disappearing)
(volume) (time) (1)

( note that this is an

-the minus sign
intensive measure means disappearance

In addition, the rates of reaction of all materials are related by

Experience shows that the rate of reaction is influenced by the composition and
the energy of the material. By energy we mean the temperature (random kinetic
energy of the molecules), the light intensity within the system (this may affect
13
14 Chapter 2 Kinetics of Homogeneous Reactions

Steady-state flow

Batch Plug flow Mixed flow

Uniform composition Fluid passes through the reactor Uniformly mixed, same
everywhere in the reactor, with no mixing of earlier and later composition everywhere,
but of course the entering fluid, and with no overtaking. within the reactor and
composition changes It is as if the fluid moved in single at the exit.
with time. file through the reactor.
Figure 2.1 Ideal reactor types.

the bond energy between atoms), the magnetic field intensity, etc. Ordinarily
we only need to consider the temperature, so let us focus on this factor. Thus,
we can write

activation

(2)
terms terms
reaction (temperature
order dependent term

Here are a few words about the concentration-dependent and the temperature-
dependent terms of the rate.

2.1 CONCENTRATION-DEPENDENT TERM OF A RATE EQUATION

Before we can find the form of the concentration term in a rate expression, we
must distinguish between different types of reactions. This distinction is based
on the form and number of kinetic equations used to describe the progress of
reaction. Also, since we are concerned with the concentration-dependent term
of the rate equation, we hold the temperature of the system constant.

Single and Multiple Reactions

First of all, when materials react to form products it is usually easy to decide
after examining the stoichiometry, preferably at more than one temperature,
whether we should consider a single reaction or a number of reactions to be oc-
curring.
When a single stoichiometric equation and single rate equation are chosen to
represent the progress of the reaction, we have a single reaction. When more
than one stoichiometric equation is chosen to represent the observed changes,
 2.1 Concentration-Dependent Term of a Rate Equation 15
then more than one kinetic expression is needed to follow the changing composi-
tion of all the reaction components, and we have multiple reactions.
Multiple reactions may be classified as:
series reactions,

parallel reactions, which are of two types

competitive side by side

and more complicated schemes, an example of which is

Here, reaction proceeds in parallel with respect to B, but in series with respect
to A, R, and S.

Elementary and Nonelementary Reactions

Consider a single reaction with stoichiometric equation

If we postulate that the rate-controlling mechanism involves the collision or

interaction of a single molecule of A with a single molecule of B, then the number
of collisions of molecules A with B is proportional to the rate of reaction. But
at a given temperature the number of collisions is proportional to the concentra-
tion of reactants in the mixture; hence, the rate of disappearance of A is given by

Such reactions in which the rate equation corresponds to a stoichiometric equa-

tion are called elementary reactions.
When there is no direct correspondence between stoichiometry and rate, then
we have nonelementary reactions. The classical example of a nonelementary
reaction is that between hydrogen and bromine,

H, + Br, +2HBr
16 Chapter 2 Kinetics of Homogeneous Reactions

which has a rate expression*

Nonelementary reactions are explained by assuming that what we observe as

a single reaction is in reality the overall effect of a sequence of elementary
reactions. The reason for observing only a single reaction rather than two or
more elementary reactions is that the amount of intermediates formed is negligi-
bly small and, therefore, escapes detection. We take up these explanations later.

Molecularity and Order of Reaction

The inolecularity of an elementary reaction is the number of molecules involved
in the reaction, and this has been found to have the values of one, two, or
occasionally three. Note that the molecularity refers only to an elementary re-
action.
Often we find that the rate of progress of a reaction, involving, say, materials
A, B, . . . , D, can be approximated by an expression of the following type:

where a, b, . . . , d are not necessarily related to the stoichiometric coefficients.

We call the powers to which the concentrations are raised the order of the
reaction. Thus, the reaction is

ath order with respect to A

bth order with respect to B
nth order overall

Since the order refers to the empirically found rate expression, it can have a
fractional value and need not be an integer. However, the molecularity of a
reaction must be an integer because it refers to the mechanism of the reaction,
and can only apply to an elementary reaction.
For rate expressions not of the form of Eq, 4, such as Eq. 3, it makes no sense
to use the term reaction order.

Rate Constant k
When the rate expression for a homogeneous chemical reaction is written in the
form of Eq. 4, the dimensions of the rate constant k for the nth-order reaction are

* To eliminate much writing, in this chapter we use square brackets to indicate concentrations. Thus,
, ,C = [HBr]
 2.1 Concentration-Dependent Term of a Rate Equation 17

which for a first-order reaction becomes simply

Representation of an Elementary Reaction

In expressing a rate we may use any measure equivalent to concentration (for
example, partial pressure), in which case

Whatever measure we use leaves the order unchanged; however, it will affect
the rate constant k.
For brevity, elementary reactions are often represented by an equation showing
both the molecularity and the rate constant. For example,

represents a biomolecular irreversible reaction with second-order rate constant

k,, implying that the rate of reaction is

It would not be proper to write Eq. 7 as

for this would imply that the rate expression is

Thus, we must be careful to distinguish between the one particular equation that
represents the elementary reaction and the many possible representations of the
stoichiometry.
We should note that writing the elementary reaction with the rate constant,
as shown by Eq. 7, may not be sufficient to avoid ambiguity. At times it may be
necessary to specify the component in the reaction to which the rate constant is
referred. For example, consider the reaction,

If the rate is measured in terms of B, the rate equation is

18 Chapter 2 Kinetics of Homogeneous Reactions

If it refers to D, the rate equation is

Or if it refers to the product T, then

But from the stoichiometry

hence,

In Eq. 8, which of these three k values are we referring to? We cannot tell.
Hence, to avoid ambiguity when the stoichiometry involves different numbers
of molecules of the various components, we must specify the component be-
ing considered.
To sum up, the condensed form of expressing the rate can be ambiguous. To
eliminate any possible confusion, write the stoichiometric equation followed by
the complete rate expression, and give the units of the rate constant.

Representation of a Nonelementary Reaction

A nonelementary reaction is one whose stoichiometry does not match its kinetics.
For example,

Stoichiometry: N2 + 3H2 2NH3

Rate:

This nonmatch shows that we must try to develop a multistep reaction model
to explain the kinetics.

Kinetic Models for Nonelementary Reactions

To explain the kinetics of nonelementary reactions we assume that a sequence
of elementary reactions is actually occurring but that we cannot measure or
observe the intermediates formed because they are only present in very minute
quantities. Thus, we observe only the initial reactants and final products, or what
appears to be a single reaction. For example, if the kinetics of the reaction
 2.1 Concentration-Dependent Term of a Rate Equation 19
indicates that the reaction is nonelementary, we may postulate a series of elemen-
tary steps to explain the kinetics, such as

where the asterisks refer to the unobserved intermediates. To test our postulation
scheme, we must see whether its predicted kinetic expression corresponds to ex-
periment.
The types of intermediates we may postulate are suggested by the chemistry
of the materials. These may be grouped as follows.

Free Radicals. Free atoms or larger fragments of stable molecules that contain
one or more unpaired electrons are called free radicals. The unpaired electron
is designated by a dot in the chemical symbol for the substance. Some free
radicals are relatively stable, such as triphenylmethyl,

but as a rule they are unstable and highly reactive, such as

Ions and Polar Substances. Electrically charged atoms, molecules, or fragments

of molecules. such as

N;, Nat, OH-, H30t, NH;, CH,OH;, I-

are called ions. These may act as active intermediates in reactions.

Molecules. Consider the consecutive reactions

Ordinarily these are treated as multiple reactions. However, if the intermediate

R is highly reactive its mean lifetime will be very small and its concentration in
the reacting mixture can become too small to measure. In such a situation R
may not be observed and can be considered to be a reactive intermediate.
20 Chapter 2 Kinetics of Homogeneous Reactions

Transition Complexes. The numerous collisions between reactant molecules

result in a wide distribution of energies among the individual molecules. This
can result in strained bonds, unstable forms of molecules, or unstable association
of molecules which can then either decompose to give products, or by further
collisions return to molecules in the normal state. Such unstable forms are called
transition complexes.
Postulated reaction schemes involving these four kinds of intermediates can
be of two types.

Nonchain Reactions. In the nonchain reaction the intermediate is formed in

the first reaction and then disappears as it reacts further to give the product. Thus,

Reactants -,(Intermediates)"
(Intermediates)" +Products

Chain Reactions. In chain reactions the intermediate is formed in a first reaction,

called the chain initiation step. It then combines with reactant to form product and
more intermediate in the chain propagation step. Occasionally the intermediate is
destroyed in the chain termination step. Thus,

Reactant +(Intermediate)" Initiation

(Intermediate)" + Reactant + (Intermediate)" + Product Propagation
(Intermediate)" -,Product Termination

The essential feature of the chain reaction is the propagation step. In this step
the intermediate is not consumed but acts simply as a catalyst for the conversion
of material. Thus, each molecule of intermediate can catalyze a long chain of
reactions, even thousands, before being finally destroyed.
The following are examples of mechanisms of various kinds.
1. Free radicals, chain reaction mechanism. The reaction

Hz + Br, +2HBr

with experimental rate

can be explained by the following scheme which introduces and involves

the intermediates Ha and Bra,

Br, *2Br. Initiation and termination

Bra+ H,*HBr + Ha Propagation
He + Br, HBr + Bra
+ Propagation
 2.1 Concentration-Dependent Term of a Rate Equation 21
2. Molecular intermediates, nonchain mechanism. The general class of en-
zyme-catalyzed fermentation reactions

with
A-R enzyme

with experimental rate

C constant

is viewed to proceed with intermediate (A. enzyme)* as follows:

A + enzyme +(A. enzyme)*

(A. enzyme)* -.R + enzyme
In such reactions the concentration of intermediate may become more than
negligible, in which case a special analysis, first proposed by Michaelis and
Menten (1913), is required.

3. Transition complex, nonchain mechanism. The spontaneous decomposition

of azomethane

exhibits under various conditions first-order, second-order, or intermediate

kinetics. This type of behavior can be explained by postulating the existence
of an energized and unstable form for the reactant, A*. Thus,

A + A +A* + A Formation of energized molecule

A* + A + A + A Return to stable form by collision
A*+R+S Spontaneous decomposition into products

Lindemann (1922) first suggested this type of intermediate.

Testing Kinetic Models

Two problems make the search for the correct mechanism of reaction difficult.
First, the reaction may proceed by more than one mechanism, say free radical
and ionic, with relative rates that change with conditions. Second, more than
one mechanism can be consistent with kinetic data. Resolving these problems
is difficult and requires an extensive knowledge of the chemistry of the substances
involved. Leaving these aside, let us see how to test the correspondence between
experiment and a proposed mechanism that involves a sequence of elemen-
tary reactions.
 22 Chapter 2 Kinetics of Homogeneous Reactions

In these elementary reactions we hypothesize the existence of either of two

types of intermediates.

Type 1. An unseen and unmeasured intermediate X usually present at such

small concentration that its rate of change in the mixture can be taken to be
zero. Thus, we assume

d [ X I-
[XI is small and - _0
dt

This is called the steady-state approximation. Mechanism types 1 and 2, above,

adopt this type of intermediate, and Example 2.1 shows how to use it.

Type 2. Where a homogeneous catalyst of initial concentration Cois present in

two forms, either as free catalyst C or combined in an appreciable extent to
form intermediate X, an accounting for the catalyst gives

[Col = [CI + [XI

We then also assume that either

dX
- --0
dt

or that the intermediate is in equilibrium with its reactants; thus,

( r e a ~ t ) (cat2st)
+ +(intermediate
X

where

Example 2.2 and Problem 2.23 deal with this type of intermediate.
The trial-and-error procedure involved in searching for a mechanism is illus-
trated in the following two examples.

, SEARCH FOR THE REACTION MECHANISM

The irreversible reaction

has been studied kinetically, and the rate of formation of product has been found
to be well correlated by the following rate equation:

I rAB= kC2,. . . independent of C A . (11)

 2.1 Concentration-Dependent Term of a Rate Equation 23

What reaction mechanism is suggested by this rate expression if the chemistry

of the reaction suggests that the intermediate consists of an association of reactant
molecules and that a chain reaction does not occur?

If this were an elementary reaction, the rate would be given by

Since Eqs. 11and 12 are not of the same type, the reaction evidently is nonelemen-
tary. Consequently, let us try various mechanisms and see which gives a rate
expression similar in form to the experimentally found expression. We start with
simple two-step models, and if these are unsuccessful we will try more complicated
three-, four-, or five-step models.

Model 1. Hypothesize a two-step reversible scheme involving the formation of

an intermediate substance A;, not actually seen and hence thought to be present
only in small amounts. Thus,

which really involves four elementary reactions

Let the k values refer to the components disappearing; thus, k, refers to A, k,

refers to A,*,etc.
Now write the expression for the rate of formation of AB. Since this component
is involved in Eqs. 16 and 17, its overall rate of change is the sum of the individual
rates. Thus,
24 Chapter 2 Kinetics of Homogeneous Reactions

Because the concentration of intermediate AT is so small and not measurable,

the above rate expression cannot be tested in its present form. So, replace [A:]
by concentrations that can be measured, such as [A], [B], or [AB]. This is done
in the following manner. From the four elementary reactions that all involve A:
we find

1
YA* = - kl[AI2- k2[A;] - k3[Af][B] + k4[A][AB] (19)
2

Because the concentration of AT is always extremely small we may assume that

its rate of change is zero or

This is the steady-state approximation. Combining Eqs. 19 and 20 we then find

which, when replaced in Eq. 18, simplifying and cancelling two terms (two terms
will always cancel if you are doing it right), gives the rate of formation of AB
in terms of measurable quantities. Thus,

In searching for a model consistent with observed kinetics we may, if we wish,

restrict a more general model by arbitrarily selecting the magnitude of the various
rate constants. Since Eq. 22 does not match Eq. 11, let us see if any of its
simplified forms will. Thus, if k, is very small, this expression reduces to

If k, is very small, r , reduces to

Neither of these special forms, Eqs. 23 and 24, matches the experimentally found
rate, Eq. 11. Thus, the hypothesized mechanism, Eq. 13, is incorrect, so another
needs to be tried.
 2.1 Concentration-Dependent Term of a Rate Equation 25
Model 2. First note that the stoichiometry of Eq. 10 is symmetrical in A and
B, so just interchange A and B in Model 1, put k, = 0 and we will get r,, =
k[BI2, which is what we want. So the mechanism that will match the second-
order rate equation is

B+B+B?
1
1

We are fortunate in this example to have represented our data by a form of

equation which happened to match exactly that obtained from the theoretical
mechanism. Often a number of different equation types will fit a set of experimen-
tal data equally well, especially for somewhat scattered data. Hence, to avoid
rejecting the correct mechanism, it is advisable to test the fit of the various
theoretically derived equations to the raw data using statistical criteria whenever
possible, rather than just matching equation forms:

SEARCH FOR A MECHANISM FOR THE ENZYME-

SUBSTRATE REACTION

Here, a reactant, called the substrate, is converted to product by the action of

an enzyme, a high molecular weight (mw > 10 000) protein-like substance. An
enzyme is highly specific, catalyzing one particular reaction, or one group of
reactions. Thus,

enzyme
A-R

I Many of these reactions exhibit the following behavior:

1. A rate proportional to the concentration of enzyme introduced into the
mixture [E,].
2. At low reactant concentration the rate is proportional to the reactant con-
centration, [A].
3. At high reactant concentration the rate levels off and becomes independent
of reactant concentration.

I Propose a mechanism to account for this behavior.

1 SOLUTION
Michaelis and Menten (1913) were the first to solve this puzzle. (By the way,
Michaelis received the Nobel prize in chemistry.) They guessed that the reaction
proceeded as follows
26 Chapter 2 Kinetics of Homogeneous Reactions

with the two assumptions

and

First write the rates for the pertinent reaction components of Eq. 26. This gives

and

Eliminating [El from Eqs. 27 and 30 gives

and when Eq. 31 is introduced into Eq. 29 we find

([MI = (lei)is called

the Michaelis constant

By comparing with experiment, we see that this equation fits the three re-
ported facts:

[A] when [A] 4 [MI

dt
is independent of [A] when [A] S [MI

For more discussion about this reaction, see Problem 2.23.

m
 2.2 Temperature-Dependent Term of a Rate Equation 27

2.2TEMPERATURE-DEPENDENT TERM OF A RATE EQUATION

Temperature Dependency from Arrhenius' Law
For many reactions, and particularly elementary reactions, the rate expression
can be written as a product of a temperature-dependent term and a composition-
dependent term, or

For such reactions the temperature-dependent term, the reaction rate constant,
has been found in practically all cases to be well represented by Arrhenius' law:

where k, is called the frequency or pre-exponential factor and E is called the

activation energy of the reaction." This expression fits experiment well over wide
temperature ranges and is strongly suggested from various standpoints as being
a very good approximation to the true temperature dependency.
At the same concentration, but at two different temperatures, Arrhenius' law
indicates that

provided that E stays constant.

Comparison of Theories with Arrhenius' Law

The expression

summarizes the predictions of the simpler versions of the collision and transition
state theories for the temperature dependency of the rate constant. For more
complicated versions m can be as great as 3 or 4. Now, because the exponential
term is so much more temperature-sensitive than the pre-exponential term, the
variation of the latter with temperature is effectively masked, and we have
in effect

* There seems to be a disagreement in the dimensions used to report the activation energy; some
authors use joules and others use joules per mole. However, joules per mole are clearly indicated
in Eq. 34.
But what moles are we referring to in the units of E? This is unclear. However, since E and R
always appear together, and since they both refer to the same number of moles, this bypasses the
problem. This whole question can be avoided by using the ratio E/R throughout.
28 Chapter 2 Kinetics of Homogeneous Reactions

5- ! A T = ioooo/ / AT= 870 I

,+-
7 do:i;ing
I of rate
I+ -1, for
doubling I
of rate
t
at 2 0 0 0 K at lOOOK at 463K at 376K
1/T
Figure 2.2 Sketch showing temperature dependency of the reaction rate.

This shows that Arrhenius' law is a good approximation to the temperature

dependency of both collision and transition-state theories.

Activation Energy and Temperature Dependency

The temperature dependency of reactions is determined by the activation energy
and temperature level of the reaction, as illustrated in Fig. 2.2 and Table 2.1.
These findings are summarized as follows:
1. From Arrhenius' law a plot of In k vs 1IT gives a straight line, with large
slope for large E and small slope for small E.
2. Reactions with high activation energies are very temperature-sensitive;reac-
tions with low activation energies are relatively temperature-insensitive.

Table 2.1 Temperature Rise Needed to Double the Rate of Reaction for
Activation Energies and Average Temperatures Showna
Activation Energy E
Average
Temperature 40 kJ/mol 160 kJ/mol 280 kJ/mol 400 kJ/mol
0°C 11°C 2.7"C 1.5"C l.l°C
400°C 65 16 9.3 6.5
1000°C 233 58 33 23
2000°C 744 185 106 74
"Shows temperature sensitivity of reactions.
 2.3 Searching for a Mechanism 29
3. Any given reaction is much more temperature-sensitive at a low temperature
than at a high temperature.
4. From the Arrhenius law, the value of the frequency factor k, does not affect
the temperature sensitivity.

SEARCH FOR THE ACTIVATION ENERGY OF A

PASTEURIZATION PROCESS

Milk is pasteurized if it is heated to 63OC for 30 min, but if it is heated to

74°C it only needs 15 s for the same result. Find the activation energy of this
sterilization process.

To ask for the activation energy of a process means assuming an Arrhenius

temperature dependency for the process. Here we are told that

t1= 30 min at a TI = 336 K

t2= 15 sec at a T2= 347 K

Now the rate is inversely proportional to the reaction time, or rate lltime so
Eq. 35 becomes

from which the activation energy

2.3 SEARCHING FOR A MECHANISM

The more we know about what is occurring during reaction, what the reacting
materials are, and how they react, the more assurance we have for proper design.
This is the incentive to find out as much as we can about the factors influencing
a reaction within the limitations of time and effort set by the economic optimiza-
tion of the process.
There are three areas of investigation of a reaction, the stoichiometry, the
kinetics, and the mechanism. In general, the stoichiometry is studied first, and
when this is far enough along, the kinetics is then investigated. With empirical
rate expressions available, the mechanism is then looked into. In any investigative
30 Chapter 2 Kinetics of Homogeneous Reactions

program considerable feedback of information occurs from area to area. For

example, our ideas about the stoichiometry of the reaction may change on the
basis of kinetic data obtained, and the form of the kinetic equations themselves
may be suggested by mechanism studies. With this kind of relationship of the
many factors, no straightforward experimental program can be formulated for
the study of reactions. Thus, it becomes a matter of shrewd scientific detective
work, with carefully planned experimental programs especially designed to dis-
criminate between rival hypotheses, which in turn have been suggested and
formulated on the basis of all available pertinent information.
Although we cannot delve into the many aspects of this problem, a number
of clues that are often used in such experimentation can be mentioned.
1. Stoichiometry can tell whether we have a single reaction or not. Thus, a
complicated stoichiometry

or one that changes with reaction conditions or extent of reaction is clear

evidence of multiple reactions.
2. Stoichiometry can suggest whether a single reaction is elementary or not
because no elementary reactions with molecularity greater than three have
been observed to date. As an example, the reaction

is not elementary.
3. A comparison of the stoichiometric equation with the experimental kinetic
expression can suggest whether or not we are dealing with an elementary re-
action.
4. A large difference in the order of magnitude between the experimentally
found frequency factor of a reaction and that calculated from collision
theory or transition-state theory may suggest a nonelementary reaction;
however, this is not necessarily true. For example, certain isomerizations
have very low frequency factors and are still elementary.
5. Consider two alternative paths for a simple reversible reaction. If one of
these paths is preferred for the forward reaction, the same path must also be
preferred for the reverse reaction. This is called the principle of microscopic
reversibility. Consider, for example, the forward reaction of

At a first sight this could very well be an elementary biomolecular reaction

with two molecules of ammonia combining to yield directly the four product
molecules. From this principle, however, the reverse reaction would then
also have to be an elementary reaction involving the direct combination of
three molecules of hydrogen with one of nitrogen. Because such a process
is rejected as improbable, the bimolecular forward mechanism must also
be rejected.
6. The principle of microreversibility also indicates that changes involving
bond rupture, molecular syntheses, or splitting are likely to occur one at a
 2.3 Searching for a Mechanism 31
time, each then being an elementary step in the mechanism. From this point
of view, the simultaneous splitting of the complex into the four product
molecules in the reaction

is very unlikely. This rule does not apply to changes that involve a shift in
electron density along a molecule, which may take place in a cascade-like
manner. For example, the transformation

CH2=CH-CH2-0-CH=CH2 +CH,=CH-CH2-CH2-CHO
......................... .......................
vinyl ally1 ether n-pentaldehyde-ene 4

can be expressed in terms of the following shifts in electron density:

H H
/ /
CH,fC
F-c\\0
- - Ilf\0
CH
v-L-
'?
CH-CH,
\,/ - 2"
CH= CH,

H H
/ /

,*-?- CF-c
CH, =C
'.-A \\
C\-H
2,:'' p 0
CH~CH, CH= CH,

7. For multiple reactions a change in the observed activation energy with

temperature indicates a shift in the controlling mechanism of reaction. Thus,
for an increase in temperature E,,, rises for reactions or steps in parallel,
E,,, falls for reactions or steps in series. Conversely, for a decrease in
temperature E,,, falls for reactions in parallel, E,,, rises for reactions in
series. These findings are illustrated in Fig. 2.3.
Mech. 1
Mech. 1 Mech. 2

LQR
A d X d R

High E

I High T Low T High T Low T

*
1/T
Figure 2.3 A change in activation energy indicates a shift in
controlling mechanism of reaction.
32 Chapter 2 Kinetics of Homogeneous Reactions

2.4 PREDICTABILITY OF REACTION RATE FROM THEORY

Concentration-Dependent Term
If a reaction has available a number of competing paths (e.g., noncatalytic and
catalytic), it will in fact proceed by all of these paths, although primarily by the
one of least resistance. This path usually dominates. Only a knowledge of the
energies of all possible intermediates will allow prediction of the dominant path
and its corresponding rate expression. As such information cannot be found in
the present state of knowledge, a priori prediction of the form of the concentra-
tion term is not possible. Actually, the form of the experimentally found rate
expression is often the clue used to investigate the energies of the intermediates
of a reaction.

Temperature-Dependent Term
Assuming that we already know the mechanism of reaction and whether or not
it is elementary, we may then proceed to the prediction of the frequency factor
and activation energy terms of the rate constant.
If we are lucky, frequency factor predictions from either collision or transition-
state theory may come within a factor of 100 of the correct value; however, in
specific cases predictions may be much further off.
Though activation energies can be estimated from transition-state theory,
reliability is poor, and it is probably best to estimate them from the experimental
findings for reactions of similar compounds. For example, the activation energies
of the following homologous series of reactions

RI + C6H50Na -
ethanol
C6H50R+ NaI

where R is

CH, GH,, iso-C3H7 sec-C4H,

C,H5 C,H17 iso-C,H, sec-C,H,,
C,H, C16H3, iso-C,H,, sec-C,H17
C,H9

all lie between 90 and 98 kJ.

Use of Predicted Values in Design

The frequent order-of-magnitude predictions of the theories tend to confirm the
correctness of their representations, help find the form and the energies of various
intermediates, and give us a better understanding of chemical structure. However,
theoretical predictions rarely match experiment by a factor of two. In addition,
we can never tell beforehand whether the predicted rate will be in the order of
magnitude of experiment or will be off by a factor of lo6. Therefore, for engi-
 Problems 33

neering design, this kind of information should not be relied on and experimen-
tally found rates should be used in all cases. Thus, theoretical studies may be
used as a supplementary aid to suggest the temperature sensitivity of a given
reaction from a similar type of reaction, to suggest the upper limits of reaction
rate, etc. Design invariably relies on experimentally determined rates.

RELATED READING

Jungers, J. C., et al., Cinttique chimique appliqute, Technip, Paris, 1958.

Laidler, K. J., Chemical Kinetics, 2nd ed., Harper and Row, New York, 1987.
Moore, W. J., Basic Physical Chemistry, Prentice-Hall, Upper Saddle River, NJ, 1983.

REFERENCES

Lindemann, F. A., Trans. Faraday Soc., 17,598 (1922).

Michaelis, L., and Menten, M. L., Biochem. Z., 49,333 (1913). This treatment is discussed
by Laidler (1987), see Related Readings.

PROBLEMS

2.1. A reaction has the stoichiometric equation A + B = 2R. What is the order
of reaction?

2.2. Given the reaction 2N0, + -21 0, = N205,what is the relation between
the rates of formation and disappearance of the three reaction components?

1
2.3. A reaction with stoichiometric equation - A
2
+B = R + -21 S has the
following rate expression

What is the rate expression for this reaction if the stoichiometric equation
is written as A + 2B = 2R + S?

2.4. For the enzyme-substrate reaction of Example 2, the rate of disappearance

of substrate is given by

What are the units of the two constants?

34 Chapter 2 Kinetics of Homogeneous Reactions

2.5. For the complex reaction with stoichiometry A + 3B + 2R + S and with

second-order rate expression

are the reaction rates related as follows: r, = r, = r,? If the rates are not
so related, then how are they related? Please account for the signs, + or -.

2.6. A certain reaction has a rate given by

-r, = 0.005C2, mol/cm3. min

If the concentration is to be expressed in mollliter and time in hours, what

would be the value and units of the rate constant?

2.7. For a gas reaction at 400 K the rate is reported as

(a) What are the units of the rate constant?

(b) What is the value of the rate constant for this reaction if the rate
equation is expressed as

2.8. The decomposition of nitrous oxide is found to proceed as follows:

What is the order of this reaction with respect to N,O, and overall?

2.9. The pyrolysis of ethane proceeds with an activation energy of about 300
kJImol. How much faster is the decomposition at 650°C than at 500"C?

2.10. A 1100-K n-nonane thermally cracks (breaks down into smaller molecules)
20 times as rapidly as at 1000 K. Find the activation energy for this decompo-
sition.

2.11. In the mid-nineteenth century the entomologist Henri Fabre noted that
French ants (garden variety) busily bustled about their business on hot
 Problems 35

days but were rather sluggish on cool days. Checking his results with Oregon
ants, I find

Runningspeed,m/hr 150 160 230 295 370

Temperature, "C 1 3 16 22 24 28

What activation energy represents this change in bustliness?

2.12. The maximum allowable temperature for a reactor is 800 K. At present

our operating set point is 780 K, the 20-K margin of safety to account for
fluctuating feed, sluggish controls, etc. Now, with a more sophisticated
control system we would be able to raise our set point to 792 K with the
same margin of safety that we now have. By how much can the reaction
rate, hence, production rate, be raised by this change if the reaction taking
place in the reactor has an activation energy of 175 kJ/mol?

2.13. Every May 22 I plant one watermelon seed. I water it, I fight slugs, I pray,
I watch my beauty grow, and finally the day comes when the melon ripens.
I then harvest and feast. Of course, some years are sad, like 1980, when a
bluejay flew off with the seed. Anyway, six summers were a pure joy and
for these I've tabulated the number of growing days versus the mean
daytime temperature during the growing season. Does the temperature
affect the growth rate? If so, represent this by an activation energy.

I
1976 1977 1982 1984 1985 1988
Growing days 87 85 74 78 90 84
Mean temp, "C
liar 22.0 23.4 26.3 24.3 21.1 22.7

2.14. On typical summer days, field crickets nibble, jump, and chirp now and
then. But at a night when great numbers congregate, chirping seems to
become a serious business and tends to be in unison. In 1897,A. E. Dolbear
(Am. Naturalist, 31,970) reported that this social chirping rate was depen-
dent on the temperature as given by

(number of chirps in 15 s) + 40 = (temperature, OF)

Assuming that the chirping rate is a direct measure of the metabolic rate,
find the activation energy in kJ/mol of these crickets in the temperature
range 60430°F.

2.15. On doubling the concentration of reactant, the rate of reaction triples. Find
the reaction order.

For the stoichiometry A +B -+ (products) find the reaction orders with

respect to A and B.
36 Chapter 2 Kinetics of Homogeneous Reactions

2.18. Show that the following scheme

NO* + NO: A 2 ~ 0 ~

proposed by R. Ogg, J. Chem. Phys., 15,337 (1947) is consistent with, and

can explain, the observed first-order decomposition of N205.

2.19. The decomposition of reactant A at 400°C for pressures between 1 and

10 atm follows a first-order rate law.
(a) Show that a mechanism similar to azomethane decomposition, p. 21,

is consistent with the observed kinetics.

Different mechanisms can be proposed to explain first-order kinetics. To
claim that this mechanism is correct in the face of the other alternatives
requires additional evidence.
(b) For this purpose, what further experiments would you suggest we run
and what results would you expect to find?

2.20. Experiment shows that the homogeneous decomposition of ozone proceeds

with a rate

(a) What is the overall order of reaction?

(b) Suggest a two-step mechanism to explain this rate and state how you
would further test this mechanism.

2.21. Under the influence of oxidizing agents, hypophosphorous acid is trans-

formed into phosphorous acid:

H,P02 -
oxidizing agent
H3PO3
 Problems 37
The kinetics of this transformation present the following features. At a
low concentration of oxidizing agent,

rH3P03
= k [oxidizing agent] [H,PO,]

At a high concentration of oxidizing agent,

To explain the observed kinetics, it has been postulated that, with hydrogen
ions as catalyst, normal unreactive H3P02is transformed reversibly into
an active form, the nature of which is unknown. This intermediate then
reacts with the oxidizing agent to give [Link] that this scheme does
explain the observed kinetics.

2.22. Come up with (guess and then verify) a mechanism that is consistent with
the experimentally found rate equation for the following reaction

2A + B -+A2B with +rAzB= k[A] [B]

2.23. Mechanism for enzyme catalyzed reactions. To explain the kinetics of en-
zyme-substrate reactions, Michaelis and Menten (1913) came up with the
following mechanism, which uses an equilibrium assumption

kl
A+E-X
with K = - and with
k2
[Eo] = [El + [XI
k [A1[El '
XL+R
E

and where [E,] represents the total enzyme and [El represents the free
unattached enzyme.
G. E. Briggs and J. B. S. Haldane, Biochem J., 19, 338 (1925), on the
other hand, employed a steady-state assumption in place of the equilibrium
assumption

kl
A+E-X
k2 with d[Xl = 0, and [Eo] = [El + [XI
dt
xk3-R+E
What final rate form -rA in terms of [A], [E,], k,, k,, and k, does
(a) the Michaelis-Menten mechanism give?
(b) the Briggs-Haldane mechanism give?
Chapter 3
Interpretation of Batch
Reactor Data

A rate equation characterizes the rate of reaction, and its form may either be
suggested by theoretical considerations or simply be the result of an empirical
curve-fitting procedure. In any case, the value of the constants of the equation
can only be found by experiment; predictive methods are inadequate at present.
The determination of the rate equation is usually a two-step procedure; first
the concentration dependency is found at fixed temperature and then the temper-
ature dependence of the rate constants is found, yielding the complete rate
equation.
Equipment by which empirical information is obtained can be divided into
two types, the batch and flow reactors. The batch reactor is simply a container
to hold the contents while they react. All that has to be determined is the extent
of reaction at various times, and this can be followed in a number of ways,
for example:
1. By following the concentration of a given component.
2. By following the change in some physical property of the fluid, such as the
electrical conductivity or refractive index.
3. By following the change in total pressure of a constant-volume system.
4. By following the change in volume of a constant-pressure system.
The experimental batch reactor is usually operated isothermally and at constant
volume because it is easy to interpret the results of such runs. This reactor is a
relatively simple device adaptable to small-scale laboratory set-ups, and it needs
but little auxiliary equipment or instrumentation. Thus, it is used whenever
possible for obtaining homogeneous kinetic data. This chapter deals with the
batch reactor.
The flow reactor is used primarily in the study of the kinetics of heterogeneous
reactions. Planning of experiments and interpretation of data obtained in flow
reactors are considered in later chapters.
There are two procedures for analyzing kinetic data, the integral and the
differential methods. In the integral method of analysis we guess a particular
form of rate equation and, after appropriate integration and mathematical manip-
ulation, predict that the plot of a certain concentration function versus time
 3.1 Constant-Volume Batch Reactor 39

should yield a straight line. The data are plotted, and if a reasonably good straight
line is obtained, then the rate equation is said to satisfactorily fit the data.
In the differential method of analysis we test the fit of the rate expression to
the data directly and without any integration. However, since the rate expression
is a differential equation, we must first find ( l / V ) ( d N l d t ) from the data before
attempting the fitting procedure.
There are advantages and disadvantages to each method. The integral method
is easy to use and is recommended when testing specific mechanisms, or relatively
simple rate expressions, or when the data are so scattered that we cannot reliably
find the derivatives needed in the differential method. The differential method
is useful in more complicated situations but requires more accurate or larger
amounts of data. The integral method can only test this or that particular mecha-
nism or rate form; the differential method can be used to develop or build up
a rate equation to fit the data.
In general, it is suggested that integral analysis be attempted first, and, if not
successful, that the differential method be tried.

3.1 CONSTANT-VOLUME BATCH REACTOR

When we mention the constant-volume batch reactor we are really referring to
the volume of reaction mixture, and not the volume of reactor. Thus, this term
actually means a constant-density reaction system. Most liquid-phase reactions
as well as all gas-phase reactions occurring in a constant-volume bomb fall in
this class.
In a constant-volume system the measure of reaction rate of component i be-
comes

or for ideal gases, where C = p/RT,

Thus, the rate of reaction of any component is given by the rate of change of
its concentration or partial pressure; so no matter how we choose to follow
the progress of the reaction, we must eventually relate this measure to the
concentration or partial pressure if we are to follow the rate of reaction.
For gas reactions with changing numbers of moles, a simple way of finding
the reaction rate is to follow the change in total pressure .n of the system. Let
us see how this is done.

Analysis of Total Pressure Data Obtained in a Constant-Volume System. For

isothermal gas reactions where the number of moles of material changes during
reaction, let us develop the general expression which relates the changing total
pressure of the system TI to the changing concentration or partial pressure of
any of the reaction components.
40 Chapter 3 Interpretation of Batch Reactor Data

Write the general stoichiometric equation, and under each term indicate the
number of moles of that component:
aA + bB +... = rR + sS +...

At time 0: N,, NBO N ~ o Nso Nineit

At time t: N , = NAo- ax NB = NBo- bx NR = NRO+ rx NS = NsO+ sx Nine,

Initially the total number of moles present in the system is

but at time t it is

where

Assuming that the ideal gas law holds, we may write for any reactant, say Ain
the system of volume V

Combining Eqs. 3 and 4 we obtain

Equation 5 gives the concentration or partial pressure of reactant A as a function

of the total pressure n at time t, initial partial pressure of A, pAo, and initial total
pressure of the system, .rr,.
Similarly, for any product R we can find

Equations 5 and 6 are the desired relationships between total pressure of the
system and the partial pressure of reacting materials.
It should be emphasized that if the precise stoichiometry is not known, or if
more than one stoichiometric equation is needed to represent the reaction, then
the "total pressure" procedure cannot be used.
 3.1 Constant-Volume Batch Reactor 41
The Conversion. Let us introduce one other useful term, the fractional conver-
sion, or the fraction of any reactant, say A, converted to something else, or the
fraction of A reacted away. We call this, simply, the conversion of A, with
symbol X A .
Suppose that NAois the initial amount of A in the reactor at time t = 0, and
that NA is the amount present at time t. Then the conversion of A in the constant
volume system is given by

and

We will develop the equations in this chapter in terms of concentration of reaction

components and also in terms of conversions.
Later we will relate X A and CA for the more general case where the volume
of the system does not stay constant.

Integral Method of Analysis of Data

General Procedure. The integral method of analysis always puts a particular
rate equation to the test by integrating and comparing the predicted C versus t
curve with the experimental C versus t data. If the fit is unsatisfactory, another
rate equation is guessed and tested. This procedure is shown and used in the
cases next treated. It should be noted that the integral method is especially useful
for fitting simple reaction types corresponding to elementary reactions. Let us
take up these kinetic forms.

Irreversible Unimolecular-Type First-Order Reactions. Consider the reaction

Suppose we wish to test the first-order rate equation of the following type,

for this reaction. Separating and integrating we obtain

42 Chapter 3 Interpretation of Batch Reactor Data

In terms of conversion (see Eqs. 7 and 8), the rate equation, Eq. 10, becomes

which on rearranging and integrating gives

A plot of In (1 - XA) or In (CA/CAo)vs. t, as shown in Fig. 3.1, gives a straight

line through the origin for this form of rate of equation. If the experimental data
seems to be better fitted by a curve than by a straight line, try another rate form
because the first-order reaction does not satisfactorily fit the data.

Caution. We should point out that equations such as

are first order but are not am'enable to this kind of analysis; hence, not all first-
order reactions can be treated as shown above.

Irreversible Bimolecular-Type Second-Order Reactions. Consider the re-

action

u f

Figure 3.1 Test for the first-order rate equation, Eq. 10.
 3.1 Constant-Volume Batch Reactor 43

with corresponding rate equation

Noting that the amounts of A and B that have reacted at any time t are equal
and given by CAoXA,we may write Eqs. 13a and b in terms of XA as

Letting M = CBoICA0
be the initial molar ratio of reactants, we obtain

which on separation and formal integration becomes

After breakdown into partial fractions, integration, and rearrangement, the final
result in a number of different forms is

Figure 3.2 shows two equivalent ways of obtaining a linear plot between the
concentration function and time for this second-order rate law.

0" 13 -e
-C &
Slope = (Ceo- CAo)k

C ~ O
Intercept = In - = In M Slope = (CBO- CAO)k
c ~ O
-
0 * 0 t
0 0
t t
Figure 3.2 Test for the bimolecular mechanism A + B -+ R with CAo# C,,,
or for the second-order reaction, Eq. 13.
44 Chapter 3 Interpretation of Batch Reactor Data

If C,, is much larger than CAo,C, remains approximately constant at all times,
and Eq. 14 approaches Eq. 11or 12 for the first-order reaction. Thus, the second-
order reaction becomes a pseudo first-order reaction.

Caution 1. In the special case where reactants are introduced in their stoichio-
metric ratio, the integrated rate expression becomes indeterminate and this
requires taking limits of quotients for evaluation. This difficulty is avoided if we
go back to the original differential rate expression and solve it for this particular
reactant ratio. Thus, for the second-order reaction with equal initial concentra-
tions of A and B, or for the reaction

the defining second-order differential equation becomes

which on integration yields

Plotting the variables as shown in Fig. 3.3 provides a test for this rate expression.
In practice we should choose reactant ratios either equal to or widely different
from the stoichiometric ratio.

Caution 2. The integrated expression depends on the stoichiometry as well as

the kinetics. To illustrate, if the reaction

0
t

Figure 3.3 Test for the bimolecular mechanisms, A +B + R with CAo =

or for the second-order reaction of Eq. 15.
 3.1 Constant-Volume Batch Reactor 45

is first order with respect to both A and B, hence second order overall, or

The integrated form is

When a stoichiometric reactant ratio is used the integrated form is

These two cautions apply to all reaction types. Thus, special forms for the
integrated expressions appear whenever reactants are used in stoichiometric
ratios, or when the reaction is not elementary.

Irreversible Trimolecular-Type Third-Order Reactions. For the reaction

A + B + D +products (204

let the rate equation be

or in terms of XA

On separation of variables, breakdown into partial fractions, and integration,

we obtain after manipulation

Now if CDois much larger than both CAoand CBo,the reaction becomes second
order and Eq. 21 reduces to Eq. 14.
46 Chapter 3 Interpretation of Batch Reactor Data

All trimolecular reactions found so far are of the form of Eq. 22 or 25. Thus

A+2B-R with d -C -
-rA= - - ~ k c C?
dt A B (22)

In terms of conversions the rate of reaction becomes

dX
- -A - k C i o (1 - XA)(M - 2XA)'
dt

where M = [Link] integration this gives

Similarly, for the reaction

A+B+R with -r,= dcA - kCACi

---
dt

integration gives

Empirical Rate Equations of nth Order. When the mechanism of reaction is

not known, we often attempt to fit the data with an nth-order rate equation of
the form

which on separation and integration yields

 3.1 Constant-Volume Batch Reactor 47

The order n cannot be found explicitly from Eq. 29, so a trial-and-error solution
must be made. This is not too difficult, however. Just select a value for n and
calculate k. The value of n which minimizes the variation in k is the desired
value of n.
One curious feature of this rate form is that reactions with order n > 1 can
never go to completion in finite time. On the other hand, for orders n < 1 this
rate form predicts that the reactant concentration will fall to zero and then
become negative at some finite time, found from Eq. 29, so

Since the real concentration cannot fall below zero we should not carry out
the integration beyond this time for n < [Link], as a consequence of this feature,
in real systems the observed fractional order will shift upward to unity as reactant
is depleted.

Zero-Order Reactions. A reaction is of zero order when the rate of reaction

is independent of the concentration of materials; thus

Integrating and noting that CAcan never become negative, we obtain directly

= kt for t <
CAo- CA= CAoXA

CA=O for t z -

which means that the conversion is proportional to time, as shown in Fig. 3.4.
As a rule, reactions are of zero order only in certain concentration ranges-the
higher concentrations. If the concentration is lowered far enough, we usually

Figure 3.4 Test for a zero-order reaction, or rate equation, Eq. 30.
48 Chapter 3 Interpretation of Batch Reactor Data

find that the reaction becomes concentration-dependent, in which case the order
rises from zero.
In general, zero-order reactions are those whose rates are determined by some
factor other than the concentration of the reacting materials, e.g., the intensity
of radiation within the vat for photochemical reactions, or the surface available
in certain solid catalyzed gas reactions. It is important, then, to define the rate
of zero-order reactions so that this other factor is included and properly ac-
counted for.

Overall Order of Irreversible Reactions from the Half-Life t,,. Sometimes, for
the irreversible reaction

we may write

If the reactants are present in their stoichiometric ratios, they will remain at
that ratio throughout the reaction. Thus, for reactants A and B at any time
CB/CA= Pla, and we may write

Integrating for n # 1 gives

Defining the half-life of the reaction, t,, , as the time needed for the concentration
of reactants to drop to one-half the original value, we obtain

This expression shows that a plot of log t,, vs. log CAogives a straight line of
slope 1 - n, as shown in Fig. 3.5.
The half-life method requires making a series of runs, each at a different initial
concentration, and shows that the fractional conversion in a given time rises
with increased concentration for orders greater than one, drops with increased
 3.1 Constant-Volume Batch Reactor 49

Eq33a>qSlope= 1- n
2 Y
00 ~ ~ ~ 0 r< 1
d e r Order > 1
9
J r =

log CAO

Figure 3.5 Overall order of reaction from

a series of half-life experiments, each at a
different initial concentration of reactant.

concentration for orders less than one, and is independent of initial concentration
for reactions of first order.
Numerous variations of this procedure are possible. For instance, by having
all but one component, say Ain large excess, we can find the order with respect
to that one component. For this situation the general expression reduces to

where

.
= k ( C k O e .) and CB CBo

And here is another variation of the half-life method.

Fractional Life Method t,. The half-life method can be extended to any frac-
tional life method in which the concentration of reactant drops to any fractional
value F = CA/CAoin time t,. The derivation is a direct extension of the half-life
method giving

Thus, a plot of log tF versus log CAo,as shown in Fig. 3.5, will give the reac-
tion order.
Example E3.1 illustrates this approach.

Irreversible Reactions in Parallel. Consider the simplest case, A decomposing

by two competing paths, both elementary reactions:
50 Chapter 3 Interpretation of Batch Reactor Data

The rates of change of the three components are given by

This is the first time we have encountered multiple reactions. For these in general,
if it is necessary to write N stoichiometric equations to describe what is happening,
then it is necessary to follow the decomposition of N reaction components to
describe the kinetics. Thus, in this system following C,, or CR,or C, alone will
not give both k , and k,. At least two components must be followed. Then,
from the stoichiometry, noting that C, + CR + Cs is constant, we can find the
concentration of the third component.
The k values are found using all three differential rate equations. First of all,
Eq. 34, which is of simple first order, is integrated to give

-In -= ( k , + k,)t

When plotted as in Fig. 3.6, the slope is k , + k,. Then dividing Eq. 35 by Eq.
36 we obtain the following (Fig. 3.6).

which when integrated gives simply

This result is shown in Fig. 3.6. Thus, the slope of a plot of CR versus Cs
gives the ratio kllk2. Knowing k,lk2 as well as k , + k, gives k , and k,. Typical
concentration-time curves of the three components in a batch reactor for the
case where CRo= Cso = 0 and k, > k, are shown in Fig. 3.7.
Reactions in parallel are considered in detail in Chapter 7.

Homogeneous Catalyzed Reactions. Suppose the reaction rate for a homoge-

neous catalyzed system is the sum of rates of both the uncatalyzed and cata-
lyzed reactions,
 3.1 Constant-Volume Batch Reactor 51

Figure 3.6 Evaluation of the rate constants for two competing elementary

first-order reactions of the type A \f R

.
S

with corresponding reaction rates

This means that the reaction would proceed even without a catalyst present and
that the rate of the catalyzed reaction is directly proportional to the catalyst
concentration. The overall rate of disappearance of reactant A is then

On integration, noting that the catalyst concentration remains unchanged, we

have

cA = -In (1 - X,) = ( k , + k2Cc)t= kobserved

-In - t
C~~

U
t
Figure 3.7 Typical concentration-time curves for competing reactions.
52 Chapter 3 Interpretation of Batch Reactor Data

Figure 3.8 Rate constants for a homogeneous catalyzed reaction

from a series of runs with different catalyst concentrations.

Making a series of runs with different catalyst concentrations allows us to find

k, and k,. This is done by plotting the observed k value against the catalyst
concentrations as shown in Fig. 3.8. The slope of such a plot is k, and the
intercept k,.

Autocatalytic Reactions. A reaction in which one of the products of reaction

acts as a catalyst is called an autocatalytic reaction. The simplest such reaction is

for which the rate equation is

Because the total number of moles of A and R remain unchanged as A is

consumed, we may write that at any time

Co = CA + CR = CAO+ CRO= constant

Thus, the rate equation becomes

Rearranging and breaking into partial fractions, we obtain

which on integration gives

 3.1 Constant-Volume Batch Reactor 53

(og~essin ti
I---------------

Low rate Start with

XA
High rate Parabolic
I I I
Low rate I I I

Figure 3.9 Conversion-time and rate-concentration curves for autocatalytic reaction

of Eq. 41. This shape is typical for this type of reaction.

In terms of the initial reactant ratio M = CRdCAOand fractional conversion of

A, this can be written as

For an autocatalytic reaction in a batch reactor some product R must be present

if the reaction is to proceed at all. Starting with a very small concentration of
R, we see qualitatively that the rate will rise as R is formed. At the other extreme,
when A is just about used up the rate must drop to zero. This result is given in
Fig. 3.9, which shows that the rate follows a parabola, with a maximum where
the concentrations of A and R are equal.
To test for an autocatalytic reaction, plot the .time and concentration coordi-
nates of Eq. 42 or 43, as shown in Fig. 3.10 and see whether a straight line passing
through zero is obtained.
Autocatalytic reactions are considered in more detail in Chapter 6.

Irreversible Reactions in Series. We first consider consecutive unimolecular-

type first-order reactions such as

Figure 3.10 Test for the autocatalytic reaction of Eq. 41.

54 Chapter 3 Interpretation of Batch Reactor Data

whose rate equations for the three components are

Let us start with a concentration CAOof A, no R or S present, and see how the
concentrations of the components change with time. By integration of Eq. 44
we find the concentration of A to be

To find the changing concentration of R, substitute the concentration of A from

Eq. 47 into the differential equation governing the rate of change of R, Eq. 45; thus

which is a first-order linear differential equation of the form

By multiplying through with the integrating factor elPdnthe solution is

\
yeJPdn= QeJPdxd r + constant
Applying this general procedure to the integration of Eq. 48, we find that the
integrating factor is ekzt. The constant of integration is found to be -klCAol
( k , - k,) from the initial conditions C,, = 0 at t = 0, and the final expression
for the changing concentration of R is

Noting that there is no change in total number of moles, the stoichiometry relates
the concentrations of reacting components by
 3.1 Constant-Volume Batch Reactor 55

which with Eqs. 47 and 49 gives

Thus, we have found how the concentrations of components A, R, and S vary

with time.
Now if k2 is much larger than k,, Eq. 50 reduces to

Cs = C,, (1 - e-ki'), k2 9 k1

In other words, the rate is determined by k , or the first step of the two-step
reaction.
If k , is much larger than k,, then

which is a first-order reaction governed by k,, the slower step in the two-step
reaction. Thus, in general, for any number of reactions in series it is the slowest
step that has the greatest influence on the overall reaction rate.
As may be expected, the values of k, and k2 also govern the location and
maximum concentration of R. This may be found by differentiating Eq. 49 and
setting dCRldt = 0. The time at which the maximum concentration of R occurs
is thus

t,, = --
h o e mean k2 - kl

The maximum concentration of R is found by combining Eqs. 49 and 51 to give

Figure 3.11 shows the general characteristics of the concentration-time curves

for the three components; A decreases exponentially, R rises to a maximum and
then falls, and S rises continuously, the greatest rate of increase of S occurring
where R is a maximum. In particular, this figure shows that one can evaluate k1
and k, by noting the maximum concentration of intermediate and the time when
this maximum is reached. Chapter 8 covers series reactions in more detail.
For a longer chain of reactions, say

the treatment is similar, though more cumbersome than the two-step reaction
just considered. Figure 3.12 illustrates typical concentration-time curves for
this situation.
56 Chapter 3 Interpretation o f Batch Reactor Data

Eq. 52

0
+Eq. 51-I
t

Figure 3.11 Typical concentration-time curves

for consecutive first-order reactions.

First-OrderReversible Reactions. Though no reaction ever goes to completion,

we can consider many reactions to be essentially irreversible because of the large
value of the equilibrium constant. These are the situations we have examined
up to this point. Let us now consider reactions for which complete conversion
cannot be assumed. The simplest case is the opposed unimolecular-type reaction

kl
A R, Kc = K = equilibrium constant (534
k2

Starting with a concentration ratio M = CRdCAothe rate equation is

Maximum of T curve

Figure 3.12 Concentration-time curves for a chain of successive first-

order reactions. Only for the last two compounds do the maximum and
the inflection points occur at the same time.
 3.1 Constant-Volume Batch Reactor 57

Now at equilibrium dC,/dt = 0. Hence from Eq. 53 we find the fractional

conversion of A at equilibrium conditions to be

and the equilibrium constant to be

Combining the above three equations we obtain, in terms of the equilibrium con-
version,

With conversions measured in terms of XAe,this may be looked on as a pseudo

first-order irreversible reaction which on integration gives

A plot of -In (1 - XA/XAe) vs. t, as shown in Fig. 3.13, gives a straight line.
The similarity between equations for the first-order irreversible and reversible
reactions can be seen by comparing Eq. 12 with Eq. 54 or by comparing Fig. 3.1
with Fig. 3.13. Thus, the irreversible reaction is simply the special case of the
reversible reaction in which CA, = 0, or XAe= 1, or Kc = m.

t
Figure 3.13 Test for the unimolecular type
reversible reactions of Eq. 53.
58 Chapter 3 Interpretation of Batch Reactor Data

Second-Order Reversible Reactions. For the bimolecular-type second-order re-

actions

with the restrictions that CAo= C,, and CRo= Cso = 0, the integrated rate
equations for A and B are all identical, as follows

A plot as shown in Fig. 3.14 can then be used to test the adequacy of these kinetics.

Reversible Reactions in General. For orders other than one or two, integration
of the rate equation becomes cumbersome. So if Eq. 54 or 56 is not able to fit
the data, then the search for an adequate rate equation is best done by the
differential method.

Figure 3.14 Test for the reversible bimolecular reactions of Eq. 55.
 3.1 Constant-Volume Batch Reactor 59

Figure 3.15 Behavior of a reaction that follows Eq. 57.

Reactions of Shifting Order. In searching for a kinetic equation it may be found

that the data are well fitted by one reaction order at high concentrations but by
another order at low concentrations. Consider the reaction

A-*R with d C-
- r A = -- -~ k l C ~
dt 1 + k2CA

From this rate equation we see

At high CA-the reaction is of zero order with rate constant kllk2
(or kzCA 1)

At low CA-the reaction is of first order with rate constant k,

(or kzCA 1)

This behavior is shown in Fig. 3.15.

To apply the integral method, separate variables and integrate Eq. 57. This gives

To linearize, rearrange Eq. 58a to give

Two ways to test this rate form are then shown in Fig. 3.16.
60 Chapter 3 Interpretation of Batch Reactor Data

I Ea. 5 8 c 7 --
-2 lu
a a
3'
-
215 I Slope = k l

Y
'

I 'Intercept
l = -k2 Intercept = - -1
k2
Figure 3.16 Test of the rate equation, Eq. 57, by integral analysis.

By similar reasoning to the above we can show that the general rate form

shifts from order m - n at high concentration to order m at low concentration,

the transition taking place where k2Ca = 1. This type of equation can then be
used to fit data of any two orders.
Another form which can account for this shift is

Mechanism studies may suggest which form to use. In any case, if one of these
equation forms will fit the data, so will the other.
The rate form of Eq. 57 and some of its generalizations are used to represent
a number of widely different kinds of reactions. For example, in homogeneous
systems this form is used for enzyme-catalyzed reactions where it is suggested
by mechanistic studies (see the Michaelis-Menten mechanism in Chap. 2 and in
Chap. 27). It is also used to represent the kinetics of surface-catalyzed reactions.
In mechanistic studies, this form of equation appears whenever the rate-con-
trolling step of a reaction is viewed to involve the association of reactant with
some quantity that is present in limited but fixed amounts; for example, the
association of reactant with enzyme to form a complex, or the association of
gaseous reactant with an active site on the catalyst surface.

FIND A RATE EQUATION USING THE INTEGRAL METHOD

Reactant A decomposes in a batch reactor

 3.1 Constant-Volume Batch Reactor 61
The composition of A in the reactor is measured at various times with results
shown in the following columns 1and 2. Find a rate equation to represent the data.

Column 1 Column 2 Column 3 Column 4

Time Concentration
t, S CA,mollliter

V Y

Reported data Calculated

SOLUTION

Guess First-Order Kinetics. Start by guessing the simplest rate form, or first-
order kinetics. This means that In CAo/CAvs, t should give a straight line, see
Eq. 11 or 12, or Fig. 3.1. So column 3 is calculated and the plot of Fig. [Link] is
made. Unfortunately, this does not give a straight line, so first-order kinetics
cannot reasonably represent the data, and we must guess another rate form.

First-order kinetics should

fall on a straight line.
These data do not,

In -
c~~
CA

columns 1 and 3

0 50 100 200 300

Time t, s
Figure [Link]

Guess Second-Order Kinetics. Equation 16 tells that 1/C, vs. t should give a
straight line. So calculate column 4, plot column 1vs, column 4, as shown in Fig.
[Link]. Again, this does not give a straight line, so a second-order kinetic form
is rejected.
 62 Chapter 3 Interpretation of Batch Reactor Data
h
1.0 -
Second-order kinetics should

0.8 - These data do not,

0.6 -
-
1
cA
0.4 -

columns 1 and 4

0 1 I I I I *
0 50 100 200 300
Time t, s
Figure [Link]

Guess nth-Order Kinetics. Let's plan to use the fractional life method with
F = 80%. Then Eq. 33b becomes

Next take logarithms

The procedure is as follows. First accurately plot the C, vs. t data, draw a smooth
curve to represent the data (most important) as shown in Fig. [Link], then pick
CAo= 10, 5, and 2 and fill in the following table from this figure.

A' end Time needed

A
'0 (= o.8cAo) t~,?3 log t~ log CAo

From the curve, not the data

I Next, plot log t, vs. log CAo,as shown in Fig. [Link] and find the slope.
 3.1 Constant-Volume Batch Reactor 63

A smooth curve does not

necessarily pass through
all the data points
A
1.75 -

:. n = 1.4
1.5 -
M
-
1.25 -

180 215
0 I I * 1.o I I *
0 100 200 300 0 0.5 1.0

Figures [Link] and E3.1d

We now have the reaction order. To evaluate the rate constant, take any point
on the CA vs. t curve. Pick CAo = 10, for which t, = 18.5 s. Replacing all values
into Eq. (i) gives

from which

k = 0.005

Therefore, the rate equation that represents this reaction is

mol
.s
rno10.~ liter. s

Differential Method of Analysis of Data

The differential method of analysis deals directly with the differential rate equa-
tion to be tested, evaluating all terms in the equation including the derivative
dC,ldt, and testing the goodness of fit of the equation with experiment.
The procedure is as follows.
1. Plot the CA vs. t data, and then by eye carefully draw a smooth curve to
represent the data. This curve most likely will not pass through all the
experimental points.
2. Determine the slope of this curve at suitably selected concentration values.
These slopes dC,ldt = r, are the rates of reaction at these compositions.
64 Chapter 3 Interpretation of Batch Reactor Data

-5
w
-d
-
b-
-8
,4

-
* *
Guess f (CA) log CA

Figure 3.17 Test for the particular rate Figure 3.18 Test for an nth-order rate
form -r, = kf(C,) by the differential form by the differential method.
method.

3. Now search for a rate expression to represent this rA vs. CAdata, either by
(a) picking and testing a particular rate form, -rA = kf (C,), see Fig. 17, or
(b) testing an nth-order form -rA = k c : by taking logarithms of the
rate equation (see Fig. 3.18).
With certain simpler rate equations, however, mathematical manipulation may
be able to yield an expression suitable for graphical testing. As an example,
consider a set of C, vs. t data to which we want to fit the M-M equation

which has already been treated by the integral method of analysis. By the differen-
tial method we can obtain -rA vs. C A . However, how do we make a straight-
line plot to evaluate k , and k,? As suggested, let us manipulate Eq. 57 to obtain
a more useful expression. Thus, taking reciprocals we obtain

and a plot of I/(-?-,) vs. l / C Ais linear, as shown in Fig. 3.19.

Alternatively, a different manipulation (multiply Eq. 61 by k,(-rA)lk2)yields
another form, also suitable for testing, thus

A plot of -rA vs. ( - r A ) / C Ais linear, as shown in Fig. 3.19.

Whenever a rate equation can be manipulated to give a linear plot, this becomes
a simple way of testing the equation. So, with any given problem we must use
good judgment in planning our experimental program.
 3.1 Constant-Volume Batch Reactor 65

-
b\. -
/
k;
Eq. 61 Eq. 62

-rA

Slope = -
kl

Figure 3.19 Two ways of testing the rate equation -rA = k,C,/(l + k,CA)
by differential analysis.

FIND A RATE EQUATION TO FIT A SET OF DATA USING

THE DIFFERENTIAL METHOD

Try to fit an nth-order rate equation to the concentration vs. time data of Exam-
ple 3.1.

The data are tabulated in the following columns 1 and 2 and are plotted in
Fig. E 3 . 2 ~ .

Smooth curve to
8

Reported experimental
*
6 data, seven points
E
3

0
0 100 200 300

I Figure E3.2a
Time r, s
66 Chapter 3 Interpretation of Batch Reactor Data

Column 1 Column 2 Column 3 Column 4 Column 5

Time Concentration Slope, from Fig. E 3 . 2 ~
t, s C A , mollliter (dC,ldt) log10 ( - d C ~ / d t ) log10 CA
0 10 (10 - O)/(O - 7 5 ) = -0.1333 -0.875 [Link]
20 8 (10 - O)/(-3 -94) = -0.1031 -0.987 0.903
40 6 (10 - O)/(-21 - 131) = -0.0658 -1.182 0.773
60 5 (8 - O)/(-15 - 180) = -0.0410 -1.387 0.699
120 3 (6 - O)/(-10 - 252) = -0.0238 -1.623 0.477
180 2 ( 4 - 1)/(24 - 255) = -0.0108 - 1.967 0.301
300 1 ( 3 - I ) / ( - 1 0 - 300) = -0.0065 -2.187 0.000

Now carefully draw a smooth curve to represent the data and at CA = 10, 8,
6, 5, 3, 2, 1, draw tangents to the curve, and evaluate them (see column 3).
Next, to fit an nth-order rate equation to this data, or

dC* - kc",
-rA= dt

take logarithms of both sides (see columns 3 and 4), or

loglo
(-x = loglOk+ nlogl0CA
- - - - - -dCA)
Y
-----

intercept J Lop:
and plot as in Fig. E3.2b. The slope and intercept of the best line gives n and k
(see Fig.E3.2b). So the rate equation is

~ C =A(0.005 literO," ~ i 4 3 -
-rA = - -
dt mo1°,43.s , mol )
liter. s

Warning. In step 1, if you use a computer to fit a polynomial to the data it could
lead to disaster. For example, consider fitting a sixth-degree polynomial to the
seven data points, or an (n - 1) degree polynomial to n points.
Fitting by eye you'd get a smooth curve, as shown in Fig. E3.2~.But if a
computer is used to obtain a polynomial that would pass through all the points
the result would very well be as shown in Fig. E3.2d.
Now, which of these curves makes more sense and which would you use? This
is why we say "draw a smooth curve by eye to represent the data." But beware,
to draw such a curve is not that simple. Take care.
 3.2 Varying-Volume Batch Reactor 67

I \intercept = loglo k = -2.305

k = 0.005
Figure E3.2b

This curve does This curve passes

not pass through through all points

Figure E3.2c, d The same seven points fitted by curves two different ways.

3.2 VARYING-VOLUME BATCH REACTOR

These reactors are much more complex than the simple constant-volume batch
reactor. Their main use would be in the microprocessing field where a capillary
tube with a movable bead would represent the reactor (see Fig. 3.20).
The progress of the reaction is followed by noting the movement of the bead
with time, a much simpler procedure than trying to measure the composition of
the mixture, especially for microreactors. Thus,

V , = initial volume of the reactor

V = the volume at time t.
68 Chapter 3 Interpretation of Batch Reactor Data

I Reactor, [Movable bead /

Figure 3.20 A varying-volume batch reactor.

This kind of reactor can be used for isothermal constant pressure operations, of
reactions having a single stoichiometry. For such systems the volume is linearly
related to the conversion, or

where E A is the fractional change in volume of the system between no conversion

and complete conversion of reactant A. Thus

As an example of the use of E,, consider the isothermal gas-phase reaction

By starting with pure reactant A,

but with 50% inerts present at the start, two volumes of reactant mixture yield,
on complete conversion, five volumes of product mixture. In this case

We see, then, that E A accounts for both the reaction stoichiometry and the
presence of inerts. Noting that
 3.2 Varying-VolumeBatch Reactor 69

we have, on combining with Eq. 63,

Thus

which is the relationship between conversion and concentration for isothermal

varying-volume (or varying-density) systems satisfying the linearity assumption
of Eq. 63.
The rate of reaction (disappearance of component A), is, in general

Replacing V from Eq. 63a and NAfrom Eq. 65 we end up with the rate in terms
of the conversion

or in terms of volume, from Eqs. 63

Differential Method of Analysis

The procedure for differential analysis of isothermal varying volume data is the
same as for the constant-volume situation except that we replace

dC, with -- c~~

C ~ odV or - d (In V)
-
dt VEA dt EA dt

This means plot In V vs. t and take slopes.

Integral Method of Analysis

Unfortunately, only a few of the simpler rate forms integrate to give manageable
V vs, t expressions. Let us look at these.
70 Chapter 3 Interpretation of Batch Reactor Data

Zero-Order Reactions For a homogeneous zero-order reaction the rate of

change of any reactant A is independent of the concentration of materials, or

Integrating gives

As shown in Fig. 3.21, the logarithm of the fractional change in volume versus
time yields a straight line of slope ksA/CAO.

First-Order Reactions. For a unimolecular-type first-order reaction the rate of

change of reactant A is

Replacing X , by V from Eqs. 63 and integrating gives

A semilogarithmic plot of Eq. 72, as shown in Fig. 3.22, yields a straight line of
slope k.

Second-Order Reactions. For a bimolecular-type second-order reaction

2A +products

Figure 3.21 Test for a homogeneous zero-order reaction, Eq. 69,

in a constant-pressure, varying volume reactor.
 3.2 Varying-VolumeBatch Reactor 71

Eq. 72

c Slope = k
I

0 t

Figure 3.22 Test for a first-order reaction, Eq. 71,

in a constant-pressure, varying-volume reactor.

or
A + B -+products, with CAO= CBO

the rate is given by

Replacing XAby V from Eqs. 63 and then integrating gives, after much algebraic
manipulation,

Figure 3.23 shows how to test for those kinetics.

Figure 3.23 Test for the second-order

reaction, Eq. 73, in a constant-pressure,
varying-volume reactor.
72 Chapter 3 Interpretation of Batch Reactor Data

nth-Order and Other Reactions. For all rate forms other than zero-, first-, and
second-order the integral method of analysis is not useful.

3.3 TEMPERATURE AND REACTION RATE

So far we have examined the effect of concentration of reactants and products
on the rate of reaction, all at a given temperature level. To obtain the complete
rate equation, we also need to know the role of temperature on reaction rate.
Now in a typical rate equation we have

and it is the reaction rate constant, the concentration-independent term, which

is affected by the temperature, whereas the concentration-dependent terms f(C)
usually remain unchanged at different temperatures.
Chemical theory predicts that the rate constant should be temperature-depen-
dent in the following manner:

However, since the exponential term is much more temperature-sensitive than

the power term, we can reasonably consider the rate constants to vary approxi-
mately as e-EIRT.
Thus, after finding the concentration dependency of the reaction rate, we can
then examine for the variation of the rate constant with temperature by an
Arrhenius-type relationship -
-

This is conveniently determined by plotting In k versus 1/T,as shown in Fig. 3.24.

If the rate constant is found at two different temperatures we have from
Chap. 2,

1IT
Figure 3.24 Temperature dependency of a reaction
according to Arrhenius' law.
 3.3 Temperature and Reaction Rate 73
Finally, as mentioned in Chap. 2, a shift in E with temperature reflects a change
in controlling mechanism of reaction. Since this is likely to be accompanied by
a change in concentration dependency, this possibility should also be examined.

Warning on Using Pressure Measures. When dealing with gases, engineers and
chemists often measure compositions in terms of partial and total pressures, and
then develop their rate equations in terms of pressures, without realizing that
this can lead to problems. The reason is that the activation energy calculated
when using these units is incorrect. Let us illustrate.

CORRECT AND INCORRECT E VALUES

Experimental studies of a specific decomposition of A in a batch reactor using

pressure units show exactly the same rate at two different temperatures:

at 400 K -rA = 2.3 p i

where
at 500 K -rA = 2.3pi
PA = [atml

(a) Evaluate the activation using these units

(b) Transform the rate expressions into concentration units and then evaluate
the activation energy.
The pressure is not excessive, so the ideal gas law can be used.

/ SOLUTION

(a) Using Pressure Units. We see right away that a change in temperature
does not affect the rate of reaction. This means that

E =0

Alternatively, we can find E by going through the calculations. Thus

I hence replacing in Eq. 75 shows that

E =0

(b) Transform pA into C,, then $nd E . First write the rate equations with
all units shown:
 74 Chapter 3 Interpretation of Batch Reactor Data

Next change p, to C,. From ideal gas law

1 Combining the two preceding equations

mol m3.
- atm
-'A1 = 2.3 m3. s . atm2 82.06 X )2 (400 K)2
mole K

where

At 500 K, similarly

where

Here we see that in concentration units the rate constants are not indepen-
dent of temperature. Evaluating the activation energy from Eq. 75, and
replacing numbers gives

E = 7394
mol

I This example shows that E values differ when either p or C used to

measure concentrations of materials.

Final Notes
1. Chemistry (collision theory or transition state theory) has developed the
equations for reaction rates and activation energies in terms of concen-
tration.
2. Literature tabulations for E and -rA for homogeneous reactions are nor-
mally based on concentrations. The clue to this is that the units for the
rate constant are often s-l, literlmol. s, etc., without pressure appearing in
the units.
 3.4 The Search for a Rate Equation 75

3. It is a good idea when making runs at different temperatures first to change

all p values to C values by using the relationships

pA = CART for ideal gases

pA = zCART for nonideal gases, where z = compressibility factor

and then proceed to solve the problem. This will avoid confusion on units
later on, especially if the reaction is reversible or involves liquids and/or
solids as well as gases.

3.4 THE SEARCH FOR A RATE EQUATION

In searching for a rate equation and mechanism to fit a set of experimental data,
we would like answers to two questions:
1. Have we the correct mechanism and corresponding type of rate equation?
2. Once we have the right form of rate equation, do we have the best values
for the rate constants in the equation?
The difficult question to answer is the first of the preceding two questions. Let
us see why this is so.
Suppose we have a set of data and we wish to find out whether any one of
the families of curves-parabolas, cubics, hyperbolas, exponentials, etc., each
representing a different rate family-really fits these data better than any other.
This question cannot be answered simply; neither can high-powered mathemati-
cal or statistical methods help in deciding for us. The one exception to this
conclusion occurs when one of the families being compared is a straight line.
For this situation we can simply, consistently, and fairly reliably tell whether the
straight line does not reasonably fit the data. Thus, we have what is essentially
a negative test, one that allows us to reject a straight line family when there is
sufficient evidence against it.
All the rate equations in this chapter were manipulated mathematically into
a linearized form because of this particular property of the family of straight
lines that allows it to be tested and rejected.
Three methods are commonly used to test for the linearity of a set of points.
These are as follows:

Calculation of k from Individual Data Points. With a rate equation at hand,

the rate constant can be found for each experimental point by either the integral
or differential method. If no trend in k values is discernible, the rate equation
is considered to be satisfactory and the k values are averaged.
Now the k values calculated this way are the slopes of lines joining the individ-
ual points to the origin. So for the same magnitude of scatter on the graph the
k values calculated for points near the origin (low conversion) will vary widely,
whereas those calculated for points far from the origin will show little variation
(Fig. 3.25). This fact can make it difficult to decide whether k is constant and,
if so, what is its best mean value.
76 Chapter 3 Interpretation of Batch Reactor Data

Figure 3.25 How the location of the experimental points influences the
scatter in calculated k values.

Calculation of k from Pairs of Data Points. k values can be calculated from

successive pairs of experimental points. For large data scatter, however, or for
points close together, this procedure will give widely different k values from
which k,,,, will be difficult to determine. In fact, finding k,,,, by this procedure
for points located at equal intervals on the x-axis is equivalent to considering
only the two extreme data points while ignoring all the data points in between.
This fact can easily be verified. Figure 3.26 illustrates this procedure.
This is a poor method in all respects and is not recommended for testing the
linearity of data or for finding mean values of rate constants.

Graphical Method of Fitting Data. Actually, the preceding methods do not

require making a plot of the data to obtain k values. With the graphical method
the data are plotted and then examined for deviations from linearity. The decision
whether a straight line gives a satisfactory fit is usually made intuitively by
using good judgment when looking at the data. When in doubt we should take
more data.
The graphical procedure is probably the safest, soundest, and most reliable
method for evaluating the fit of rate equations to the data, and should be used
whenever possible. For this reason we stress this method here.

Figure 3.26 Calculated k values from successive

experimental points are likely to fluctuate
widely.
 Problems 77

RELATED READINGS

Frost, A. A., and Pearson, R. G., Kinetics and Mechanism, 2nd ed., John Wiley & Sons,
New York, 1961.
Laidler, K. J., Chemical Kinetics, 2nd ed., McGraw-Hill, New York, 1965.

PROBLEMS

3.1. If -rA = -(dC,ldt) = 0.2 mol/[Link] when CA = 1 mollliter, what is

the rate of reaction when CA = 10 mollliter?
Note: the order of reaction is not known.

3.2. Liquid A decomposes by first-order kinetics, and in a batch reactor 50%

of A is converted in a 5-minute run. How much longer would it take to
reach 75% conversion?

3.3. Repeat the previous problem for second-order kinetics.

3.4. A 10-minute experimental run shows that 75% of liquid reactant is con-
verted to product by a %-order rate. What would be the fraction converted
in a half-hour run?

3.5. In a homogeneous isothermal liquid polymerization, 20% of the monomer

disappears in 34 minutes for initial monomer concentration of 0.04 and
also for 0.8 mollliter. What rate equation represents the disappearance of
the monomer?

3.6. After 8 minutes in a batch reactor, reactant (CAO= 1 mollliter) is 80%

converted; after 18 minutes, conversion is 90%. Find a rate equation to
represent this reaction.

Snake-Eyes Magoo is a man of habit. For instance, his Friday evenings are
all alike-into the joint with his week's salary of $180, steady gambling at
"2-up" for two hours, then home to his family leaving $45 behind. Snake
Eyes's betting pattern is predictable. He always bets in amounts propor-
tional to his cash at hand, and his losses are also predictable-at a rate
proportional to his cash at hand. This week Snake-Eyes received a raise
in salary, so he played for three hours, but as usual went home with $135.
How much was his raise?

3.8. Find the overall order of the irreversible reaction

78 Chapter 3 Interpretation of Batch Reactor Data

from the following constant-volume data using equimolar amounts of hy-

drogen and nitric oxide:

Total pressure, mm Hg
Half-life, sec 1 200 240 280 320 360
265 186 11.5 104 67

3.9. The first-order reversible liquid reaction

takes place in a batch reactor. After 8 minutes, conversion of A is 33.3%

while equilibrium conversion is 66.7%. Find the rate equation for this re-
action.

3.10. Aqueous A reacts to form R (A -,R) and in the first minute in a batch
reactor its concentration drops from CAo= 2.03 mollliter to CAf = 1.97
mollliter. Find the rate equation for the reaction if the kinetics are second-
order with respect to A.

3.11. Aqueous A at a concentration C,, = 1 mollliter is introduced into a batch

reactor where it reacts away to form product R according to stoichiometry
A -,R. The concentration of A in the reactor is monitored at various
times, as shown below:
t, min 0 100 200 300 400
CA,mol/m3 1000 500 333 250 200
For CAo= 500 mol/m3 find the conversion of reactant after 5 hours in the
batch reactor.

3.12. Find the rate for the reaction of Problem 11.

3.13. Betahundert Bashby likes to play the gaming tables for relaxation. He does
not expect to win, and he doesn't, so he picks games in which losses are
a given small fraction of the money bet. He plays steadily without a break,
and the sizes of his bets are proportional to the money he has. If at
"galloping dominoes" it takes him 4 hours to lose half of his money and
it takes him 2 hours to lose half of his money at "chuk-a-luck," how long
can he play both games simultaneously if he starts with $1000 and quits
when he has $10 left, which is just enough for a quick nip and carfare home?

3.14. For the elementary reactions in series

kl k2 CA = CAO,
A-R-S, k, = k,, att = 0
C,, = Cso= 0

find the maximum concentration of R and when it is reached.

 Problems 79

3.15. At room temperature sucrose is hydrolyzed by the catalytic action of the

enzyme sucrase as follows:

sucrase
sucrose -products

Starting with a sucrose concentration CAo= 1.0 millimol/liter and an en-

zyme concentration CEO= 0.01 millimollliter, the following kinetic data
are obtained in a batch reactor (concentrations calculated from optical
rotation measurements):

Determine whether these data can be reasonably fitted by a kinetic equation

of the Michaelis-Menten type, or

-YA =
C~
k3
where CM= Michaelis constant
CA+ Chl

If the fit is reasonable, evaluate the constants k3 and [Link] by the

integral method.

3.16. Repeat the above problem, except this time solve by the differential
method.

3.17. An ampoule of radioactive Kr-89 (half life = 76 minutes) is set aside for
a day. What does this do to the activity of the ampoule? Note that radioac-
tive decay is a first-order process.

3.18. Enzyme E catalyzes the transformation of reactant A to product R as

follows:

enzyme 200CACEO mol

A-R, -rA=
2 + CA liter. min

If we introduce enzyme (CEO= 0.001 mollliter) and reactant (CAo= 10

mollliter) into a batch reactor and let the reaction proceed, find the time
needed for the concentration of reactant to drop to 0.025 mollliter. Note
that the concentration of enzyme remains unchanged during the reaction.

3.19. Find the conversion after 1 hour in a batch reactor for

mol
A-R, -rA = 3Ci5- CAo= I mollliter
liter hr '
80 Chapter 3 Interpretation of Batch Reactor Data

Table P3.20
C,H,SO,H, C,H,S04H,
t, min mollliter t, min mollliter

3.20. M. Hellin and J. C. Jungers, Bull. soc. chim. France, 386 (1957), present
the data in Table P3.20 on the reaction of sulfuric acid with diethylsulfate
in aqueous solution at 22.9OC:

Initial concentrations of H2S04and (C,H5),SO4 are each 5.5 mollliter. Find

a rate equation for this reaction.

3.21. A small reaction bomb fitted with a sensitive pressure-measuring device

is flushed out and then filled with pure reactant A at 1-atm pressure. The
operation is carried out at 25OC, a temperature low enough that the reaction
does not proceed to any appreciable extent. The temperature is then raised
as rapidly as possible to 100°C by plunging the bomb into boiling water,
and the readings in Table P3.21 are obtained. The stoichiometry of the
reaction is 2A -+B, and after leaving the bomb in the bath over the
weekend the contents are analyzed for A; none can be found. Find a rate
equation in units of moles, liters, and minutes which will satisfactorily fit
the data.

Table P3.21
T, min n, atm T, min n-, atm
1 1.14 7 0.850
2 1.04 8 0.832
3 0.982 9 0.815
4 0.940 10 0.800
5 0.905 15 0.754
6 0.870 20 0.728
 Problems 81
3.22. For the reaction A -+R, second-order kinetics and C,, = 1 mollliter, we
get 50% conversion after 1 hour in a batch reactor. What will be the
conversion and concentration of A after 1 hour if C,, = 10 mollliter?

3.23. For the decomposition A R, C,, = 1 mollliter, in a batch reactor

-+

conversion is 75% after 1 hour, and is just complete after 2 hours. Find a
rate equation to represent these kinetics.

3.24. In the presence of a homogeneous catalyst of given concentration, aqueous

reactant A is converted to product at the following rates, and C, alone
determines this rate:
C,, mollliter 1 2 4 6 7 9 1 2
-r,,rnol/[Link] 0.06 0.1 0.25 1.0 2.0 1.0 0.5

We plan to run this reaction in a batch reactor at the same catalyst concen-
tration as used in getting the above data. Find the time needed to lower
the concentration of A from CAo= 10 mollliter to CAf= 2 mollliter.

3.25. The following data are obtained at O°C in a constant-volume batch reactor
using pure gaseous A:
Time,min
Partial pressure of A , m / 0
760
2
600
4
175
6
390
8
320
10
275
12
240
14
215
m
150

The stoichiometry of the decomposition is A -.2.5R. Find a rate equation

which satisfactorily represents this decomposition.

3.26. Example [Link] showed how to find a rate equation by using the fractional
life method where F = 80%. Take the data from that example and find
the rate equation by using the half-life method. As a suggestion, why not
take C, = 10, 6, and 2?

3.27. When a concentrated urea solution is stored it slowly condenses to biuret

by the following elementary reaction:

To study the rate of condensation a sample of urea (C = 20 mollliter) is

stored at 100°C and after 7 hr 40 min we find that 1 mol% has turned into
biuret. Find the rate equation for this condensation reaction. [Data from
W. M. Butt, Pak. I. Ch. E., 1, 99 (1973).]

3.28. The presence of substance C seems to increase the rate of reaction of A

+
and B, A B + AB. We suspect that C acts catalytically by combining
with one of the reactants to form an intermediate, which then reacts further.
From the rate data in Table P3.28 suggest a mechanism and rate equation
for this reaction.
82 Chapter 3 Interpretation of Batch Reactor Data

Table P3.28

3.29. Find the first-order rate constant for the disappearance of A in the gas
reaction 2A + R if, on holding the pressure constant, the volume of the
reaction mixture, starting with 80% A, decreases by 20% in 3 min.

3.30. Find the first-order rate constant for the disappearance of A in the gas
reaction A -+ 1.6R if the volume of the reaction mixture, starting with
pure Aincreases by 50% in 4 min. The total pressure within the system
stays constant at 1.2 atm, and the temperature is 25°C.

3.31. The thermal decomposition of hydrogen iodide

is reported by M. Bodenstein [[Link]. chem., 29, 295 (1899)l as follows:

Find the complete rate equation for this reaction. Use units of joules, moles,
cm3. and seconds.
 Introduction to Reactor Design

4.1 GENERAL DISCUSSION

So far we have considered the mathematical expression called the rate equation
which describes the progress of a homogeneous reaction. The rate equation for
a reacting component i is an intensive measure, and it tells how rapidly compo-
nent i forms or disappears in a given environment as a function of the conditions
there, or

= f (conditions within the region of volume V)

This is a differential expression.

In reactor design we want to know what size and type of reactor and method
of operation are best for a given job. Because this may require that the conditions
in the reactor vary with position as well as time, this question can only be
answered by a proper integration of the rate equation for the operation. This
may pose difficulties because the temperature and composition of the reacting
fluid may vary from point to point within the reactor, depending on the endother-
mic or exothermic character of the reaction, the rate of heat addition or removal
from the system, and the flow pattern of fluid through the vessel. In effect, then,
many factors must be accounted for in predicting the performance of a reactor.
How best to treat these factors is the main problem of reactor design.
Equipment in which homogeneous reactions are effected can be one of three
general types; the batch, the steady-state flow, and the unsteady-state flow or
semibatch reactor. The last classification includes all reactors that do not fall into
the first two categories. These types are shown in Fig. 4.1.
Let us briefly indicate the particular features and the main areas of application
of these reactor types. Naturally these remarks will be amplified further along
in the text. The batch reactor is simple, needs little supporting equipment, and
is therefore ideal for small-scale experimental studies on reaction kinetics. Indus-
trially it is used when relatively small amounts of material are to be treated. The
steady-state flow reactor is ideal for industrial purposes when large quantities
of material are to be processed and when the rate of reaction is fairly high to
extremely high. Supporting equipment needs are great; however, extremely good
84 Chapter 4 Introduction to Reactor Design

Composltlon at any
polnt IS unchanged
wlth tlme

Composltlon
changes

Volume and Volume IS constan

composltlon but composltlon
change IS unchanged changes
(c) (4 (4
Figure 4.1 Broad classification of reactor types. (a) The batch reactor. (b) The steady-state
flow reactor. (c), ( d ) ,and (e) Various forms of the semibatch reactor.

product quality control can be obtained. As may be expected, this is the reactor
that is widely used in the oil industry. The semibatch reactor is a flexible system
but is more difficult to analyze than the other reactor types. It offers good control
of reaction speed because the reaction proceeds as reactants are added. Such
reactors are used in a variety of applications from the calorimetric titrations in
the laboratory to the large open hearth furnaces for steel production.
The starting point for all design is the material balance expressed for any

[ [ [
reactant (or product). Thus, as illustrated in Fig. 4.2, we have

rate of reactant rate of

flow
rzz:itinto = rz:t loss due to + ~ccumulation]
flow out + chemical reaction of reactant (1)
element of element within the element in element
of volume of volume of volume of volume

Element of reactor volume

/

Reactant disappears by reaction

within the element
!- Reactant accumulates
within the element
Figure 4.2 Material balance for an element of volume
of the reactor.
 4.1 General Discussion 85

r Element of reactor volume

Heat disappears by reaction

within the element
Heat accumulates
within the element
Figure 4.3 Energy balance for an element of volume
of the reactor.

Where the composition within the reactor is uniform (independent of position),

the accounting may be made over the whole reactor. Where the composition is
not uniform, it must be made over a differential element of volume and then
integrated across the whole reactor for the appropriate flow and concentration
conditions. For the various reactor types this equation simplifies one way or
another, and the resultant expression when integrated gives the basic performance
equation for that type of unit. Thus, in the batch reactor the first two terms are
zero; in the steady-state flow reactor the fourth term disappears; for the semibatch
reactor all four terms may have to be considered.
In nonisothermal operations energy balances must be used in conjunction with
material balances. Thus, as illustrated in Fig. 4.3, we have

rate of
rate of
rate of heat rate of heat disappearance
accumulation
flow into flow out of

volume

Again, depending on circumstances, this accounting may be made either about

a differential element of reactor or about the reactor as a whole.
The material balance of Eq. 1and the energy balance of Eq. 2 are tied together
by their third terms because the heat effect is produced by the reaction itself.
Since Eqs. 1 and 2 are the starting points for all design, we consider their
integration for a variety of situations of increasing complexity in the chapters
to follow.
When we can predict the response of the reacting system to changes in op-
erating conditions (how rates and equilibrium conversion change with tempera-
ture and pressure), when we are able to compare yields for alternative designs
(adiabatic versus isothermal operations, single versus multiple reactor units, flow
versus batch system), and when we can estimate the economics of these various
alternatives, then and only then will we feel sure that we can arrive at the
design well fitted for the purpose at hand. Unfortunately, real situations are
rarely simple.
86 Chapter 4 Introduction to Reactor Design

At the start Later

t=O time = t

Constant pressure -..

-
Constant volume

Let time pass

Or

Figure 4.4 Symbols used for batch reactors.

Should we explore all reasonable design alternatives? How sophisticated

should our analysis be? What simplifying assumptions should we make? What
shortcuts should we take? Which factors should we ignore and which should we
consider? And how should the reliability and completeness of the data at hand
influence our decisions? Good engineering judgment, which only comes with
experience, will suggest the appropriate course of action.

Symbols and Relationship between CA and X ,

For the reaction aA + bB -t rR, with inerts iI, Figs. 4.4 and 4.5 show the symbols
commonly used to tell what is happening in the batch and flow reactors. These
figures show that there are two related measures of the extent of reaction, the
concentration CA and the conversion XA. However, the relationship between
CAand XAis often not obvious but depends on a number of factors. This leads
to three special cases, as follows.

Special Case 1. Constant Density Batch and Flow Systems. This includes most
liquid reactions and also those gas reactions run at constant temperature and
density. Here CA and XA are related as follows:

X A = 1 - -cA and d X A = --
C~~ v ~ A =-
l h A = O
for EA = =0 (3)
vxA=o
-- - 1 - xA and dCA = -CAodXA
C~~

-t

-
FA= FAo(1 - XA)
U P CAI XA

i-
FAO = moles fedlhr
If there is any
uo = m3 fluid enteringthr
ambiguity call these
CAO= concentration of A
F A f l uf,CAP XAf
in the feed stream

V = volume
Figure 4.5 Symbols used for flow reactors.
 4.1 General Discussion 87
To relate the changes in B and R to A we have

Special Case 2. Batch and Flow Systems of Gases of Changing Density but
with T and .rr Constant. Here the density changes because of the change in
number of moles during reaction. In addition, we require that the volume of a
fluid element changes linearly with conversion, or V = Vo (1 + &AXA).

X -

CA
-
A

-
CAO- CA
- CAO+ eACA
- 1 - XA
CAO 1 + EAXA
and dXA = - 'AO(' + 'A) dcA

~ C-
and - -
(CAO+ EACA)'
A- 1 + EA
~ X A
CAO (1 + E A X A ) ~
] for

To follow changes in the other components we have

\% = ( ~ / Q ) X+AC R O / ~ A O
between for products 1 + &AXA
and inerts 1
CIO 1 + EAXA

Special Case 3. Batch and Flow Systems for Gases in General (varying p, T,
.rr) which react according to

Pick one reactant as the basis for determining the conversion. We call this the
key reactant. Let A be the key. Then for ideal gas behavior,
88 Chapter 4 Introduction to Reactor Design

For high-pressure nonideal gas behavior replace (z)(-) ,

by
zoTor
where z is
the compressibility factor. To change to another key reactant, say B, note that

CAOXA
as, - ~ E B and - - CBOXB
--
c~o c~o a b

For liquids or isothermal gases with no change pressure and density

Tor
sA+O and (%) -1

and the preceding expressions simplify greatly.

A BALANCE FROM STOICHIOMETRY

Consider a feed CAo= 100, C,, = 200, Cio = 100 to a steady-flow reactor. The
isothermal gas-phase reaction is

If CA = 40 at the reactor exit, what is C,, XA, and XB there?

First sketch what is known (see Fig. E4.1).

Figure E4.1

Next recognize that this problem concerns Special Case 2. So evaluate sAand
E ~ For
. this take 400 volumes of gas
 Problems 89

Then from the equations in the text

PROBLEMS

The following four problems consider an isothermal single-phase flow reactor

operating at steady-state and constant pressure.

4.1. Given a gaseous feed, CAo= 100, CBO= 200, A + B + R + S, XA = 0.8.

Find XB, CA, CB.

4.2. Given a dilute aqueous feed, CA, = CBo= 100, A + 2B + R + S, CA= 20.
Find XA, XB, CB.

4.3. Given a gaseous feed, CAO= 200, CBO= 100, A + B +R, CA = 50. Find
XA,XB, CB.

4.4. Given a gaseous feed, CAO= CBO= 100, A + 2B +R, CB = 20. Find XA,
XB, CA.

In the following two problems a continuous stream of fluid enters a vessel at

temperature To and pressure no, reacts there, and leaves at T and n.
4.5. Given a gaseous feed, To = 400 K, no = 4 atm, CAo= 100, CBO= 200,
A+B+2R,T=300K,n=3atm,CA=[Link],XB,CB.

4.6. Given a gaseous feed, To = 1000 K, no = 5 atm, CAo= 100, CBO= 200,
+
A B +5R, T = 400 K, n = 4 atm, CA = 20. Find XA,XB, CB.

4.7. A Commercial Popcorn Popping Popcorn Popper. We are constructing a

1-liter popcorn popper to be operated in steady flow. First tests in this unit
show that 1literlmin of raw corn feed stream produces 28 literlmin of mixed
exit stream. Independent tests show that when raw corn pops its volume
goes from 1 to 31. With this information determine what fraction of raw
corn is popped in the unit.
Chapter 3

Ideal Reactors for a Single

Reaction

In this chapter we develop the performance equations for a single fluid reacting
in the three ideal reactors shown in Fig. 5.1. We call these homogeneous reactions.
Applications and extensions of these equations to various isothermal and noniso-
thermal operations are considered in the following four chapters.
In the batch reactor, or BR, of Fig. [Link] the reactants are initially charged into
a container, are well mixed, and are left to react for a certain period. The resultant
mixture is then discharged. This is an unsteady-state operation where composition
changes with time; however, at any instant the composition throughout the
reactor is uniform.
The first of the two ideal steady-state flow reactors is variously known as the
plug flow, slug flow, piston flow, ideal tubular, and unmixed flow reactor, and it
is shown in Fig. [Link]. We refer to it as the plugpow reactor, or PFR, and to this
pattern of flow as plugpow. It is characterized by the fact that the flow of fluid
through the reactor is orderly with no element of fluid overtaking or mixing with
any other element ahead or behind. Actually, there may be lateral mixing of
fluid in a plug flow reactor; however, there must be no mixing or diffusion along

Feed Product

Uniformly
rnixed Product
(a) (b) (c)

Figure 5.1 The three types of ideal reactors: (a) batch reactor, or BR; (b) plug flow
reactor, or PFR; and ( c ) mixed flow reactor, or MFR.
 5.1 Ideal Batch Reactor 91

the flow path. The necessary and sufficient condition for plug flow is for the
residence time in the reactor to be the same for all elements of fluid.'
The other ideal steady-state flow reactor is called the mixed reactor, the
backmix reactor, the ideal stirred tank reactor, the C* (meaning C-star), CSTR,
or the CFSTR (constant flow stirred tank reactor), and, as its names suggest, it
is a reactor in which the contents are well stirred and uniform throughout. Thus,
the exit stream from this reactor has the same composition as the fluid within
the reactor. We refer to this type of flow as mixed pow, and the corresponding
reactor the mixed pow reactor, or MFR.
These three ideals are relatively easy to treat. In addition, one or other usually
represents the best way of contacting the reactants-no matter what the opera-
tion. For these reasons, we often try to design real reactors so that their flows
approach these ideals, and much of the development in this book centers
about them.
In the treatment to follow it should be understood that the term V , called the
reactor volume, really refers to the volume of fluid in the reactor. When this
differs from the internal volume of reactor, then V ,designates the internal volume
of reactor while V designates the volume of reacting fluid. For example, in solid
catalyzed reactors with voidage E we have

For homogeneous systems, however, we usually use the term V alone.

5.1 IDEAL BATCH REACTOR

Make a material balance for any component A. For such an accounting we
usually select the limiting component. In a batch reactor, since the composition
is uniform throughout at any instant of time, we may make the accounting about
the whole reactor. Noting that no fluid enters or leaves the reaction mixture
during reaction, Eq. 4.1, which was written for component A, becomes
=o =O
i d t = o d u t + disappearance + accumulation
rate of loss of reactant A rate of accumulation
within reactor due to
chemical reaction
Evaluating the terms of Eq. 1, we find
=- (of reactant A
within the reactor

disappearance of A
moles A reacting
by reaction, = (-rA)V = (volume of fluid)
molesltime (time)(volume of fluid)

accumulation of A, - dNA - d[NAo(1 - XA)] - ~ X A

molesltime dt dt - -NAO 7
* The necessary condition follows directly from the definition of plug flow. However, the sufficient
condition-that the same residence times implies plug flow-can be established only from the second
law of thermodynamics.
92 Chapter 5 Ideal Reactors for a Single Reaction

By replacing these two terms in Eq. 1, we obtain

Rearranging and integrating then gives

This is the general equation showing the time required to achieve a conversion
XA for either isothermal or nonisothermal operation. The volume of reacting
fluid and the reaction rate remain under the integral sign, for in general they
both change as reaction proceeds.
This equation may be simplified for a number of situations. If the density of
the fluid remains constant, we obtain

For all reactions in which the volume of reacting mixture changes proportionately
with conversion, such as in single gas-phase reactions with significant density
changes, Eq. 3 becomes

In one form or another, Eqs. 2 to 5 have all been encountered in Chapter 3.

They are applicable to both isothermal and nonisothermal operations. For the
latter the variation of rate with temperature, and the variation of temperature
with conversion, must be known before solution is possible. Figure 5.2 is a
graphical representation of two of these equations.

Constant-density systems only

Figure 5.2 Graphical representation of the performance equations for batch reactors,
isothermal or nonisothermal.
 5.1 Ideal Batch Reactor 93

Space-Time and Space-Velocity

Just as the reaction time t is the natural performance measure for a batch reactor,
so are the space-time and space-velocity the proper performance measures of
flow reactors. These terms are defined as follows:

Space-time:

time required to process one

reactor volume of feed measured (6)
S
at specified conditions

number of reactor volumes of

feed at specified conditions which (7)
7
can be treated in unit time

Thus, a space-velocity of 5 hr-l means that five reactor volumes of feed at

specified conditions are being fed into the reactor per hour. A space-time of 2
min means that every 2 min one reactor volume of feed at specified conditions
is being treated by the reactor.
Now we may arbitrarily select the temperature, pressure, and state (gas, liquid,
or solid) at which we choose to measure the volume of material being fed to
the reactor. Certainly, then, the value for space-velocity or space-time depends
on the conditions selected. If they are of the stream entering the reactor, the
relation between s and r and the other pertinent variables is

moles A entering
(volume of reactor)
r=-=--
moles A entering
(8)
time
-V --
-- (reactor volume)
v, (volumetric feed rate)

It may be more convenient to measure the volumetric feed rate at some

standard state, especially when the reactor is to operate at a number of tempera-
tures. If, for example, the material is gaseous when fed to the reactor at high
temperature but is liquid at the standard state, care must be taken to specify
precisely what state has been chosen. The relation between the space-velocity
and space-time for actual feed conditions (unprimed symbols) and at standard
conditions (designated by primes) is given by
94 Chapter 5 Ideal Reactors for a Single Reaction

In most of what follows, we deal with the space-velocity and space-time based
on feed at actual entering conditions; however, the change to any other basis is
easily made.

5.2 STEADY-STATE MIXED FLOW REACTOR

The performance equation for the mixed flow reactor is obtained from Eq. 4.1,
which makes an accounting of a given component within an element of volume
of the system. But since the composition is uniform throughout, the accounting
may be made about the reactor as a whole. By selecting reactant A for consider-
ation, Eq. 4.1 becomes
=0
input = output + disappearance by reaction + accudation (10)
As shown in Fig. 5.3, if FA,= voCAois the molar feed rate of component A to
the reactor, then considering the reactor as a whole we have

input of A, molesltime = FAo(l- XAo)= FAo

output of A, molesltime = FA= FAo(l- XA)

disappearance of A
moles A reacting volume of
by reaction,
molesltime
= (-rA)V =
(
((time)(volume of fluid)) reactor )
Introducing these three terms into Eq. 10, we obtain

which on rearrangement becomes

Uniform 1
throughout
Figure 5.3 Notation for a mixed reactor.
 5.2 Steady-State Mixed Flow Reactor 95

where XA and rA are measured at exit stream conditions, which are the same as
the conditions within the reactor.
More generally, if the feed on which conversion is based, subscript 0, enters
the reactor partially converted, subscript i, and leaves at conditions given by
subscript f,we have

For the special case of constant-density systems XA = 1 - CA/CAo, in which

case the performance equation for mixed reactors can also be written in terms
of concentrations or

These expressions relate in a simple way the four terms XA,-YA, V ,FAO;thus,
knowing any three allows the fourth to be found directly. In design, then, the
size of reactor needed for a given duty or the extent of conversion in a reactor
of given size is found directly. In kinetic studies each steady-state run gives,
without integration, the reaction rate for the conditions within the reactor. The
ease of interpretation of data from a mixed flow reactor makes its use very
attractive in kinetic studies, in particular with messy reactions (e.g., multiple
reactions and solid catalyzed reactions).
Figure 5.4 is a graphical representation of these mixed flow performance
equations. For any specific kinetic form the equations can be written out directly.

rGz&q Constant-density systems only

Figure 5.4 Graphical representation of the design equations for mixed flow reactor.
96 Chapter 5 Ideal Reactors for a Single Reaction

As an example, for constant density systems C,IC, = 1 - X,, thus the perfor-
mance expression for first-order reaction becomes

k T = - -x* - c*o - CA for E* = 0

1 - x* c*
On the other hand, for linear expansion

C, - 1 - x*
V = VO(l+ and --
CAO 1 + &AX*

thus for first-order reaction the performance expression of Eq. 11 becomes

for any E A
1- X A

For second-order reaction, A +products, -r, = kc:, EA = 0, the performance

equation of Eq. 11 becomes

Similar expressions can be written for any other form of rate equation. These
expressions can be written either in terms of concentrations or conversions.
Using conversions is simpler for systems of changing density, while either form
can be used for sytems of constant density.

;I REACTION RATE IN A MIXED FLOW REACTOR

One liter per minute of liquid containing A and B (C,, = 0.10 mollliter, CBo=
0.01 mollliter) flow into a mixed reactor of volume V = 1 liter. The materials
react in a complex manner for which the stoichiometry is unknown. The outlet
stream from the reactor contains A, B, and C (CAf= 0.02 mollliter, CBf= 0.03
mollliter, Ccf = 0.04 mollliter), as shown in Fig. E5.1. Find the rate of reaction
of A, B, and C for the conditions within the reactor.

CA = CAS=0.02 rnol/lit
CB = 0.03 molllit
Cc = 0.04 mol/lit
Liquid
1V = 1 lit
Figure E5.1
 5.2 Steady-State Mixed Flow Reactor 97

For a liquid in a mixed flow reactor E, = 0 and Eq. 13 applies to each of the
reacting components, giving for the rate of disappearance:

CAo - CA= CAO- CA- = 0.08 mollliter. min

-rA =
7 Vlv 111

-rB=
cBo
- CB - 0.01 - 0.03
- = -0.02 mollliter. min
7 1
C 0 - Cc - 0 - 0.04 -
-rc = -0.04 mollliter. min
7 1

L Thus A is disappearing while B and C are being formed.

KINETICS FROM A MIXED FLOW REACTOR

Pure gaseous reactant A (CAo= 100 millimollliter) is fed at a steady rate into a
mixed flow reactor (V = 0.1 liter) where it dimerizes (2A +R). For different
gas feed rates the following data are obtained:

Run number 1 2 3 4
vo,literlhr 10.0 3.0 1.2 0.5
CAf,millimol/liter 85.7 66.7 50 33.4

Find a rate equation for this reaction.

For this stoichiometry, 2A -.R, the expansion factor is

and the corresponding relation between concentration and conversion is

 98 Chapter 5 Ideal Reactors for a Single Reaction

The conversion for each run is then calculated and tabulated in column 4 of
Table E5.2.

Table E5.2
Calculated
Given
VOCAOXA
Run Vo cA XA (-TA) =
V log CA log (-7~)

From the performance equation, Eq. 11, the rate of reaction for each run is
given by

(-rA) =
VOCAOXA
V ' 1 1
millimol
liter. hr

These values are tabulated in column 5 of Table E5.2.

Having paired values of rA and C, (see Table E5.2) we are ready to test
various kinetic expressions. Instead of separately testing for first-order (plot rA
VS.C,), second-order (plot rA vs. C a ) , etc., let us test directly for nth-order
kinetics. For this take logarithms of -rA = k C 1 , giving

+ n log C A
I log(-r,) = log k

For nth-order kinetics this data should give a straight line on a log (-r,) vs. log
CA plot. From columns 6 and 7 of Table E5.2 and as shown in Fig. E5.2, the

-
Slope =

-
3.398 - 2.602
1.933 - 1.522
= 1.93 2

Fit the lowest point

400 = k (33.3)'

log CA
Figure E5.2
 5.2 Steady-State Mixed Flow Reactor 99

four data points are reasonably represented by a straight line of slope 2, so the
rate equation for this dimerization is

liter millimol
hr . millimol) " [-]
Comment. If we ignore the density change in our analysis (or put E, = 0
and use CA/CAo= 1 - XA)we end up with an incorrect rate equation (reaction
order n = 1.6) which when used in design would give wrong performance predic-

I tions.

MIXED FLOW REACTOR PERFORMANCE

lementary liquid-phase reaction

with rate equation

-rA = -1.
2
= (12.5 liter2/[Link])C,C; - (1.5 minW1)CR,
liter. min

is to take place in a 6-liter steady-state mixed flow reactor. Two feed streams,
one containing 2.8 mol Alliter and the other containing 1.6 mol Blliter, are to
be introduced at equal volumetric flow rates into the reactor, and 75% conversion
of limiting component is desired (see Fig. E5.3). What should be the flow rate
of each stream? Assume a constant density throughout.

Cia= 2.8 mol Nliter

UA = UB =U

Figure E5.3
100 Chapter 5 Ideal Reactors for a Single Reaction

SOLUTION

The concentration of components in the mixed feed stream is

C,, = 1.4 mollliter

CBo= 0.8 mollliter
C,, = 0

These numbers show that B is the limiting component, so for 75% conversion
of B and E = 0, the composition in the reactor and in the exit stream is

CA = 1.4 - 0.612 = 1.1mollliter

CB = 0.8 - 0.6 = 0.2 mollliter or 75% conversion
C, = 0.3 mollliter

Writing the rate and solving the problem in terms of B we have at the conditions
within the reactor

= (25 mo12. ) g)(0.2 g)' (3 min-l)

liter'min (1.1 -

mol mol
= (1.1 - 0.9) = 0.2
liter. min liter min

For no density change, the performance equation of Eq. 13 gives

Hence the volumentric flow rate into and out of the reactor is

(6 liter)(0.2 mollliter. min)

U =
-
-
6 literlmin
=-
(0.8 - 0.6) mollliter

3 literlmin of each of the two feed streams

 5.3 Steady-State Plug Flow Reactor 101

5.3 STEADY-STATE PLUG FLOW REACTOR

In a plug flow reactor the composition of the fluid varies from point to point
along a flow path; consequently, the material balance for a reaction component
must be made for a differential element of volume dV. Thus for reactant A, Eq.
4.1 becomes
=0
input = output + disappearance by reaction + accuydation (10)
Referring to Fig. 5.5, we see for volume dV that

input of A, molesltime = FA
output of A, molesltime = FA + dFA
disappearance of A by
reaction, molesltime = (-r,)dV
moles A reacting)
= ((time)(volume of fluid)

Introducing these three terms in Eq. 10, we obtain

Noting that

We obtain on replacement

Distance through reactor

Figure 5.5 Notation for a plug flow reactor.
102 Chapter 5 Ideal Reactors for a Single Reaction

This, then, is the equation which accounts for A in the differential section of
reactor of volume dV. For the reactor as a whole the expression must be inte-
grated. Now FA,, the feed rate, is constant, but r, is certainly dependent on
the concentration or conversion of materials. Grouping the terms accordingly,
we obtain

Thus

Equation 17 allows the determination of reactor size for a given feed rate and
required conversion. Compare Eqs. 11 and 17. The difference is that in plug flow
rAvaries, whereas in mixed flow rA is constant.
As a more general expression for plug flow reactors, if the feed on which
conversion is based, subscript 0, enters the reactor partially converted, subscript
i, and leaves at a conversion designated by subscript f,we have

For the special case of constant-density systems

XA=l-- ~ C A
and dXA= - -
C ~ o C ~ o

in which case the performance equation can be expressed in terms of concentra-

tions, or
 5.3 Steady-State Plug Flow Reactor 103

pzziq Constant-density systems only

Figure 5.6 Graphical representation of the performance equations for plug flow reactors.

These performance equations, Eqs. 17 to 19, can be written either in terms of

concentrations or conversions. For systems of changing density it is more conve-
nient to use conversions; however, there is no particular preference for constant
density systems. Whatever its form, the performance equations interrelate the
rate of reaction, the extent of reaction, the reactor volume, and the feed rate, and
if any one of these quantities is unknown it can be found from the other three.
Figure 5.6 displays these performance equations and shows that the space-
time needed for any particular duty can always be found by numerical or graphical
integration. However, for certain simple kinetic forms analytic integration is
possible-and convenient. To do this, insert the kinetic expression for r, in Eq.
17 and integrate. Some of the simpler integrated forms for plug flow are as follows:
Zero-order homogeneous reaction, any constant E,

First-order irreversible reaction, A +products, any constant &A,

k7=-(1+ ln (1 - x,) - 1 (21)

*
First-order reversible reaction, A rR, CRdCAo= M , kinetics approximated or
fitted by -r, = klCA - k2CRwith an observed equilibrium conversion XAe,any
constant &A,

Second-order irreversible reaction, A + B +products with equimolar feed or

2A +products, any constant E,,
 104 Chapter 5 Ideal Reactors for a Single Reaction

Where the density is constant, put E , = 0 to obtain the simplified perfor-

mance equation.
By comparing the batch expressions of Chapter 3 with these plug flow expressions
we find:
(1) For systems of constant density (constant-volume batch and constant-den-
sity plug flow) the performance equations are identical, T for plug flow
is equivalent to t for the batch reactor, and the equations can be used
interchangeably.
(2) For systems of changing density there is no direct correspondence between
the batch and the plug flow equations and the correct equation must be
used for each particular situation. In this case the performance equations
cannot be used interchangeably.
The following illustrative examples show how to use these expressions.

PLUG FLOW REACTOR PERFORMANCE

I A homogeneous gas reaction A -+3R has a reported rate at 215OC

-rA = 10-2C~2, [mollliter sec]

Find the space-time needed for 80% conversion of a 50% A-50% inert feed to
a plug flow reactor operating at 215OC and 5 atm (CAo= 0.0625 mollliter).

Figure E5.4a

For this stoichiometry and with 50% inerts, two volumes of feed gas would give
four volumes of completely converted product gas; thus

in which case the plug flow performance equation, Eq. 17, becomes

1 The integral can be evaluated in any one of three ways: graphically, numerically,
or analytically. Let us illustrate these methods.
 5.3 Steady-State Plug Flow Reactor 105
Table E5.4

Graphical Integration. First evaluate the function to be integrated at selected

values (see Table E5.4) and plot this function (see Fig. E5.4b).

Average height = 1.7

Area 1.7(0.8) = 1.36

XA

Figure E5.4b

Counting squares or estimating by eye we find

Area = lo(-)1 + xA dXA

0.8 'I2
= (1.70)(0.8) = -
1.36
-

Numerical Integration. Using Simpson's rule, applicable to an even number

of uniformly spaced intervals on the XAaxis, we find for the data of Table E5.4,

( 2 ) ' l 2 dXA= (average height)(total width)

l(1) + 4(1.227) + 2(1.528) + 4(2) + l(3)

12 I
 106 Chapter 5 Ideal Reactors for a Single Reaction

Analytical Integration. From a table of integrals

0.8 1+ x*
dX*
0-

= (arc sin XA-

The method of integration recommended depends on the situation. In this prob-

lem probably the numerical method is the quickest and simplest and gives a
good enough answer for most purposes.
So with the integral evaluated, Eq. (i) becomes

(0.0625 m ~ l l l i t e r ) ~ '(1.33)
~ = 33.2 sec
7=
= moll"lliterl". sec)

PLUG FLOW REACTOR VOLUME

The homogeneous gas decomposition of phosphine

proceeds at 649OC with the first-order rate

What size of plug flow reactor operating at 649°C and 460 kPa can produce 80%
conversion of a feed consisting of 40 mol of pure phosphine per hour?

Lv=?
Figure E5.5

I SOLUTION

Let A = pH3, R = P4, S = H,. Then the reaction becomes

 5.3 Steady-State Plug Flow Reactor 107

with

I The volume of plug flow reactor is given by Eq. 21

Evaluating the individual terms in this expression gives

FA, = 40 mollhr
k = lOlhr

= PAO 460 000 Pa = 60 mol/m3

= (8.314 Pa. m31mol.K)(922 K)

hence the volume of reactor

= 148 liters

TEST OF A KINETIC EQUATION IN A PLUG FLOW

REACTOR

We suspect that the gas reaction between A, B, and R is an elementary revers-

ible reaction

and we plan to test this with experiments in an isothermal plug flow reactor.

(a) Develop the isothermal performance equation for these kinetics for a feed
of A, B, R, and inerts.
(b) Show how to test this equation for an equimolar feed of A and B.
108 Chapter 5 Ideal Reactors for a Single Reaction

(a) Feed of A, B, R, and inerts. For this elementary reaction the rate is

At constant pressure, basing expansion and conversion on substance A,

Letting M = CBdCAo,M' = CRdCAO,

we obtain

Hence, the design equation for plug flow, Eq. 17, becomes

In this expression accounts for the stoichiometry and for inerts present in
the feed.
(b) Equimolar feed of A and B. For CAo= CBo,CRo= 0, and no inerts, we
have M = 1, M' = 0, = -0.5; hence the expression for part a reduces to

(1 - 0.5XA)2dXA j:f(xA)dxA (i)

o klCAo(l- XA)' - kZXA(I- 0.5XA) this

Having V , v,, and XA data from a series of experiments, separately evaluate

the left side and the right side of Eq. (i). For the right side, at various XA

t = V/uo

Figure E5.6
 5.3 Steady-State Plug Flow Reactor 109
evaluate f (X,), then integrate graphically to give Sf (XA)dXAand then make
the plot of Fig. E5.6. If the data fall on a reasonably straight line, then the
suggested kinetic scheme can be said to be satisfactory in that it fits the data.

Holding Time and Space Time for Flow Reactors

We should be clearly aware of the distinction between these two measures of
time, 2 and 7. They are defined as follows:

time needed to

mean residence time

of flowing material ~ X A [hrl (24)
in the reactor 0 ( - Y A ) (+
~ EAXA)'

For constant density systems (all liquids and constant density gases)

For changing density systems 7 # 7 and 2 # Vlu, in which case it becomes difficult
to find how these terms are related.
As a simple illustration of the difference between 2 and 7, consider two cases
of the steady-flow popcorn popper of Problem 4.7 which takes in 1 literlmin of
raw corn and produces 28 literslmin of product popcorn.
Consider three cases, called X, Y, and Z, which are shown in Fig. 5.7. In the
first case (case X) all the popping occurs at the back end of the reactor. In the

Case X Case Y Case Z

+28 literlmin 28 literlmin

1 literlmin 1 literlmin, 1 literlmin,

of unpopped corn unpopped unpopped
Figure 5.7 For the same 7 value the i values differ in these three cases.
110 Chapter 5 Ideal Reactors for a Single Reaction

second case (case Y) all the popping occurs at the front end of the reactor. In
the third case (case Z) the popping occurs somewhere between entrance and exit.
In all three cases

rx=ry=rz=-V
- - 1liter = I min
v, 1literlmin

irrespective of where the popping occurs. However, we see that the residence
time in the three cases is very different, or

1liter
t - = 1min
- 1literlmin
1liter
t -
- 28 literlmin
= 2 sec
,7 is somewhere between 2 and 60 s, depending on the kinetics

Note that the value o f t depends on what happens in the reactor, while the value
of r is independent of what happens in the reactor.
This example shows that 7 and r are not, in general, identical. Now which is
the natural performance measure for reactors? For batch systems Chapter 3
shows that it is the time of reaction; however, holding time does not appear
anywhere in the performance equations for flow systems developed in this chap-
ter, Eqs. 13 to 19, while it is seen that space-time or VIF,, does naturally appear.
Hence, r or VIFAOis the proper performance measure for flow systems.
The above simple example shows that in the special case of constant fluid
density the space-time is equivalent to the holding time; hence, these terms can
be used interchangeably. This special case includes practically all liquid phase
reactions. However, for fluids of changing density, e.g., nonisothermal gas reac-
tions or gas reactions with changing number of moles, a distinction should be
made between rand 7 and the correct measure should be used in each situation.

Summary of Performance Equations

Tables 5.1 and 5.2 present the integrated performance equations for single
ideal reactors.

REFERENCES

Corcoran, W. H., and Lacey, W. N., Introduction to Chemical Engineering Problems,

McGraw Hill, New York, 1970, p. 103.
Pease, R. N., J. Am. Chem. Soc., 51,3470 (1929).
 Problems 113
PROBLEMS

5.1. Consider a gas-phase reaction 2A-R + 2s with unknown kinetics. If a

space velocity of llmin is needed for 90% conversion of A in a plug flow
reactor, find the corresponding space-time and mean residence time or
holding time of fluid in the plug flow reactor.

5.2. In an isothermal batch reactor 70% of a liquid reactant is converted in 13

min. What space-time and space-velocity are needed to effect this conver-
sion in a plug flow reactor and in a mixed flow reactor?

5.3. A stream of aqueous monomer A (1 mollliter, 4 literlmin) enters a 2-liter

mixed flow reactor, is radiated therein, and polymerizes as follows:

+A tA tA
A-R-S-T.. .
In the exit stream CA = 0.01 mollliter, and for a particular reaction product
W, C, = 0.0002 mollliter. Find the rate of reaction of A and the rate of
formation of W.

5.4. We plan to replace our present mixed flow reactor with one having double
the volume. For the same aqueous feed (10 mol Alliter) and the same feed
rate find the new conversion. The reaction kinetics are represented by

and present conversion is 70%.

5.5. An aqueous feed of A and B (400 literlmin, 100 mmol Alliter, 200 mmol
Blliter) is to be converted to product in a plug flow reactor. The kinetics
of the reaction is represented by

mol
A+B+R, -rAZ200C C
A liter .min

Find the volume of reactor needed for 99.9% conversion of A to product.

5.6. A plug flow reactor (2 m3) processes an aqueous feed (100 literlmin) con-
taining reactant A (CAo= 100 mmollliter). This reaction is reversible and
represented by

First find the equilibrium conversion and then find the actual conversion
of A in the reactor.
114 Chapter 5 Ideal Reactors for a Single Reaction

5.7. The off gas from a boiling water nuclear power reactor contains a whole
variety of radioactive trash, one of the most troublesome being Xe-133
(half life = 5.2 days). This off gas flows continuously through a large holdup
tank in which its mean residence time is 30 days, and where we can assume
that the contents are well mixed. Find the fraction of activity removed in
the tank.

5.8. A mixed flow reactor (2 m3) processes an aqueous feed (100 literlrnin)
containing reactant A (C,, = 100 mmollliter). The reaction is reversible
and represented by

mol
A R, -7, = 0.04 CA - 0.01 CR
liter. min

What is the equilibrium conversion and the actual conversion in the reactor?

5.9. A specific enzyme acts as catalyst in the fermentation of reactant A. At a

given enzyme concentration in the aqueous feed stream (25 literlmin) find
the volume of plug flow reactor needed for 95% conversion of reactant
A (C,, = 2 mollliter). The kinetics of the fermentation at this enzyme
concentration is given by

enzyme 0.1 CA m01

A-R, -rA =
1 + 0.5 C, liter. min

5.10. A gaseous feed of pure A (2 mollliter, 100 mollmin) decomposes to give

a variety of products in a plug flow reactor. The kinetics of the conversion
is represented by

A -t 2.5 (products), -rA = (10 min-')C,

Find the expected conversion in a 22-liter reactor.

5.11. Enzyme E catalyses the fermentation of substrate A (the reactant) to

product R. Find the size of mixed flow reactor needed for 95% conversion
of reactant in a feed stream (25 literlmin) of reactant (2 mollliter) and
enzyme. The kinetics of the fermentation at this enzyme concentration are
given by

enzyme 0.1 cA m01

A-R, -?'A=
+
1 0.5 C, liter min

5.12. An aqueous feed of A and B (400 literlmin, 100 mmol Alliter, 200 mmol
Blliter) is to be converted to product in a mixed flow reactor. The kinetics
 Problems 115

of the reaction are represented by

mol
A+B+R, -rA=200CACB
liter min

Find the volume of reactor needed for 99.9% conversion of A to product.

5.13. At 650°C phosphine vapor decomposes as follows:

What size of plug flow reactor operating at 649°C and 11.4 atm is needed
for 75% conversion of 10 mollhr of phosphine in a 213 phosphine-113
inert feed?

5.14. A stream of pure gaseous reactant A (CAo= 660 mmollliter) enters a plug
flow reactor at a flow rate of FA,= 540 mmollmin and polymerizes there
as follows

mmol
3A+R, -r,=54
liter min

How large a reactor is needed to lower the concentration of A in the exit

stream to CAf= 330 mmollliter?

5.15. A gaseous feed of pure A (1 mollliter) enters a mixed flow reactor (2 liters)
and reacts as follows:

mol
2A +R, -rA = 0.05 C i
liter. sec

Find what feed rate (literlmin) will give an outlet concentration CA = 0.5
mollliter.

5.16. Gaseous reactant A decomposes as follows:

Find the conversion of A in a 50% A-50% inert feed (v, = 180 literlmin,
CAo= 300 mmollliter) to a 1 m3 mixed flow reactor.

5.17. 1 literls of a 20% ozone-80% air mixture at 1.5 atm and 93°C passes
through a plug flow reactor. Under these conditions ozone decomposes by
homogeneous reaction

liter
20, +30,, -r,,,,, = kC~,on,, k = 0.05 -
mol . s

What size reactor is needed for 50% decomposition of ozone? This problem
is a modification of a problem given by Corcoran and Lacey (1970).
116 Chapter 5 Ideal Reactors for a Single Reaction

5.18. An aqueous feed containing A (1 mollliter) enters a 2-liter plug flow reactor
and reacts away (2A +R, - r , = 0.05 C2, mollliter s). Find the outlet con-
centration of A for a feed rate of 0.5 literlmin.

5.19. Pure gaseous A at about 3 atm and 30°C (120 mmollliter) is fed into a 1-
liter mixed flow reactor at various flow rates. There it decomposes, and
the exit concentration of A is measured for each flow rate. From the
following data find a rate equation to represent the kinetics of the decompo-
sition of A. Assume that reactant A alone affects the rate.
v,, literlmin
C, , mmollliter 1 0.06
30
0.48
60
1.5
80
8.1 A +3R
105

5.20. A mixed flow reactor is being used to determine the kinetics of a reaction
whose stoichiometry is A +R. For this purpose various flow rates of an
aqueous solution of 100 mmol Alliter are fed to a 1-liter reactor, and for
each run the outlet concentration of A is measured. Find a rate equation
to represent the following data. Also assume that reactant alone affects
the rate.
v, literlmin 1 6 24
C,, mmollliter 4 20 50

5.21. We are planning to operate a batch reactor to convert A into R. This is a

liquid reaction, the stoichiometry is A + R , and the rate of reaction is
given in Table P5.21. How long must we react each batch for the concentra-
tion to drop from C,, = 1.3 mollliter to CAf= 0.3 mollliter?

Table P5.21
C, , mollliter - rA, mollliter min
0.1 0.1
0.2 0.3
0.3 0.5
0.4 0.6
0.5 0.5
0.6 0.25
0.7 0.10
0.8 0.06
1.0 0.05
1.3 0.045
2.0 0.042

5.22. For the reaction of Problem 5.21, what size of plug flow reactor would be
needed for 80% conversion of a feed stream of 1000 mol Alhr at CAo=
1.5 mollliter?
 Problems 117
5.23. (a) For the reaction of Problem 5.21, what size of mixed flow reactor is
needed for 75% conversion of a feed stream of 1000 mol Alhr at CAO
= 1.2 mollliter?
(b) Repeat part (a) with the modification that the feed rate is doubled,
thus 2000 mol Alhr at CAo= 1.2 mollliter are to be treated.
( c ) Repeat part (a) with the modification that CAo= 2.4 mollliter; however,
1000 mol Alhr are still to be treated down to CAf= 0.3 mollliter.

5.24. A high molecular weight hydrocarbon gas A is fed continuously to a heated

high temperature mixed flow reactor where it thermally cracks (homoge-
neous gas reaction) into lower molecular weight materials, collectively
called R, by a stoichiometry approximated by A -5R. By changing the
feed rate different extents of cracking are obtained as follows:

The internal void volume of the reactor is V = 0.1 liter, and at the tempera-
ture of the reactor the feed concentration is CAo= 100 millimol/liter. Find
a rate equation to represent the cracking reaction.

5.25. The aqueous decomposition of A is studied in an experimental mixed flow

reactor. The results in Table P5.25 are obtained in steady-state runs. To
obtain 75% conversion of reactant in a feed, CA, = 0.8 mollliter, what
holding time is needed in a plug flow reactor?

Table P5.25
Concentration of A, mollliter
Holding Time,
In Feed In Exit Stream sec

5.26. Repeat the previous problem but for a mixed flow reactor.

5.27. HOLMES: You say he was last seen tending this vat . . . .
SIR BOSS: You mean "overflow stirred tank reactor," Mr. Holmes.
HOLMES: You must excuse my ignorance of your particular technical
jargon, Sir Boss.
SIR BOSS: That's all right; however, you must find him, Mr. Holmes.
118 Chapter 5 Ideal Reactors for a Single Reaction

Imbibit was a queer chap; always staring into the reactor, taking deep
breaths, and licking his lips, but he was our very best operator. Why,
since he left, our conversion of googliox has dropped from 80% to 75%.
HOLMES (tapping the side of the vat idly): By the way, what goes on in
the vat?
SIR BOSS: Just an elementary second-order reaction, between ethanol
and googliox, if you know what I mean. Of course, we maintain a large
excess of alcohol, about 100 to 1 and . . . .
HOLMES (interrupting): Intriguing, we checked every possible lead in
town and found not a single clue.
SIR BOSS (wiping away the tears): We'll give the old chap a raise-about
twopence per week-if only he'll come back.
DR. WATSON: Pardon me, but may I ask a question?
HOLMES: Why certainly, Watson.
WATSON: What is the capacity of this vat, Sir Boss?
SIR BOSS: A hundred Imperial gallons, and we always keep it filled to
the brim. That is why we call it an overflow reactor. You see we are
running at full capacity-profitable operation you know.
HOLMES: Well, my dear Watson, we must admit that we're stumped, for
without clues deductive powers are of no avail.
WATSON: Ahh, but there is where you are wrong, Holmes. (Then, turning
to the manager): Imbibit was a largish fellow-say about 18 stone-was
he not?
SIR BOSS: Why yes, how did you know?
HOLMES (with awe): Amazing, my dear Watson!
WATSON (modestly): Why it's quite elementary, Holmes. We have all the
clues necessary to deduce what happened to the happy fellow. But first
of all, would someone fetch me some dill?

With Sherlock Holmes and Sir Boss impatiently waiting, Dr. Watson
casually leaned against the vat, slowly and carefully filled his pipe, and-
with the keen sense of the dramatic-lit it. There our story ends.
(a) What momentous revelation was Dr. Watson planning to make, and
how did he arrive at this conclusion?
(b) Why did he never make it?

5.28. The data in Table P5.28 have been obtained on the decomposition of
gaseous reactant A in a constant volume batch reactor at 100°C. The

Table P5.28
t, sec PA, atm t, sec P A , atm
0 1.00 140 0.25
20 0.80 200 0.14
40 0.68 260 0.08
60 0.56 330 0.04
80 0.45 420 0.02
100 0.37
 Problems 119

stoichiometry of the reaction is 2A -+ R + S. What size plug flow reactor

(in liters) operating at 100°C and 1 atm can treat 100 mol Alhr in a feed
consisting of 20% inerts to obtain 95% converson of A?

5.29. Repeat the previous problem for a mixed flow reactor.

5.30. The aqueous decomposition of A produces R as follows:

The following results are obtained in a series of steady state runs, all having
no R in the feed stream.

Space Time, r, , , In Feed,

,C CAf,In Exit Stream,
sec mollliter mollliter

From this kinetic information, find the size of reactor needed to achieve
75% conversion of a feed stream of v = 1 literlsec and ,C
, = 0.8 mollliter.
In the reactor the fluid follows
(a) plug flow
(b) mixed flow
Chapter 6
Design for Single Reactions

There are many ways of processing a fluid: in a single batch or flow reactor, in
a chain of reactors possibly with interstage feed injection or heating, in a reactor
with recycle of the product stream using various feed ratios and conditions, and
so on. Which scheme should we use? Unfortunately, numerous factors may have
to be considered in answering this question; for example, the reaction type,
planned scale of production, cost of equipment and operations, safety, stability
and flexibility of operation, equipment life expectancy, length of time that the
product is expected to be manufactured, ease of convertibility of the equipment
to modified operating conditions or to new and different processes. With the
wide choice of systems available and with the many factors to be considered, no
neat formula can be expected to give the optimum setup. Experience, engineering
judgment, and a sound knowledge of the characteristics of the various reactor
systems are all needed in selecting a reasonably good design and, needless to
say, the choice in the last analysis will be dictated by the economics of the
overall process.
The reactor system selected will influence the economics of the process by
dictating the size of the units needed and by fixing the ratio of products formed.
The first factor, reactor size, may well vary a hundredfold among competing
designs while the second factor, product distribution, is usually of prime consider-
ation where it can be varied and controlled.
In this chapter we deal with single reactions. These are reactions whose progress
can be described and followed adequately by using one and only one rate expres-
sion coupled with the necessary stoichiometric and equilibrium expressions. For
such reactions product distribution is fixed; hence, the important factor in com-
paring designs is the reactor size. We consider in turn the size comparison of
various single and multiple ideal reactor systems. Then we introduce the recycle
reactor and develop its performance equations. Finally, we treat a rather unique
type of reaction, the autocatalytic reaction, and show how to apply our findings
to it.
Design for multiple reactions, for which the primary consideration is product
distribution, is treated in the next two chapters.
 6.1 Size Comparison of Single Reactors 121
6.1 SIZE COMPARISON OF SINGLE REACTORS
Batch Reactor
First of all, before we compare flow reactors, let us mention the batch reactor
briefly. The batch reactor has the advantage of small instrumentation cost and
flexibility of operation (may be shut down easily and quickly). It has the disadvan-
tage of high labor and handling cost, often considerable shutdown time to empty,
clean out, and refill, and poorer quality control of the product. Hence we may
generalize to state that the batch reactor is well suited to produce small amounts
of material and to produce many different products from one piece of equipment.
On the other hand, for the chemical treatment of materials in large amounts the
continuous process is nearly always found to be more economical.
Regarding reactor sizes, a comparison of Eqs. 5.4 and 5.19 for a given duty
and for E = 0 shows that an element of fluid reacts for the same length of time
in the batch and in the plug flow reactor. Thus, the same volume of these reactors
is needed to do a given job. Of course, on a long-term production basis we must
correct the size requirement estimate to account for the shutdown time between
batches. Still, it is easy to relate the performance capabilities of the batch reactor
with the plug flow reactor.

Mixed Versus Plug Flow Reactors, First- and Second-Order Reactions

For a given duty the ratio of sizes of mixed and plug flow reactors will depend
on the extent of reaction, the stoichiometry, and the form of the rate equation.
For the general case, a comparison of Eqs. 5.11 and 5.17 will give this size ratio.
Let us make this comparison for the large class of reactions approximated by
the simple nth-order rate law

where n varies anywhere from zero to three. For mixed flow Eq. 5.11 gives

whereas for plug flow Eq. 5.17 gives

Dividing we find that

122 Chapter 6 Design for Single Reactions

With constant density, or E = 0, this expression integrates to

Equations 1 and 2 are displayed in graphical form in Fig. 6.1 to provide a

quick comparison of the performance of plug flow with mixed flow reactors. For

Figure 6.1 Comparison of performance of single mixed flow and plug flow reactors
for the nth-order reactions

The ordinate becomes the volume ratio V,,,/Vpor space-time ratio T,,,/T~ if the
same quantities of identical feed are used.
 6.1 Size Comparison of Single Reactors n3

identical feed composition CAoand flow rate FA, the ordinate of this figure gives
directly the volume ratio required for any specified conversion. Figure 6.1 shows
the following.

1. For any particular duty and for all positive reaction orders the mixed reactor
is always larger than the plug flow reactor. The ratio of volumes increases
with reaction order.
2. When conversion is small, the reactor performance is only slightly affected
by flow type. The performance ratio increases very rapidly at high conver-
sion; consequently, a proper representation of the flow becomes very impor-
tant in this range of conversion.
3. Density variation during reaction affects design; however, it is normally of
secondary importance compared to the difference in flow type.

Figures 6.5 and 6.6 show the same first- and second-order curves for E = 0
but also include dashed lines which represent fixed values of the dimensionless
reaction rate group, defined as

k~ for first-order reaction

kCA07for second-order reaction

With these lines we can compare different reactor types, reactor sizes, and
conversion levels.
Example 6.1 illustrates the use of these charts.

Variation of Reactant Ratio for Second-Order Reactions

Second-order reactions of two components and of the type

behave as second-order reactions of one component when the reactant ratio is

unity. Thus

-rA = k c A c B= k c ; when M = 1 (3)

On the other hand, when a large excess of reactant B is used then its concentration
does not change appreciably (CB= CBO) and the reaction approaches first-order
behavior with respect to the limiting component A, or

-YA = kCACB = (kCBO)CA

= krCA when M +1 (4)

Thus in Fig. 6.1, and in terms of the limiting component A, the size ratio of
mixed to plug flow reactors is represented by the region between the first-order
and the second-order curves.
u 4 Chapter 6 Design for Single Reactions

Any rate curve

t

Figure 6.2 Comparison of performance of

mixed flow and plug flow reactors for any reac-
tion kinetics.

General Graphical Comparison

For reactions with arbitrary but known rate the performance capabilities of
mixed and plug flow reactors are best illustrated in Fig. 6.2. The ratio of shaded
and of hatched areas gives the ratio of space-times needed in these two reactors.
The rate curve drawn in Fig. 6.2 is typical of the large class of reactions whose
rate decreases continually on approach to equilibrium (this includes all nth-order
reactions, n > 0). For such reactions it can be seen that mixed flow always needs
a larger volume than does plug flow for any given duty.

6.2 MULTIPLE-REACTOR SYSTEMS

Plug Flow Reactors in Series andlor in Parallel
Consider N plug flow reactors connected in series, and let XI, X,, . . . , X, be
the fractional conversion of component A leaving reactor 1, 2, . . . , N. Basing
the material balance on the feed rate of A to the first reactor, we find for the
ith reactor from Eq. 5.18

or for the N reactors in series

Hence, N plug flow reactors in series with a total volume V gives the same
conversion as a single plug flow reactor of volume V.
 6.2 Multiple-Reactor Systems 125

For the optimum hook up of plug flow reactors connected in parallel or in

any parallel-series combination, we can treat the whole system as a single plug
flow reactor of volume equal to the total volume of the individual units if the
feed is distributed in such a manner that fluid streams that meet have the same
composition. Thus, for reactors in parallel VIF or T must be the same for each
parallel line. Any other way of feeding is less efficient.

OPERATING A NUMBER OF PLUG FLOW REACTORS

e reactor setup shown in Fig. E6.1 consists of three plug flow reactors in two
parallel branches. Branch D has a reactor of volume 50 liters followed by a
reactor of volume 30 liters. Branch E has a reactor of volume 40 liters. What
fraction of the feed should go to branch D?

Figure E6.1

Branch D consists of two reactors in series; hence, it may be considered to be

a single reactor of volume

VD = 50 + 30 = 80 liters
Now for reactors in parallel VIF must be identical if the conversion is to be the
same in each branch. Therefore,

I Therefore, two-thirds of the feed must be fed to branch D.

126 Chapter 6 Design for Single Reactions

CA in

\
\
I \

Single mixed flow reactor, N = 1

I

* C~ out

Figure 6.3 Concentration profile through an N-stage mixed flow reactor system
compared with single flow reactors.

Equal-Size Mixed Flow Reactors in Series

In plug flow, the concentration of reactant decreases progressively through the
system; in mixed flow, the concentration drops immediately to a low value.
Because of this fact, a plug flow reactor is more efficient than a mixed flow
reactor for reactions whose rates increase with reactant concentration, such as
nth-order irreversible reactions, n > 0.
Consider a system of N mixed flow reactors connected in series. Though the
concentration is uniform in each reactor, there is, nevertheless, a change in
concentration as fluid moves from reactor to reactor. This stepwise drop in
concentration, illustrated in Fig. 6.3, suggests that the larger the number of units
in series, the closer should the behavior of the system approach plug flow. This
will be shown to be so.
Let us now quantitatively evaluate the behavior of a series of N equal-size
mixed flow reactors. Density changes will be assumed to be negligible; hence
E = 0 and t = 7. As a rule, with mixed flow reactors it is more convenient to
develop the necessary equations in terms of concentrations rather than fractional
conversions; therefore, we use this approach. The nomenclature used is shown
in Fig. 6.4 with subscript i referring to the ith vessel.

Figure 6.4 Notation for a system of N equal-size mixed reactors in series.

 6.2 Multiple-Reactor Systems n7
First-Order Reactions. From Eq. 5.12 a material balance for component A
about vessel i gives

Because E = 0 this may be written in terms of concentrations. Hence

Now the space-time r (or mean residence time t ) is the same in all the equal-
size reactors of volume [Link],

Rearranging, we find for the system as a whole

In the limit, for N + m , this equation reduces to the plug flow equation

With Eqs. 6b and 7 we can compare performance of N reactors in series with a

plug flow reactor or with a single mixed flow reactor. This comparison is shown
in Fig. 6.5 for first-order reactions in which density variations are negligible.

Second-Order Reactions. We may evaluate the performance of a series of

mixed flow reactors for a second-order, bimolecular-type reaction, no excess of
either reactant, by a procedure similar to that of a first-order reaction. Thus, for
128 Chapter 6 Design for Single Reactions

Figure 6.5 Comparison of performance of a series of N equal-size mixed flow

reactors with a plug flow reactor for the first-order reaction

For the same processing rate of identical feed the ordinate measures the volume
ratio V,IV, directly.

N reactors in series we find

whereas for plug flow

A comparison of the performance of these reactors is shown in Fig. 6.6.

Figures 6.5 and 6.6 support our intuition by showing that the volume of system
required for a given conversion decreases to plug flow volume as the number of
reactors in series is increased, the greatest change taking place with the addition
of a second vessel to a one-vessel system.
 6.2 Multiple-Reactor Systems f29

Figure 6.6 Comparison of performance of a series of N equal-size mixed flow

reactors with a plug flow reactor for elementary second-order reactions

with negligible expansion. For the same processing rate of identical feed the
ordinate measures the volume ratio v N / V por space-time ratio rN/rpdirectly.

MIXED FLOW REACTORS IN SERIES

At present 90% of reactant A is converted into product by a second-order reaction

in a single mixed flow reactor. We plan to place a second reactor similar to the
one being used in series with it.
(a) For the same treatment rate as that used at present, how will this addition
affect the conversion of reactant?
(b) For the same 90% conversion, by how much can the treatment rate be in-
creased?
130 Chapter 6 Design for Single Reactions

The sketch of Fig. E6.2 shows how the performance chart of Fig. 6.6 can be used
to help solve this problem.

100

1
0.01

Figure E6.2

(a) Find the conversion for the same treatment rate. For the single reactor at
90% conversion we have from Fig. 6.6

kCor = 90

For the two reactors the space-time or holding time is doubled; hence, the
operation will be represented by the dashed line of Fig. 6.6 where

kCor = 180

This line cuts the N = 2 line at a conversion X = 97.4%, point a.

(b) Find the treatment rate for the same conversion. Staying on the 90% conver-
sion line, we find for N = 2 that

kC,r = 27.5, point b

Comparing the value of the reaction rate group for N = 1 and N = 2, we find

2 ( V / V ) N = 27.5
(kC07),=, - 7 ~ = - ~
(kc,~),=, T N = ~ ( V ) 90
 6.2 Multiple-Reactor Systems 131

I Since V,=, = 2V,,, the ratio of flow rates becomes

Thus, the treatment rate can be raised to 6.6 times the original.

Note. If the second reactor had been operated in parallel with the original unit
then the treatment rate could only be doubled. Thus, there is a definite advantage
in operating these two units in series. This advantage becomes more pronounced
at higher conversions.

Mixed Flow Reactors of Different Sizes in Series

For arbitrary kinetics in mixed flow reactors of different size, two types of
questions may be asked: how to find the outlet conversion from a given reactor
system, and the inverse question, how to find the best setup to achieve a given
conversion. Different procedures are used for these two problems. We treat
them in turn.

Finding the Conversion in a Given System A graphical procedure for finding

the outlet composition from a series of mixed flow reactors of various sizes for
reactions with negligible density change has been presented by Jones (1951). All
that is needed is an r versus C curve for component A to represent the reaction
rate at various concentrations.
Let us illustrate the use of this method by considering three mixed flow reactors
in series with volumes, feed rates, concentrations, space-times (equal to residence
times because E = O), and volumetric flow rates as shown in Fig. 6.7. Now from
Eq. 5.11, noting that E = 0, we may write for component A in the first reactor

Figure 6.7 Notation for a series of unequal-size mixed flow reactors.

Chapter 6 Design for Single Reactions

Rate versus

(from Eq. 9)

(from Eq. 10)

C3 c2 c1 co
Reactant concentration
Figure 6.8 Graphical procedure for finding compositions in a series
of mixed flow reactors.

Similarly, from Eq. 5.12 for the ith reactor we may write

Plot the C versus r curve for component A and suppose that it is as shown in
Fig. 6.8. To find the conditions in the first reactor note that the inlet concentration
Co is known (point L), that C, and ( - r ) , correspond to a point on the curve to
be found (point M), and that the slope of the line LM = MNINL = (-r),l
(C, - Co) = -(l/rl) from Eq. 6.9. Hence, from Co draw a line of slope -(l/r,)
until it cuts the rate curve; this gives C1. Similarly, we find from Eq. 6.10 that a
line of slope -(11r2) from point N cuts the curve at P, giving the concentration
C2 of material leaving the second reactor. This procedure is then repeated as
many times as needed.
With slight modification this graphical method can be extended to reactions
in which density changes are appreciable.

Determining the Best System for a Given Conversion. Suppose we want to

find the minimum size of two mixed flow reactors in series to achieve a specified
conversion of feed which reacts with arbitrary but known kinetics. The basic
performance expressions, Eqs. 5.11 and 5.12, then give, in turn, for the first reactor

and for the second reactor

These relationships are displayed in Fig. 6.9 for two alternative reactor arrange-
ments, both giving the same final conversion X2. Note, as the intermediate
 6.2 Multiple-Reactor Systems 133

Area measures
size of first unit

Of second unit

Figure 6.9 Graphical representation of the variables for two mixed flow reactors in series.

conversion X I changes, so does the size ratio of the units (represented by the
two shaded areas) as well as the total volume of the two vessels required (the
total area shaded).
Figure 6.9 shows that the total reactor volume is as small as possible (total
shaded area is minimized) when the rectangle KLMN is as large as possible.
This brings us to the problem of choosing X I (or point M on the curve) so as
to maximize the area of this rectangle. Consider this general problem.

Maximization of Rectangles. In Fig. 6.10, construct a rectangle between the

x-y axes and touching the arbitrary curve at point M(x, y). The area of the
rectangle is then

Figure 6.10 Graphical procedure for maximizing the area of a rectangle.

134 Chapter 6 Design for Single Reactions

This area is maximized when

or when

In words, this condition means that the area is maximized when M is at that
point where the slope of the curve equals the slope of the diagonal NL of the
rectangle. Depending on the shape of the curve, there may be more than one
or there may be no "best" point. However, for nth-order kinetics, n > 0, there
always is just one "best" point.
We will use this method of maximizing a rectangle in later chapters. But let
us return to our problem.
The optimum size ratio of the two reactors is achieved where the slope of the
rate curve at M equals the diagonal NL. The best value of M is shown in Fig.
6.11, and this determines the intermediate conversion X , as well as the size of
units needed.
The optimum size ratio for two mixed flow reactors in series is found in general
to be dependent on the kinetics of the reaction and on the conversion level. For
the special case of first-order reactions equal-size reactors are best; for reaction
orders n > 1 the smaller reactor should come first; for n < 1 the larger should
come first (see Problem 6.3). However, Szepe and Levenspiel (1964) show that
the advantage of the minimum size system over the equal-size system is quite
small, only a few percent at most. Hence, overall economic consideration would
nearly always recommend using equal-size units.
The above procedure can be extended directly to multistage operations; how-
ever, here the argument for equal-size units is stronger still than for the two-
stage system.

Diagonal of
rectangle\ ,
f slope of
curve at
M

Figure 6.11 Maximization of rectan-

gles applied to find the optimum in-
termediate conversion and optimum
sizes of two mixed flow reactors in
series.
 6.2 Multiple-Reactor Systems 135

Figure 6.12 Graphical design procedure for reactors in series.

Reactors of Different Types in Series

If reactors of different types are put in series, such as a mixed flow reactor
followed by a plug flow reactor which in turn is followed by another mixed flow
reactor, we may write for the three reactors

These relationships are represented in graphical form in Fig. 6.12. This allows us
to predict the overall conversions for such systems, or conversions at intermediate
points between the individual reactors. These intermediate conversions may be
needed to determine the duty of interstage heat exchangers.

Best Arrangement of a Set of Ideal Reactors. For the most effective use of a
given set of ideal reactors we have the following general rules:

1. For a reaction whose rate-concentration curve rises monotonically (any

nth-order reaction, n > 0) the reactors should be connected in series. They
should be ordered so as to keep the concentration of reactant as high as
possible if the rate-concentration curve is concave (n > I), and as low as
possible if the curve is convex (n < 1). As an example, for the case of Fig.
6.12 the ordering of units should be plug, small mixed, large mixed, for
n > 1; the reverse order should be used when n < 1.
2. For reactions where the rate-concentration curve passes through a maximum
or minimum the arrangement of units depends on the actual shape of curve,
the conversion level desired, and the units available. No simple rules can
be suggested.
3. Whatever may be the kinetics and the reactor system, an examination of
the ll(-r,) vs. CA curve is a good way to find the best arrangement of
units.
The problems at the end of this chapter illustrate these findings.
136 Chapter 6 Design for Single Reactions

6.3 RECYCLE REACTOR

In certain situations it is found to be advantageous to divide the product stream
from a plug flow reactor and return a portion of it to the entrance of the reactor.
Let the recycle ratio R be defined as

volume of fluid returned to the reactor entrance

R = (15)
volume leaving the system

This recycle ratio can be made to vary from zero to infinity. Reflection suggests
that as the recycle ratio is raised the behavior shifts from plug flow (R = 0) to
mixed flow (R = m). Thus, recycling provides a means for obtaining various
degrees of backmixing with a plug flow reactor. Let us develop the performance
equation for the recycle reactor.
Consider a recycle reactor with nomenclature as shown in Fig. 6.13. Across
the reactor itself Eq. 5.18 for plug flow gives

v
- - - \;=xw 5
FA, -'.A

where FAo would be the feed rate of A if the stream entering the reactor (fresh
feed plus recycle) were unconverted. Since FAo and X,, are not known directly,
they must be written in terms of known quantities before Eq. 16 can be used.
Let us now do this.
The flow entering the reactor includes both fresh feed and the recycle stream.
Measuring the flow split at point L (point K will not do if E f 0) we then have

(
A which would enter in an
FA0 = unconverted recycle stream
A entering in
fresh feed
(17)

Now to the evaluation of XA1:from Eq. 4.5 we may write

Because the pressure is taken to be constant, the streams meeting at point K

may be added directly. This gives

F ~ 3

Figure 6.13 Nomenclature for the recycle reactor.

 6.3 Recycle Reactor 137

Combining Eqs. 18 and 19 gives XA, in terms of measured quantities, or

Finally, on replacing Eqs. 17 and 20 in Eq. 16 we obtain the useful form for
the performance equation for recycle reactors, good for any kinetics, any F value
and for X A , = 0.

For the special case where density changes are negligible we may write this
equation in terms of concentrations, or

These expressions are represented graphically in Fig. 6.14.

for any 6 only for E =0

Figure 6.14 Representation of the performance equation for recy-

cle reactors.
138 Chapter 6 Design for Single Reactions

For the extremes of negligible and infinite recycle the system approaches plug
flow and mixed flow, or

------------- - - - - - - - -- -
-,
plug flow mixed flow

The approach to these extremes is shown in Fig. 6.15.

Integration of the recycle equation gives, for first-order reaction, E , = 0,

and for second-order reaction, 2A + products, -r, = kc;, E, = 0,

(24)

The expressions for E , f 0 and for other reaction orders can be evaluated, but
are more cumbersome.

Figure 6.15 The recycle extremes approach plug flow (R +0) and
mixed flow (R + a).
 6.3 Recycle Reactor 139

Figure 6.16 Comparison of performance of recycle and plug flow for first-order re-
actions

Figures 6.16 and 6.17 show the transition from plug to mixed flow as R increases,
and a match of these curves with those for N tanks in series (Figs. 6.5 and 6.6)
gives the following rough comparison for equal performance:

R for first-order R for second-order

No. reaction reaction
of
tanks at XA = 0.5 0.90 0.99 at X A = 0.5 0.90 0.99
1 w w CC] CO CO CO

2 1.0 2.2 5.4 1.0 2.8 7.5

3 0.5 1.1 2.1 0.5 1.4 2.9
4 0.33 0.68 1.3 0.33 0.90 1.7
10 0.11 0.22 0.36 0.11 0.29 0.5
CQ 0 0 0 0 0 0

The recycle reactor is a convenient way for approaching mixed flow with what
is essentially a plug flow device. Its particular usefulness is with solid catalyzed
reactions with their fixed bed contactors. We meet this and other applications
of recycle reactors in later chapters.
140 Chapter 6 Design for Single Reactions

Figure 6.17 Comparison of performance of recycle reactors with plug flow

reactors for elementary second-order reactions (Personal communication,
from T. J. Fitzgerald and P. Fillesi):

2A -+products, E =0

A + B +products, C,, = C,, with E = 0

6.4 AUTOCATALYTIC REACTIONS

When a material reacts away by any nth order rate (n > 0) in a batch reactor,
its rate of disappearance is rapid at the start when the concentration of reactant
is high. This rate then slows progressively as reactant is consumed. In an autocata-
lytic reaction, however, the rate at the start is low because little product is present;
it increases to a maximum as product is formed and then drops again to a low
value as reactant is consumed. Figure 6.18 shows a typical situation.
Reactions with such rate-concentration curves lead to interesting optimization
problems. In addition, they provide a good illustration of the general design
method presented in this chapter. For these reasons let us examine these reactions
in some detail. In our approach we deal exclusively with their ll(-r,) versus
X, curves with their characteristic minima, as shown in Fig. 6.18.
 6.4 Autocatalytic Reactions 141

Point of
maximum

Progress
of reaction

CA

Figure 6.18 Typical rate-concentration curve for autocatalytic reactions,

for example:

Plug Flow Versus Mixed Flow Reactor, N o Recycle. For any particular rate-
concentration curve a comparison of areas in Fig. 6.19 will show which reactor
is superior (which requires a smaller volume) for a given job. We thus find
1. At low conversion the mixed reactor is superior to the plug flow reactor.
2. At high enough conversions the plug flow reactor is superior.
These findings differ from ordinary nth-order reactions (n > 0) where the plug
flow reactor is always more efficient than the mixed flow reactor. In addition,
we should note that a plug flow reactor will not operate at all with a feed of
pure reactant. In such a situation the feed must be continually primed with
product, an ideal opportunity for using a recycle reactor.

Figure 6.19 For autocatalytic reactions mixed flow is more efficient at low conversions, plug
flow is more efficient at high conversions.
142 Chapter 6 Design for Single Reactions

Optimum Recycle Operations. When material is to be processed to some fixed

final conversion XAf in a recycle reactor, reflection suggests that there must be
a particular recycle ratio which is optimum in that it minimizes the reactor volume
or space-time. Let us determine this value of R.
The optimum recycle ratio is found by differentiating Eq. 21 with respect to
R and setting to zero, thus

take
7 R +1
dR ~ X A (25)
Ai ~ t l

This operation requires differentiating under an integral sign. From the theorems
of calculus, if

then

For our case, Eq. 25, we then find

where

Combining and rearranging then gives for the optimum

In words, the optimum recycle ratio introduces to the reactor a feed whose
ll(-r,) value (KL in Fig. 6.20) equals the average l l ( - r A ) value in the reactor
as a whole (PQ in Fig. 6.20). Figure 6.20 compares this optimum with conditions
where the recycle is either too high or too low.
 6.4 Autocatalytic Reactions 143

I Recycle just right I

Figure 6.20 Correct recycle ratio for an autocatalytic reaction compared with recycle ratios which
are too high and too low.

Occurrence of Autocatalytic Reactions. The most important examples of auto-

catalytic reactions are the broad class of fermentation reactions which result
from the reaction of microorganism on an organic feed. When they can be treated
as single reactions, the methods of this chapter can be applied directly. Another
type of reaction which has autocatalytic behavior is the exothermic reaction (say,
the combustion of fuel gas) proceeding in an adiabatic manner with cool reactants
entering the system. In such a reaction, called autothermal, heat may be consid-
ered to be the product which sustains the reaction. Thus, with plug flow the
reaction will die. With backmixing the reaction will be self-sustaining because
the heat generated by the reaction can raise fresh reactants to a temperature at
which they will react. Autothermal reactions are of great importance in solid
catalyzed gas-phase systems and are treated later in the book.

Reactor Combinations
For autocatalytic reactions all sorts of reactor arrangements are to be considered
if product recycle or product separation with recycle is allowable. In general,
for a rate-concentration curve as shown in Fig. 6.21 one should always try to
reach point M in one step (using mixed flow in a single reactor), then follow
with plug flow or as close to plug flow as possible. This procedure is shown as
the shaded area in Fig. 6.21a.
When separation and reuse of unconverted reactant is possible, operate at
point M (see Fig. 6.21b).
The volume required is now the very minimum, less than any of the previous
ways of operating. However, the overall economics, including the cost of separa-
tion and of recycle, will determine which scheme is the optimum overall.
144 Chapter 6 Design for Single Reactions

Product
' ~ f

C ~ o CAO Separator

recycle

(a) (b)
Figure 6.21 ( a ) The best multiple reactor scheme. (b) The best scheme when uncon-
verted reactant can be separated and recycled.

FINDING THE BEST REACTOR SETUP

In the presence of a specific enzyme E, which acts as a homogeneous catalyst,

a harmful organic A present in industrial waste water degrades into harmless
chemicals. At a given enzyme concentration C, tests in a laboratory mixed flow
reactor give the following results:
, , mmol/m3
,C 2 5 6 6 11 14 16 24
C, , mmol/m3 0.53 1 2 6 1 0 8 4
r, min 30 1 50 8 4 20 20 4
We wish to treat 0.1 m3/min of this waste water having ,C
, = 10 mmol/m3 to
90% conversion with this enzyme at concentration C.,
(a) One possibility is to use a long tubular reactor (assume plug flow) with
possible recycle of exit fluid. What design do you recommend? Give the
size of the reactor, tell if it should be used with recycle, and if so determine
the recycle flow rate in cubic meters per minute (m3/min). Sketch your
recommended design.
(b) Another possibility is to use one or two stirred tanks (assume ideal). What
two-tank design do you recommend, and how much better is it than the
one-tank arrangement?
(c) What arrangement of plug flow and mixed flow reactors would you use to
minimize the total volume of reactors needed? Sketch your recommended
design and show the size of units selected. We should mention that separa-
tion and recycle of part of the product stream is not allowed.
 6.4 Autocatalytic Reactions 145

SOLUTION

First calculate and tabulate 11-r, at the measured C,. This is shown as the last
line of Table E6.3. Next, draw the 11-r, vs. C, curve. This is seen to be U-
shaped (see Figs. E6.3a, b, c) so we must prepare to deal with an autocatalytic
type reacting system.

Table E6.3
C,, mmol/m3 2 5 6 6 11 14 16 24
C,, mmol/m3 0.5 3 1 2 6 1 0 8 4
7,min 30 1 50 8 4 20 20 4
-1- -
7 m y min
20 0.5 10 2 0.8 5 2.5 0.2
-r, C,, - C, ' mmol

Part (a) Solution. From the -l/r, vs. C, curve we see that we should use plug
flow with recycle. From Fig. E 6 . 3 ~we find

V = 7U0 = area (u,) = [(lo - 1)(1.2)](0.1) = 1.08 m3

v, = u,R = O.l(O.607) = 0.0607 mVmin

rn mol
-
m3

I Figure E6.3a Plug flow with recycle.

146 Chapter 6 Design for Single Reactions

Part (b) Solution. Drawing slopes and diagonals according to the method of
maximization of rectangles we end up with Fig. E6.3b.
"2
Area2 = .r2= = (2.6 - 1)10 = 16 min
A /

- 1)10 = 90 min

(10 - 2.6)0.8 = 5.92 min

\ m mol
\ C A ! , ~

Figure E6.3b One and two mixed flow reactors in series.

For 1tank V = n, = 90(0.1) = 9.0 m3

Vl = T ~ V= 5.92(0.1) = 0.59
For 2 tanks Vtotal= 2.19 m3
V, = T2V = 16(0.1) = 1.6 m3

Part ( c )Solution. Following the reasoning in this chapter we should use a mixed
flow reactor followed by a plug flow reactor. So with Fig. E 6 . 3 we
~ find

By graphical integration

m mol
CA,-
m3
Figure E 6 . 3 ~Arrangement with smallest volume.
 Problems 1147

For the MFR V,n= U T , = O.l(l.2) = 0.12 m3

Vtotal = 0.7 m3
For the PFR V, = u?, = 0.1(5.8) = 0.58 m3

Note which scheme (a) or (b) or (c) gives the smallest size of reactors.

REFERENCES

Jones, R. W., Chem. Eng. Progr., 47,46 (1951).

Szepe, S., and 0. Levenspiel, Ind. Eng. Chem. Process Design Deuelop., 3, 214 (1964).

PROBLEMS

6.1. A liquid reactant stream (1 mollliter) passes through two mixed flow reac-
tors in a series. The concentration of A in the exit of the first reactor is
0.5 mollliter. Find the concentration in the exit stream of the second reactor.
The reaction is second-order with respect to A and V2/V,= 2.

6.2. Water containing a short-lived radioactive species flows continuously

through a well-mixed holdup tank. This gives time for the radioactive
material to decay into harmless waste. As it now operates, the activity of
the exit stream is 117 of the feed stream. This is not bad, but we'd like to
lower it still more.
One of our office secretaries suggests that we insert a baffle down the
middle of the tank so that the holdup tank acts as two well-mixed tanks
in series. Do you think this would help? If not, tell why; if so, calculate
the expected activity of the exit stream compared to the entering stream.

6.3. An aqueous reactant stream (4 mol Alliter) passes through a mixed flow
reactor followed by a plug flow reactor. Find the concentration at the exit
of the plug flow reactor if in the mixed flow reactor C, = 1 mollliter. The
reaction is second-order with respect to A, and the volume of the plug
flow unit is three times that of the mixed flow unit.

6.4. Reactant A (A -t R, C,, = 26 mol/m3)passes in steady flow through four

equal-size mixed flow reactors in series (T,,,,, = 2 min). When steady state
is achieved the concentration of A is found to be 11, 5, 2, 1 mol/m3in the
four units. For this reaction, what must be rplUg SO as to reduce C, from
C,, = 26 to CAf= 1 mol/m3?

6.5. Originally we had planned to lower the activity of a gas stream containing
radioactive Xe-138 (half-life = 14 min) by having it pass through two
holdup tanks in series, both well mixed and of such size that the mean
residence time of gas is 2 weeks in each tank. It has been suggested that
we replace the two tanks with a long tube (assume plug flow). What must
be the size of this tube compared to the two original stirred tanks, and
148 Chapter 6 Design ,for Single Reactions

what should be the mean residence time of gas in this tube for the same
extent of radioactive decay?

6.6. At 100°C pure gaseous A reacts away with stoichiometry 2A + R + S in

a constant volume batch reactor as follows:
sec
t,
p,,atm 1 0
1.00
20
0.96
40
0.80
60
0.56
80
0.32
100
0.18
120
0.08
140
0.04
160
0.02
What size of plug flow reactor operating at 100°C and 1atm can treat 100
moles Alhr in a feed consisting of 20% inerts to obtain 95% conversion of A?

6.7. We wish to treat 10 literstmin of liquid feed containing 1 mol Alliter to

99% conversion. The stoichiometry and kinetics of the reaction are given
by

A+R, -rA= CA mot

0.2 + C, liter. min

Suggest a good arrangement for doing this using two mixed flow reactors,
and find the size of the two units needed. Sketch the final design chosen.

6.8. From steady-state kinetics runs in a mixed flow reactor, we obtain the
following data on the reaction A + R.
- - -
- --- -

7, sec ,C
,, mmollliter C, mmollliter
60 50 20
35 100 40
11 100 60
20 200 80
11 200 100

Find the space time needed to treat a feed of ,C

, = 100 mmollliter to
80% conversion
(a) in a plug flow reactor.
(b) in a mixed flow reactor.

6.9. At present we have 90% conversion of a liquid feed (n = 1,,C , = 10 moll

liter) to our plug flow reactor with recycle of product (R = 2). If we shut
off the recycle stream, by how much will this lower the processing rate of
our feed to the same 90% conversion?

6.10. Aqueous feed containing reactant A (C,, = 2 mol/liter) enters a plug flow
reactor (10 liter) which has a provision for recycling a portion of the flowing
stream. The reaction kinetics and stoichiometry are

mol
A-tR, -rA=lC C
liter emin
 Problems 149
and we wish to get 96% conversion. Should we use the recycle stream? If
so, at what value should we set the recycle flow rate so as to obtain the
highest production rate, and what volumetric feed rate can we process to
this conversion in the reactor?

6.11. Consider the autocatalytic reaction A --+ R, with -rA = 0.001 CACRmoll
1iter.s. We wish to process 1.5 litersls of a CAo = 10 mollliter feed to the
highest conversion possible in the reactor system consisting of four 100-
liter mixed flow reactors connected as you wish and any feed arrangement.
Sketch your recommended design and feed arrangement and determine
CAf from this system.

6.12. A first-order liquid-phase reaction, 92% conversion, is taking place in a

mixed flow reactor. It has been suggested that a fraction of the product
stream, with no additional treatment, be recycled. If the feed rate remains
unchanged, in what way would this affect conversion?

6.13. 100 literslhr of radioactive fluid having a half-life of 20 hr is to be treated

by passing it through two ideal stirred tanks in series, V = 40 000 liters
each. In passing through this system, how much will the activity decay?

6.14. At present the elementary liquid-phase reaction A + B + R + S takes

place in a plug flow reactor using equimolar quantities of A and B. Conver-
sion is 96%, C,, = C,, = 1 mollliter. If a mixed flow reactor ten times as
large as the plug flow reactor were hooked up in series with the existing
unit, which unit should come first and by what fraction could production
be increased for that setup?

.IS. The kinetics of the aqueous-phase decomposition of A is investigated in

two mixed flow reactors in series, the second having twice the volume of
the first reactor. At steady state with a feed concentration of 1mol Alliter
and mean residence time of 96 sec in the first reactor, the concentration
in the first reactor is 0.5 rnol Alliter and in the second is 0.25 rnol Alliter.
Find the kinetic equation for the decomposition.

6.16. Using a color indicator which shows when the concentration of A falls
below 0.1 mollliter, the following scheme is devised to explore the kinetics
of the decomposition of A. A feed of 0.6 mol Alliter is introduced into
the first of the two mixed flow reactors in series, each having a volume of
400 [Link] color change occurs in the first reactor for a steady-state feed
rate of 10 cmymin, and in the second reactor for a steady-state feed rate
of 50 cm3/min. Find the rate equation for the decomposition of A from
this information.

6.17. The elementary irreversible aqueous-phase reaction A + B -,R + S is

carried out isothermally as follows. Equal volumetric flow rates of two
liquid streams are introduced into a 4-liter mixing tank. One stream contains
0.020 mol Alliter, the other 1.400 rnol Blliter. The mixed stream is then
150 Chapter 6 Design for Single Reactions

passed through a 16-liter plug flow reactor. We find that some R is formed
in the mixing tank, its concentration being 0.002 mollliter. Assuming that
the mixing tank acts as a mixed flow reactor, find the concentration of R
at the exit of the plug flow reactor as well as the fraction of initial A that
has been converted in the system.

6.18. At present conversion is 213 for our elementary second-order liquid reaction
2A i.2R when operating in an isothermal plug flow reactor with a recycle
ratio of unity. What will be the conversion if the recycle stream is shut off?

6.19. We wish to explore various reactor setups for the transformation of A into
R. The feed contains 99% A, 1% R; the desired product is to consist of
10% A, 90% R. The transformation takes place by means of the elementary
reaction

with rate constant k = 1literlmol . min. The concentration of active materi-

als is

throughout.
What reactor holding time will yield a product in which C, = 0.9 moll
liter (a) in a plug flow reactor, (b) in a mixed flow reactor, and (c) in a
minimum-size setup without recycle?

6.20. Reactant A decomposes with stoichiometry A-R and with rate dependent
only on C,. The following data on this aqueous decomposition are obtained
in a mixed flow reactor:

T, sec c ~ o cA
14 200 100
25 190 90
29 180 80
30 170 70
29 160 60
27 150 50
24 140 40
19 130 30
15 120 20
12 110 10
20 101 1

Determine which setup, plug flow, mixed flow, or any two-reactor combina-
tion gives minimum T for 90% conversion of a feed consisting of ,C, =
100. Also find this T minimum. If a two-reactor scheme is found to be
optimum, give C, between stages and T for each stage.
 Problems 151

6.21. For an irreversible first-order liquid-phase reaction (CAO= 10 mollliter)

conversion is 90% in a plug flow reactor. If two-thirds of the stream leaving
the reactor is recycled to the reactor entrance, and if the throughput to
the whole reactor-recycle system is kept unchanged, what does this do to
the concentration of reactant leaving the system?

6.22. At room temperature the second-order irreversible liquid-phase reaction

proceeds as follows:

A batch reactor takes 18 min to fill and empty. What percent conversion
and reaction time should we use so as to maximize the daily output of
product R?
Chapter 7
Design for Parallel Reactions

Introduction to Multiple Reactions

The preceding chapter on single reactions showed that the performance (size)
of a reactor was influenced by the pattern of flow within the vessel. In this and
the next chapter, we extend the discussion to multiple reactions and show that
for these, both the size requirement and the distribution of reaction products
are affected by the pattern of flow within the vessel. We may recall at this point
that the distinction between a single reaction and multiple reactions is that the
single reaction requires only one rate expression to describe its kinetic behavior
whereas multiple reactions require more than one rate expression.
Since multiple reactions are so varied in type and seem to have so little
in common, we may despair of finding general guiding principles for design.
Fortunately, this is not so because many multiple reactions can be considered
to be combinations of two primary types: parallel reactions and series
reactions.
In this chapter we treat parallel reactions. In the next chapter we treat series
reactions as well as all sorts of series-parallel combinations.
Let us consider the general approach and nomenclature. First of all, we find
it more convenient to deal with concentrations rather than conversions. Second,
in examining product distribution the procedure is to eliminate the time variable
by dividing one rate equation by another. We end up then with equations relating
the rates of change of certain components with respect to other components of
the systems. Such relationships are relatively easy to treat. Thus, we use two
distinct analyses, one for determination of reactor size and the other for the
study of product distribution.
The two requirements, small reactor size and maximization of desired product,
may run counter to each other. In such a situation an economic analysis will
yield the best compromise. In general, however, product distribution controls;
consequently, this chapter concerns primarily optimization with respect to prod-
uct distribution, a factor which plays no role in single reactions.
Finally, we ignore expansion effects in this chapter; thus, we take E = 0
throughout. This means that we may use the terms mean residence time, reactor
holding time, space time, and reciprocal space velocity interchangeably.
 Chapter 7 Design for Parallel Reactions 153

Qualitative Discussion About Product Distribution. Consider the decomposi-

tion of A by either one of two paths:
R (desired product) (la)
k'f

S (unwanted product)

with corresponding rate equations

Dividing Eq. 2a by Eq. 2b gives a measure of the relative rates of formation of

R and S. Thus

and we wish this ratio to be as large as possible.

Now C, is the only factor in this equation which we can adjust and control
(k,, k,, a,, and a, are all constant for a specific system at a given temperature)
and we can keep C, low throughout the reactor by any of the following means:
by using a mixed flow reactor, maintaining high conversions, increasing inerts
in the feed, or decreasing the pressure in gas-phase systems. On the other hand,
we can keep CA high by using a batch or plug flow reactor, maintaining low
conversions, removing inerts from the feed, or increasing the pressure in gas-
phase systems.
For the reactions of Eq. 1 let us see whether the concentration of A should
be kept high or low.
If a, > a,, or the desired reaction is of higher order than the unwanted reaction,
Eq. 3 shows that a high reactant concentration is desirable since it increases the
RIS ratio. As a result, a batch or plug flow reactor would favor formation of
product R and would require a minimum reactor size.
If a, < a,, or the desired reaction is of lower order than the unwanted reaction,
we need a low reactant concentration to favor formation of R. But this would
also require large mixed flow reactor.
If a, = a,, or the two reactions are of the same order, Eq. 3 becomes

Y~ -
dCR - k1 - constant
Ys dCs k,

Hence, product distribution is fixed by k,lk, alone and is unaffected by type of

reactor used.
154 Chapter 7 Design for Parallel Reactions

We also may control product distribution by varying k21k,. This can be done
in two ways:
1. By changing the temperature level of operation. If the activation energies
of the two reactions are different, k,lk2 can be made to vary. Chapter 9
considers this problem.
2. By using a catalyst. One of the most important features of a catalyst is its
selectivity in depressing or accelerating specific reactions. This may be a
much more effective way of controlling product distribution than any of
the methods discussed so far.
We summarize our qualitative findings as follows:

For reactions in parallel, the concentration level of reactants is

the key to proper control ofproduct distribution. A high
reactant concentration favors the reaction of higher order, a
low concentration favors the reaction of lower order, while the
concentration level has n o effect o n the product distribution for
reactions of the same order.

When you have two or more reactants, combinations of high and low reactant
concentrations can be obtained by controlling the concentration of feed materials,
by having certain components in excess, and by using the correct contacting
pattern of reacting fluids. Figures 7.1 and 7.2 illustrate methods of contacting
two reacting fluids in continuous and noncontinuous operations that keep the
concentrations of these components both high, both low, or one high and the
other low. In general, the number of reacting fluids involved, the possibility of
recycle, and the cost of possible alternative setups must all be considered before
the most desirable contacting pattern can be achieved.
In any case, the use of the proper contacting pattern is the critical factor in
obtaining a favorable distribution of products for multiple reactions.

m CAPCB both high

n CA high, CB low

Add A and B \
all at one time
Add A and B 1
slowly; level rises
Start with A,
add B slowly
1
Figure 7.1 Contacting patterns for various combinations of high and low concentration
of reactants in noncontinuous operations.
 Chapter 7 Design for Parallel Reactions 155

Figure 7.2 Contacting patterns for various combinations of high and low concentration
of reactants in continuous flow operations.

CONTACTING PATTERNS FOR REACTIONS IN PARALLEL

The desired liquid-phase reaction

is accompanied by the unwanted side reaction

From the standpoint of favorable product distribution, order the contacting

schemes of Fig. 7.2, from the most desirable to the least desirable.

SOLUTION

Dividing Eq. (5) by Eq. (6) gives the ratio

which is to be kept as large as possible. According to the rule for reactions in

parallel, we want to keep CAhigh, CBlow, and since the concentration depen-
dency of B is more pronounced than of A, it is more important to have low CB
than high [Link] contacting schemes are therefore ordered as shown in Fig. E7.1.
156 Chapter 7 Design for Parallel Reactions

Most - Least
desirable desirable

CA high CA low CA high C, low

CB low CB low CB high CB high

(a) (b) (c) (4

Figure E7.1

Comment. Example 7.2 verifies these qualitative findings. We should also note
that there are still other contacting schemes which are superior to the best found
in this example. For example, if we can use an excess of a reactant, or if it is
practical to separate and recycle unconverted reactant, then vastly improved
hroduct distribution is possible.

Quantitative Treatment of Product Distribution and of Reactor Size. If rate

equations are known for the individual reactions, we can quantitatively determine
product distribution and reactor-size requirements. For convenience in evaluating
product distribution we introduce two terms, p and [Link], consider the decom-
position of reactant A, and let cp be the fraction of A disappearing at any instant
which is transformed into desired product R. We call this the instantaneous
fractional yield of R. Thus at any CA

.=( I--
moles Rformed - dCR
moles A reacted -dCA

For any particular set of reactions and rate equations cp is a function of CA,
and since CA in general varies through the reactor, p will also change with
position in the reactor. So let us define @ as the fraction of all the reacted A
that has been converted into R, and let us call this the overall fractional yield
of R. The overall fractional yield is then the mean of the instantaneous fractional
yields at all points within the reactor; thus we may write

all R formed - = --
C ~ f-
@ = C ~ f
(all A reacted) - CAo- CAf (-ACA) in. reactor

It is the overall fractional yield that really concerns us for it represents the
product distribution at the reactor outlet. Now the proper averaging for cp depends
 Chapter 7 Design for Parallel Reactions 157

on the type of flow within the reactor. Thus for plug flow,where C, changes
progressively through the reactor, we have from Eq. 7:

For PFR: @, =
-1
CAO
- CAf
I"
CAO
pdC, =-
1
jcAf
4C.A c*,
qdC,

For mixed flow,the composition is C,, everywhere, so cp is likewise constant

throughout the reactor, and we have

The over-all fractional yields from mixed and plug flow reactors processing A
from C,, to CAfare related by

a,,,= (2) at and QP= -

AC,
I"
c,
@,dCA

These expressions allow us to predict the yields from one type of reactor given
the yields from the other.
For a series of 1, 2, . . . , N mixed flow reactors in which the concentration of
A is CAI,C,,, . . . , C, the overall fractional yield is obtained by summing the
fractional yields in each of the N vessels and weighting these values by the
amount of reaction occurring in each vessel. Thus

from which

(~l(C,o - CAI)+ ~ C A- ICA,) + ... + ( P N ( ~ A -

, NCAN)
-I
@N mixed = (12)
c,, - CM
For any reactor type the exit concentration of R is obtained directly from Eq.
8. Thus

and Fig. 7.3 shows how C, is found for different types of reactors. For mixed
flow reactors, or mixed flow reactors in series, the best outlet concentration to
use, that which maximizes C,, may have to be found by maximization of rectangles
(see Chapter 6).
Now the shape of the cp versus C, curve determines which type of flow gives
the best product distribution, and Fig. 7.4 shows typical shapes of these curves
for which plug flow, mixed flow, and mixed followed by plug flow are best.
These fractional yield expressions allow us to relate the product distribution
from different types of reactors and to search for the best contacting scheme.
However, one condition must be satisfied before we can safely use these relation-
ships: We must truly have parallel reactions in which no product influences the
158 Chapter 7 Design for Parallel Reactions

Plug flow I 1 Staged reactors I

CRffrom CRfflOrn CRf from
4 Eqs. 13 and 9 1 4 Eqs. 13 and 10 ! Eqs. 13 and 12

Figure 7.3 Shaded and dashed area gives total R formed.

rate to change the product distribution. The easiest way to test this is to add
products to the feed and verify that the product distribution is in no way altered.
So far, the fractional yield of R has been taken as a function of C, alone and
has been defined on the basis of the amount of this component consumed. More
generally, when there are two or more reactants involved, the fractional yield
can be based on one of the reactants consumed, on all reactants consumed, or
on products formed. It is simply a matter of convenience which definition is
used. Thus, in general, we define p(M/N) as the instantaneous fractional yield
of M, based on the disappearance or formation of N.
The use of fractional yields to determine the product distribution for parallel
reactions was developed by Denbigh (1944, 1961).

The Selectivity. Another term, the selectivity,is often used in place of fractional
yield. It is usually defined as follows:

moles of desired product formed

selectivity =
moles of undesired material formed

Figure 7.4 The contacting pattern with the largest area produces most R: (a) plug flow
is best, (b) mixed flow is best, ( c ) mixed flow up to CAI followed by plug flow is best.
 Chapter 7 Design for Parallel Reactions 159

This definition may lead to difficulties. For example, if the reaction is the partial
oxidation of a hydrocarbon, as follows:

A (reactant) -to2
R (desired) +
a goulash of undesired
materials (CO, CO,, H 2 0 ,
CH,OH, etc.

Here the selectivity is hard to evaluate, and not very useful. Thus, we stay away
from selectivity and use the clearly defined and useful fractional yield, p(R/A).

flpi ' 'I II

PRODUCT DISTRIBUTION FOR PARALLEL REACTIONS

Consider the aqueous reactions

~ C =R1.0 C: Ci3,mol/[Link]
-
R, desired dt
k'f

For 90% conversion of A find the concentration of R in the product stream.

Equal volumetric flow rates of the A and of B streams are fed to the reactor,
and each stream has a concentration of 20 mollliter of reactant.
The flow in the reactor follows.
(a) Plug flow
(b) Mixed flow
( c ) The best of the four plug-mixed contacting schemes of Example 7.1.

As a warning, be careful to get the concentrations right when you mix streams.
We show this in the three sketches for this problem. Also, the instantaneous
fractional yield of the desired compound is

Now let us proceed.

(a) Plug Flow
Referring to Fig. E7.2a, noting that the starting concentration of each reactant
in the combined feed is CAO= C,, = 10 mollliter and that CA = CB everywhere,
we find from Eq. 9 that
 160 Chapter 7 Design for Parallel Reactions

In separate streams

cAf= cBf= 1
ego = 20 CR+, ~C~f =

Figure E7.2a

Let C y = x, then C, = x2 and dC, = 2xdx. Replacing C, by x in the above

expression gives

(b) Mixed Flow

Referring to Fig. E7.2b, we have from Eq. 10, for C, = CB

Therefore Eq. 13 gives

5x2: I CA0= c,, = 10

-Cb,
J = 20

Figure E7.2b

(c) Plug Flow A-Mixed Flow B

Assuming that B is introduced into the reactor in such a way that CB = 1 moll
liter throughout, we find concentrations as shown in Fig. E 7 . 2 ~Then
. accounting

cg, = 20 f f f) ~ s s u r n eC,= 1 everywhere

I Figure E 7 . 2 ~
 Chapter 7 Design for Parallel Reactions 161

I for the changing C, in the reactor, we find

I:(
@ =
-1
CA" - c
JCAf
~f
A
'"
'pic* =-
-1
19 - 1
j 1 CAdCA
19 CA f (I)'.'

Therefore

For plug flow: @ = 0.32 and CRf= 2.86 mollliter

For mixed flow: @ = 0.50 and CRf= 4.5 mollliter

For the optimum: @ = 0.87 and CRf= 7.85 mollliter

Note. These results verify the qualitative findings of Example 7.1.

The Side Entry Reactor

To evaluate how best to distribute the side entry feed and how to calculate the
corresponding conversion equation is a rather complex problem. For this type
of contacting in general see Westerterp et al. (1984).
To actually build a commercial-sized side entry reactor is another problem.
Chem. Eng. News (1997) reports on how this was cleverly done by using a reactor
somewhat like a shell-and-tube heat exchanger which used porous walled tubes.
Reactant A flowed through the tubes which contained baffles to encourage
lateral mixing of fluid and approach to plug flow. Reactant B, to be kept at a
close-to-constant low concentration in the tubes, enters the exchanger through
the shell side at a somewhat higher pressure than that in the tubes. Thus, B
diffused into the tubes along the whole length of the tubes.

GOOD OPERATING CONDITIONS FOR PARALLEL

REACTIONS

Often a desired reaction is accompanied by a variety of undesired side reactions,

some of higher order, some of lower order. To see which type of single reactor
gives the best product distribution, consider the simplest typical case, the parallel
 162 Chapter 7 Design for Parallel Reactions

decompositions of A, C,, = 2,

Find the maximum expected C, for isothermal operations

(a) in a mixed flow reactor
(b) in a plug flow reactor
( c ) in a reactor of your choice if unreacted A can be separated from the
product stream and returned to the feed at CAo= 2.

SOLUTION

Since S is the desired product, let us write fractional yields in terms of S. Thus

Plotting this function we find the curve of Fig. E7.3 whose maximum occurs where

CA
(a)

Figure E7.3a, b, c

I Solving we find

(a) Mixed Reactor. Most S is formed when the rectangle under the cp versus
CA curve has the largest area. The required conditions can be found either
by graphical maximization of rectangles or analytically. Since simple explicit
expressions are available in this problem, let us use the latter approach. Then
from Eqs. 10 and 13 we find for the area of the rectangle
 Chapter 7 Design for Parallel Reactions 163

Differentiating and setting to zero to find the conditions at which most S

is formed

Evaluating this quantity gives the optimum operating conditions of a mixed

reactor as

(b) Plug Flow Reactor. The production of S is maximum when the area under
the q versus C, curve is maximum. This occurs at 100% conversion of A, as seen
in Fig. E7.3b. Thus, from Eqs. 9 and 13

Evaluating this integral gives, for the plug flow optimum,

(c) Any Reactor with Separation and Recycle of Unused Reactant. Since no
reactant leaves the system unconverted, what is important is to operate at condi-
tions of highest fractional yield. This is at CA= 1,where cp (SIA) = 0.5, as shown
in Fig. E 7 . 3 ~ Thus,
. we should use a mixed flow reactor operating at CA = 1.
We would then have 50% of reactant A forming product S.

Comment. Summarizing, we find

moles S formed
= 0.33 for MFR
moles A fed
= 0.43 for a PFR

= 0.50 for an MFR with separation and recycle

Thus, a mixed flow reactor operating at conditions of highest cp with separation

and recycle of unused reactant gives the best product distribution. This result is
quite general for a set of parallel reactions of different order.

#
.I:, i , , , . l l l i
$
l
BEST OPERATING CONDITIONS FOR PARALLEL
REACTIONS

For the reaction of Example 7.3 determine the arrangement of reactors which
would produce most S in a flow system where recycle and reconcentration of
unreacted feed is not possible. Find C,,t,,t,, for this arrangement of reactors.
164 Chapter 7 Design for Parallel Reactions

From Fig. E7.4 we see that mixed flow followed by plug flow would be best.

Mixed
/- flow

cA
Figure E7.4

Thus, for mixed flow, from Example 7.3

C, = 1, p = 0.5, thus C, = p(ACA) = 0.5(2 - 1 ) = 0.5 mollliter

For plug flow, from Example 7.3

cs =- /: pdc, /,,(12+cc,)~
=
, 1
dC, = 0.386 mollliter

Therefore, the total amount of C, formed is

This is only slightly better than for plug flow alone, calculated in Example 7.3.

REFERENCES

Chem. and Eng. News, 51, April 7, 1997.

Denbigh, K. G., Trans. Faraday Soc., 40, 352 (1944).
Denbigh, K. G., Chem. Eng. Sci., 14, 25 (1961).
Westerterp, K. R., van Swaaij, W. P. M., and Beenackers, A. A. C. M., Chemical Reactor
Design and Operation, Wiley, New York, 1984.

PROBLEMS

7.1. For a given feed stream having C,, should we use a PFR or a MFR and
should we use a high or low or some intermediate conversion level for the
 Problems 165
exit stream if we wish to maximize (p(SIA)? The reaction system is
R
A 1.;" S, desired
S
-+

where n,, n,, and n, are the reaction orders of reactions 1,2, and 3.
(a) n, = 1, n, = 2, n, = 3
(b) n, = 2, n, = 3, n3 = 1
(c) n1 = 3, n, = 1, n3 = 2

Using separate feeds of A and B sketch the contacting pattern and reactor
conditions which would best promote the formation of product R for the
following systems of elementary reactions.

A+B+R
A+
7.2. +
A+S
R} HOW system
2B + T

7.4. A + tR}
A+S
Batch system 7.5. A +
2A+S
+
R} Flow system

7.6. Substance A in a liquid reacts to produce R and S as follows:

R first order
f
A
\i
S first order
A feed (CAo = 1, C,, = 0, C,, = 0 ) enters two mixed flow reactors in
series, (7, = 2.5 min, 7, = 5 min). Knowing the composition in the first
reactor (CAI= 0.4, CRl = 0.4, CSl = 0.2), find the composition leaving the
second reactor.

7.7. Substance A in the liquid phase produces R and S by the following reactions:

The feed (CAo= 1.0, CRo= 0, Cso = 0.3) enters two mixed flow reactors
in series ( 7 1 = 2.5 min, 7, = 10 min). Knowing the composition in the first
reactor (CAI = 0.4, CRl = 0.2, C,, = 0.7), find the composition leaving the
second reactor.
166 Chapter 7 Design for Parallel Reactions

Liquid reactant A decomposes as follows:

A feed of aqueous A (C,, = 40 mol/m3) enters a reactor, decomposes, and

a mixture of A, R, and S leaves.

7.8. Find C,, and C, and r for X , = 0.9 in a mixed flow reactor.

7.9. Find C,, and C, and r for X , = 0.9 in a plug flow reactor.

7.10. Find the operating condition (X,,7; and C,) which maximizes Cs in a
mixed flow reactor.

7.11. Find the operating condition (X,, r, and C,) which maximizes C, in a
mixed flow reactor.

7.12. Reactant A in a liquid either isomerizes or dimerizes as follows:

(a) Write P(RIA) and (p[RI(R + S)].

With a feed stream of concentration C,, find C,,,,, which can be formed
(b) in a plug flow reactor,
(c) in a mixed flow reactor.
A quantity of A of initial concentration C,, = 1 mollliter is dumped into
a batch reactor and is reacted to completion.
(d) If C, = 0.18 mollliter in the resultant mixture, what does this tell of
the kinetics of the reaction?

7.13. In a reactive environment, chemical A decomposes as follows

For a feed stream C,, = 4 mollliter what size ratio of two mixed flow
reactors will maximize the production rate of R? Also give the composition
of A and R leaving these two reactors.
 Problems 167

Consider the parallel decomposition of A of different orders

Determine the maximum concentration of desired product obtainable in

(a) plug flow,
(b) mixed flow.

7.14, R is the desired product and CAo= 2.

7.15, S is the desired product and CAo= 4.

7.16, T is the desired product and CAo= 5.

Under ultraviolet radiation, reactant A of CA, = 10 kmol/m3 in a process

stream (v = lm3/min) decomposes as follows.

r, = 16~:' kmol/[Link]
rR3
A- S, rS = 12CA kmol/[Link]
I
T, r, = C; [Link]

We wish to design a reactor setup for a specific duty. Sketch the scheme
selected, and calculate the fraction of feed transformed into desired product
as well as the volume of reactor needed.

7.17. Product R is the desired material.

7.18, Product S is the desired material.

7.19, Product T is the desired material,

The stoichiometry of a liquid-phase decomposition is known to be

In a series of steady-state flow experiments (CAo= 100, CRO= CSO= 0) in

a laboratory mixed flow reactor the following results are obtained:

Further experiments indicate that the level of C , and Cs have no effect

on the progress of the reaction.
168 Chapter 7 Design for Parallel Reactions

7.20. With a feed C,, = 100 and exit concentration CAf= 20, find CRat the exit
from a plug flow reactor.

7.21. With C,, = 200 and CAf= 20, find C , at the exit from a mixed flow reactor.

7.22. How should we operate a mixed flow reactor so as to maximize the produc-
tion of R? Separation and recycle of unused reactant is not practical.

When aqueous A and aqueous B (C,, = CBo)are brought together they

react in two possible ways:
mol
R +T. r, = 5 0 C ~-
f [Link]
1 mol
S + U, rs = l0OCB -
[Link]
to give a mixture whose concentration of active components (A, B, R, S,
T, U) is C,,,,, = C,, +
CB, = 60 mol/m3. Find the size of reactor needed
and the R/S ratio produced for 90% conversion of an equimolar feed of
FA,= FBo= 300 mollhr:

7.23. in a mixed flow reactor;

7.24. in a plug flow reactor;

7.25. in a reactor that gives highest CR. Chapter 6 tells that this should be plug
flow for A and side entry for B. In such a reactor introduce B in such a
way that CBis constant throughout the reactor.

7.26. Reactant A decomposes in an isothermal batch reactor (CAo= 100 ) to

produce wanted R and unwanted S, and the following progressive concen-
tration readings are recorded:

Additional runs show that adding R or S does not affect the distribution
of products formed and that only A does. Also, it is noted that the total
number of moles of A, R, and S is constant.
(a) Find the cp versus C, curve for this reaction.
With a feed of CAo= 100 and CAf = 10, find C,
(b) from a mixed flow reactor,
(c) from a plug flow reactor,
(d) and (e): Repeat parts (b) and (c) with the modification that CAo= 70.

7.27. The great naval battle, to be known to history as the Battle of Trafalgar
(1805), was soon to be joined. Admiral Villeneuve proudly surveyed his
powerful fleet of 33 ships stately sailing in single file in the light breeze.
 Problems 169

The British fleet under Lord Nelson was now in sight, 27 ships strong.
Estimating that it would still be 2 hours before battle, Villeneuve popped
open another bottle of burgundy and point by point reviewed his carefully
thought-out battle strategy. As was the custom of naval battles at that time,
the two fleets would sail in single file parallel to each other and in the same
direction, firing their cannons madly. Now, by long experience in battles
of this kind, it was a well-known fact that the rate of destruction of a fleet
is proportional to the fire power of the opposing fleet. Considering his
ships to be on a par, one for one, with the British, Villeneuve was confident
of victory. Looking at his sundial, Villeneuve sighed and cursed the light
wind-he'd never get it over with in time for his afternoon snooze. "Oh
well," he sighed, "c'est la vie." He could see the headlines next morning:
"British fleet annihilated, Villeneuve's losses are . . ." Villeneuve stopped
short. How many ships would he lose? Villeneuve called over his chief
bottle-cork popper, Monsieur Dubois, and asked this question. What an-
swer does he get?
At this very moment, Nelson, who was enjoying the air on the poop
deck of the Victory, was stuck with the realization that all was ready except
for one detail-he had forgotten to formulate his battle plan. Commodore
Archibald Forsythe-Smythe, his trusty trusty, was hurriedly called over for
a conference. Being familiar with the firepower law, Nelson was loathe to
fight the whole French fleet (he could see the headlines too). Now certainly
it was no disgrace for Nelson to be defeated in battle by superior forces,
so long as he did his best and played the game; however, he had a sneaking
suspicion that maybe he could pull a fast one. With a nagging conscience as
to whether it was cricket or not, he proceeded to investigate this possibility.
It was possible to "break the line3'-that is, to start parallel to the French
fleet, and then cut in and divide the enemy fleet into two sections. The
rear section could be engaged and disposed of before the front section
could turn around and rejoin the fray. Now to the question. Should he
split the French fleet and if so, then where? Commodore Forsythe-Smythe,
who was so rudely taken from his grog, grumpily agreed to consider this
question and to advise Nelson at what point to split the French fleet so as
to maximize their chance for success. He also agreed to predict the outcome
of the battle using this strategy. What did he come up with?

7.28 Find the size of the two reactors needed in Example 7.4 (see Figure E7.4)
for a feed flow rate of 100 literls, and for reaction rates given in units of
mollliter. s.
Chapter 8
Potpourri of Multiple Reactions

Chapter 7 considered reactions in parallel. These are reactions where the product
does not react further. This chapter considers all sorts of reactions where the
product formed may react further. Here are some examples:

A-R-S
\1\1
T U
Series Series parallel, or Denbigh system
consecutive-competitive

AZR-S
Reversible and Reversible Reversible
irreversible network

We develop or present the performance equations of some of the simpler systems

and point out their special features such as maxima of intermediates.

8.1 IRREVERSIBLE FIRST-ORDER REACTIONS IN SERIES

For easy visualization consider that the reactions

proceed only in the presence of light, that they stop the instant the light is turned
off, and that for a given intensity of radiation, the rate equations are

Our discussion centers about these reactions.

 8.1 Irreversible First-Order Reactions in Series 171

Time since light was turned on

Figure 8.1 Concentration-time curves if
the contents of the beaker are irradi-
ated uniformly.

Qualitative Discussion About Product Distribution. Consider the following two

ways of treating a beaker containing A: First, the contents are uniformly irradi-
ated; second, a small stream is continuously withdrawn from the beaker, irradi-
ated, and returned to the beaker; the rate of absorption of radiant energy is the
same in the two cases. The two schemes are shown in Figs. 8.1 and 8.2. During
this process A disappears and products are formed. Is the product distribution
of R and S different in the two beakers? Let us see whether we can answer this
question qualitatively for all values of the rate constants.
In the first beaker, when the contents are being irradiated all at the same time,
the first bit of light will attack A alone because only A is present at the start.
The result is that R is formed. With the next bit of light both A and R will
compete; however, A is in very large excess so it will preferentially absorb the
radiant energy to decompose and form more R. Thus, the concentration of R
will rise while the concentration of A will fall. This process will continue until
R is present in high enough concentration so that it can compete favorably with
A for the radiant energy. When this happens, a maximum R concentration is
reached. After this the decomposition of R becomes more rapid than its rate of

Essentially
A-R-S

Y
b
m
a
.-C
C
.-
-
C1
E
al
C
8

Time since light was turned on

Figure 8.2 Concentration-time curves for the contents of the beaker if
only a small portion of the fluid is irradiated at any instant.
172 Chapter 8 Potpourri of Multiple Reactions

formation and its concentration drops. A typical concentration time curve is

shown in Fig. 8.1.
In the alternative way of treating A, a small fraction of the beaker's contents
is continuously removed, irradiated, and returned to the beaker. Although the
total absorption rate is the same in the two cases, the intensity of radiation
received by the removed fluid is greater, and it could well be, if the flow rate is
not too high, that the fluid being irradiated reacts essentially to completion. In
this case, then, A is removed and S is returned to the beaker. So, as time passes
the concentration of A slowly decreases in the beaker, S rises, while R is absent.
This progressive change is shown in Fig. 8.2.
These two methods of reacting the contents of the beaker yield different
product distributions and represent the two extremes in possible operations, one
with a maximum possible formation of R and the other with a minimum, or no
formation, of R. How can we best characterize this behavior? We note in the
first method that the contents of the beaker remain homogeneous throughout,
all changing slowly with time, whereas in the second a stream of highly reacted
fluid is continually being mixed with fresh fluid. In other words, we are mixing
two streams of different compositions. This discussion suggests the following
rule governing product distribution for reactions in series.

1 For irreversible reactions in series the mixing of fluid of

different composition is the key to the formation of
intermediate. The maximum possible amount of any and all
intermediates is obtained if fluids of different compositions
and at different stages of conversion are not allowed to mix.

As the intermediate is frequently the desired reaction product, this rule allows
us to evaluate the effectiveness of various reactor systems. For example, plug
flow and batch operations should both give a maximum R yield because here
there is no mixing of fluid streams of different compositions. On the other hand,
the mixed reactor should not give as high a yield of R as possible because a
fresh stream of pure A is being mixed continually with an already reacted fluid
in the reactor.
The following examples illustrate the point just made. We then give a quantita-
tive treatment which will verify these qualitative findings.

FAVORABLE CONTACTING PATTERNS FOR ANY SET

OF IRREVERSIBLE REACTIONS IN SERIES, NOT JUST
A+R-S

Which contacting pattern of Figs. E8.1, when properly operated, can give a higher
concentration of any intermediate, the contacting pattern on the left or the one
on the right?
 8.1 Irreversible First-Order Reactions in Series 173

(dl
Figure [Link], b, c, d

I SOLUTION
Focusing on the mixing rule for reactions in series, that the extent of mixing of
streams of different composition should be minimized, we reason
for part (a): The left pattern is better; in fact it is the best possible flow
scheme.
for part (b): Looking at Figs. 6.5,6.6, and 6.16,6.17 of Chapter 6 we see that
the left is closer to plug flow for both first- and for second-order
reactions. So we generalize this to any positive order reaction.
for part (c): The right pattern is better because it is closer to plug flow.
for part (d): Turbulent flow has less intermixing of fluids of different ages,
less bypassing; hence, the right scheme is better.

Note. In the quantitative analysis that follows we verify this general and impor-
tant rule.

Quantitative Treatment, Plug Flow or Batch Reactor. In Chapter 3 we devel-

oped the equations relating concentration with time for all components of the
unimolecular-type reactions
174 Chapter 8 Potpourri of Multiple Reactions

in batch reactors. The derivations assumed that the feed contained no reaction
products R or S. If we replace reaction time by the space time, these equations
apply equally well for plug flow reactors, thus

The maximum concentration of intermediate and the time at which it occurs is

given by

rp,opt - --
h o g mean k2 - kl

This is also the point at which the rate of formation of S is most rapid.
Figure 8.3a, prepared for various k,lk, values, illustrates how this ratio governs

Figure 8.3a, b Behavior of unimolecular-type reactions

in a plug flow reactor: (a) concentration-time curves, and (b) relative concentration of the
reaction components. See Fig. 8.13 for a more detailed figure.
 8.1 Irreversible First-Order Reactions in Series 175

Figure 8.4 Variables for reactions in series

(no R or S in the feed) occurring in a mixed
flow reactor.

the concentration-time curves of the intermediate R. Figure 8.3b, a time-indepen-

dent plot, relates the concentration of all reaction components; also see Eq. 37.

Quantitative Treatment, Mixed Flow Reactor. Let us develop the concentra-

tion-time curves for this reaction when it takes place in a mixed flow reactor.
This may be done by referring to Fig. 8.4. Again, the derivation will be limited
to a feed which contains no reaction product R or S.
By the steady-state material balance we obtain for any component

input = output + disappearance by reaction (4.1) or (10)

which for reactant A becomes

Noting that

we obtain for A, on rearranging,

For component R the material balance, Eq. 10, becomes

176 Chapter 8 Potpourri of Multiple Reactions

With Eqs. 11 and 12 we obtain, on rearranging,

C, is found by simply noting that at any time

C, + CR + C, = C,, = constant

hence

The location and maximum concentration of R are found by determining

dCR/d.r, = 0. Thus

which simplifies neatly to give

The corresponding concentration of R is given by replacing Eq. 15 in Eq. 13.

On rearranging, this becomes

Typical concentration-time curves for various k21k, values are shown in Fig.
8.5a. A time-independent plot, Fig. 8.5b, relates the concentrations of reactant
and products.

Remarks on Performance Characteristics, Kinetic Studies, and Design. Figures

8.3a and 8.5a show the general time-concentration behavior for plug and mixed
flow reactors and are an aid in visualizing the actual progress of the reaction.
Comparison of these figures shows that except when k1 = k2 the plug flow
reactor always requires a smaller time than does the mixed reactor to achieve
the maximum concentration of R, the difference in times becoming progressively
 8.1 Irreversible First-Order Reactions in Series 177

k17 xA= 1 - cA/cAo

(a) (b)
Figure 8.5a, b Behavior of unimolecular-type reactions

in a mixed flow reactor: (a) concentration-time curves, and (b) relative concentration of the
reaction components. See Fig. 8.14 for a more detailed figure.

larger as k21k,departs from unity (see Eqs. 15 and 9). In addition, for any reaction
the maximum obtainable concentration of R in a plug flow reactor is always
higher than the maximum obtainable in a mixed reactor (see Eqs. 16 and 8).
This verifies the conclusions arrived at by qualitative reasoning.
Figures 8.3b and 8.5b, time-independent plots, show the distribution of materi-
als during reaction. Such plots find most use in kinetic studies because they allow
the determination of k21k, by matching the experimental points with one of the
family of curves on the appropriate graph. Figures 8.13 and 8.14 are more detailed
representations of these two figures. Though not shown in the figures, C, can
be found by difference between CA, and CA + C,.
Figure 8.6 presents the fractional yield curves for intermediate R as a function
of the conversion level and the rate constant ratio. These curves clearly show
that the fractional yield of R is always higher for plug flow than for mixed flow
for any conversion level. A second important observation in this figure concerns
the extent of conversion of A we should plan for. If for the reaction considered
k,lkl is much smaller than unity, we should design for a high conversion of A
and probably dispense with recycle of unused reactant. However, if k,lk, is
greater than unity, the fractional yield drops very sharply even at low conversion.
Hence, to avoid obtaining unwanted S instead of R we must design for a very
small conversion of A per pass, separation of R, and recycle of unused reactant.
In such a case large quantities of material will have to be treated in the A-R
separator and recycled, and this part of the process will figure prominently in
cost considerations.
178 Chapter 8 Potpourri of Multiple Reactions

XA = 1 - cA/cAo

Figure 8.6 Comparison of the fractional yields of R in mixed flow and

plug flow reactors for the unimolecular-type reactions

8.2 FIRST-ORDER FOLLOWED BY ZERO-ORDER REACTION

Let the reactions be

-rA = klCA
AAR-s
nl=l n2=0
k2
where K = -
k21C~0
rR= k l C A - k2 kl

For batch or plug flow with CRo= Cso = 0 integration gives

and
 8.3 Zero-Order Followed by First-Order Reaction 179
Concentration

Initial slope
klCA0 - k2 '

Slope = k 2 1 fcR,max (Eq. 21) Get this t from

k2 + e-kl' =1
C~~

Figure 8.7 Product distribution for the reactions A %R -%s

The maximum concentration of intermediate, CR,m,,,and the time when this

occurs is found to be

and

Graphically we show these findings in Fig. 8.7.

8.3 ZERO-ORDER FOLLOWED BY FIRST-ORDER REACTION

Let the reactions be

kl k2
A-R-Snl=O nz-1 = kl - k,cR A present
= - k,CR A absent

for batch or plug pow with CRO= CSO= 0 integration gives

180 Chapter 8 Potpourri of Multiple Reactions

Initial slop; = 0

Figure 8.8 Product distribution for the reactions A S R ~ S

and

The maximum concentration of intermediate, C,,,,,, and the time when this
occurs is found to be

and

Graphically we show these findings in Fig. 8.8.

8.4 SUCCESSIVE IRREVERSIBLE REACTIONS OF DIFFERENT

ORDERS
In principle, concentration-time curves can be constructed for successive reac-
tions of different orders. For the plug flow or batch reactor and for the mixed
reactor explicit solutions are difficult to obtain; thus, numerical methods provide
the best tool for treating such reactions.
 8.6 Irreversible Series-Parallel Reactions 181
For these reactions the concentration-time curves are of little generality for
they are dependent on the concentration of reactant in the feed. As with reactions
in parallel, a rise in reactant concentration favors the higher-order reaction; a
lower concentration favors the lower-order reaction. This causes a shift in C*, ,
and this property can be used to improve the product distribution.

8.5 REVERSIBLE REACTIONS

Solution of the equations for successive reversible reactions is quite formidable
even for the first-order case; thus, we illustrate only the general characteristics
for a few typical cases. Consider the reversible first-order reactions

and

Figures 8.9 and 8.10 display the concentration time curves for the components
in batch or plug flow for different values of the rate constants.
Figure 8.9 shows that the concentration of intermediate in reversible series
reactions need not pass through a maximum, while Fig. 8.10 shows that a product
may pass through a maximum concentration typical of an intermediate in the
irreversible series reaction; however, the reactions may be of a different kind.
A comparison of these figures shows that many of the curves are similar in shape,
making it difficult to select a mechanism of reaction by experiment, especially
if the kinetic data are somewhat scattered. Probably the best clue to distinguishing
between parallel and series reactions is to examine initial rate data-data ob-
tained for very small conversion of reactant. For series reactions the time-concen-
tration curve for S has a zero initial slope, whereas for parallel reactions this is
not so.

8.6 IRREVERSIBLE SERIES-PARALLEL REACTIONS

Multiple reactions that consist of steps in series and steps in parallel are called
series-parallel reactions. From the point of view of proper contacting, these
reactions are more interesting than the simpler types already considered because
a larger choice of contacting is usually possible, leading to much wider differences
in product distribution. Thus, design engineers are dealing with a more flexible
system and this affords them the opportunity to display their talents in devising
the best of the wide variety of possible contacting patterns. Let us develop our
ideas with a reaction type that represents a broad class of industrially important
reactions. We will then generalize our findings to other series-parallel reactions.
182 Chapter 8 Potpourri of Multiple Reactions

1 1
A-R-S

Figure 8.9 Concentration-time curves for the ele-

mentary reversible reactions

From Jungers et al. (1958), p. 207.

 8.6 Irreversible Series-Parallel Reactions I83

Figure 8.10 Concentration-time curves

for the elementary reversible parallel re-
actions

From Jungers et al. (1958), p. 207.

184 Chapter 8 Potpourri of Multiple Reactions

For the reaction set consider the successive attack of a compound by a reactive
material. The general representation of these reactions is

A+B&R

R+B*S

s+~k3-T
etc.

t B , k, t B , k, +B,k,
A-R-S-T

where A is the compound to be attacked, B is the reactive material, and R, S,

T, etc., are the polysubstituted materials formed during reaction. Examples of
such reactions may be found in the successive substitutive halogenation (or
nitration) of hydrocarbons, say benzene or methane, to form monohalo, dihalo,
trihalo, etc., derivatives as shown below:

Another important example is the addition of alkene oxides, say ethylene oxide,
to compounds of the proton donor class such as amines, alcohols, water, and
hydrazine to form monoalkoxy, dialkoxy, trialkoxy, etc., derivatives, some exam-
ples of which are shown below:

0
/ \ CH2-CH20H ,,0
lH +CH2-CH, / +CH,-C H ,
N,- H N-H ... * N (CH2-CH20H),
\ \ triethanolamine, TEA
H H
monoethanolamine, MEA

H 0
/ \ CH2-CH20H 0 ,,
+CH,-CH, / +CH,-C H ,
0 + 0 * O= (CH, -CH20H),
\ \ diethylene glycol
H H
ethylene glycol

Such processes are frequently bimolecular, irreversible, hence second-order ki-

netically. When occurring in the liquid phase they are also essentially constant-
density reactions.
 8.6 Irreversible Series-Parallel Reactions 185

Two-Step Irreversible Series-Parallel Reactions

We first consider the two-step reaction where the first substitution product is
desired. Actually for an n-step reaction the third and succeeding reactions do
not occur to any appreciable extent and may be ignored if the mole ratio of A
to B is high (see qualitative treatment given next). The reaction set considered
is thus

With the assumption that the reaction is irreversible, bimolecular, and of constant
density, the rate expressions are given by

Qualitative Discussion About Product Distribution. To get the "feel" for what
takes place when A and B react according to Eq. 31, imagine that we have two
beakers, one containing A and the other containing B. Should it make any
difference in the product distribution how we mix A and B? To find out, consider
the following ways of mixing the reactants: (a) add A slowly to B, (b) add B
slowly to A, and finally (c) mix A and B together rapidly.

(a) Add A Slowly to B. For the first alternative pour A a little at a time into
the beaker containing B, stirring thoroughly and making sure that all the A is
used up and that the reaction stops before the next bit is added. With each
addition a bit of R is produced in the beaker. But this R finds itself in an excess
of B so it will react further to form S. The result is that at no time during the
slow addition will A and R be present in any appreciable amount. The mixture
becomes progressively richer in S and poorer in B. This continues until the
beaker contains only S. Figure 8.11 shows this progressive change.

(b) Add B Slowly to A. Now pour B a little at a time into the beaker containing
A, again stirring thoroughly. The first bit of B will be used up, reacting with A
to form R. This R cannot react further for there is now no B present in the
186 Chapter 8 Potpourri of Multiple Reactions

Moles A added differentially

(all A is consumed with each addition)
Figure 8.11 Distribution of materials in the B
beaker for the method of mixing shown.

mixture. With the next addition of B, both A and R will compete with each
other for the B added, and since A is in very large excess it will react with most
of the B, producing even more R. This process will be repeated with progressive
buildup of R and depletion of A until the concentration of R is high enough so
that it can compete favorably with A for the B added. When this happens, the
concentration of R reaches a maximum, then decreases. Finally, after addition
of 2 moles of B for each mole of A, we end up with a solution containing only
S. This progressive change is shown in Fig. 8.12.

(c) Mix A and B Rapidly. Now consider the third alternative where the contents
of the two beakers are rapidly mixed together, the reaction being slow enough
so that it does not proceed to any appreciable extent before the mixture becomes
uniform. During the first few reaction increments R finds itself competing with
a large excess of A for B and hence it is at a disadvantage. Carrying through
this line of reasoning, we find the same type of distribution curve as for the
mixture in which B is added slowly to A. This situation is shown in Fig. 8.12.
The product distribution of Figs. 8.11 and 8.12 are completely different. Thus,
when A is kept uniform in composition as it reacts, as in Fig. 8.12, then R is
formed. However, when fresh A is mixed with partly reacted mixture, as in Fig.
8.11, then no intermediate R forms. But this is precisely the behavior of reactions
in series. Thus, as far as A, R, and S are concerned, we may view the reactions
of Eq. 31 as

+B tB
A-R-S

A second observation of Fig. 8.12 is that the concentration level of B, whether

high or low, has no effect on the path of the reaction and on the distribution of
 8.6 Irreversible Series-Parallel Reactions 187

-
01
A , Contents'uniformly
mixed before
.-L
lo
reaction starts
22 L

F:
+
0 n
a,
c M
.5
.z
R + B-tS
b E
5U '-

8
*
0 1 2
AB = B, - 8, moles B consumed during reaction
with either method of mixing reactants
Figure 8.12 Distribution of materials in the mixing beaker for either of
the methods of mixing shown.

products. But this is precisely the behavior of parallel reactions of the same
order. So with respect to B, Eq. 31 can be looked at as

From this discussion we propose the general rule:

Irreversible series-parallel reactions can be analyzed in terms of their

constituent series reactions and parallel reactions in that optimum
contacting for favorable product distribution is the same as for the
constituent reactions.

For the reactions of Eq. 31 where R is desired this rule shows that the best
way of contacting A and B is to react A uniformly, while adding B in any
convenient way.
This is a powerful generalization which, without needing specific values for
the rate constants, can already show in many cases which are the favorable
contacting patterns. It is essential, however, to have the proper representation
of the stoichiometry and form of rate equation. Example 8.6 and many of the
problems of Chapter 10 apply these generalizations.
188 Chapter 8 Potpourri of Multiple Reactions

Quantitative Treatment, Plug Flow or Batch Reactor. Here we quantitatively

treat the reactions of Eq. 31 with the understanding that R, the intermediate, is
the desired product, and that the reaction is slow enough so that we may ignore
the problems of partial reaction during the mixing of reactants.
In general, taking the ratio of two rate equations eliminates the time variable
and gives information on the product distribution. So dividing Eq. 34 by Eq. 32
we obtain the first-order linear differential equation

whose method of solution is shown in Chapter 3. With no R present in the feed

the limits of integration are C,, to C, for A and C,, = 0 for R, and the solution
of this differential equation is

with maximum CR at

This gives the relationship between C, and C, in a batch or in a plug flow

reactor. To find the concentrations of the other components, simply make a
material balance. An A balance gives

from which C, can be found as a function of C, and C,. Finally, a balance about
B gives

from which C, can be found.

 8.6 Irreversible Series-Parallel Reactions 189

Quantitative Treatment, Mixed Flow. Writing the design equation for mixed
flow in terms of A and R gives

Rearranging, we obtain

which is the difference equation corresponding to the differential equation, Eq.

36. Writing CR in terms of CAthen gives

with

Equations 39 and 40, material balances about A and B in plug flow, hold equally
well for mixed flow and serve to complete the set of equations giving complete
product distribution in this reactor.

Graphical Representation. Figures 8.13 and 8.14, time-independent plots, show

the distribution of materials in plug and mixed flow and are prepared from Eqs.
37 to 41. As mentioned earlier, A, R, and S behave like the components in first-
order reactions in series. Comparing Figs. 8.13 and 8.14 with Figs. 8.3b and 8.5b,
we see that the distribution of these materials is the same in both cases, plug
flow again giving a higher concentration of intermediate than mixed flow. The
lines of slope 2 on these charts show the amount of B consumed to reach any
particular point on the curve. It makes no difference whether B is added all at
one time as in a batch reactor or a little at a time as in a semibatch reactor; in
either case the same point on the chart will be reached when the same total
amount of B is consumed.
These figures indicate that no matter what reactor system is selected, when
the fractional conversion of A is low the fractional yield of R is high. Thus, if
it is possible to separate R cheaply from a product stream, the optimum setup
for producing R is to have small conversions per pass coupled with a separation
of R and recycle of unused A. The actual mode of operation will, as usual,
depend on the economics of the system under study.
190 Chapter 8 Potpourri o f Multiple Reactions

Figure 8.13 Distribution of materials in a batch or plug flow reactor for the elementary series-
parallel reactions

Experimental Determination of the Kinetics of Reaction. The ratio k21kl may

be found by analyzing the products of reaction from an experiment, and locating
the corresponding point on the appropriate design chart. The simplest way to
do this is to use different ratios of B to A in a batch reactor, allowing the reaction
to go to completion each time. For each run a value of k,lkl can be determined.
 8.6 Irreversible Series-Parallel Reactions 191

Figure 8.14 Distribution of materials in a mixed flow reactor for the elementary series-parallel reac-
tions

The best mole ratios to use are those where the lines of constant k21k,are furthest
apart, or when - ( A B / A , ) = 1.0, or close to equimolar ratios.
With k 2 / k , known, all that is needed is k, which must be found by kinetic
experiments. The simplest procedure is to use a large excess of B in which case
the disappearance of A follows first-order kinetics.
192 Chapter 8 Potpourri of Multiple Reactions

KINETICS OF SERIES-PARALLEL REACTION

From each of the following experiments, what can we say about the rate constants
of the multiple reactions

(a) Half a mole of B is poured bit by bit, with stirring, into a flask containing
a mole of A. The reaction proceeds slowly, and when B is entirely consumed
0.67 mole of A remains unreacted.
(b) One mole of A and 1.25 moles of B are rapidly mixed, and the reaction is
slow enough so that it does not proceed to any appreciable extent before
homogeneity in composition is achieved. On completion of the reaction, 0.5
mole of R is found to be present in the mixture.
(c) One mole of A and 1.25 moles of B are rapidly brought together. The
reaction is slow enough so that it does not proceed to any appreciable extent
before homogeneity in A and B is achieved. At the time when 0.9 mole of
B is consumed, 0.3 mole of S is present in the mixture.

1 SOLUTION
The sketches in Fig. E8.2 show how Fig. 8.13 is used to find the desired informa-
tion. Thus we find:

(a) k21k, = 4, (b) k21k, = 0.4, (c) k21k, = 1.45

A = 0.67
A0
(a)

Figure E8.2
 8.6 Irreversible Series-Parallel Reactions 193

Extensions and Applications

Three or More Reactions. Analysis of three or more reactions can be made
by procedures analogous to those presented. Of course, the mathematics becomes
more involved; however, much of the extra labor can be avoided by selecting
experimental conditions in which only two reactions need be considered at any
time. Figure 8.15 shows product distribution curves for one such reaction set,
the progressive chlorination of benzene.
Catipovic and Levenspiel (1979) have developed charts to represent the first
three steps of an n-step reaction sequence. Beyond three steps, simple perfor-
mance charts cannot be prepared.
Again, as with the two-reaction set, we find that a plug flow reactor yields a
higher maximum concentration of any intermediate than does a mixed flow re-
actor.

Polymerization. The field of polymerization affords an opportunity for a fruitful

application of these ideas. Often hundreds or even thousands of reactions in
series occur in the formation of polymers, and the type of crosslinking and
molecular weight distribution of these products are what gives these materials
their particular physical properties of solubility, density, flexibility, etc.
Since the mode of mixing of monomers with their catalysts profoundly affects
product distribution, great importance must be paid to this factor if the product
is to have the desired physical and chemical properties. Denbigh (1947, 1951)
considered some of the many aspects of this problem, and Fig. 8.16 shows for
various kinetics how reactor type influences the molecular weight distribution
of products.

- I
Batch or plug flow
Mixed I
I

Figure 8.15 Product distribution in the progressive chlorination of benzene:

A S B ~ R + U C6H6+ C12A C,H,C~+ HCI

With k,lk, = 118 and k,Ik, = 11240; from R. B. MacMullin (1948).

194 Chapter 8 Potpourri of Multiple Reactions

Number of monomer units in polymer

Figure 8.16 Type of flow and kinetics influence the molecular weight distribution
of polymer: (a) duration of polymerization reaction (life of active polymer) is
short compared to the reactor holding time; (b) duration of polymerization reaction
is long compared to the reactor holding time, or where polymerization has no
termination reaction. Adapted from Denbigh (1947).

8.7 THE DENBIGH REACTIONS AND THEIR SPECIAL CASES

Denbigh (1958) was the first to treat the following rather general reaction scheme

with

The performance equations for this reaction scheme reduce directly to all the
special cases, for example

This scheme has wide application to a whole host of real reacting systems.
 8.7 The Denhigh Reactions and its Special Cases 195

These rate equations are all of first order and so to develop the performance
expressions does not involve complex mathematics, although it may be a tedious
task. In our treatment we will not present the detailed calculation steps, but will
just present the final results.

Batch or Plug Flow Reactors. Integration gives the performance equations for
this system

-
Cu . c s but with k3-- k4 and C,
. . same as - -- Cuo
c ~ o c ~ o

For the special case where CRo= Cso = CTo= Cuo = 0 the above expressions
simplify. We also can find CR = f(CA), thus

and

For C,, = C, = CTo= Cuo= 0 the behavior of the system is displayed as shown
in Fig. 8.17. This figure also shows CR,,,, and the time when this occurs in terms
of the rate constants.
196 Chapter 8 Potpourri of Multiple Reactions

Zero 'slope
Figure 8.17 Progress of the Denbigh reaction scheme in a plug flow reactor
for C,, = Cso = C,, = C,, = 0.

Mixed Flow Reactors. Using the mixed flow performance equation with these
rates gives

c u ... same as -
- c s but with k3c+ k4 and Csoc+ C,,
(55)
c ~ o c~o
At the optimum

Graphically, for C,, = Cso = CTo = Cuo = 0, Fig. 8.18 shows the behavior of
this system.
 8.7 The Denbigh Reactions and its Special Cases 197

Ln
.-
0
4-
e
.,-
w
K

0
C
0

* T = -v= - c ~ O v
~ m R,, max (Eq, 57) F~~

Figure 8.18 Progress of the Denbigh reaction scheme in a mixed flow reactor
for C, = C,, = C,, = C,, = 0.

Comments, Suggestions, and Extensions

The equations of this chapter can be extended and applied directly to many
other reaction schemes, for example

Some problems at the end of this chapter consider these extensions.

A careful examination of the shape of the C versus 7 curves gives much useful
information on the rate constants. Here are some clues:
Look at the initial slope of the curves; measure the slopes. Are the initial
slopes of produced materials zero or not?
Measure the final concentration of all the reaction components.
@ Find when an intermediate reaches its maximum concentration and measure
this concentration.
In searching for a model or mechanism for the reaction scheme make runs
at different C,, and different CBoICA0.
If possible, also make runs starting with intermediate. For example, for the
reaction A + R --+ S start with R alone and follow its disappearance.
If the two steps of first-order reactions in series have very different values
for their rate constants, we can approximate the overall behavior as follows:

k -1 k
k,=100
A - R ~ S =+A-S, wherek = -
1 . 1
- 0.99
 198 Chapter 8 Potpourri of Multiple Reactions

For schemes involving different reaction orders, for reversible reactions, and
for a very-many-step scheme typical of polymerizations, analysis becomes
complicated.
The key to optimum design for multiple reactions is proper contacting and
proper flow pattern of fluids within the reactor. These requirements are
determined by the stoichiometry and observed kinetics. Usually qualitative
reasoning alone can already determine the correct contacting scheme. This
is discussed further in Chapter 10. However, to determine the actual equip-
ment size requires quantitative considerations.

i;d , ,
' l ~ ~ ~ & & t ~ ~ # t
,
, a , , \ , 2 ,, , , , , b, , , ,
EVALUATE THE KINETICS FROM A BATCH EXPERIMENT

Japanese researchers very carefully followed the oxidation of sodium sulfide

Na2S (A) to sodium thiosulfate Na2S203(R) in a batch reactor. Intermediates
were measured and the results found were sketched in Fig. E8.3.
(a) Think up a simple network of reactions, all of first-order, to represent this oxi-
dation.
(b) Evaluate the rate constants of this network.

Initial slope
of R curve =
2.0 mmollliter min

Initial slope min

of S curve =
1.3 mmollliter min
Initial Slope of T curve is
difficult to evaluate, but is z 0
Figure E8.3
 8.7 The Denbigh Reactions and its Special Cases 199

I This problem draws on a number of techniques.

(a) First, Look for Clues in the Graph. We first note that the initial slopes of
the R, S, and T curves are all nonzero, suggesting that these compounds are
formed directly from A, or

Next, since there is no S and T in the final product, our suggested reaction
scheme is

(b) Evaluate the Rate Constants. Noting that A disappears by first-order kinet-
ics, we have

CAO 185
In -= In - = ( k , + k ,
CA 100
+ k3)t= k1,,t = kIz3(30)
from which

k123= 0.0205 min-l

From the initial rate for formation of R

Similarly for S

So by difference

k3 = klZ3- kl - k2 = 0.0205 - 0.0108 - 0.0070 = 0.0027 min-I

200 Chapter 8 Potpourri of Multiple Reactions

Next, let us look at the maxima of the S and T curves. For S, by an extension
of Eq. 49, we can write

Solving for k4 by trial and error gives

Similarly for T

So we end up with the following kinetic scheme

kl = 0.0070 min-I

A
7s.e with
k2 = 0.0108 min-I
k, = 0.0027 min-I
u 9"T k4 = 0.0099 min-I
k, = 0.0163 min-I

This problem was prepared by David L. Cates (1988).

REFERENCES

Catipovic, N., and Levenspiel, O., Ind. Eng. Chem. Process Des. Dev., 18, 558 (1979).
Denbigh, K. G., Trans. Faraday Soc., 43, 648 (1947).
Denbigh, K. G., J. Appl. Chem., 1, 227 (1951).
Denbigh, K. G., Chem. Eng. Sci., 8, 125 (1958).
Jungers, J. C., et al., Cinktique Chimique Appliqute, Technip, Paris, 1958.
MacMullin, R. B., Chem. Eng. Prog., 44, 183 (1948).
 Problems 201
PROBLEMS

8.1. Starting with separate feeds of reactant A and B of given concentration

(no dilution with inerts permitted), for the competitive-consecutive reac-
tions with stoichiometry and rate as shown

sketch the best contacting patterns for both continuous and noncontinuous
operations:

(a) r1 = k1CAC$j (b) r1 = ~ I C A C B

r2 = k2CRCB Y2 = k2cRc$j

(c) ~1 = klCACB (d) r1 = klCiCB

r2 = k2C$CB r2 = k2CRCB

8.2. Under appropriate conditions A decomposes as follows:

kl=O.l/rnin
AR
-

R is to be produced from 1000 literlhr of feed in which CAo= 1 mollliter,

cRo= cso= 0.
(a) What size of plug flow reactor will maximize the concentration of R,
and what is that concentration in the effluent stream from this reactor?
(b) What size of mixed flow reactor will maximize the concentration of
R, and what is CR,,,, in the effluent stream from this reactor?

Pure A (CAo= 100) is fed to a mixed flow reactor, R and S are formed,
and the following outlet concentrations are recorded. Find a kinetic scheme
to fit this data.

8.3. Run cA CR cs

8.4. Run cA CR cs

8.5. Run cA CR cs t, min

1 50 40 10 5
2 20 40 40 20
202 Chapter 8 Potpourri of Multiple Reactions

8.6. In grinding a continuous flow of pigments for paint, our company finds
that too many too-small and too many too-large particles exit from our
well-mixed grinder. A multistage grinder, approximating plug flow, could
also have been used, but wasn't. Anyway, in either type of grinder the
pigments are progressively ground into smaller and smaller particles.
At present, the exit stream from our well-mixed grinder contains 10%
too-large (d, > 147 ,urn); 32% just-right (d, = 38 - 14'7 ,urn); and 58% too-
small (d, < 38 ,urn) particles.
(a) Can you suggest a better grinding scheme for our present unit, and
what would it give?
(b) How about the multistage grinder? How would it do?
By "better" we mean giving more just right-sized pigment in the product
stream. Also, separation and recycle of the size cuts is not practical.

8.7. Consider the following elementary reactions:

(a) One mole A and 3 moles B are rapidly mixed together. The reaction
is very slow, allowing analysis of compositions at various times. When
2.2 moles B remain unreacted, 0.2 mole S is present in the mixture.
What should be the composition of the mixture (A, B, R, and S) when
the amount of S present is 0.6 mole?
(b) One mole A is added bit by bit with constant stirring to 1 mole B.
Left overnight and then analyzed, 0.5 mole S is found. What can we
say about k21k,?
(c) One mole A and 1 mole B are thrown together and mixed in a flask.
The reaction is very rapid and goes to completion before any rate
measurements can be made. On analysis of the products of reaction
0.25 mole S is found to be present. What can we say about k21k,?

8.8. The liquid phase reaction of aniline with ethanol produces wanted mono-
ethylaniline and unwanted diethylaniline

(a) An equimolar feed is introduced into a batch reactor, and reaction is

allowed to proceed to completion. Find the concentration of reactants
and products at the end of the run.
(b) Find the ratio of mono- to diethylaniline produced in a mixed flow
 Problems 203
reactor for an alcohol-to-aniline feed ratio of 2 to 1for 70% conversion
of alcohol.
( c ) For an equimolar feed to a plug flow reactor, what will be the conversion
of the two reactants when the concentration of monoethylaniline is at
its highest?

8.9. Monoethylaniline can also be produced in the vapor phase in a fluidized

bed using natural bauxite as the solid catalyst. The elementary reactions
are shown in the previous problem. Using an equimolar feed of aniline
and ethanol, the fluidized bed produces 3 parts monoethylaniline to 2 parts
diethylaniline for a 40% conversion of aniline. Assuming mixed flow of gas
in the fluidized bed, find k21k, and the concentration ratio of reactants and
products at the exit of the reactor.

Under the action of mixed enzymes reactant A is converted to products

as follows:

A -
t enzyme

kl
t enzyme
R ----+ S,
k2

where the rate constants are dependent on the pH of the system.

(a) What reactor set-up (plug flow, mixed flow, or staged mixed flow units)
and what uniform pH level would you use?
(b) If it were possible to change the pH level along the plug flow reactor,
or from stage to stage in mixed flow units, in what direction would
you change the pH level?

8.10. k, = pH2 - 8 pH + 23 with 2 < pH < 6, and R the desired product

k2 = pH + 1

8.11. k, = pH + 1 with 2 < pH < 6, and S the desired product

k2=pH2-8pH+23

8.12. The progressive chlorination of o- and p-dichlorobenzene proceed with

second-order rate, as shown in Fig. P8.12.

Figure P8.12
204 Chapter 8 Potpourri of Multiple Reactions

For a single-feed stream having CAo= 2, C,, = 1,and 1,2,3-trichloroben-

zene as the desired product
(a) Tell which kind of flow reactor is best.
(b) In this reactor find C,,,,, .
See T. E. Corrigan, Chem. Eng., 201, March 1956, for discussion of this
process.
8.13. Consider the following first order decompositions with rate constants as
shown
40 0.1 0 02 10
(a) A-R-S
'92 (b)A-R-S
F0l

T U T U

If a colleague reports that Cs = 0.2 CAoin the exit stream of a plug flow
reactor,what can you say about the concentration of the other reactor
components, A, R, T, and U in the exit stream?
8.14. Chemicals A and B are thrown into a vat and react away according to the
following elementary reactions:

What can you say about these six rate constants if an analysis of the mixture
shows that

C, = 5 mollliter C, = 1mollliter
C, = 9 mollliter C, = 3 mollliter

at the time
(a) when reaction is incomplete?
(b) when reaction is complete?
8.15. With a particular catalyst and at a given temperature, the oxidation of
naphthalene to phthalic anhydride proceeds as follows:

D A = naphthalene kl = 0.21 s-l

<Y
A-
; 4
S--T
R = naphthaquinone k2 = 0.20 s-I
S = phthalic anhydride k3 = 4.2 s-I
T = oxidation products k4 = 0.004 s-I
 Problems 205

What reactor type gives the maximum yield of phthalic anhydride? Roughly
estimate this yield and the fractional conversion of naphthalene which will
give this yield. Note the word "roughly".

8.16. Sandy's Rock and Gravel Company wants to shift a mountain of gravel,
estimated at about 20 000 tons, from one side of their yard to the other.
For this they intend to use a power shovel to fill a hopper, which in turn
feeds a belt conveyor. The latter then transports the gravel to the new
location.
The shovel scoops up large amounts of gravel at first; however, as the
gravel supply decreases, the handling capacity of the shovel also decreases
because of the increased time required to move away from the hopper for
a load and then return and dump. Roughly, then, we may estimate that
the shovel's gravel handling rate is proportional to the size of the pile still
to be moved, its initial rate being 10 tonlmin. The conveyor, on the other
hand, transports the gravel at a uniform 5 tonlmin. At first the shovel will
work faster than the conveyor, then slower. Hence, the storage bin will
first accumulate material, then empty.
(a) What will be the largest amount of gravel in the bin?
(b) When will this occur?
(c) When will the rates of bin input and output be equal?
(d) When will the bin become empty?

8.17. A large fully automated municipal incinerator is being designed. A survey

estimates the garbage load to be 1440 tonslday. This will be harvested
by a fleet of compaction trucks which will disgorge their loads into an
underground storage bin. A conveyor will then feed the garbage to the
incinerator.
The proposed daily collection route is such that at the beginning of the
working day (6 A.M. sharp!) relatively large quantities of garbage (average
of 6 tonslmin) are returned from nearby commercial areas. Subsequently,
the supply will diminish as more remote suburban areas are serviced. It is
assumed that the collection rate is proportional to the amount of garbage
still to be collected, the initial rate being one truckloadlmin. The conveyor,
on the other hand, will transport garbage at a uniform 1 tonlmin to the
incinerator. At the beginning of the working day, the trucks will work
faster than the conveyor; later in the day, slower. Thus, each day the bin
will accumulate material, then lose material.
To evaluate this operation, we need information. Please help us with this.
(a) At what time of day will the trucks have collected 95% of the day's
garbage?
(b) How much garbage whould the storage bin be designed for?
(c) At what time of day will the bin be fullest?
(d) At what time of day will the bin be empty?

8.18. Upper Slobbovians and Lower Slobbovians are always at i t . . . , crushing

skulls, slitting throats, and so on. . . . At any gathering the rate at which
Upper Slobs are killed is directly proportional to the number of Lower
206 Chapter 8 Potpourri of Multiple Reactions

Slobs around, and vice versa. And at the end of any meeting of these
friendly people, either Upper Slobs leave or Lower Slobs, but never both.
Last week ten Upper Slobs happened upon three Lower Slobs, and when
it was over eight Upper Slobs lived to tell of their exciting victory.
(a) From this encounter, how would you rate Upper and Lower Slobs as
fighters? For example, would you say that they were equally good or
that one Upper Slob is as good as 2.3 Lower Slobs, or what?
(b) What would be the outcome of a friendly meeting of ten Upper Slobs
with ten Lower Slobs?

8.19. Chemical X, a powdered solid, is slowly and continuously fed for half an
hour into a well-stirred vat of water. The solid quickly dissolves and hydroly-
ses to Y, which then slowly decomposes to Z as follows

The volume of liquid in the vat stays close to 3 m3throughout this operation,
and if no reaction of Y to Z occurred, the concentration of Y in the vat
would be 100 mol/m3 at the end of the half-hour addition of X.
(a) What is the maximum concentration of Y in the vat and at what time
is this maximum reached?
(b) What is the concentration of product Z in the vat after 1 hour?
Problem prepared by Bhaskar Chandan (1990).

8.20. When oxygen is bubbled through a high-temperature batch of A-containing

liquid material, A oxidizes slowly to give a slowly decomposing intermediate
X and final product R. Here are the results of an experiment:

t, min CA,mol/m3 CR,mol/m3

100 0
95.8 1.4
35 26
12 41
4.0 52
1.5 60
negligible 80
0 100

We have no way of analyzing for X; however, we are safe in assuming that

at any time CA + CR + C, = [Link] can we say about the mechanism
and kinetics of this oxidation? Hint: plot the data and examine the plot.

8.21. Chemical A reacts to form R (k, = 6 hr-') and R reacts away to form S
(k, = 3 hr-l). In addition R slowly decomposes to form T (k, = 1 hr-I).
If a solution containing 1.0 mollliter of A is introduced into a batch reactor,
how long would it take to reach CR,,,,, and what would be CR,,,,?
Chapter 9
Temperature and Pressure
Effects

In our search for favorable conditions for reaction we have considered how
reactor type and size influence the extent of conversion and distribution of
products. The reaction temperature and pressure also influence the progress of
reactions, and it is the role of these variables that we now consider.
We follow a three-step procedure: First, we must find how equilibrium composi-
tion, rate of reaction, and product distribution are affected by changes in op-
erating temperatures and pressures. This will allow us to determine the optimum
temperature progression, and it is this that we strive to approximate with a real
design. Second, chemical reactions are usually accompanied by heat effects, and
we must know how these will change the temperature of the reacting mixture.
With this information we are able to propose a number of favorable reactor and
heat exchange systems-those which closely approach the optimum. Finally,
economic considerations will select one of these favorable systems as the best.
So, with the emphasis on finding the optimum conditions and then seeing how
best to approach them in actual design rather than determining what specific
reactors will do, let us start with discussions of single reactions and follow this
with the special considerations of multiple reactions.

9.1 SINGLE REACTIONS

With single reactions we are concerned with conversion level and reactor stability.
Questions of product distribution do not occur.
Thermodynamics gives two important pieces of information, the first being
the heat liberated or absorbed for a given extent of reaction, the second being
the maximum possible conversion. Let us briefly summarize these findings. A
justification of the expressions to follow can be found in any standard thermody-
namics texts for chemical engineers.

Heats of Reaction from Thermodynamics

The heat liberated or absorbed during reaction at temperature T, depends on
the nature of the reacting system, the amount of material reacting, and the
207
208 Chapter 9 Temperature and Pressure Effects

temperature and pressure of the reacting system, and is calculated from the heat
of reaction AH,, for the reaction in question. When this is not known, it can in
most cases be calculated from known and tabulated thermochemical data on
heats of formation AH,. or heats of combustion AH, of the reacting materials.
These are tabulated at some standard temperature, T I ,usually 25°C. As a brief
reminder, consider the reaction

By convention we define the heat of reaction at temperature T as the heat

transferred to the reacting system from the surroundings when a moles of A
disappear to produce r moles of R and s moles of S with the system measured
at the same temperature and pressure before and after the change. Thus

aA +rR + sS, {
positive, endothermic
AHrT negative, exothermic

Heats of Reaction and Temperature. The first problem is to evaluate the heat
of reaction at temperature T2 knowing the heat of reaction at temperature T,.
This is found by the law of conservation of energy as follows:

heat added to
heat absorbed
reactants to

In terms of enthalpies of reactants and products this becomes

where subscripts 1 and 2 refer to quantities measured at temperatures TI and

T,, respectively. In terms of specific heats

AHr2= AH,, + l2
TI
VC, dT

where

When the molar specific heats are functions of temperatures as follows,

 9.1 Single Reactions 209

we obtain

AHr2= AH,, +:
j (Va + VPT + VyT2)d T
(7)
VP VY
= AHrl + Va(T2 - TI) + -(Ti - Ti) + -(T: - T:)
2 3

where

Knowing the heat of reaction at any one temperature as well as the specific
heats of reactants and products in the temperature range concerned allows us
to calculate the heat of reaction at any other temperature. From this the heat
effects of the reaction can be found.

, AHr AT VARIOUS TEMPERATURES

From the AH, and AHf tables, I've calculated that the standard heat of my gas-
phase reaction at 25OC is as follows:

At 25°C the reaction is strongly exothermic. But this doesn't interest me because
I plan to run the reaction at 1025OC. What is the AHr at that temperature, and
is the reaction still exothermic at that temperature?

Data. Between 25°C and 1025°C the average Cp values for the various reaction
components are
-
Cp,=35J/mol.K G = 4 5 J l m o l . ~~ = 7 0 J l m o l . K

SOLUTION

First, prepare a reaction map as shown in Fig. E9.1. Then an enthalpy balance
for 1 mol A, 1 mol B, and 2 mol R gives

AH, =AH2 + AH, + AH,

1A - 1B + A H r , 2 5 ~ + (ncpAT)products
= (n~pAT)reactants
2R

= 1(35)(25 - 1025) + 1(45)(25 - 1025) + (-50 000) + 2(70)(1025 - 25)

210 Chapter 9 Temperature and Pressure Effects

;1:
Heat the
a reactants @ products
AH4
the

Figure E9.1

exothermic at 25OC
The reaction is
endothermic at 1025OC

Equilibrium Constants from Thermodynamics

From the second law of thermodynamics equilibrium constants, hence equilib-
rium compositions of reacting systems, may be calculated. We must remember,
however, that real systems do not necessarily achieve this conversion; therefore,
the conversions calculated from thermodynamics are only suggested attainable
values.
As a brief reminder, the standard free energy AGO for the reaction of Eq. 1
at temperature T is defined as

where f is the fugacity of the component at equilibrium conditions; f O is the

fugacity of the component at the arbitrarily selected standard state at temperature
T, the same one used in calculating AGO; Go is the standard free energy of a
reacting component, tabulated for many compounds; and K is the thermodynamic
equilibrium constant for the reaction. Standard states at given temperature are
commonly chosen as follows:
Gases-pure component at one atmosphere, at which pressure ideal gas behav-
ior is closely approximated
Solid-pure solid component at unit pressure
Liquid-pure liquid at its vapor pressure
Solute in liquid-1 molar solution; or at such dilute concentrations that the
activity is unity.
 9.1 Single Reactions 211
For convenience define

and
An=r+s-a

Simplified forms of Eq. 9 can be obtained for various systems. For gas reactions
standard states are usually chosen at a pressure of 1 atm. At this low pressure
the deviation from ideality invariably is small; hence fugacity and pressure are
identical and f" = p0 = 1 atm. Thus

The term in braces in this equation and in Eq. 13 is always unity but is retained
to keep the equations dimensionally correct.
For any component i of an ideal gas

Hence

Kf = Kp
and

For a solid component taking part in a reaction, fugacity variations with pres-
sure are small and can usually be ignored. Hence

=3
( ~ ) s o l i c component
i

Equilibrium Conversion. The equilibrium composition, as governed by the

equilibrium constant, changes with temperature, and from thermodynamics the
rate of change is given by

d(ln K ) --AH,
--
dT RT2

On integrating Eq. 15,we see how the equilibrium constant changes with tempera-
ture. When the heat of reaction AH, can be considered to be constant in the
temperature interval, integration yields
212 Chapter 9 Temperature and Pressure Effects

When the variation of AH, must be accounted for in the integration we have

where AH, is given by a special form of Eq. 4 in which subscript 0 refers to the
base temperature

Replacing Eq. 18 in Eq. 17 and integrating, while using the temperature depen-
dency for Cp given by Eq. 8, gives

These expressions allow us to find the variation of the equilibrium constant,

hence, equilibrium conversion, with temperature.
The following conclusions may be drawn from thermodynamics. These are
illustrated in part by Fig. 9.1.
1. The thermodynamic equilibrium constant is unaffected by the pressure of
the system, by the presence or absence of inerts, or by the kinetics of the
reaction, but is affected by the temperature of the system.
2. Though the thermodynamic equilibrium constant is unaffected by pressure
or inerts, the equilibrium concentration of materials and equilibrium conver-
sion of reactants can be influenced by these variables.

approximates
irreversible
reaction

only very low

conversions
possible

Figure 9.1 Effect of temperature on equilibrium conversion as pre-

dicted by thermodynamics (pressure fixed.)
 9.1 Single Reactions 213
3. K S 1 indicates that practically complete conversion may be possible and
that the reaction can be considered to be irreversible. K < 1 indicates that
reaction will not proceed to any appreciable extent.
4. For an increase in temperature, equilibrium conversion rises for endother-
mic reactions and drops for exothermic reactions.
5. For an increase in pressure in gas reactions, conversion rises when the
number of moles decreases with reaction; conversion drops when the num-
ber of moles increases with reaction.
6. A decrease in inerts for all reactions acts in the way that an increase in
pressure acts for gas reactions.

EQUILIBRIUM CONVERSION AT DIFFERENT

TEMPERATURES

(a) Between 0°C and 100°C determine the equilibrium conversion for the
elementary aqueous reaction

AG;,, = -14 130 Jlmol

AaR C,, = C,, = constant
{AH& = 7 . 5 300 Jim01

I Present the results in the form of a plot of temperature versus conversion.

(b) What restrictions should be placed on the reactor operating isothermally
if we are to obtain a conversion of 75% or higher?

I SOLUTION
(a) With all specific heats alike, VC, = 0. Then from Eq. 4 the heat of reaction
is independent of temperature and is given by

AH, = AH,,,, = -75 300 Jlmol 6)

From Eq. 9 the equilibrium constant at 25OC is given by

14 130 Jlmol
(ii)
.
= exp((8.314 Jlrnol K) (298K) = 300

Since the heat of reaction does not change with temperature, the equilibrium
constant K at any temperature T is now found from Eq. 16. Thus
214 Chapter 9 Temperature and Pressure Effects

Rearranging gives

Replacing K,,, and AH, from Eqs. (i) and (ii) gives on rearranging,

(iii)

But at equilibrium

Putting T values into Eq. (iii), then K into Eq. (iv), as shown in Table E9.2
gives the changing equilibrium conversion as a function of temperature in the
range of O"C to 100°C. This result is displayed in Fig. E9.2.
(b) From the graph we see that the temperature must stay below 78°C if
conversion of 75% or higher may be expected.

Table E9.2 Calculating XAe(T)from Eqs. (iii) and (iv)

Selected Temperature K = exp[= R T - 24.71

XA,
"C K from Eq. (iii) from Eq. (iv)
 9.1 Single Reactions 215

Temperature, "C
Figure E9.2

General Graphical Design Procedure

Temperature, composition, and reaction rate are uniquely related for any single
homogeneous reaction, and this may be represented graphically in one of three
ways, as shown in Fig. 9.2. The first of these, the composition-temperature plot,
is the most convenient so we will use it throughout to represent data, to calculate
reactor sizes, and to compare design alternatives.
For a given feed (fixed CAo,C ,, . . .) and using conversion of key component
as a measure of the composition and extent of reaction, the XA versus T plot
has the general shape shown in Fig. 9.3. This plot can be prepared either from
a thermodynamically consistent rate expression for the reaction (the rate must
be zero at equilibrium) or by interpolating from a given set of kinetic data in
conjunction with thermodynamic information on the equilibrium. Naturally, the
reliability of all the calculations and predictions that follow are directly dependent
on the accuracy of this chart. Hence, it is imperative to obtain good kinetic data
to construct this chart.

T T Composition, C
Figure 9.2 Different ways of representing the relationship of temperature, composition,
and rate for a single homogeneous reaction.
216 Chapter 9 Temperature and Pressure Effects

Reversible exothermic Reversible endothermic

Equilibrium,

Figure 9.3 General shape of the temperature-conversion plot for different reaction types.

The size of reactor required for a given duty and for a given temperature
progression is found as follows:
1. Draw the reaction path on the XA versus T plot. This is the operating line
for the operation.
2. Find the rates at various XA along this path.

Along path AB

XA arbitrary
temperature
0.05

T1 0.3 0.5 0.8 1

Recycle flow,

Figure 9.4 Finding the reactor size for different types of flow and for a feed temperature T I .
 9.1 Single Reactions 217
3. Plot the l/(-rA) versus XA curve for this path.
4. Find the area under this curve. This gives VIFAo.
For exothermic reactions we illustrate this procedure in Fig. 9.4 for three paths:
path AB for plug flow with an arbitrary temperature profile, path CD for noniso-
thermal plug flow with 50% recycle, and point E for mixed flow. Note that for
mixed flow the operating line reduces to a single point.
This procedure is quite general, applicable for any kinetics, any temperature
progression, and any reactor type or any series of reactors. So, once the operating
line is known. the reactor size can be found.

CONSTRUCTION OF THE RATE-CONVERSION-

TEMPERATURE CHART FROM KINETIC DATA

With the system of Example 9.2 and starting with an R-free solution, kinetic
experiments in a batch reactor give 58.1% conversion in 1 min at 65OC, 60%
conversion in 10 min at 25OC. Assuming reversible first-order kinetics, find the
rate expression for this reaction and prepare the conversion-temperature chart
with reaction rate as parameter.

Integrate the Performance Equation. For a reversible first-order reaction, the

performance equation for a batch reactor is

According to Eq. 3.54 this integrates to give

Calculate the Forward Rate Constant. From the batch run at 65OC, noting
from Example 9.2 that XAe= 0.89, we find with Eq. (i)

k , (1 min)
0.89

or

kl,338 = 0.942 min-I (ii)

Similarly, for the batch run at 25°C we find

kl,298 = 0.0909 min-I (iii)

218 Chapter 9 Temperature and Pressure Effects

Assuming an Arrhenius temperature dependency, the ratio of forward rate con-

stants at these two temperatures gives

from which the activation energy of the forward reaction can be evaluated, giving

Note that there are two activation energies for this reaction, one for the forward
reaction, another for the reverse.
Now for the complete rate constant for the forward reaction. From either
the numerator or the denominator of Eq. (iv) first evaluate k,,, then use it as
shown below:

[-:TI
k, = 34 X lo6exp - = exp [17.34 - -
RT

kl thus k,
Noting that K = -, =-
kl where K is given by Eq. (iii) of Example 9.2,
k2 K'
we can find the value of k,.

Summary. For the reversible first-order reaction of Example 9.2 we have

-- --

1
AGR; K=-.CR,, -Y,= r R = k l C A - k2cR
2 c,,

Equilibrium: K = exp [E
RT
- 24.71

Rate constants: k, = exp , min-I

, min-l
RT

From these values the X, versus T chart for any specific C,, can be prepared
and for this purpose the electronic computer is a great timesaver. Figure E9.3
is such a plot prepared for C,, = 1 mollliter and CRo= 0.
Since we are dealing with first-order reactions this plot can be used for any
C,, value by properly relabeling the rate curves. Thus, for C , = 10 mollliter
simply multiply all the rate values on this graph by a factor of 10.
 9.1 Single Reactions 219

Temperature, "C +
Figure E9.3

Optimum Temperature Progression

We define the optimum temperature progression to be that progression which
minimizes VIF,, for a given conversion of reactant. This optimum may be an
isothermal or it may be a changing temperature: in time for a batch reactor,
along the length of a plug flow reactor, or from stage to stage for a series of
mixed flow reactors. It is important to know this progression because it is the
ideal which we try to approach with a real system. It also allows us to estimate
how far any real system departs from this ideal.
The optimum temperature progression in any type of reactor is as follows: At
any composition, it will always be at the temperature where the rate is a maximum.
The locus of maximum rates is found by examining the r(T, C) curves of Fig.
9.4; Fig. 9.5 shows this progression.
For irreversible reactions, the rate always increases with temperature at any
composition, so the highest rate occurs at the highest allowable temperature. This
220

XA
r(f+ki(
Chapter 9 Temperature and Pressure Effects

E l
lrreversi ble

T
I
Best path,
isothermal

Tmax, allowable
Reversible
endothermic

T
1
Best path,

isothermal

Tmax, allowable
Reversible
exothermic

T
rates, best

I \

Trnax, allowable

Figure 9.5 Operating lines for minimum reactor size.

temperature is set by the materials of construction or by the possible increasing

importance of side reactions.
For endothermic reactions a rise in temperature increases both the equilibrium
conversion and the rate of reaction. Thus, as with irreversible reactions, the
highest allowable temperature should be used.
For exothermic reversible reactions the situation is different, for here two
opposing factors are at work when the temperature is raised-the rate of forward
reaction speeds up but the maximum attainable conversion decreases. Thus, in
general, a reversible exothermic reaction starts at a high temperature which
decreases as conversion rises. Figure 9.5 shows this progression, and its precise
values are found by connecting the maxima of the different rate curves. We call
this line the locus of maximum rates.

Heat Effects
When the heat absorbed or released by reaction can markedly change the temper-
ature of the reacting fluid, this factor must be accounted for in design. Thus we
need to use both the material and energy balance expressions, Eqs. 4.1 and 4.2,
rather than the material balance alone, which was the starting point of all the
analyses of isothermal operations of Chapters 5 through 8.
First of all, if the reaction is exothermic and if heat transfer is unable to remove
all of the liberated heat, then the temperature of the reacting fluid will rise as
conversion rises. By similar arguments, for endothermic reactions the fluid cools
as conversion rises. Let us relate this temperature change with extent of con-
version.
We start with adiabatic operations, later extending the treatment to account
for heat interchange with the surroundings.

Adiabatic Operations
Consider either a mixed flow reactor, a plug flow reactor, or a section of plug
flow reactor, in which the conversion is X,, as shown in Fig. 9.6. In Chapters 5
and 6 one component, usually the limiting reactant, was selected as the basis for
 9.1 Single Reactions 221

Heat released or absorbed Plug flow or section of

plug flow reactor
----
\-,,x
I
----J

' Insulated, 1
adiabatic operations

Figure 9.6 Adiabatic operations with large enough heat effect to cause a
rise in temperature (exothermic) or drop in temperature (endothermic) in
the reacting fluid.

all material balance calculations. The same procedure is used here, with limiting
reactant A taken as the basis. Let
Subscripts 1, 2 refer to temperatures of entering and leaving streams.

C;, Ci = Mean specific heat of unreacted feed stream and of completely

converted product stream per mole of entering reactant A.
H', H" = enthalpy of unreacted feed stream and of completely converted
product stream per mole of entering reactant A.
A H = heat of reaction per mole of entering reacting A, and at tempera-
ture Ti.

With TI as the reference temperature on which enthalpies and heats of reaction

are based we have
Enthalpy of entering feed:

Enthalpy of leaving stream:

Energy absorbed by reaction:

Replacing these quantities in the energy balance,

input = output + accumulation + disappearance by reaction (4.2)

we obtain at steady state

222 Chapter 9 Temperature and Pressure Effects

By rearranging,

or, with Eq. 18

/heat needed to raise\

-:=*x CiA T
AHr, - 1\feed stream to T,
heat released by
reaction at T,
I
which for complete conversion becomes

-AHr2 = CLAT, for X, = 1 (23)

The latter form of the equation simply states that the heat released by reaction
just balances the heat necessary to raise the reactants from TI to T2.
The relation between temperature and conversion, as given by the energy
balances of Eq. 21 or 22, is shown in Fig. 9.7. The resulting lines are straight for
all practical purposes since the variation of the denominator term of these equa-

Complete conversion
f of limiting reactant

T
Figure 9.7 Graphical representation of energy balance equa-
tion for adiabatic operation. These are adiabatic operating
lines.
 9.1 Single Reactions 223
tions is relatively small. When Cg - C; = 0, the heat of reaction is independent
of temperature and Eqs. 21 and 22 reduce to

which are straight lines in Fig. 9.7.

This figure illustrates the shape of the energy balance curve for both endother-
mic and exothermic reactions for both mixed flow and plug flow reactors. This
representation shows that whatever is the conversion at any point in the reactor,
the temperature is at its corresponding value on the curve. For plug flow the
fluid in the reactor moves progressively along the curve, for mixed flow the fluid
immediately jumps to its final value on the curve. These are the adiabatic operating
lines for the reactor. With increased inerts Cp rises, and these curves become
more closely vertical. A vertical line indicates that temperature is unchanged as
reaction proceeds. This then is the special case of isothermal reactions treated
in Chapters 5 through 7.
The size of reactor needed for a given duty is found as follows. For plug flow
tabulate the rate for various XA along this adiabatic operating line, prepare the
ll(-r,) versus XAplot and integrate. For mixed flow simply use the rate at the
conditions within the reactor. Figure 9.8 illustrates this procedure.
The best adiabatic operations of a single plug flow reactor are found by shifting
the operating line (changing the inlet temperature) to where the rates have the
highest mean value. For endothermic operations this means starting at the highest
allowable temperature. For exothermic reactions this means straddling the locus
of maximum rates as shown in Fig. 9.9. A few trials will locate the best inlet
temperature, that which minimizes VIF,,. For mixed flow the reactor should
operate on the locus of maximum rates, again shown in Fig. 9.9.
The best reactor type, that which minimizes VIF,,, is found directly from this
X, versus T graph. If the rate progressively decreases with conversion, then use
plug flow. This is the case for endothermic reactions (Fig. 9 . 8 ~ )and close to
isothermal exothermic reactions. For exothermic reactions that have a large
temperature rise during reaction, the rate rises from a very low value to a
maximum at some intermediate XA,then falls. This behavior is characteristic of
autocatalytic reactions, thus recycle operations are best. Figure 9.10 illustrates
two situations, one where plug flow is best, the other where large recycle or
mixed flow is best. The slope of the operating line, CpI-AH,, will determine
which case one has at hand. Thus
1. for small CpI-AH, (pure gaseous reactants) mixed flow is best.
2. for large Cp/-AH, (gas with much inerts, or liquid systems) plug flow is best.

Nonadiabatic Operations
For the adiabatic operating line of Fig. 9.7 to more closely approach the ideals
of Fig. 9.5, we may want deliberately to introduce or remove heat from the
reactor. In addition, there are heat losses to the surroundings to account for.
224 Chapter 9 Temperature and Pressure Effects

Along path

Adiabatic
line

[2
0

point Ei c l

XA

Figure 9.8 Finding reactor size for adiabatic operations of plug flow and mixed
flow reactors.

,-Locus of maximum rates

7

/ / /
/ I /
/ I /
/ I /
/ / /
Ti Ti' Ti"
%-'
b o low 1 w
Too high

Correct inlet Correct inlet

temperature temperature
Figure 9.9 Best location for the adiabatic operating line. For plug flow, a trial
and error search is needed to find this line; for mixed flow, no search is needed.
 9.1 Single Reactions 225

Operating line:
Pure or rich gas slope =

With small slope

XA

. Liquid or
dilute gas

+
Very low With large slope the rate
rates here continually decreases,
T use plug flow

Figure 9.10 For exothermic reactions mixed flow is best where the temperature rise
is large; plug flow is best for close to isothermal systems.

Let us see how these forms of heat interchange modify the shape of the adiabatic
operating line.
Let Q be the total heat added to a reactor per mole of entering reactant A,
and let this heat also include the losses to the surroundings. Then Eq. 20, the
energy balance about the system, is modified to

which on rearrangement and with Eq. 18 gives

=I i
net heat still needed after heat
CLAT - Q transfer to raise feed to T,
XA=
-AH,, heat released by reaction at T,

and for C; = C;, which often is a reasonable approximation

With heat input proportional to AT = T, - TI the energy balance line rotates

about TI. This change is shown in Fig. 9.11. Other modes of heat addition or
removal yield corresponding shifts in the energy balance line.
Using this modified operating line, the procedure for finding the reactor size
and optimum operations follows directly from the discussion on adiabatic opera-
tions.

Comments and Extensions

Adiabatic operations of an exothermic reaction give a rising temperature with
conversion. However, the desired progression is one of falling temperature. Thus,
226 Chapter 9 Temperature and Pressure Effects

Energy balance line for

adiabatic operations
Heat removal

Heat addition

TI = temperature of feed
T

Figure 9.11 Sketch of the energy balance equation show-

ing the shift in adiabatic line caused by heat exchange
with surroundings.

very drastic heat removal may be needed to make the operating line approach
the ideal, and many schemes may be proposed to do this. As an example, we
may have heat exchange with the incoming fluid (see Fig. 9.12a), a case treated
by van Heerden (1953,1958).Another alternative is to have multistage operations
with interstage cooling between adiabatic sections (see Fig. 9.12b). In general
multistaging is used when it is impractical to effect the necessary heat exchange
within the reactor itself. This is usually the case with gas-phase reactions with
their relatively poor heat transfer characteristics. For endothermic reactions,
multistaging with reheat between stages is commonly used to keep the tempera-
ture from dropping too low (see Fig. 9.12~).
Since the main use of these and many other forms of multistage operations is
with solid catalyzed gas-phase reactions we discuss these operations in Chapter
19. Design for homogeneous reactions parallels that for catalytic reactions, so
the reader is referred to Chapter 19 for the development.

Exothermic Reactions in Mixed Flow Reactors-A Special Problem

For exothermic reactions in mixed flow (or close to mixed flow) an interesting
situation may develop in that more than one reactor composition may satisfy
the governing material and energy balance equations. This means that we may
not know which conversion level to expect. van Heerden (1953, 1958) was the
first to treat this problem. Let us examine it.
First, consider reactant fluid fed at a given rate (fixed r or VIF,,) to a mixed
flow reactor. At each reactor temperature there will be some particular conver-
sion which satisfies the material balance equation, Eq. 5.11. At low temperature
the rate is low so this conversion is low. At higher temperature the conversion
 227

Tb
9.1 Single Reactions

,/Desiredtemperature
falling

Reaction and
cooling
B \ feed of
\

Adiabatic

Hot
feed

:b
out out

Desired constant
Adiabatic /m perature

,; allowable

Distance
(c)

Figure 9.12 Ways of approaching the ideal temperature profile with heat exchange: (a)
and (b) exothermic reaction; (c) endothermic reaction.

rises and approaches the equilibrium. At a still higher temperature we enter the
region of falling equilibrium so the conversion for given Twill likewise fall. Figure
9.13 illustrates this behavior for different T values. Note that these lines do not
represent an operating line or a reaction path. Actually any point on these
curves represents a particular solution of the material balance equations; thus,
it represents an operating point for the mixed flow reactor.
Now, for a given feed temperature TI the intersection of the energy balance
line with the S-shaped material balance line for the operating T gives the condi-
tions within the reactor. Here three cases may be distinguished. These are shown
in Fig. 9.14 for irreversible reactions.
First, the energy balance line T,A represents the situation where insufficient
heat is liberated by reaction to raise the temperature to a high enough level for
the reaction to be self-sustaining. Hence, conversion is negligible. At the other
extreme, if we have more than enough heat liberated, the fluid will be hot and
conversion essentially complete. This is shown as line TIB. Finally, line T I C
indicates an intermediate situation which has three solutions to the material and
228 Chapter 9 Temperature and Pressure Effects

Locus of operating
t points for this T 7

Irreversible in Reversible in
this T range this T range
Figure 9.13 Conversion in a mixed flow reactor as a function of T and r, from
the material balance equation, Eq. 5.11.

energy balance equations, points M ' , M", and M"'. However, point M" is an
unstable state because with a small rise in temperature the heat produced (with
the rapidly rising material balance curve) is greater than the heat consumed by
the reacting mixture (energy balance curve). The excess heat produced will make
the temperature rise until Mu' is reached. By similar reasoning, if the temperature
drops slightly below MI' it will continue to drop until M' is reached. Thus, we
look upon M" as the ignition point. If the mixture can be raised above this
temperature, then the reaction will be self-sustaining.
For reversible exothermic reactions the same three cases occur, as shown in
Fig. 9.15. However, it can be seen that here there is an optimum operating
temperature for the given T value where conversion is maximized. Above or
below this temperature the conversion drops; thus, proper control of heat removal
is essential.
The type of behavior described here occurs in systems where the slope of the
energy balance line, C,I-AH,, is small; thus, large liberation of heat and pure
reactants which leads to far from isothermal operations. van Heerden (1953,
1958) discusses and gives examples of this type of reacting system. In addition,
though it is a much more complex situation, a gas flame illustrates well the

Practically
complete
. conversion7 A

TI Ignition

Figure 9.14 Three types of solutions to the energy and material balances for
exothermic irreversible reactions.
 9.1 Single Reactions 229

Energy balance
1, Optimum point of
intersection of energy and
material balance curves

Material balance

- -

T1 = feed temperature

Figure 9.15 Solution of energy and material balances for reversible

exothermic reaction.

multiple solutions discussed here: the unreacted state, the reacted state, and the
ignition point.
Reactor dynamics, stability, and start-up procedures are particularly important
for auto-induced reactions such as these. For example, a small change in feed
rate (7value), feed composition or temperature, or heat transfer rate may cause
the reactor output to jump from one operating point to the other.

PERFORMANCE FOR THE OPTIMAL

TEMPERATURE PROGRESSION

Using the optimal temperature progression in a plug flow reactor for the reaction
of Examples 9.2 and 9.3,

(a) calculate the space time and volume needed for 80% conversion of a feed
of FA, = 1000 mollmin where CAo= 4 mollliter.
(b) plot the temperature and conversion profile along the length of the reactor.

Let the maximum allowable operating temperature be 95°C.

Note that Fig. E9.3 was prepared for CAo = 1 mollliter, not 4 mollliter.

(a) Minimum Space-Time. On the conversion-temperature graph (Fig. E9.3)

draw the locus of maximum rates. Then, remembering the temperature restric-
tion, draw the optimum path for this system (line ABCDE in Fig E 9 . 4 ~ )and
230 Chapter 9 Temperature and Pressure Effects

Path ABCDE from

Fig.E4a or E3

1
-
XA
0.4 - (-rA) -
10% of area

0.2 -

0
0
,
20
,40
25T 1
60
Temperature, "C
(a)
1
80
A

T,,,
/lo0
=95OC

Figure E9.4a
Figure E9.4b

integrate graphically along this path to obtain

= 0.405 literlmol . min

Therefore

and

V = FA,(area) = (1000 mollmin) (0.405 liter

-
-
405 liters

(b) T and X, Profiles Through the Reactor. Let us take 10% increments
through the reactor by taking 10% increments in area under the curve of Fig.
E9.4b. This procedure gives X, = 0.34 at the 10% point, X, = 0.485 at the 20%
point, etc. The corresponding temperatures are then 362 K at XA = 0.34 (point
C), 354 K at X , = 0.485 (point D), etc.
In addition we note that the temperature starts at 95"C, and at X, = 0.27
(point B) it drops. Measuring areas in Fig. E9.4b we see that this happens after
the fluid has passed 7% of the distance through the reactor.
In this manner the temperature and conversion profiles are found. The result
is shown in Fig. E9.4~.
 9.1 Single Reactions 231

Figure E9.4~
Distance through reactor -
OPTIMUM MIXED FLOW REACTOR PERFORMANCE

A concentrated aqueous A-solution of the previous examples (C, = 4 mollliter,

FA, = 1000 mol/min) is to be 80% converted in a mixed flow reactor.
(a) What size of reactor is needed?
(b) What is the heat duty if feed enters at 25°C and product is to be withdrawn
at this temperature?
Note that
1000 cal 1kg 1liter - cal
C p ~ = ~ G i & 2. 5-0 -m o l ~ . ~

(a) Reactor Volume. For CAo= 4 mollliter we may use the X A versus T chart
of Fig. E9.3 as long as we multiply all rate values on this chart by 4.
Following Fig. 9.9 the mixed flow operating point should be located where
the locus of optima intersects the 80% conversion line (point C on Fig. E 9 . 5 ~ ) .
Here the reaction rate has the value

I -rA = 0.4 mol A convertedlmin .liter

From the performance equation for mixed flow reactors, Eq. 5.11, the volume
required is given by

v=- FAOX, -(
- 1 0 0 0- 2000 liters molimin)

= ( r ) 0.4 mollmin liter -

 232 Chapter 9 Temperature and Pressure Effects

/ ' ~ n e r balance,
~~
/' slope = 1/72

0.2

Feed, 25°C
0
40 60 80 100
4:4 Temperature, "C

Figure E9.5a

(b) Heat Duty. Of course we can use joules in our calculations; however, since
we are dealing with aqueous solutions it is simpler to use calories. Let us use
calories. Then the slope of the energy balance line is

C, - (250 callmol A . K) = -
1 K-1
slope = --
A H 18 000 callmol A) 72

Drawing this line through point C (line BCD) we see that the feed must be
cooled 20" (from point A to point B) before it enters and reacts adiabatically.
Also, the product must be cooled 37°C (from point C to point E). Thus the heat
duty is

Precooler: QAB = (250 callmol A . K)(20 K) = 5000 callmol A fed

= (5000 callmol A)(1000 molA1min) = 5000 000 callmin
= 348.7 kW
Postcooler: Q,, = (250)(37) = 9250 callmol A fed

I Figure E9.5b shows two reasonable arrangements for the coolers.

 9.1 Single Reactions 233

Qout = 3 4 9 KW Qout = 645 KW Qout = 3 4 9 KW

Figure E9.5b

ADIABATIC PLUG FLOW REACTOR PERFORMANCE

e size of adiabatic plug flow reactor to react the feed of Example 9.5 (FA,
= 1000 mollmin and C,, = 4 mollliter) to 80% conversion.

I SOLUTION

Following the procedure of Fig. 9.9 draw trial operating lines (see Fig. E 9 . 6 ~ )
with a slope of (from Example 9.5), and for each evaluate the integral

to find which is smallest. Figures E 9 . 6 ~and b show this procedure for lines AB
and CD. Line CD has the smaller area, is in fact closer to the minimum, and is

Note: These rates are 4 times

those shown in Fig. E3

6 For operating line AB

For operating line CD
5 Area under curve CD

4
-
1
(-rA) 3

0
0 0.2 0.4 0.6 0.8
l4 Temperature, "C XA

Figure E9.6
 234 Chapter 9 Temperature and Pressure Effects

I therefore the desired adiabatic operating line. So

V = FA,
-
-
1, 7
0.8 dXA
= FAo (area under curve CD)

I = (1000 mol/min)(1.72 liter. minlmol)

I = 1720 liters

This volume is somewhat smaller than the volume of mixed flow reactor (from
Example 9.5) but it is still four times as large as the minimum possible (405
liters, from Example 9.4).
Regarding temperatures: Figure E6a shows that the feed must first be cooled
to 16.0°C, it then passes through the adiabatic reactor and leaves at 73.6OC and
80% conversion.

ADIABATIC PLUG FLOW REACTOR WITH RECYCLE

I Repeat Example 9.6 but now allow recycle to product stream.

For operating line CD of Example 9.6 we find an optimum recycle area, shown
in Fig. E9.7 as rectangle EFGH.
Area = (0.8 - 0)(15 liter minlmol) = 1.2 liter. minlmol
V = FA, (area) = (1000 mol/min)(l.2 literlmol . min)
-
-
= 1200 liter

5
For operating line CD

4
1
-
A-' 3 Area = 1.5 (0.8-0)

2
1.5
1

0
0 0.2 0.4 0.6 0.8
XA
Figure E9.7
 9.2 Multiple Reactions 235

To summarize the results of these four examples, all of which are performing
the same duty with the same feed

for the plug flow with optimal T progression V = 405 liter (Example 4)
for the mixed flow reactor V = 2000 liter (Example 5 )
for the adiabatic plug flow reactor V = 1720 liter (Example 6)
for the adiabatic recycle reactor V = 1200 liter (Example 7)

9.2 MULTIPLE REACTIONS

As pointed out in the introduction to Chapter 7, in multiple reactions both
reactor size and product distribution are influenced by the processing conditions.
Since the problems of reactor size are no different in principle than those for
single reactions and are usually less important than the problems connected with
obtaining the desired product material, let us concentrate on the latter problem.
Thus, we examine how to manipulate the temperature so as to obtain, first, a
desirable product distribution, and second, the maximum production of desired
product in a reactor with given space-time.
In our development we ignore the effect of concentration level by assuming
that the competing reactions are all of the same order. This effect is studied in
Chapter 7.

Product Distribution and Temperature

If two competing steps in multiple reactions have rate constants k , and k,, then
the relative rates of these steps are given by

kl - k,,,e-E1lRT - k l ( )
k, -
k e-EzlRT--e
20 k20
'
(E E R T
- E - E,)IRi

This ratio changes with temperature depending on whether E, is greater or

smaller than E,, so

k l / k 2increases if E, > E2
when T rises
k,lk, decreases if E, < E,

Thus the reaction with larger activation energy is the more temperature-sensitive
of the two reactions. This finding leads to the following general rule on the
influence of temperature on the relative rates of competing reactions:

A high temperature favors the reaction of higher activation energy,

a low temperature favors the reaction of lower activation energy.
236 Chapter 9 Temperature and Pressure Effects

Let us apply this rule to find the proper temperature of operations for various
types of multiple reactions.
For parallel reactions

Step 1 is to be promoted, step 2 depressed, so kllk, is to be made as large as

possible. Thus, from the above rule

For reactions in series

if El > E, use high T
if El < E, use low T
I
2
A Rdeiired

the production of R is favored if k,lk, is increased. Thus

if El > E, use high TI

if El < E, use low T
1
For the general series-parallel reaction we introduce two additional considera-
tions. First of all for parallel steps if one requirement is for a high temperature
and another is for a low temperature, then a particular intermediate temperature
is best in that it gives the most favorable product distribution. As an example,
consider the reactions

'
A-S 2 where El > E2, El < E3 (31)
\T
Now, El > E, requires a high T, El < E3 requires a low T, and it can be shown
that the most favorable product distribution is obtained when the temperature
satisfies the following condition
 Problems 237

Second, for steps in series if an early step needs a high temperature and a later
step needs a low temperature, then a falling progression of temperatures should
be used. Analogous arguments hold for other progressions.
The problems at the end of this chapter verify some of the qualitative findings
on To,, and also show some possible extension.

Comments
This discussion of multiple reactions shows that the relative size of the activation
energies will tell which temperature level or progression is favored, just as Chap-
ter 7 showed what concentration level or progression and what state of mixing
is best. Although the general pattern of low, high, falling, or rising temperature
can usually be determined without much difficulty, calculation of the optimum
is not easy.
In experimentation we usually meet with the inverse of the situation outlined
here in that we observe product distributions from experiment, and from this
we wish to find the stoichiometry, kinetics, and the most favorable operating
conditions. The generalizations of this chapter should be helpful in this induc-
tive search.
Finally, when the reactions are of different order and of different activation
energies, we must combine the methods of Chapters. 7, 8, and 9. Jackson et al.
(1971) treat a particular system of this type and find that the optimum policy
requires adjusting only one of the two factors, temperature or concentration,
while keeping the other at its extreme. Which factor to adjust depends on whether
the change in product distribution is more temperature dependent or concentra-
tion dependent. It would be interesting to know whether this finding represents
a general conclusion.

REFERENCES

Jackson, R., Obando, R., and Senior, M. G., Chem. Eng. Sci.,26,853 (1971).
van Heerden, C., Ind. Eng. Chem., 45,1242 (1953).
van Heerden, C., Chem. Eng. Sci.,8,133 (1958).

PROBLEMS

Examples 9.4 through 9.7 illustrate the approach to problems dealing with noniso-
thermal reactors. Chapter 19 extends this approach to multistage operations of
solid catalyzed reactions.
To reinforce these concepts, Problems 9.1 through 9.9 ask the reader to redo
these examples with one or more changes. In many of these problems there is
no need to redo the whole problem, just indicate where changes are needed in
text and graphs.
238 Chapter 9 Temperature and Pressure Effects

9.1. For the reaction system of Example 9.4

(a) find needed for 60% conversion of reactant using the optimal tempera-
ture progression in the plug flow reactor.
(b) also find the exit temperature of fluid from the reactor.
Use whatever information you need from Example 9.4.

9.2. For the mixed flow reactor system of Example 9.5, we wish to get 70%
conversion in the smallest size of reactor. Sketch your recommended system
and on it indicate the temperature of the flowing stream entering and
leaving the reactor as well as T , the space time needed.

9.3. For the optimum temperature progression in a plug flow reactor in Example
9.4 (C,, = 4 mollliter, FA, = 1000 mol Almin, XA = 0.8, T,,, = 5"C, T,,,
= 95°C) and feed and product both at 25"C, how much heating and cooling
would be needed
(a) for the feed stream?
(b) in the reactor itself?
( c ) for the stream leaving the reactor?

9.4. We plan to run the reaction of Example 9.4 (C,, = 4 mollliter, FA, = 1000
mollmin) in a plug flow reactor kept at 40°C throughout to 90% conversion.
Find the volume of reactor needed.

9.5. Redo Example 9.4

9.6. Redo Example 9.5

I
where CAo= 4 mollliter is replaced by CAo= 1
> mollliter, and
9.7. Redo Example 9.6
where FA, remains unchanged at 1000 mol Almin.
9.8. Redo Example 9.7
1
9.9. We wish to run the reaction of Example 9.4 in a mixed flow reactor to
95% conversion for a feed concentration C,, = 10 mollliter and feed rate
of u = 100 literlmin. What size of reactor would we need?

9.10. Qualitatively find the optimum temperature progression to maximize C,

for the reaction scheme
 Problems 239
9.11. The first-order reactions

A-LRLS .
desired
k2 -
-
107e-4000/T
I T' Y U k - 108e-9000/T
3-
k, = 1012e-12,000/T

are to be run in two mixed flow reactors in series anywhere between 10

and 90°C. If the reactors may be kept at different temperatures, what
should these temperatures be for maximum fractional yield of S? Find this
fractional yield.

9.12. The reversible first-order gas reaction

is to be carried out in a mixed flow reactor. For operations at 300 K the

volume of reactor required is 100 liters for 60% conversion of A. What
should be the volume of the reactor for the same feed rate and conversion
but with operations at 400 K?
Data: kl = lo3 exp [-2416/T]
ACp = CPR- CpA= 0
AH, = -8000 callmol at 300 K
K = 10 at 300 K
Feed consists of pure A
Total pressure stays constant
Chapter 10
Choosing the Right Kind
of Reactor

So far we have concentrated on homogeneous reactions in ideal reactors. The

reason is two-fold; because this is the simplest of systems to analyze and is the
easiest to understand and master; also because the rules for good reactor behavior
for homogeneous systems can often be applied directly to heterogeneous systems.
The important lessons learned in the first nine chapters of this book should
guide us right away, or with a very minimum of calculations, to the optimum
reactor system. Previously we have come up with six general rules. Let us present
them and then practice using them.

Rule 1. For Single Reactions

To minimize the reactor volume, keep the concentration as high as possible
for a reactant whose order is n > 0. For components where n < 0 keep the
concentration low.

Rule 2. For Reactions in Series

Consider reactions in series, as shown:

To maximize any intermediate, do not mix fluids that have different concentra-
tions of the active ingredients-reactant or intermediates. See Fig. 10.1.

Rule 3. For Parallel Reactions

Consider the parallel reactions with reaction orders ni

xRdesir
A-S ed
nl low order
n2 ... intermediate
't T ng ... high order
 Chapter 10 Choosing the Right Kind of Reactor 241

Recycle
/This fresh feed
I mixes with partly,
reacted f l u ~ d

0
0
(a) (h)
Figure 10.1 (a) Plug flow (no intermixing) gives the most of all the intermedi-
ates. (b) Intermixing depresses the formation of all intermediates.

To get the best product distribution,

low C, favors the reaction of lowest order

high C, favors the reaction of highest order
If the desired reaction is of intermediate order then some intermediate C,
will give the best product distribution.
For reactions all of the same order the product distribution is not affected
by the concentration level.

Rule 4. Complex Reactions

These networks can be analyzed by breaking them down into their simple series
and simple parallel components. For example, for the following elementary
reactions, where R is the desired product, the breakdown is as follows:

This breakdown means that A and R should be in plug flow, without any recycle,
while B can be introduced as you wish, at any concentration level, since it will
not affect the product distribution.

Rule 5. Continuous versus Noncontinuous Operations

Any product distribution that can be obtained in continuous steady-state flow
operations can be gotten in a non-flow operation and vice versa. Figure 10.2
illustrates this.
242 Chapter 10 Choosing the Right Kind of Reactor

Flow reactor Nonflow reactor

Add all f l u ~ d
Equ~valent at one t ~ m e

, Fill slowly
Keep cornposlt~on

Figure 10.2 Correspondence between the residence time distribution of steady flow
and either non-flow, batch or semibatch systems.

Rule 6. Effect of Temperature on Product Distribution

Given

a high temperature favors the reaction with larger E, while a low temperature
favors the reaction with smaller E.
Let us now see how these six rules can be used to guide us to the optimum.

Optimum Operation of Reactors

In reactor operations the word "optimum" can have different meanings. Let us
look at two definitions which are particularly useful.
Feed a stream containing reactant A to a reactor and let R, S, T, . . . be formed,
with R being the desired product. Then by optimum

1. we could mean maximizing the overall fractional yield of R, or

@(;) (moles
=moles R formed
of A consumed
 Chapter 10 Choosing the Right Kind of Reactor 243

2. we could mean running the reactor system so that the production of R is

maximized, or

(Prod R),,,,, = ( moles of R formed

moles of A fed to the system

For reactions in series we calculate the maximum production rate of R directly,

as shown in Chapter 8. However, for reactions in parallel we find it useful to
first evaluate the instantaneous fractional yield of R, or

moles R formed
moles A consumed

and then proceed to find the optimum. This procedure is shown in Chapter 7.
If unused reactant can be separated from the exit stream, reconcentrated to
feed conditions and then recycled, then

(Prod R),,, = @(RIA),,, (4)

THE TRAMBOUZE REACTIONS (1958).

The elementary reactions

are to be run in four equal-size MFR's (mixed flow reactors), connected any
way you wish. The feed is C,, = 1, the feed flow rate is u = 100 literslmin.
The best scheme that the computer could come up with to maximize the
fractional yield of S, or @(S/A) [see problem 5, Chem. Eng. Sci., 45, 595-614
(1990)], is shown in Fig. [Link].
(a) How would you arrange a four-MFR system?
(b) With your best system what should be the volume of your four reactors?

I Figure [Link]
244 Chapter 10 Choosing the Right Kind of Reactor

(a) First of all, the computer solution looks somewhat complicated from the
engineering point of view. But never mind, let us proceed with our calcula-
tions. The instantaneous fractional yield, q(S/A), is

To maximize (p(SIA) put

Solving gives

--
C~opt-
- 0.25
-

So from Eq. (i), at C,,,,,

Thus the best way of running these four reactors is to keep the conditions
at the optimum in all four units. One such design is shown in Fig. [Link].
Problem P20 shows another design, and so does Fig. [Link].

- C~ opt

CS opt

V = 187.5 liters
Figure [Link]
 Chapter 10 Choosing the Right Kind of Reactor 245
(b) The volume per MFR comes from the performance equation

= 187.5 liters

Therefore, for the four reactor system

I&,,
-
= 187.5 X 4 = 750 liters

TEMPERATURE PROGRESSION FOR MULTIPLE

REACTIONS
I
I Consider the following scheme of elementary reactions:

What temperature progression would you recommend if the desired product is:

and if reactor size is not important?

This industrially important reaction scheme is reported by Binns et al. (1969)
and is used by Husain and Gangiah (1976, p. 245). In this problem we inter-
changed two of the reported E values to make the problem more interesting.

( a ) Intermediate R is Desired We want step 1 fast compared to step 2, and

we want step 1 fast compared to step 3.

I Since El < E2 and El < E, use a low temperature and plug flow.
 246 Chapter 10 Choosing the Right Kind of Reactor

I ( b ) Final Product S is Desired Here speed is all that matters.

So use a high temperature and plug flow.

(c) Intermediate T is Desired We want step 2 fast compared to step 1, and

we want step 2 fast compared to step 4.

Since E, > E, and E, < E4 use a falling temperature and plug flow

(d) Intermediate U is Desired We want step 1 fast compared to step 2, and

step 3 fast compared to step 5.
Since E, < E, and E, > E5 use a rising temperature and plug flow

REFERENCES

Binns, D.T., Kantyka, T.A., and Welland, R.C., Trans. I. Chem. E., 47, T53 (1969).
Husain, A., and Gangiah, K., Optimization Techniques for Chemical Engineers, Macmillan
of India, Delhi (1976).
Trambouze, P.J., and Piret, E.L., AIChE J, 5, 384 (1959).
van der Vusse, J.G., Chem. Eng. Sci., 19, 994 (1964).

PROBLEMS

10.1. Given the two reactions

where R is the desired product and is to be maximized. Rate the four

schemes shown in Fig. P10.1-either "good" or "not so good," Please,
no complicated calculations, just reason it out.

A A A
(c)

Figure P1O.l
 Problems 247

10.2. Repeat Problem 1 with just one change

10.3. Repeat Problem 1 with just one change

10.4. For the reactions

where R is the desired product, which of the following ways of running

a batch reactor is favorable, which is not? See Fig. P10.4.

Dump A & B lnto

the reactor
all at one t ~ m e
Add B drop Add A drop
by drop by drop

Figure P10.4

10.5. The Oxydation of Xylene. The violent oxidation of xylene simply produces
CO,and H,O;however, when oxidation is gentle and carefully controlled,
it can also produce useful quantities of valuable phthalic anhydride as
shown in Fig. P10.5. Also, because of the danger of explosion, the fraction

CO,,H20

anhydride

xylene
CO,,H20
Figure P10.5
248 Chapter 10 Choosing the Right Kind of Reactor

of xylene in the reacting mixture must be kept below 1%. Naturally, the
problem in this process is to obtain a favorable product distribution.
(a) In a plug flow reactor what values of the three activation energies
would require that we operate at the maximum allowable temper-
ature?
(b) Under what circumstances should the plug flow reactor have a falling
temperature progression?
10.6. The Trambouze Reactions-Reactions in parallel. Given the set of ele-
mentary reactions with a feed of CAO= 1 mollliter and v = 100 literslmin
we wish to maximize the fractional yield, not the production of S, in a
reactor arrangement of your choice.

$%eked r~ = ko ko = 0.025 mol/[Link]

A-S ... rS = klCA kl = 0.2 min-I
Y T
...
r~ = k2c; k2 = 0.4 [Link]

The computer, going through a multidimensional search [see problem

3, Chem. Eng. Sci.,45,595-614 (1990)l came up with the arrangement of
Fig. P10.6, which the authors claim is a LOCAL optimum, or a STATION-
ARY POINT. We are not interested in LOCAL optima, if such things
exist. We are interested in finding the GLOBAL optimum. So with this
in mind,
(a) do you judge that the arrangement of Fig. P10.6 is the best set up?
(b) if not, suggest a better scheme. Sketch your scheme and calculate the
volume of the reactors you plan to use.

vo = 100 literslmin 20 literslrnin

liters

= 600.63 liters

Figure P10.6

10.7. For the set of elementary reactions of Problem 10.6, with a feed of
CAo= 1 mollliter and v = 100 literslmin we now wish to maximize the
production rate of intermediate S (not the fractional yield) in a reactor
arrangement of your choice. Sketch your chosen reactor scheme and
determine C,,,,, obtainable.

10.8. Automobile Antifreeze. Ethylene glycol and diethylene glycol are used
as automobile antifreeze, and are produced by the reactions of ethylene
oxide with water, as follows:
 Problems 249

/CH2-CH20H / CH2-CH20H
--+
O\H + CH2-CH2 O\CH, -CH,OH
------------------
diethylene glycol

A mole of either glycol in water is as effective as the other in reducing

the freezing point of water; however, on a molar basis the diethylene
glycol is twice as expensive as the ethylene glycol. So we want to maximize
ethylene glycol, and minimize the diethylene glycol in the mixture.
One of the country's largest suppliers produced millions of kilograms of
antifreeze annually in reactors shown in Fig. P10.8~.One of the company's
engineers suggested that they replace their reactors with one of the type
shown in Fig. P10.8b. What do you think of this suggestion?

,- 500 m of 10 cm ID pipe
EtOx
+ -
water

Fluid moves
lazily

(a)

Figure P 1 0 . 8 ~& P10.8b

10.9. The Homogeneous Catalytic Reaction. Consider the elementary reaction

A + B -2B,
k
-r, = kCACB with k = 0.4 literlmol . min

For the following feed and reactor space time

Flow rate v = 100 literslmin

C,, = 0.45 mollliter
Feed composition
CBo= 0.55 mollliter
Space time T= 1min

we want to maximize the concentration of B in the product stream. Our

clever computer [see problem 8, Chem. Eng. Sci., 45, 595-614 (1990)l
gives the design of Fig. P10.9 as its best try.
250 Chapter 10 Choosing the Right Kind of Reactor

Figure P10.9

Do you think that this is the best way to run this reaction? If not,
suggest a better scheme. Do not bother to calculate reactor size, recycle
rate, etc. Just indicate a better scheme.

10.10. To Color Cola Drinks. When viscous corn syrup is heated it caramelizes
(turns a deep dark brown). However, if it is heated a bit too long it
transforms into carbon

corn heat more carbon

----)caramel
syrup particles

Very long tube of just the right

length so as to maximize

154°C
Corn
syrup
Heat

(a) Present reactor (b) Proposed reactor

Figure P1O.10 (a) Present reactor; ( b ) Proposed design.

The caramelized liquid is sent by railroad tank cars to the cola syrup
formulators, who then test the solution for quality. If it is too light in
color-penalty; if it has too many carbon particles per unit volume, then
the whole tank car is rejected. There is thus a delicate balance between
underreacting and overreacting.
At present a batch of corn syrup is heated at 154°Cin a vat for a precise
time. Then it is rapidly discharged and cooled, the vat is thoroughly
cleaned (very labor intensive), and then is recharged.
The company wants to reduce costs and replace this costly labor inten-
sive batch operation with a continuous flow system. Of course it will be
a tubular reactor (rule 2). What do you think of their idea? Comment
please, as you sit and sip your Coke or Pepsi.
 Problems 251
10.11. The Denbigh Reactions. We intend to run the reactions below:

kl = 1.0 1iterImol.s 2nd order

k2 = k, = 0.6 s- 1 1st order
k4 = 0.1 1iterlrnol.s 2nd order
T U

in a flow system under the following conditions

Feed flow rate v = 100 litersls
CAo= 6 mollliter
Feed composition
C,, = 0.6 mollliter

We want to maximize the concentration ratio of CR/& in the product

stream.
As reported [see problem 7, Chem. Eng. Sci., 45, 595-614 (1990)],
the attack on this problem used 2077 continuous variables, 204 integer
variables, 2108 constraints, and gave as an optimal solution the design
shown in Fig. [Link].
(a) Do you think you could do better? If so, what reactor design would
you suggest we use, and what C,/C, would you expect to obtain?
(b) If you wished to minimize the ratio of CR/&, how would you go
about it?

(2) is maximized ?

Figure [Link]

10.12. For the homogeneous catalytic reaction

and with a feed of CAo= 90 mol/m3, CBo= 10 mol/m3 we want about

44% conversion of reactant A. What flow reactor or combination of flow
reactors is best in that it gives the smallest total volume of reactors needed?
There is no need to try to calculate the size of reactors needed; just
determine the type of reactor system that is best and the type of flow that
should be used.

10.13. Repeat Problem 12 with just one change. We need 90% conversion of
reactant A.
252 Chapter 10 Choosing the Right Kind of Reactor

10.14. Repeat Problem 12 with just one change. We only want about 20% conver-
sion of reactant A.

10.15. We want to produce R from A in a batch reactor with a run time no

greater than 2 hours and at a temperature somewhere between 5 and
90°C. The kinetics of this liquid first-order reaction system is as follows:

Determine the optimum temperature (to give C,) and run time to use,
and the corresponding conversion of A to R.

10.16. Reactor-Separator-RecycleSystem-Benzene Chlorination. Here the el-

ementary reactions are

C6H6+ C1,
C6H5Cl+ C12
- 1
C6H,C1 + HC1
C6H,C12+ HC1
k1 = 0.412 literlkmol . hr
k2 = 0.055 literlkmol . hr

The desired product is monochlorobenzene. Also assume that any unre-

acted benzene in the product stream can be cleanly separated and reused
as desired.
With the requirement that we only use PFRs, a minimum of three, in
any arrangement, plus separator and recycle of unused reactant, the best
of which was determined by the computer [see case 3, Chem. Eng. Sci.,
46, 1361-1383 (1991)l is shown in Fig. P10.16.
Can you do better? There is no need to calculate volumes and flow
rates. Just come up with an improved scheme.

Benzene stream

Benzene
separator

Mono and
dichlorobenzene
Figure P10.16

10.17. Acrolein Production. Adams et al. [J. Catalysis, 3,379 (1964)l studied the
catalytic oxidation of propylene on bismuth molybdate catalyst to form
acrolein. With a feed of propylene and oxygen and reaction at 460"C, the
following three reactions occur.
 Problems 253

The reactions are all of first order in olefin and independent of oxygen
and of reaction products, and with reaction rate ratios

If no cooling is needed to keep the reaction close to 460°C and if no

separation and recycle of unused C,H6 is allowed, what kind of contactor
would you suggest be used and what should be the maximum expected
production rate of acrolein from this reactor?

10.18. Nonisothermal van der Vusse Reactions (1964). Consider the following
reactions:

where the Arrhenius activation energy is given in units of Jlmol, C, is to

be maximized and CAo= 1 mol/liter.
Insisting on using three MFR's with T~between 0.1 and 20 s [see example
2, AIChE J, 40,849 (1994)], with possible intercooling and a temperature
range between 360 K and 396 K, the best scheme calculated by the com-
puter is shown in Fig. P10.18.
(a) Do you like this design? If not, what do you suggest we do with this
three-reactor system? Please retain the three MFR's.
(b) What CR/CAocould be obtained and what T should be used with
the best reactor scheme (plug, mixed, or combined) and with ideal
heat transfer?

Figure P10.18
254 Chapter 10 Choosing the Right Kind of Reactor

10.19. Phthalic Anhydride from Naphthalene. The accepted mechanism for the
highly exothermic solid catalyzed oxidation of naphthalene to produce
phthalic anhydride is

k, = k, = 2 X 1013exp(-159000lRT) [hr-'1
where k, = 8.15 X 1017 exp(-209000lRT) [hill
k4 = 2.1 X lo5 exp(-83600lRT) [hr-'1

and where
A = naphthalene (reactant)
R = naphthaquinone (postulated intermediate)
S = phthalic anhydride (desired product)
T = CO, + H,O (waste products)
and the Arrenhius activation energy is given in units of Jlmol. This reaction
is to be run somewhere between 900 K and 1200 K.
A local optimum reactor setup discovered by the computer [see example
1, Chem. Eng. Sci., 49, 1037-1051 (1994)l is shown in Fig. P10.19.

Heat
/exchangers\

Figure P10.19

(a) Do you like this design? Could you do better? If so, how?
(b) If you could keep the whole of your reactors at whatever temperature
and 7 value desired, and if recyle is allowed, how much phthalic
anhydride could be produced per mole of naphthalene reacted?
Suggestion: Why not determine the values of k,, k,, k,, and k, for both
extremes of temperature, look at the values, and then proceed with the so-
lution?

10.20. Professor Turton dislikes using reactors in parallel, and he cringed when
he saw my recommended "best" design for Example 10.1. He much prefers
using reactors in series, and so for that example he suggests using the
design of Figure [Link], but without any recycle of fluid.
Determine the fractional yield of S, cP (SIA), obtainable with Turton's
design, and see if it matches that found in Example 10.1.
Part I1

Flow Pattern, Contacting, and

Non-Ideal Flow

Chapter 11 Basics of Non-Ideal Flow 1257

Chapter 12 Compartment Models 1283
Chapter 13 The Dispersion Model 1293
Chapter 14 The Tanks-in-Series Model 1321
Chapter 15 The Convection Model for Laminar Flow 1339
Chapter 16 Earliness of Mixing, Segregation, and RTD 1350
Chapter 11
Basics of Non-Ideal Flow

So far we have treated two flow patterns, plug flow and mixed flow. These can
give very different behavior (size of reactor, distribution of products). We like
these flow patterns and in most cases we try to design equipment to approach
one or the other because

one or the other often is optimum no matter what we are designing for.
these two patterns are simple to treat.

But real equipment always deviates from these ideals. How to account for this?
That is what this and the following chapters are about.
Overall three somewhat interrelated factors make up the contacting or flow
pattern:

1. the RTD or residence time distribution of material which is flowing through

the vessel
2. the state of aggregation of the flowing material, its tendency to clump and
for a group of molecules to move about together
3. the earliness and lateness of mixing of material in the vessel.

Let us discuss these three factors in a qualitative way at first. Then, this and the
next few chapters treat these factors and show how they affect reactor behavior.

The Residence Time Distribution, RTD

Deviation from the two ideal flow patterns can be caused by channeling of fluid,
by recycling of fluid, or by creation of stagnant regions in the vessel. Figure 11.1
shows this behavior. In all types of process equipment, such as heat exchangers,
packed columns, and reactors, this type of flow should be avoided since it always
lowers the performance of the unit.
If we know precisely what is happening within the vessel, thus if we have a
complete velocity distribution map for the fluid in the vessel, then we should,
in principle, be able to predict the behavior of a vessel as a reactor. Unfortunately,
this approach is impractical, even in today's computer age.
257
258 Chapter I 1 Basics of Non-Ideal Flow

Stagnant regions -/

Packed bed

Channeling, especially
serious in countercurrent
two-phase operations

Figure 11.1 Nonideal flow patterns which may exist in process

equipment.

Setting aside this goal of complete knowledge about the flow, let us be less
ambitious and see what it is that we actually need to know. In many cases we
really do not need to know very much, simply how long the individual molecules
stay in the vessel, or more precisely, the distribution of residence times of the
flowing fluid. This information can be determined easily and directly by a widely
used method of inquiry, the stimulus-response experiment.
This chapter deals in large part with the residence time distribution (or RTD)
approach to nonideal flow. We show when it may legitimately be used, how to
use it, and when it is not applicable what alternatives to turn to.
In developing the "language" for this treatment of nonideal flow (see Danck-
werts, 1953), we will only consider the steady-state flow, without reaction and
without density change, of a single fluid through a vessel.

State of Aggregation of the Flowing Stream

Flowing material is in some particular state of aggregation, depending on its
nature. In the extremes these states can be called microfluids and macrofluids,
as sketched in Fig. 11.2.

Single-Phase Systems. These lie somewhere between the extremes of macro-

and microfluids.
 Ch,apter 11 Basics of Non-Ideal Flow 259

Gases and ordinary Noncoalescing droplets

not very viscous Solid particles
liquids Very viscous liquids

Molecules are kept

[Individual molecules are free
grouped together in
to move about and intermix
aggregates or packets
Figure 11.2 Two extremes of aggregation of fluid.

Two-Phase Systems. A stream of solids always behaves as a macrofluid, but

for gas reacting with liquid, either phase can be a macro- or microfluid depending
on the contacting scheme being used. The sketches of Fig. 11.3 show completely
opposite behavior. We treat these two phase reactors in later chapters.

Earliness of Mixing
The fluid elements of a single flowing stream can mix with each other either
early or late in their flow through the vessel. For example, see Fig. 11.4.
Usually this factor has little effect on overall behavior for a single flowing
fluid. However, for a system with two entering reactant streams it can be very
important. For example, see Fig. 11.5.

Role of RTD, State of Aggregation, and Earliness of Mixing in Determining

Reactor Behavior
In some situations one of these three factors can be ignored; in others it can
become crucial. Often, much depends on the time for reaction, t,,the time for

Spray tower
Sparger reactor
reactor
G

,
- in liauid
/ $ . .
Liquid droplets
in gas

L
G
L e r e gas is a macrofluid
while liquid is a rnicrofluid Here gas us a rnicrofluid
while liquid is a macrofluid
Figure 11.3 Examples of macro- and microfluid behavior.
264) Chapter 11 Basics of Non-Ideal Flow

Early mixing Uniform mixing Late mixing

Lots of mixing
of young and
old fluid
Flat1
velocity
profile
l S a m e along
the whole
vessel
NO mixing
of young and
old fluid
' Well-mixed
region

Figure 11.4 Examples of early and of late mixing of fluid.

mixing t,,,and the time for stay in the vessel t,,,,. In many cases <,, has a
meaning somewhat like t,,but somewhat broader.

11.1 E, THE AGE DISTRIBUTION OF FLUID, THE RTD

It is evident that elements of fluid taking different routes through the reactor
may take different lengths of time to pass through the vessel. The distribution
of these times for the stream of fluid leaving the vessel is called the exit age
distribution E, or the residence time distribution RTD of fluid. E has the units
of time-l.
We find it convenient to represent the RTD in such a way that the area under
the curve is unity, or

This procedure is called normalizing the distribution, and Fig. 11.6 shows this.
We should note one restriction on the E curve-that the fluid only enters and
only leaves the vessel one time. This means that there should be no flow or
diffusion or upflow eddies at the entrance or at the vessel exit. We call this the
closed vessel boundary condition. Where elements of fluid can cross the vessel
boundary more than one time we call this the open vessel boundary condition.
With this representation the fraction of exit stream of age* between t and t +
dt is

Well mixed at entry, Separate parallel L Mixing only occurs

so A and B have flows so A and B at exit leaving no
lots of time for reaction are unable to react time for reaction
Figure 11.5 Early or late mixing affects reactor behavior.

* The t e r m "age" for a n element of the exit stream refers t o the time spent by that element in the vessel.
 11.1 E, The Age Distribution of Fluid, The RTD 261

Figure 11.6 The exit age distribution curve E for

fluid flowing through a vessel; also called the resi-
dence time distribution, or RTD.

the fraction younger than age t, is

whereas the fraction of material older than t,, shown as the shaded area in Fig.
11.6, is

The E curve is the distribution needed to account for nonideal flow.

Experimental Methods (Nonchemical) for Finding E

The simplest and most direct way of finding the E curve uses a physical or
nonreactive tracer. For special purposes, however, we may want to use a reactive
tracer. This chapter deals in detail with the nonreactive tracer, and for this all
sorts of experiments can be used. Figure 11.7 shows some of these. Because the
pulse and the step experiments are easier to interpret, the periodic and random
harder, here we only consider the pulse and the step experiment.

Periodic
input

Figure 11.7 Various ways of studying the flow pattern in vessels.

262 Chapter 11 Basics of Non-Ideal Flow

We next discuss these two experimental methods for finding the E curve. We
then show how to find reactor behavior knowing the E curve for the reactor.

The Pulse Experiment

Let us find the E curve for a vessel of volume V m3 through which flows v m3/s
of fluid. For this instantaneously introduce M units of tracer (kg or moles) into
the fluid entering the vessel, and record the concentration-time of tracer leaving
the vessel. This is the C,,,,, curve. From the material balance for the vessel we find

Area under the M

CpulseCurve i u

=-
v [s]
i

All this is shown in Fig. 11.8.

To find the E curve from the C,,,,, curve simply change the concentration scale
such that the area under the curve is unity. Thus, simply divide the concentration
readings by M l v , as shown in Fig. 11.9.

E = -Cpulse
Mlu

Instantaneously introduce
M units of tracer into the
fluid entering the vessel
/

I/ .m.
E
t r Record the leaving traces
3
on
time, s

u , m3/s Evaluate the mean

t-= - h- -c d t - xticiAti
lurne Icdt BCiAti

Figure 11.8 The useful information obtainable from the pulse

trace experiment.
 11.1 E, The Age Distribution of Fluid, The RTD 263

Figure 11.9 Transforming an experimental C,,,,, curve into an E curve.

We have another RTD function E,. Here time is measured in terms of mean
residence time 6 = tl?. Thus

Egisa useful measure when dealing with flow models which come up in Chapters
13,14, and 15. Figure 11.10 shows how to transform E into E,.
One final reminder, the relationship between C,,,,, and the E curves only holds
exactly for vessels with closed boundary conditions.

The Step Experiment

Consider u m3/s of fluid flowing through a vessel of volume V. Now at time t =
0 switch from ordinary fluid to fluid with tracer of concentration C,, =

[kg z y l ] , and measure the outlet tracer concentration CSkpversus t, as shown

in Fig. 11.11.

Figure 11.10 Transforming an E curve into an E, curve.

264 Chapter 11 Basics of Non-Ideal Flow

Step input at time t = 0

t < 0 . .. no tracer
t > 0 ... m [kglsl of tracer,
Cstep

Figure 11.11 Information obtainable from a step tracer experiment.

A material balance relates the different measured quantities of the output

curve of a step input

mV kg.s2
and
shaded area
(ofFig.11.11)
= -
v2 [ ] m3

- 1:"" tdcstep
1
t=
I:" dCstep
=
Cmax
t dCstep

where m [kgls] is the flow rate of tracer in the entering fluid.

The dimensionless form of the CStep curve is called the F curve. It is found by
having the tracer concentration rise from zero to unity, as shown in Fig. 11.12.

Relationship between the F and E Curves

To relate E with F imagine a steady flow of white fluid. Then at time t = 0 switch
to red and record the rising concentration of red fluid in the exit stream, the F

Figure 11.12 Transforming an experimental C,,,, curve to an F curve.

 11.1 E, The Age Distribution of Fluid, The RTD 255

curve. At any time t > 0 red fluid and only red fluid in the exit stream is younger
than age t. Thus we have

fraction of red fluid

in the exit stream ) (fraction
=
of exit stream
younger than age t

But the first term is simply the F value, while the second is given by Eq. 1. So
we have, at time t,

F=
I: Edt
(7)

and on differentiating

In graphical form this relationship is shown in Fig. 11.13.

These relationships show how stimulus-response experiments, using either step
or pulse inputs can conveniently give the RTD and mean flow rate of fluid in
the vessel. We should remember that these relationships only hold for closed
vessels. When this boundary condition is not met, then the Cpuls,and E curves
differ. The Cpuls,curves of the convection model (see Chap. 15) clearly show
this.
Figure 11.14 shows the shapes of these curves for various types of flow.

Figure 11.13 Relationship between the E and F

curves.
266 Chapter I 1 Basics of Non-Ideal Flow

Mixed flow

-+

1
-
t

t
t t 5 10 15

1.5

1 1.0

0.5

0 0 0
1 t 0.5 1 1.5
Figure 11.14 Properties of the E and F curves for various flows. Curves are drawn in terms of
ordinary and dimensionless time units. Relationship between curves is given by Eqs. 7 and 8.

At any time these curves are related as follows:

E = -v- . Cpulse, v dF
m F = -.
,i ,Cstep 7
E =-
dt '

t = -v
-
,g=-,t -,g - 1 E - -
E- , 0-tE
v' t
6, E,, F . . . all dimensionless, E = [time-']
 11.1 E, The Age Distribution of Fluid, The RTD 267

FINDING THE RTD BY EXPERIMENT

The concentration readings in Table Ell.l represent a continuous response to

a pulse input into a closed vessel which is to be used as a chemical reactor.
Calculate the mean residence time of fluid in the vessel t, and tabulate and plot
the exit age distribution E.

Table El1.1
Tracer Output Concentration, C,,,,,
Time t, min gmlliter fluid

SOLUTION

The mean residence time, from Eq. 4, is

-
t=
C tiCiAti ~t = constant C tici
ci~ti C. ci

The area under the concentration-time curve,

gives the total amount of tracer introduced. To find E, the area under this curve
must be unity; hence, the concentration readings must each be divided by the
total area, giving

E=- C
area

Thus we have
t, min
0 5 10 15 20 25 30
C
E=- min-l 0 0.03 0.05 0.05 0.04 0.02 0.01
area'
268 Chapter 11 Basics of Non-Ideal Flow

0.051
Ci -
-
-
+ Smoothed
curve

5 10 15 20 25 30
t, min

Figure E l l . l

Figure E l l . l is a plot of this distribution.

FINDING THE E CURVE FOR LIQUID FLOWING

THROUGH A VESSEL

A large tank (860 liters) is used as a gas-liquid contactor. Gas bubbles up through
the vessel and out the top, liquid flows in at one part and out the other at 5
litersls. To get an idea of the flow pattern of liquid in this tank a pulse of tracer
(M = 150 gm) is injected at the liquid inlet and measured at the outlet, as shown
in Fig. E11.2a.
(a) Is this a properly done experiment?
(b) If so, find the liquid fraction in the vessel.
(c) Determine the E curve for the liquid.
(d) Qualitatively what do you think is happening in the vessel?

*
.
-
IU
.-
C-'

E
M
6

2 4 6 8
t , min

Figure E11.2~

(a) Check the material balance against the tracer curve. From the material
balance, Eq. 3, we should have

Area = -=
150gm = 3 0 e = 0 . 5 _gm - min
u 5 litersls liter liter
 11.1 E, The Age Distribution of Fluid, The RTD 269

From the tracer curve

gm .min
(
Area=A, 1+ - + - +
6:
- .
) ()
=0.375 - =0.5-
liter

These values agree. The results are consistent.

(b) For the liquid, Eq. 4 gives

Thus the liquid volume in the vessel is

V , = tlv, = 2.67(5 X 60) = 800 liters

and the volume fraction of phases is

Fraction of liquid = -= 93%

Fraction of gas = 7%
: 1 J
(c) Finally, from Eq. 5 we find the E curve, or

E = -Cpulse - -0.75
- C= 1.5C
Mlv 0.5

Thus the E curve for the liquid is as shown in Fig. E11.2b. (c)
(d) The vessel has a strong recirculation of liquid, probably induced by the
rising bubbles. (4

2 4 6
t , min

Figure E11.2b
270 Chapter I 1 Basics of Non-Ideal Flow

The Convolution Integral

Suppose we introduce into a vessel a one-shot tracer signal Cinversus t as shown
in Fig. 11.15. In passing through the vessel the signal will be modified to give an
output signal Co, versus t. Since the flow with its particular RTD is responsible
for this modification let us relate Cin,E, and C.,
Focus attention on tracer leaving at time about t. This is shown as the narrow
rectangle B in Fig. 11.15. We may then write

iin rectangle B
all the tracer entering t' seconds earlier than t,
and staying for time t' in the vessel

We show the tracer which enters t' seconds earlier than t as the narrow rectangle
A. In terms of this rectangle the above equation may be written

tracer in fraction of tracer in A

( )(
tracerleaving)
rectangle which stays for about
(in rectangle a11 rectangles
A which enter A t' seconds in the vessel
earlier then
time t

In symbols and taking limits (shrinking the rectangles) we obtain the desired
relationship, which is called the convolution integral

In what can be shown to be equivalent form we also have

We say that C
,, is the convolution of E with Cinand we write concisely

t , P i n
out

t-t' t

C t' seconds
earlier
Figure 11.15 Sketch showing derivation of the convolution integral.
 11.1 E, The Age Distribution of Fluid, The RTD 271

Figure 11.16 Modification of an input tracer signal Ci, on passing through three succes-
sive regions.

Application of These Tools. To illustrate the uses of these mathematical tools

consider three independent* flow units a, b, and c, which are closed and connected
in series (see Fig. 11.16).

Problem 1. If the input signal Ci, is measured and the exit age distribution
functions E,, E,, and E, are known, then C1 is the convolution of E, with Cin
and so on, thus

and on combining

Thus we can determine the output from a multiregion flow unit.

Problem 2. If we measure Cinand C,, and know E, and E, we can then extract the
unknown E, . This type of problem is of particular importance in experimentation
where the entrance region and collection region for tracer are both large com-
pared with the experimental section.
It is a straightforward matter to convolute; however, to deconuolute, to find
one of the distribution functions under the integral, is difficult. Thus Problem 2
is harder to treat than Problem 1 and requires using a computer.

* By independence we mean that the fluid loses its memory as it passes from vessel to vessel. Thus,
a faster-moving fluid element in one vessel does not remember this fact in the next vessel and doesn't
preferentially flow faster (or slower) there. Laminar flow often does not satisfy this requirement of
independence; however, complete (or lateral) mixing of fluid between units satisfies this condition.
 272 Chapter I 1 Basics of Non-Ideal Flow

In some cases, however, we can in essence deconvolute. This special situation

is considered at the end of Chapter 14 and is illustrated with an example there.
Example 11.3 illustrates convolution; Example 14.4 illustrates deconvolution.

1 EXAMPLE 11.3 CONVOLUTION

Let us illustrate the use of the convolution equation, Eq. 10, with a very simple
example in which we want to find C,,, given Ci, and the E curve for the vessel,
as shown in Fig. E11.3a.
I

I + Tracer input curve

RTD for the vessel
8

0.4
Cin
4 E
0.2

0
0 1 2 3 4 O 6 7 8 9 1 0
t'- t

Figure E 1 1 . 3 ~

SOLUTION

First of all, take 1 min time slices. The given data are then

tr- t Ci, t' E

0 0 5 0 Note: The area under the
1 0 6 0.05 E curve is unity.
2 8 7 0.50
3 4 8 0.35
4 6 9 0.10
5 0 10 0
Now the first bit of tracer leaves at 8 min, the last bit at 13 min. Thus, applying
the convolution integral, in discrete form, we have
 11.2 Conversion in Non-Idea2 Flow' Reactors 273

The C,,, E, and C,,, curves, in discrete and in continuous form, are shown in
Fig. E11.3b. Note that the area under the C,,, curve equals the area under the
Cincurve.

0 2 4 6 8 10 12 14
t'- t t

Figure E11.3b

11.2 CONVERSION IN NON-IDEAL FLOW REACTORS

To evaluate reactor behavior in general we have to know four factors:
1. the kinetics of the reaction
2. the RTD of fluid in the reactor
3. the earliness or lateness of fluid mixing in the reactor
4. whether the fluid is a micro or macro fluid
For microfluids in plug or mixed flow we have developed the equations in the
earlier chapters. For intermediate flow we will develop appropriate models in
Chapters 12, 13, and 14.
To consider the early and late mixing of a microfluid, consider the two flow
patterns shown in Fig. 11.17 for a reactor processing a second-order reaction. In
(a) the reactant starts at high concentration and reacts away rapidly because
n > 1. In (b) the fluid drops immediately to a low concentration. Since the rate
of reaction drops more rapidly than does the concentration you will end up with
a lower conversion. Thus, for microfluids

Early mixing favors reactions where n < 1

For macrofluids, imagine little clumps of fluid staying for different lengths of
time in the reactor (given by the E function). Each clump reacts away as a little
274 Chapter 11 Basics of Non-Ideal Flow

Figure 11.17 This shows the latest and the earliest mixing we can have for a
given RTD.

batch reactor, thus fluid elements will have different compositions. So the mean
composition in the exit stream will have to account for these two factors, the

tactant )
kinetics and the RTD. In words, then

concentration of
fraction of exit
meanofconcentration)
reactant = in an element
remaining)(stream
of which is
of age between t
in exit stream age between t
and t + dt
all elements
of exit stream
and t + dt

In symbols this becomes

-
(5)=Jm(5)
. E dt
CAO at exit 0 CAO for an element or little
batch of fluid of age t

or in terms of conversions

FA = Jm0 (X~)element ' dt

or in a form suitable for numerical integration

intervals

From Chapter 3 on batch reactors we have

for first-order reactions

1
for second-order reactions

for an nth-order reaction

(2) element
= [I + (n - l)C~;'kt]~~l-~ (16)
 11.2 Conversion in Non-Ideal Flow Reactors 275

with area = 1,
6(t - to)

to
Figure 11.18 The E function for
plug flow.

These are terms to be introduced into the performance equation, Eq. 13. Also,
further on in this chapter we will show that for first-order reactions, the macrofluid
equation is identical to the batch or to the microfluid equation.
We continue this discussion in Chapter 16.

The Dirac Delta Function, S(t - to). One E curve which may puzzle us is the
one which represents plug flow. We call this the Dirac S function, and in symbols
we show it as

which says that the pulse occurs at t = to, as seen in Fig. 11.18.
The two properties of this function which we need to know are

Area under the curve:

I,' S(t - to)dt = 1
(18)
Any integration
with a S function: I: S(t - to)f ( f ) d t= f (41)

Once we understand what this means we will see that it is easier to integrate
with a 6 function than with any other. For example,

I,' S(t - 5)t6dt = 5* ('just replace toby 5 )

I: S(t - 5)t6dt = 0

CONVERSION IN REACTORS HAVING NON-IDEAL FLOW

The vessel of Example 11.1 is to be used as a reactor for a liquid decomposing

with rate
276 Chapter 11 Basics of Non-Ideal Flow

Find the fraction of reactant unconverted in the real reactor and compare this
with the fraction unconverted in a plug flow reactor of the same size.

SOLUTION

For the plug flow reactor with negligible density change we have

and with r from Example 11.1

Thus the fraction of reactant unconverted in a plug flow reactor equals 1.0%.
For the real reactor the fraction unconverted, given by Eq. 13 for macrofluids,
is found in Table E11.4. Hence the fraction of reactant unconverted in the
real reactor

Table E11.4
t E kt e-kt e-ktEAt
5 0.03 1.53 0.2154 (0.2154)(0.03)(5) = 0.0323
10 0.05 3.07 0.0464 0.0116
15 0.05 4.60 0.0100 0.0025

-
20
25
30
given
0.04
0.02
0.01
6.14
7.68
9.21
0.0021
0.005
0.0001
-
' - -
A
'A0
e-"E At
0.0004
0.0001
0
= 0.0469

From the table we see that the unconverted material comes mostly from the
early portion of the E curve. This suggests that channeling and short-circuiting
can seriously hinder attempts to achieve high conversion in reactors.
Note that since this is a first-order reaction we can treat it as a microfluid, or
a macrofluid, whatever we wish. In this problem we solved the plug flow case
as a microfluid, and we solved the nonideal case as a macrofluid.
 11.2 Conversion in Non-Ideal Flow Reactors 277

1 EXAMPLE 11.5 REACTION OF A MACROFLUID

Dispersed noncoalescing droplets (C,, = 2 mollliter) react (A -+ R, -r, =

k c ; , k = 0.5 literlmol .min) as they pass through a contactor. Find the average
concentration of A remaining in the droplets leaving the contactor if their RTD
is given by the curve in Fig. E11.5.

1 3
t , rnin
Figure E11.5

Equation 13is the pertinent performance equation. Evaluate terms in this expres-
sion. For

-r, = k c ; , k = 0.5 literlmol . min

I The batch equation from Chapter 3 is

With E = 0.5 for 1 < t < 3, Eq. 13 becomes

-
C^ = (5)
CAO
1"
E dt = f2.
0
(0.5)dt = 0.5 in 2 = 0.347
CAObatch ll+t

REFERENCE

Danckwerts, P. V., Chem. Eng. Sci.,2, 1 (1953).

278 Chapter 11 Basics of Non-Ideal Flow

PROBLEMS

11.1. A pulse input to a vessel gives the results shown in Fig. P1l.l.
(a) Check the material balance with the tracer curve to see whether the
results are consistent.
(b) If the result is consistent, determine t, V and sketch the E curve.

t , min

Figure P1l.l

11.2. Repeat Problem P1l.l with one change: The tracer curve is now as shown
in Fig. P11.2.

t , min
Figure P11.2

11.3. A pulse input to a vessel gives the results shown in Fig. P11.3.
(a) Are the results consistent? (Check the material balance with the
experimental tracer curve.)
(b) If the results are consistent, determine the amount of tracer introduced
M, and the E curve.

t, sec

Figure P11.3

11.4. A step experiment is made on a reactor. The results are shown in Fig. P11.4.
(a) Is the material balance consistent with the tracer curve?
(b) If so, determine the vessel volume V ,7, the F curve, and the E curve.
 Problems 279

Lv,? t , rnin

Figure P11.4

11.5. A batch of radioactive material is dumped into the Columbia River at

Hanford, Washington. At Bonneville Dam, about 400 km downstream
the flowing waters (6000 m3/s) are monitored for a particular radioisotope
(t,,, > 10 yr) and the data of Fig. P11.5 are obtained.
(a) How many units of this tracer were introduced into the river?
(b) What is the volume of Columbia River waters between Bonneville
Dam and the point of introduction of tracer?

20 35 125
t, days
Figure P11.5

11.6. A pipeline (10 cm I.D., 19.1 m long) simultaneously transports gas and
liquid from here to there. The volumetric flow rate of gas and liquid are
60 000 cm3/s and 300 cm3/s, respectively. Pulse tracer tests on the fluids
flowing through the pipe give results as shown in Fig. P11.6. What fraction
of the pipe is occupied by gas and what fraction by liquid?

2 100
t , sec

Figure P11.6
280 Chapter 11 Basics of Non-Ideal Flow

A liquid macrofluid reacts according to A + R as it flows through a

vessel. Find the conversion of A for the flow patterns of Figs. P11.7 to
[Link] and kinetics as shown.

11.7. CAo= I mollliter

-rA = kCI5
k = 2 m ~ l ~ . ~ l l i tmin
er~

11.8. CAo= 2 mollliter

-rA = k C i 1
k = 2 literlmol .min t, min
0.5

11.9. CAo= 6 mollliter

-rA = k
k = 3 mollliter. min

11.11. CAo= 0.1 mollliter

-rA = k
k = 0.03 mollliter. min " .-
L,, a
Figures P11.7, P11.8, P11.9, [Link], [Link]

11.12-11.14. Hydrogen sulfide is removed form coal gas by contact with a moving
bed of iron oxide particles which convert to the sulfide as follows:

Fe203-+FeS

In our reactor the fraction of oxide converted in any particle is determined

by its residence time t and the time needed for complete conversion of
 Problems 281
the particle 7, and this is given by

and
X = 1 when t r 1 hr

Find the conversion of iron oxide to sulfide if the RTD of solids in the
contactor is approximated by the curve of Fig. P11.12, P11.13, or P11.14.

Coal gas
without
H2S

Coal
with

Solids out
Fe203 + FeS
Figures P11.12, P11.13, P11.14

11.15. Cold solids flow continuously into a fluidized bed where they disperse
rapidly enough so that they can be taken as well mixed. They then heat
up, they devolatilize slowly, and they leave. Devolatilization releases gas-
eous A which then decomposes by first-order kinetics as it passes through
the bed. When the gas leaves the bed decomposition of gaseous A stops.
From the following information determine the fraction of gaseous A which
has decomposed.
Data: Since this is a large-particle fluidized bed containing cloudless
bubbles, assume plug flow of gas through the unit. Also assume
that the volume of gases released by the solids is small compared
to the volume of carrier gas passing through the bed.
Mean residence time in the bed:
6 6
= 15 min, = 2 s for carrier gas
For the reaction: A +products, -rA = k c A ,k = 1s-l
282 Chapter 11 Basics of Non-Ideal Flow

Volatlles only react

Figure P11.15

11.16. Reactant A (C,, = 64 mol/m3) flows through a plug flow reactor ( r =

50 s), and reacts away as follows:

Determine the conversion of A if the stream is:

(a) a microfluid,
(b) a macrofluid.
Chapter 12
Compartment Models

Flow models can be of different levels of sophistication and the compartment

models of this chapter are the next stage beyond the very simplest, those that
assume the extremes of plug flow and mixed flow. In the compartment models
we consider the vessel and the flow through it as follows:

Total
I V,-plug flow region
volume - - . V,-mixed
[v,-dead
flow region
V,-active volume

or stagnant region within the vessel

v,-active flow, that through the plug and mixed flow regions
Total
throughflow ... v,-bypass flow
v
v,-recycle flow

By comparing the E curve for the real vessel with the theoretical curves for
various combinations of compartments and throughflow, we can find which model
best best fits the real vessel. Of course, the fit will not be perfect; however,
models of this kind are often a reasonable approximation to the real vessel.
Figure 12.1, on the next few pages, shows what the E curves look like for
various combinations of the above elements-certainly not all combinations.

Hints, Suggestions, and Possible Applications

(a) If we know M (kilograms of tracer introduced in the pulse) we can make

a material balance check. Remember that M = v (area of curve). However,
if we only measure the output C on an arbitrary scale, we cannot find M
or make this material balance check.
283
284 Chapter 12 Compartment Models

1 -1
Ideal pulse, height is

Area = I

h Area = 1
I A r e a
Area = 2

r Mixed

'LMixed

Figure 12.1 Various compartment flow models.

 Chapter 12 Compartment Models 285

Area = -

d
'rg

w'
U1
1ove: = v= + ' ~ 2

t mean u ul +u2

log E

This method of plotting and

fitting exponentials is only
useful when and t2
are very different.
Equation is the sum of

t
0 t
2$7=vl+v2
u1+u2

Put on semilog
scale
Figure 12.1 (Continued)
 1

1
-
e Area = -

1 Mean of shaded
portion of curve - U,
Overall-- Vp+Vm
Overall-- V~+Vm mean - 'obs = 7
s 7
mean - t ~ b =
Meanof V~+Vm
first ,oflion= '(7

1
Area = -

t
0

Figure 12.1 (Continued)

 -
Chapter 12 Compartment Models 287

log C
Initial slope = - -=
log A
A Slope = - -
Equivalent
(log A) - 1
A
-
e
0 t (log A) - 2 t

Mean of curve: i
log C

Slope = - -

Figure 12.2 Properties of exponential decay tracer curves.

(b) We must know both V and v if we want to properly evaluate all the
elements of a model, including dead spaces. If we only measure jobs,we
cannot find the size of these stagnant regions and must ignore them in
our model building. Thus

If the real vessel - -t =-v

has dead spaces: tabs < 2 v
. . - where
If the real vessel - - vw,
has no dead spaces: tabs = - tabs = -v

(c) The semilog plot is a convenient tool for evaluating the flow parameters
of a mixed flow compartment. Just draw the tracer response curve on this
plot, find the slope and intercept and this gives the quantities A, B, and
C , as shown in Fig. 12.2.

Diagnosing Reactor Ills

These combined models are useful for diagnostic purposes, to pinpoint faulty
flow and suggest causes. For example, if you expect plug flow and you know
i = Vlv, Fig. 12.3 shows what you could find.
If you expect mixed flow, Fig. 12.4 shows what you may find.
288 Chapter 12 Compartment Models

Slim trim curve means Early curve is a sure sign Multiple decaying peaks at regular
reasonably good flow of stagnant backwaters intervals indicate strong
internal recirculation

Mean is too early

C

Double peaks come from flow Late curve

in parallel paths, channeling
Late tracer is puzzling. Material balance
says it can't happen so the only

.
explanations are:
u or v are incorrectly measured
(check flow meters, etc.)
C C tracer is not inert (adsorbs on
surface? Try a different one)
the closed vessel assumption is
far from satisfied.
. -
tabs
Figure 12.3 Misbehaving plug flow reactors.

Reasonably good flow Time lag means plug flow Slow internal circulation suggests
in series with mixed flow sluggish slow turnover of fluid
Inadequate mixing
Draft tube effect

Early curve means Late curve means either incorrect Sharp early peak
v or V or noninert tracer
from inlet to outlet

. - t
t tabs

Figure 12.4 Misbehaving mixed flow reactors.

 Chapter 12 Compartment Models 289

BEHAVIOR OF A G/L CONTACTOR

From the measured pulse tracer response curves (see figure), find the fraction
of gas, of flowing liquid, and of stagnant liquid in the gas-liquid contactor shown
in Fig. E12.1.

20 rn

t, sec t, sec

Figure E12.1

To find V,, V,, and V,,,,, first calculate j, and j, from the tracer curves. Thus
from Fig. E12.1

and
-
t1 = 40 s.

Therefore

V, = j,v, = (10)(0.5) = 5m3

Vl = jlv, = 40(0.1) = 4m3

In terms of void volume

% stagnant = 10%)
290 Chapter 12 Compartment Models

At present our 6-m3tank reactor gives 75% conversion for the first order reaction
A-R. However, since the reactor is stirred with an underpowered paddle
turbine, we suspect incomplete mixing and poor flow patterns in the vessel. A
pulse tracer shows that this is so and gives the flow model sketched in Fig. E12.2.
What conversion can we expect if we replace the stirrer with one powerful
enough to ensure mixed flow?

Deadwater

Figure E12.2

Let subscript 1 represent today's reactor and subscript 2 represent the cured
reactor. At present, from Chapter 5 for the MFR, we have

But k7, = 3 kr, = 3 X 3 = 9

Therefore
 Problems 291

PROBLEMS

12.1. to 12.6. A pulse of concentrated NaCl solution is introduced as tracer into

the fluid entering a vessel (V = 1m3,v = 1m3/min)and the concentration of
tracer is measured in the fluid leaving the vessel. Develop a flow model
to represent the vessel from the tracer output data sketched in Figs. P12.1
to P12.6.

t. rnin t , rnin

t, min t , sec

r Area =

t, sec t, sec

Figures P12.1 through P12.6

12.7. to 12.10.A step input tracer test (switching from tap water to salt water,
measuring the conductivity of fluid leaving the vessel) is used to explore
the flow pattern of fluid through the vessel (V = 1 m3, v = 1 m3/min).
Devise a flow model to represent the vessel from the data of Figs. P12.7
to P12.10.
292 Chapter 12 Compartment Models

t , rnin
9
C
4

40
t, sec t, sec
Figures P12.7 through P12.10

12.11. The second order aqueous reaction A + B +R + S is run in a large tank

reactor (V = 6 m3) and for an equimolar feed stream (C,, = C,,) conver-
sion of reactants is 60%. Unfortunately, agitation in our reactor is rather
inadequate and tracer tests of the flow within the reactor give the flow
model sketched in Fig. P12.11. What size of mixed flow reactor will equal
the performance of our present unit?

Figure P12.11 Figure Pl2.12

12.12. Repeat Example 12.2 with one change: The model for the present flow
is as shown in Fig. P12.12.
Chapter 13
The Dispersion Model

Choice of Models
Models are useful for representing flow in real vessels, for scale up, and for
diagnosing poor flow. We have different kinds of models depending on whether
flow is close to plug, mixed, or somewhere in between.
Chapters 13 and 14 deal primarily with small deviations from plug flow. There
are two models for this: the dispersion model and the tanks-in-series model. Use
the one that is comfortable for you. They are roughly equivalent. These models
apply to turbulent flow in pipes, laminar flow in very long tubes, flow in packed
beds, shaft kilns, long channels, screw conveyers, etc.
For laminar flow in short tubes or laminar flow of viscous materials these
models may not apply, and it may be that the parabolic velocity profile is the
main cause of deviation from plug flow. We treat this situation, called the pure
convection model, in Chapter 15.
If you are unsure which model to use go to the chart at the beginning of
Chapter 15. It will tell you which model should be used to represent your setup.

13.1 AXIAL DISPERSION

Suppose an ideal pulse of tracer is introduced into the fluid entering a vessel.
The pulse spreads as it passes through the vessel, and to characterize the spreading
according to this model (see Fig. 13.1), we assume a diffusion-like process super-
imposed on plug flow. We call this dispersion or longitudinal dispersion to distin-
guish it from molecular diffusion. The dispersion coefficient D (m2/s)represents
this spreading process. Thus
large D means rapid spreading of the tracer curve
small D means slow spreading
D = 0 means no spreading, hence plug flow
Also

(s) is the dimensionless group characterizing the spread in the whole vessel.
294 Chapter 13 The Dispersion ,Wodel

The pulse starts spreading and this can be

A pulse of tracer caused by many things: velocity profile,
at time t = 0 turbulent mixing, molecular diffusion, etc.

Symmetrical and gaussian

Pulse input Measurement

(8-input) point

Figure 13.1 The spreading of tracer according to the dispersion model.

We evaluate D or DIuL by recording the shape of the tracer curve as it passes

the exit of the vessel. In particular, we measure

t = mean time of passage, or when the curve passes by the exit

v2 = variance, or a measure of the spread of the curve

These measures, t and v2, are directly linked by theory to D and DIuL. The
mean, for continuous or discrete data, is defined as

The variance is defined as

or in discrete form

The variance represents the square of the spread of the distribution as it passes
the vessel exit and has units of (time)2. It is particularly useful for matching
experimental curves to one of a family of theoretical curves. Figure 13.2illustrates
these terms.
 13.1 Axial Dispersion 295

Figure 13.2

Consider plug flow of a fluid, on top of which is superimposed some degree

of backmixing, the magnitude of which is independent of position within the
vessel. This condition implies that there exist no stagnant pockets and no gross
bypassing or short-circuiting of fluid in the vessel. This is called the dispersed
plug flow model, or simply the dispersion model. Figure 13.3 shows the conditions
visualized. Note that with varying intensities of turbulence or intermixing the
predictions of this model should range from plug flow at one extreme to mixed
flow at the other. As a result the reactor volume for this model will lie between
those calculated for plug and mixed flow.
Since the mixing process involves a shuffling or redistribution of material either
by slippage or eddies, and since this is repeated many, many times during the
flow of fluid through the vessel we can consider these disturbances to be statistical
in nature, somewhat as in molecular diffusion. For molecular diffusion in the
x-direction the governing differential equation is given by Fick's law:

where a, the coefficient of molecular diffusion, is a parameter which uniquely

characterizes the process. In an analogous manner we may consider all the
contributions to intermixing of fluid flowing in the x-direction to be described

x
Fluctuations due to different flow
Flat velocity velocities and due to molecular
and turbulent diffusion

Dispersed plug flow

Plug flow
Figure 13.3 Representation of the dispersion (dispersed plug
flow) model.
296 Chapter 13 The Dispersion Model

by a similar form of expression, or

where the parameter D, which we call the longitudinal or axial dispersion coefi-
cient, uniquely characterizes the degree of backmixing during flow. We use the
terms longitudinal and axial because we wish to distinguish mixing in the direction
of flow from mixing in the lateral or radial direction, which is not our primary
concern. These two quantities may be quite different in magnitude. For example,
in streamline flow of fluids through pipes, axial mixing is mainly due to fluid
velocity gradients, whereas radial mixing is due to molecular diffusion alone.
In dimensionless form where z = (ut + x)lL and I3 = tlt = tulL, the basic
differential equation representing this dispersion model becomes

where the dimensionless group (s), called the vessel dispersion number, is
the parameter that measures the extent of axial dispersion. Thus

D
-+ 0 negligible dispersion, hence plug flow
UL
D
- large dispersion, hence mixed flow
UL+

This model usually represents quite satisfactorily flow that deviates not too
greatly from plug flow, thus real packed beds and tubes (long ones if flow
is streamline).

Fitting the Dispersion Model for Small Extents of Dispersion, DIuL < 0.01
If we impose an idealized pulse onto the flowing fluid then dispersion modifies
this pulse as shown in Fig. 13.1. For small extents of dispersion (if DIuL is small)
the spreading tracer curve does not significantly change in shape as it passes the
measuring point (during the time it is being measured). Under these conditions
the solution to Eq. 6 is not difficult and gives the symmetrical curve of Eq. 7
shown in Figs. 13.1 and 13.4.

This represents a family of gaussian curves, also called error or Normal curves.
 13.1 Axial Dispersion 297

Shaded area = 0.68

At this value of DIuL the

tracer curve is becoming

0.9 1.O 1.1 t V

Figure 13.4 Relationship between DIuL and the dimensionless E, curve for small
extents of dispersion, Eq. 7.

The equations representing this family are

E,=;.E=

-
t E = Vv = uL or G E = l
4DLlu I
-mean of E curve
2
cT- - 2 or

Note that DIuL is the one parameter of this curve. Figure 13.4 shows a number
of ways to evaluate this parameter from an experimental curve: by calculating
its variance, by measuring its maximum height or its width at the point of
inflection, or by finding that width which includes 68% of the area.
298 Chapter 13 The Dispersion Model

\
-
u$verall
2
r vIa r u , 2 un 7,
I
Vessel a b
)'
c ... Vessel n
Atfa&.-
t

Figure 13.5 Illustration of additivity of means and of variances of the E curves of vessels
a , b , . . . ,n.

Also note how the tracer spreads as it moves down the vessel. From the
variance expression of Eq. 8 we find that

a2K L or
(widthcurve
of tracer

Fortunately, for small extents of dispersion numerous simplifications and ap-

proximations in the analysis of tracer curves are possible. First, the shape of the
tracer curve is insensitive to the boundary condition imposed on the vessel,
whether closed or open (see above Eq. 11.1). So for both closed and open vessels
Cpulse = E and C,,,, = F.
For a series of vessels the t and u2of the individual vessels are additive, thus,
referring to Fig. 13.5 we have

and

The additivity of times is expected, but the additivity of variance is not generally
expected. This is a useful property since it allows us to subtract for the distortion
of the measured curve caused by input lines, long measuring leads, etc.
This additivity property of variances also allows us to treat any one-shot tracer
input, no matter what its shape, and to extract from it the variance of the E
curve of the vessel. So, on referring to Fig. 13.6, if we write for a one-shot input

b ~ input
n ~

Figure 13.6 Increase in variance is the same in both cases, or u2 = a:,, - uin= Au2.
 13.1 Axial Dispersion 299

Aris (1959) has shown, for small extents of dispersion, that

a2 - 2
out-(Tin - A(+'
(tout - tin)' (Aj>2

Thus no matter what the shape of the input curve, the DIuL value for the vessel
can be found.
The goodness of fit for this simple treatment can only be evaluated by compari-
son with the more exact but much more complex solutions. From such a compari-
son we find that the maximum error in estimate of DIuL is given by

D
error < 5% when -< 0.01
uL

D
Large Deviation from Plug Flow, -> 0.01
UL
Here the pulse response is broad and it passes the measurement point slowly
enough that it changes shape-it spreads-as it is being measured. This gives a
nonsymmetrical E curve.
An additional complication enters the picture for large DIuL: What happens
right at the entrance and exit of the vessel strongly affects the shape of the tracer
curve as well as the relationship between the parameters of the curve and DIuL.
Let us consider two types of boundary conditions: either the flow is undisturbed
as it passes the entrance and exit boundaries (we call this the open b.c.), or you
have plug flow outside the vessel up to the boundaries (we call this the closed
b.c.). This leads to four combinations of boundary conditions, closed-closed,
open-open, and mixed. Figure 13.7 illustrates the closed and open extremes,
whose RTD curves are designated as E,, and E,,.
Now only one boundary condition gives a tracer curve which is identical to
the E function and which fits all the mathematics of Chapter 11,and that is the
closed vessel. For all other boundary conditions you do not get a proper RTD.
In all cases you can evaluate DIuL from the parameters of the tracer curves;
however, each curve has its own mathematics. Let us look at the tracer curves
for closed and for the open boundary conditions.

m Closed vessel

Plug flow,
D
Open vessel

Same D
D=O everywhere
f -I+
1-
-,v -
+ fl
-?--
--A++
--3

t I
Change in flow pattern
t I
Undisturbed flow at
at boundaries boundaries
Figure 13.7 Various boundary conditions used with the dispersion model.
300 Chapter 13 The Dispersion Model

"small deviation"

Figure 13.8 Tracer response curves for closed vessels and large deviations
from plug flow.

Closed Vessel. Here an analytic expression for the E curve is not available.
However, we can construct the curve by numerical methods, see Fig. 13.8, or
evaluate its mean and variance exactly, as was first done by van der Laan
(1958). Thus

Open Vessel. This represents a convenient and commonly used experimental

device, a section of long pipe (see Fig. 13.9). It also happens to be the only
physical situation (besides small DIuL) where the analytical expression for the
E curve is not too complex. The results are given by the response curves shown
 13.1 Axial Dispersion 301
Introduce Measure
I A

A squirt of tracer across the Measure the intensity of light

cross section, or a flash of by "looking through the wall"
radiation to light sensitive or measure conductivity with
fluid, etc. a small probe, etc.

Figure 13.9 The open-open vessel boundary condition.

in Fig. 13.10, and by the following equations, first derived by Levenspiel and
Smith (1957).

1
E @ ,=~V ~ T ( D / ~ L )

open-open -2
vessel

Comments

(a) For small DIuL the curves for the different boundary conditions all ap-
proach the "small deviation" curve of Eq. 8. At larger DIuL the curves
differ more and more from each other.
(b) To evaluate DIuL either match the measured tracer curve or the measured
u2to theory. Matching u2is simplest, though not necessarily best; however,
it is often used. But be sure to use the right boundary conditions.
(c) If the flow deviates greatly from plug (DIuL large) chances are that the
real vessel doesn't meet the assumption of the model (a lot of independent
random fluctuations). Here it becomes questionable whether the model
should even be used. I hesitate when DIuL > 1.
(d) You must always ask whether the model should be used. You can always
match u2values, but if the shape looks wrong, as shown in the accompa-
nying sketches, don't use this model, use some other model.
302 Chapter 13 The Dispersion Model

Figure 13.10 Tracer response curves for "open" vessels having large deviations
from plug flow.

(e) For large DIuL the literature is profuse and conflicting, primarily because
of the unstated and unclear assumptions about what is happening at the
vessel boundaries. The treatment of end conditions is full of mathematical
subtleties as noted above, and the additivity of variances is questionable.
Because of all this we should be very careful in using the dispersion model
where backmixing is large, particularly if the system is not closed.
(f) We will not discuss the equations and curves for the open-closed or closed-
open boundary conditions. These can be found in Levenspiel (1996).

Step Input of Tracer

Here the output F curve is S-shaped and is obtained by integrating the corre-
sponding E curve. Thus at any time t or 6

The shape of the F curve depends on DIuL and the boundary conditions for
the vessel. Analytical expressions are not available for any of the F curves;
 13.1 Axial Dispersion 303

8 = =t
t

Figure 13.11 Step response curves for small deviations from plug flow.

however, their graphs can be constructed. Two typical cases are displayed below,
in Figs. 13.11 and 13.13.

Small Deviation from Plug Flow, DIuL < 0.01 From Eqs. 8 and 16 we can find
the curves of Fig. 13.11, as shown. For these small deviations from plug flow we
can find DIuL directly by plotting the experimental data on probability graph
paper as indicated in Fig. 13.12. Example 13.2 shows in detail how this is done.

Step Response for Large Dispersion, DIuL > 0.01. For large deviations from
plug flow, the problem of boundary conditions must be considered, the resulting
S-shaped response curves are not symmetrical, their equations are not available,
and they are best analyzed by first differentiating them to give the corresponding
C,,,,, curve. Figure 13.13 shows an example of this family of curves.

Figure 13.12 Probability plot of a step response signal. From this we

find DIuL directly.
304 Chapter 13 The Dispersion Model

0 0.5 1 1.5 2
e= i
t
Figure 13.13 Step response curves for large deviations from plug flow in
closed vessels.

Comments
(a) One direct commercial application of the step experiment is to find the
zone of intermixing-the contaminated width-between two fluids of
somewhat similar properties flowing one after the other in a long pipeline.
Given DIuL we find this from the probability plot of Fig. 13.12. Design
charts to ease the calculation are given by Levenspiel (1958a).
(b) Should you use a pulse or step injection experiment? Sometimes one type
of experiment is naturally more convenient for one of many reasons. In
such a situation this question does not arise. But when you do have a
choice, then the pulse experiment is preferred because it gives a more
"honest" result. The reason is that the F curve integrates effects; it gives
a smooth good-looking curve which could well hide real effects. For
example, Fig. 13.14 shows the corresponding E and F curves for a given
vessel.

is clearly evident.
You can't miss it.
E F miss this?

t t

Figure 13.14 Sensitivity of the E and F curves for the same flow.
 13.1 Axial Dispersion 305

[ EXAMPLE 13.1 DluL FROM A C,,,,,,CURVE

On the assumption that the closed vessel of Example 11.1, Chapter 11, is well
represented by the dispersion model, calculate the vessel dispersion number
DIuL. The C versus t tracer response of this vessel is

SOLUTION

Since the C curve for this vessel is broad and unsymmetrical, see Fig. 11.E1, let
us guess that dispersion is too large to allow use of the simplification leading to
Fig. 13.4. We thus start with the variance matching procedure of Eq. 18. The
mean and variance of a continuous distribution measured at a finite number of
equidistant locations is given by Eqs. 3 and 4 as

and

Using the original tracer concentration-time data, we find

Cci=3+5+5+4+2+1=20
2 t i c i = (5 X 3 ) + (10 X 5) + . + (30 X 1 ) = 300 min
t:Ci = (25 X 3 ) + (100 X 5 ) + . - + (900 X 1 ) = 5450 min2

Therefore

and
306 Chapter 13 The Dispersion Model

Now for a closed vessel Eq. 13 relates the variance to DIuL. Thus

Ignoring the second term on the right, we have as a first approximation

--2: 0.106
uL

Correcting for the term ignored we find by trial and error that

Our original guess was correct: This value of DIuL is much beyond the limit
where the simple gaussian approximation should be used.

D/uL FROM AN F CURVE

von Rosenberg (1956) studied the displacement of benzene by n-butyrate in a

38 mm diameter packed column 1219 mm long, measuring the fraction of n-
butyrate in the exit stream by refractive index methods. When graphed, the
fraction of n-butyrate versus time was found to be S-shaped. This is the F curve,
and it is shown in Fig. E13.2a for von Rosenberg's run at the lowest flow rate,
where u = 0.0067 mmls, which is about 0.5 mlday.
Find the vessel dispersion number of this system.

Time, (sec x
(a)
Figure E13.2~From von Rosenberg (1956).
 13.1 Axial Dispersion 307

Instead of taking slopes of the F curve to give the E curve and then determining
the spread of this curve, let us use the probability paper method. So, plotting
the data on this paper does actually give close to a straight line, as shown in
Fig. E13.2b.

(b)
Figure E13.2b From Levenspiel and Smith (1957).

To find the variance and DIuL from a probability graph is a simple matter.
Just follow the procedure illustrated in Fig. 13.12. Thus Fig. E13.2b shows that

the 16th percentile point falls at t = 178 550 s

the 84th percentile point falls at t = 187 750 s

and this time interval represents 2u. Therefore the standard deviation is

We need this standard deviation in dimensionless time units if we are to find

D. Therefore
308 Chapter 13 The Dispersion Model

Hence the variance

and from Eq. 8

Note that the value of DIuL is well below 0.01, justifying the use of the gaussian
approximation to the tracer curve and this whole procedure.

D/uL FROM A ONE-SHOT INPUT

Find the vessel dispersion number in a fixed-bed reactor packed with 0.625-cm
catalyst pellets. For this purpose tracer experiments are run in equipment shown
in Fig. E13.3.
The catalyst is laid down in a haphazard manner above a screen to a height
of 120 cm, and fluid flows downward through this packing. A sloppy pulse of
 13.2 Correlations for Axial Dispersion 309

radioactive tracer is injected directly above the bed, and output signals are
recorded by Geiger counters at two levels in the bed 90 cm apart.
The following data apply to a specific experimental run. Bed voidage = 0.4,
superficial velocity of fluid (based on an empty tube) = 1.2 cmlsec, and variances
of output signals are found to be o: = 39 sec2 and of = 64 sec2. Find DIuL.

SOLUTION

Bischoff and Levenspiel (1962) have shown that as long as the measurements
are taken at least two or three particle diameters into the bed, then the open
vessel boundary conditions hold closely. This is the case here because the mea-
surements are made 15 cm into the bed. As a result this experiment corresponds
to a one-shot input to an open vessel for which Eq. 12 holds. Thus

I or in dimensionless form

Ao; = A o 2 (b)2 (25 sec2)[(901.2cm)(0.4)

=
cmlsec - 1
] 36 --

I from which the dispersion number is

13.2 CORRELATIONS FOR AXIAL DISPERSION

The vessel dispersion number DIuL is a product of two terms

(dispersion) ( factor )
D intensity of geometric D d
z= = ( 2 )(z)
where

fluid
properties
)(dynamics
flow ) =
no.

and where

d is a characteristic length = d,,,, or d,

Experiments show that the dispersion model well represents flow in packed
beds and in pipes. Thus theory and experiment give Dlud for these vessels. We
summarize them in the next three charts.
310 Chapter 13 The Dispersion Model

Figure 13.15 Correlation for the dispersion of fluids flowing in pipes, adapted from Levenspiel
(1958b).

Figures 13.15 and 13.16 show the findings for flow in pipes. This model repre-
sents turbulent flow, but only represents streamline flow in pipes when the pipe
is long enough to achieve radial uniformity of a pulse of tracer. For liquids this
may require a rather long pipe, and Fig. 13.16 shows these results. Note that
molecular diffusion strongly affects the rate of dispersion in laminar flow. At
low flow rate it promotes dispersion; at higher flow rate it has the opposite effect.
Correlations similar to these are available or can be obtained for flow in
beds of porous and/or adsorbing solids, in coiled tubes, in flexible channels, for
pulsating flow, for non-Newtonians, and so on. These are given in Chapter 64
of Levenspiel (1996).
Figure 13.17 shows the findings for packed beds.
 13.2 Correlations for Axial Dispersion 311

Bodenstein no.: (Re)(Sc) = * 1= 5

I*. PB 9

Figure 13.16 Correlation for dispersion for streamline flow in pipes; prepared from Taylor
(1953, 1954a) and Aris (1956).

Figure 13.17 Experimental findings on dispersion of fluids flowing with mean axial
velocity u in packed beds; prepared in part from Bischoff (1961).
312 Chapter 13 The Dispersion Model

13.3 CHEMICAL REACTION AND DISPERSION

Our discussion has led to the measure of dispersion by a dimensionless group
DIuL. Let us now see how this affects conversion in reactors.
Consider a steady-flow chemical reactor of length L through which fluid is
flowing at a constant velocity u, and in which material is mixing axially with a
dispersion coefficient D. Let an nth-order reaction be occurring.

A * products, -rA = kC2

By referring to an elementary section of reactor as shown in Fig. 13.18, the

basic material balance for any reaction component

input = output + disappearance by reaction + accumulation (4.1)

becomes for component A, at steady state,

disappearance
+ (Out-in)axialdispersion+ by reaction + accumulation = o
(Out-in)b~lkfl~~

The individual terms (in moles Altime) are as follows:

entering by bulk flow = (moles

-
volume velocity
(
A) flow (cross-sectional
area
)
= CA,,uS, [molls]
leaving by bulk flow = CAJt~,uS

entering by axial dispersion =

dt

Cross-sectional
area = S
/
Disappearance
Accurndation of A
(= 0, for steady state)
of A
Figure 13.18 Variables for a closed vessel in
which reaction and dispersion are occurring.
 13.3 Chemical Reaction and Dispersion 313

d N =~ - (DS
leaving by axial dispersion = -
dt
%) [+PI

disappearance by reaction = (-r,) V = (-rA)S Al, [molls]

Note that the difference between this material balance and that for the ideal
plug flow reactors of Chapter 5 is the inclusion of the two dispersion terms,
because material enters and leaves the differential section not only by bulk flow
but by dispersion as well. Entering all these terms into Eq. 17 and dividing by
S A1 gives

Now the basic limiting process of calculus states that for any quantity Q which
is a smooth continuous function of 1

So taking limits as A1 + 0 we obtain

In dimensionless form where z = l/L and r = t = Llu = Vlv, this expression be-
comes

or in terms of fractional conversion

This expression shows that the fractional conversion of reactant A in its passage
through the reactor is governed by three dimensionless groups: a reaction rate
group k r C;il, the dispersion group DIuL, and the reaction order n.

First-Order Reaction. Equation 18 has been solved analytically by Wehner and

Wilhelm (1956)for first-order reactions. For vessels with any kind of entrance
314 Chapter 13 The Dispersion Model

Figure 13.19 Comparison of real and plug flow reactors for the first-order
A +products, assuming negligible expansion;from Levenspiel and Bischoff
(1959, 1961).

and exit conditions the solution is

where

Figure 13.19 is a graphical representation of these results in useful form,

prepared by combining Eq. 19 and Eq. 5.17, and allows comparison of reactor
sizes for plug and dispersed plug flow.
For small deviations from plug flow DIuL becomes small, the E curve ap-
proaches gaussian; hence, on expanding the exponentials and dropping higher
order terms Eq. 19 reduces to

" It should be noted that Eq. 21 applies to any gaussian RTD with variance u2.
 13.3 Chemical Reaction and Dispersion 315

Equation 20 with Eq. 5.17 compares the performance of real reactors which are
close to plug flow with plug flow reactors. Thus the size ratio needed for identical
conversion is given by

D
- --- - - 1 + ( k 3- for same C,,.,
LP VP uL
while the exit concentration ratio for identical reactor size is given by

D
-- -1 +( k ~-) ~ for same V or r
CAP uL
nth-Order Reactions. Figure 13.20 is the graphical representation of the solu-
tion of Eq. 18 for second-order reactions in closed vessels. It is used in a manner
similar to the chart for first-order reactions. To estimate reactor performance for
reactions of order different from one and two we may extrapolate or interpolate
between Figs. 13.19 and 13.20.

Figure 13.20 Comparison of real and plug flow reactors for the second-
order reactions
A +B -t products, C,, = C,,
or. .. 2A -+ products
assuming negligible expansion; from Levenspiel and Bischoff (1959,1961).
316 Chapter 13 The Dispersion Model

CONVERSION FROM THE DISPERSION MODEL

two methods and comment.

Matching the experimentally found variance with that of the dispersion model,
we find from Example 13.1

D
-= 0.12
uL

Conversion in the real reactor is found from Fig. 13.19. Thus moving along the
k r = (0.307)(15) = 4.6 line from CICo = 0.01 to DIuL = 0.12, we find that the
fraction of reactant unconverted is approximately

- - 0.035, or 35%
-
Co

Comments. Figure E13.4 shows that except for a long tail the dispersion model
curve has for the most part a greater central tendency than the actual curve. On
the other hand, the actual curve has more short-lived material leaving the vessel.

Both curves have the sarnev2

Relatively large
contribution to
variance of the

Figure E13.4

Because this contributes most to the reactant remaining unconverted, the finding
 Problems 317

Extensions
Levenspiel (1996) Chapter 64 discusses and presents performance equations for
various extensions to this treatment. A much more detailed exposition of this
subject is given by Westerterp et al. (1984) Chapter 4.

REFERENCES
Aris, R., Proc. Roy. Soc. (London), A235, 67 (1956).
Aris, R., Chem. Eng. Sci., 9, 266 (1959).
Bischoff, K. B., Ph.D. Thesis, Chemical Engineering Department, Illinois Institute of
Technology (1961).
Bischoff, K. B., and Levenspiel, O., Chem. Eng. Sci., 17, 245 (1962).
Levenspiel, O., Petroleum Refiner, March (1958a).
Levenspiel, O., Ind. Eng. Chern., 50, 343 (1958b).
Levenspiel, O., The Chemical Reactor Omnibook, Chap. 64, OSU Bookstores, Corvallis,
OR 97339,1996.
Levenspiel, O., and Bischoff, K. B., Ind. Eng. Chem., 51,1431 (1959); 53,313 (1961).
Levenspiel, O., and Smith, W. K., Chem. Eng. Sci., 6,227 (1957).
Taylor, G. I., Proc. Roy. Soc. (London), 219A, 186 (1953); 2254,473 (1954).
van der Laan, E. T., Chem. Eng. Sci., 7, 187 (1958).
von Rosenberg, D. U., AIChE J., 2, 55 (1956).
Wehner, J. F., and Wilhelm, R. H., Chem. Eng. Sci., 6, 89 (1956).
Westerterp, K. R., van Swaaij, W. P. M., and Beenackers, A. A. C. M., Chemical Reactor
Design and Operations, John Wiley, New York 1984.

PROBLEMS
13.1. The flow pattern of gas through blast furnaces was studied by VDEh (Veren
Deutscher Eisenhiittenleute Betriebsforschungsinstitut) by injecting Kr-85
into the air stream entering the tuyeres of the 688 m3 furnace. A sketch
and listing of pertinent quantities for run 10.5 of 9.12.1969 is shown in Fig.
P13.1. Assuming that the axial dispersion model applies to the flow of gas

Iron ore
zP=,,rn,\ ~=~EFrnrn Width of
curve at 61%

Figure P13.1
318 Chapter 13 The Dispersion Model

in the blast furnace, compare Dlud for the middle section of the blast
furnace with that expected in an ordinary packed bed.
From Standish and Polthier, Blast Furnace Aerodynamics, p. 99, N.
Standish, ed., Australian I. M. M. Symp., Wollongong, 1975.
13.2. Denmark's longest and greatest river, the Gudenaa, certainly deserves
study, so pulse tracer tests were run on various stretches of the river using
radioactive Br-82. Find the axial dispersion coefficient in the upper stretch
of the river, between Torring and Udlum, 8.7 km apart, from the following
reported measurements.
C, arbitrary C, arbitrary
440
250
122
51
20
9
3
0

Data from Danish Isotope Center, report of November 1976.

13.3. RTD studies were carried out by Jagadeesh and Satyanarayana (IECIPDD
11520, 1972) in a tubular reactor (L = 1.21 m, 35 mm ID). A squirt of
NaCl solution (5 N) was rapidly injected at the reactor entrance, and
mixing cup measurements were taken at the exit. From the following results
calculate the vessel dispersion number; also the fraction of reactor volume
taken up by the baffles.
t, sec NaCl in sample

13.4. A pulse of radioactive Ba-140 was injected into a 10-in. pipeline (25.5 cm
ID) 293 km long used for pumping petroleum products (u = 81.7 cmls,
Re = 24 000) from Rangely, Colorado to Salt Lake City, Utah. Estimate
the time of passage of fluid having more than 112,C, of tracer and compare
the value you calculate with the reported time of passage of 895 sec averaged
over five runs. From the table of values for the gaussian distribution C >
C,,12 occurs between 4 t- 1.18 (T,. This may be helpful information. Data
from Hull and Kent, Ind. Eng. Chem., 44,2745 (1952).
 Problems 319

13.5. An injected slug of tracer material flows with its carrier fluid down a long,
straight pipe in dispersed plug flow. At point A in the pipe the spread of
tracer is 16 m. At point B, 1 kilometer downstream from A, its spread is
32 m. What do you estimate its spread to be at a point C, which is 2
kilometers downstream from point A?
13.6. A refinery pumps products A and B successively to receiving stations up
to 100 km away through a 10-cm ID pipeline. The average properties of
A and B are p = 850 kg/m3, p = 1.7 X kglm-s, !3 = m2/s, the
fluid flows at u = 20 cmls, and there are no reservoirs, holding tanks or
pipe loops in the line; just a few bends. Estimate the 16%-84% contaminated
width 100 km downstream. Adapted from Petroleum Refiner, 37, 191
(March 1958); Pipe Line Industry, pg. 51 (May 1958).
13.7. Kerosene and gasoline are pumped successively at 1.1 m/s through a 25.5-
cm ID pipeline 1000 km long. Calculate the 5195%-9515% contaminated
width at the exit of the pipe given that the kinematic viscosity for the 501
50% mixture is

(Data and problem from Sjenitzer, Pipeline Engineer, December 1958).

13.8. Water is drawn from a lake, flows through a pump and passes down a long
pipe in turbulent flow. A slug of tracer (not an ideal pulse input) enters
the intake line at the lake, and is recorded downstream at two locations
in the pipe L meters apart. The mean residence time of fluid between
recording points is 100 sec, and variance of the two recorded signals is

What would be the spread of an ideal pulse response for a section of this
pipe, free from end effects and of length L/5?
13.9. Last autumn our office received complaints of a large fish kill along the
Ohio River, indicating that someone had discharged highly toxic material
into the river. Our water monitoring stations at Cincinnati and Portsmouth,
Ohio (119 miles apart) report that a large slug of phenol is moving down
the river and we strongly suspect that this is the cause of the pollution.
The slug took 9 hours to pass the Portsmouth monitoring station, and its
concentration peaked at 8:00 A.M. Monday. About 24 hours later the slug
peaked at Cincinnati, taking 12 hours to pass this monitoring station.
Phenol is used at a number of locations on the Ohio River, and their
distance upriver from Cincinnati are as follows:
Ashland, KY-150 miles upstream Marietta, OH-303
Huntington, WV-168 Wheeling, WV-385
Pomeroy, OH-222 Steubenville, OH-425
Parkersburg, WV-290 Pittsburgh, PA-500
What can you say about the probable pollution source?
320 Chapter 13 The Dispersion Model

13.10. A 12-m length of pipe is packed with 1 m of 2-mm material, 9 m of 1-cm

material, and 2 m of 4-mm material. Estimate the variance in the output
C curve for a pulse input into this packed bed if the fluid takes 2 min to
flow through the bed. Assume a constant bed voidage and a constant
intensity of dispersion given by Dlud, = 2.

13.11. The kinetics of a homogeneous liquid reaction are studied in a flow reactor,
and to approximate plug flow the 48-cm long reactor is packed with 5-
mm nonporous pellets. If the conversion is 99% for a mean residence time
of 1 sec, calculate the rate constant for the first-order reaction
(a) assuming that the liquid passes in plug flow through the reactor
(b) accounting for the deviation of the actual flow from plug flow
(c) What is the error in calculated k if deviation from plug flow is not con-
sidered?
Data: Bed voidage E = 0.4
Particle Reynolds number Re, = 200

13.12. Tubular reactors for thermal cracking are designed on the assumption of
plug flow. On the suspicion that nonideal flow may be an important factor
now being ignored, let us make a rough estimate of its role. For this
assume isothermal operations in a 2.5-cm ID tubular reactor, using a
Reynolds number of 10 000 for flowing fluid. The cracking reaction is
approximately first order. If calculations show that 99% decomposition
can be obtained in a plug flow reactor 3 m long, how much longer must
the real reactor be if nonideal flow is taken into account?

13.13. Calculations show that a plug flow reactor would give 99.9% conversion
of reactant which is in aqueous solution. However, our rector has an RTD
, = 1000, what outlet concentration
somewhat as shown in Fig. P13.13. If ,C
can we expect in our reactor if reaction is first order? From mechanics
a2= a2/24for a symmetrical triangle with base a, rotating about its center
of gravity.

from mechanics

+a+
Figure P13.13
Chapter 14
The Tanks-In-Series Model

This model can be used whenever the dispersion model is used; and for not too
large a deviation from plug flow both models give identical results, for all practical
purposes. Which model you use depends on your mood and taste.
The dispersion model has the advantage in that all correlations for flow in
real reactors invariably use that model. On the other hand the tanks-in-series
model is simple, can be used with any kinetics, and it can be extended without
too much difficulty to any arrangement of compartments, with or without recycle.

14.1 PULSE RESPONSE EXPERIMENTS AND THE RTD

Figure 14.1 shows the system we are considering. We also define

t
0; = = = dimensionless time based on the mean residence time per tank 2;
ti
t
6 = = = dimensionless time based on the mean residence time in all N tanks, 7.
t

Then
-
Oi=N6 ... and ... 6 , = 1 , 3 = 1

and at any particular time, from Eq. 11 in Chapter 11

For the first tank. Consider a steady flow v m3/s of fluid into and out of the
first of these ideal mixed flow units of volume V,. At time t = 0 inject a pulse
of tracer into the vessel which when evenly distributed in the vessel (and it is)
has a concentration C,,.
At any time t after the tracer is introduced make a material balance, thus

rate of disappearance input output

of tracer ) ( ) (
= rate - rate )
322 Chapter 14 The Tanks-in-Series Model

Exit tracer curve,

the RTD, the E function

= mean timeltanks

i= N $ = mean time for all N tanks

Figure 14.1 The tanks-in-series model.

In symbols this expression becomes

dC1
vl-=o-uc, mol tracer
dt

where C1is the concentration of tracer in tank "1." Separating and integrating
then gives

Since the area under this CIC, versus t curve is i1(check this if you wish) it
allows you to find the E curve; so one may write

For the second tank where C1 enters, C2leaves, a material balance gives

Separating gives a first-order differential equation, which when integrated gives

For the Nth tank. Integration for the 3rd, 4th, . . . , Nth tank becomes more
complicated so it is simpler to do all of this by Laplace transforms.
 14.1 Pulse Response Experiments and the RTD 323

The RTD's, means and variances, both in time and dimensionless time were
first derived by MacMullin and Weber (1935) and are summarized by Eq. 3.

N-1 -2
. . . 7 = NZ, . . . g2
7. = (-&) NN e-tNli
( N - I)!
= L
N
N-1 -
e-"it t 2.,. (
.. . -t . = -. =Nt;
(N - I)! l N
-
E,=tiE= .. . (.ZL - N
0. -

e-NR.. . @2 --
1
E@= (NQ E =N
( N - I)! @-N

Graphically these equations are shown in Fig. 14.2. The properties of the RTD
curves are sketched in Fig. 14.3.

~N t=
i t
r=t %.=A
I ti

Figure 14.2 RTD curves for the tanks-in-series model, Eq. 3.

324 Chapter 14 The Tanks-in-Series Model

e
Figure 14.3 Properties of the RTD curve for the tanks-in-series model.

Comments and Extensions

1ndependence.l If M tanks are connected to N more tanks (all of the same size)
then the individual means and variances (in ordinary time units) are additive, or

%+,
-
= tM + -t,. . . and . . . CT~,,, = a% + a$ (4)

Because of this property we can join incoming streams with recycle streams.
Thus this model becomes useful for treating recirculating systems.

One-shot Tracer Input. If we introduce any one-shot tracer input into N tanks,
as shown in Fig. 14.4, then from Eqs. 3 and 4 we can write

Because of the independence of stages it is easy to evaluate what happens to

the C curve when tanks are added or subtracted. Thus this model becomes useful
in treating recycle flow and closed recirculation systems. Let us briefly look at
these applications.

By independence we mean that the fluid loses its memory as it passes from vessel to vessel. Thus
a faster moving fluid element in one vessel does not remember this fact in the next vessel and doesn't
preferentially flow faster (or slower) there. Laminar flow often does not satisfy this requirement of
independence; however, complete (or lateral) mixing of fluid between units satisfies this condition.
 14.1 Pulse Response Experiments and the RTD 325

Model for vessel

.............................

...
I----------------------------J
Figure 14.4 For any one-shot tracer input Eq. 4 relates input, output, and number of tanks.

Closed Recirculation System. If we introduce a 6 signal into an N stage system,

as shown in Fig. 14.5, the recorder will measure tracer as it flows by the first
time, the second time, and so on. In other words it measures tracer which has
passed through N tanks, 2N tanks, and so on. In fact it measures the superposition
of all these signals.
To obtain the output signal for these systems simply sum up the contributions
from the first, second, and succeeding passes. If m is the number of passes, we
then have from Eq. 3

iiCpulse

C Qpulse
~,
= e-titi

= e-oi
-
' (tlTi)rnN-l
m=l ( m N - I ) !

2 ( m0yN-1
rn=l N I)! -

CBpulse= Ne-NQ
m=l
-
C ((N6)rnN-I
mN - I)!

Figure 14.5 shows the resulting C curve. As an example of the expanded form
of Eq. 5 we have for five tanks in series

where the terms in brackets represent the tracer signal from the first, second,
and successive passes.
Recirculation systems can be represented equally well by the dispersion model
[see van der Vusse (1962),Voncken et al. (1964),and Harrell and Perona (1968)l.
Which approach one takes simply is a matter of taste, style, and mood.
326 Chapter 14 The Tanks-in-Series Model

8-input signal Output signal

' N tanks, 1 pass (Eq. 3)

N tanks, 2nd pass

I
1 2 3 4
= e

Figure 14.5 Tracer signal in a recirculating system.

Recirculation with Throughflow. For relatively rapid recirculation compared

to throughflow, the system as a whole acts as one large stirred tank; hence,
the observed tracer signal is simply the superposition of the recirculation
pattern and the exponential decay of an ideal stirred tank. This is shown in
Fig. 14.6 where C, is the concentration of tracer if it is evenly distributed in
the system.
This form of curve is encountered in closed recirculation systems in which
tracer is broken down and removed by a first-order process, or in systems using

ut

T . E , ~ Model, mean time =i

co

t
Figure 14.6 Recirculation with slow throughflow.
 14.1 Pulse Response Experiments and the R T D 327

radioactive tracer. Drug injection on living organisms give this sort of superposi-
tion because the drug is constantly being eliminated by the organism.

Step Response Experiments and the F Curve The output F curve from a series
of N ideal stirred tanks is, in its various forms, given by Eq. 8.

1+ N O + W2+ ... +(W

2!
+
( N - I)!
..I
I I

Number
=
o f tanks
ForN=3
For N tanks

This is shown in graphical form in Fig. 14.7.

Figure 14.7 The F curve for the tanks-in-series model, from MacMullin and
Weber (1935).
328 Chapter 14 The Tanks-in-Series Model

14.2 CHEMICAL CONVERSION

First-Order Reaction
Chapter 6 develops the conversion equation. Thus for first-order reactions in
one tank

for N tanks in series

A comparison with plug flow performance is given in Fig. 6.5.

For small deviations from plug flow (large N ) comparison with plug flow gives

for same CA,,,,:

c~,~tank-
- s
- ; (k?)2
for same volume V:
c~~ 2N

These equations apply to both micro- and macrofluids.

Second-Order Reaction of a Microfluid, A +R or A +B + R with C,, = C,,

For a microfluid flowing through N tanks in series Eq. 6.8 gives

and Fig. 6.6 compares the performance to that for plug flow.

All Other Reaction Kinetics of Microfluids

Either solve the mixed flow equation for tank after tank

a rather tedious process, but no problem today with our handy slave, the com-
puter. Or else we could use the graphical procedure shown in Fig. 14.8.
 14.2 Chemical Conversion 329

Parallel lines for

-'A

c ~ 4c ~ 3 c ~ 2 c ~ l

Figure 14.8 Graphical method of evaluating the performance

of N tanks in series for any kinetics.

Chemical Conversion of Macrofluids

There is rare use for macrofluid equations for homogeneous reactions. However,
if you do need them combine Eq. 11.3 with Eq. 3 for N tanks in series, to give

These equations may not be of practical use for homogeneous systems; however,
they are of primary importance for heterogeneous systems, especially for GIs
systems.

MODIFICATIONS TO A WINERY

A small diameter pipe 32 m long runs from the fermentation room of a winery
to the bottle filling cellar. Sometimes red wine is pumped through the pipe,
sometimes white, and whenever the switch is made from one to the other a
small amount of "house blend" rosC is produced (8 bottles). Because of some
construction in the winery the pipeline length will have to be increased to 50 m.
For the same flow rate of wine, how many bottles of rosC may we now expect
to get each time we switch the flow?

Figure E14.1 sketches the problem. Let the number of bottles, the spread, be
related to a.

Original: Ll=32m a , = 8 aI=64

Longer pipe: L, = 50 m a, = ? (T; =?
330 Chapter 14 The Tanks-in-Series Model

Eight bottles

How many bottles?

Figure E14.1

But for small deviations from plug flow, from Eq. 3 a2oc N or a2 L.

:. a, = 10. . . or we can expect 10 bottles of vin rosC

A FABLE ON RIVER POLLUTION

Last spring our office received complaints of a large fish kill along the Ohio
River, indicating that someone had discharged highly toxic material into the
river. Our water monitoring stations at Cincinnati and Portsmouth, Ohio (119
miles apart), report that a large slug of phenol is moving down the river, and
we strongly suspect that this is the cause of the pollution. The slug took about
10.5 hours to pass the Portsmouth monitoring station, and its concentration
peaked at 8:00 A.M. Monday. About 26 hours later the slug peaked at Cincinnati,
taking 14 hours to pass this monitoring station.
Phenol is used at a number of locations on the Ohio River, and their distance
upriver from Cincinnati are as follows:

Ashland, KY-150 miles upstream Marietta, OH-303

Huntington, WV-168 Wheeling, WV-385
Pomeroy, OH-222 Steubenville, OH-425
Parkersburg, WV-290 Pittsburgh, PA-500

What can you say about the probable pollution source?

SOLUTION

Let us first sketch what is known, as shown in Fig. E14.2. To start, assume that
a perfect pulse is injected. Then according to any reasonable flow model, either
 14.2 Chemical Conversion 331

Figure E14.2
-
L miles
119 miles

dispersion or tanks-in-series, we have

distance from
~ f r a c ecurve
r CC
point of origin

(
spread of
curve a ) J distance from
origin

:. from Cincinnati: 14 = k L1"

:. from Portsmouth: 10.5 = k(L - 119)la

Dividing one by the other gives

14 L
. . . from which L = 272 moles

Comment. Since the dumping of the toxic phenol may not have occurred instan-
taneously, any location where L 5 272 miles is suspect, or

Ashland
Huntington
1} t

Pomeroy 1
This solution assumes that different stretches of the Ohio River have the same
flow and dispersion characteristics (reasonable), and that no suspect tributary
joins the Ohio within 272 miles of Cincinnati. This is a poor assumption . . .
check a map for the location of Charleston, WV, on the Kanawah River.
 332 Chapter 14 The Tanks-in-Series Model

1 : B M 4, * - FLOW MODELS FROM RTD CURVES

Let us develop a tanks-in-series model to fit the RTD shown in Fig. E14.3~.

(a)

Figure E 1 4 . 3 ~

1 SOLUTION
As a first approximation, assume that all the tracer curves are ideal pulses. We
will later relax this assumption. Next notice that the first pulse appears early.
This suggests a model as shown in Fig. E14.3b, where v = 1 and V, + V, +
V, = 1. In Chapter 12 we see the characteristics of this model, so let us fit it.
Also it should be mentioned that we have a number of approaches. Here is one:
* Look at the ratio of areas of the first two peaks

From the location of the first peak

Vl -
- Vl - 1 . ..
( R + 1)u (1 + 1) 6
* From the time between peaks

Since V, + V, add up to 1, there is no dead volume, so at this point our model

reduces to Fig. E14.3~.Now relax the plug flow assumption and adopt the
tanks-in-series model. From Fig. 14.3

I So our model finally is shown in Fig. E14.3d.

 14.2 Chemical Corzversion 333

Figure E14.3b and c

Figure E14.3d

Bypassing the Complex Process of Deconvolution

Suppose we measure the sloppy input and output tracer curves for a process
vessel for the purpose of studying the flow through the vessel, thus to find the
E curve for the vessel. In general this requires deconvolution (see Chapter 11);
however, if we have a flow model in mind whose parameter has a one-to-one
relationship with its variance, then we can use a very simple shortcut to find the
E curve for the vessel.
Example 14.4 illustrates this method.

FINDING THE VESSEL E CURVE USING A SLOPPY

TRACER INPUT
I
Given Ci, and C,,, as well as the location and spread of these tracer curves, as
shown in Fig. E14.4~estimate the vessel E curve. We suspect that the tanks-in-
series model reasonably represents the flow in the vessel.

I From Fig. E14.4a we have, for the vessel,

334 Chapter 14 The Tanks-in-Series Model

,
C,, with '= 220s
u2=1oos2

Find the E curve for

the vessel
(a)

Figure E14.4a

Equation 3 represents the tanks-in-series model and gives

So from Eq. 3a, for N tanks-in-series we have

and for N = 4

Figure E14.4b shows the shape of this E curve.

t, S
(6)
Figure E14.4b
 Problems 335
REFERENCES

Harrell, J. E., Jr., and Perona, J. J., Znd. Eng. Chem. Process Design Develop., 7,464 (1968).
MacMullin, R. B., and Weber, M., Jr., Trans. AIChE, 31, 409 (1935).
van der Vusse, J. G., Chem. Eng. Sci., 17, 507 (1962).
Vonken, R. M., Holmes, D. B., and den Hartog, H. W., Chem. Eng. Sci., 19,209 (1964).

PROBLEMS

14.1. Fit the tanks-in-series model to the following mixing cup output data to
a pulse input.

14.2. Fluid flows at a steady rate through ten well-behaved tanks in series. A
pulse of tracer is introduced into the first tank, and at the time this tracer
leaves the system

maximum concentration = 100 mmol

tracer spread = 1min

If ten more tanks are connected in series with the original ten tanks, what
would be
(a) the maximum concentration of leaving tracer?
(b) the tracer spread?
(c) How does the relative spread change with number of tanks?

14.3. From the New York Times Magazine, December 25,1955, we read: "The
United States Treasury reported that it costs eight-tenths of a cent to
print dollar bills, and that of the billion and a quarter now in circulation,
a billion have to be replaced annually." Assume that the bills are put into
circulation at a constant rate and continuously, and that they are with-
drawn from circulation without regard to their condition, in a random
manner.
Suppose that a new series of dollar bills is put in circulation at a given
instant in place of the original bills.
(a) How many new bills will be in circulation at any time?
(b) 21 years later, how many old bills will still be in circulation?

14.4. Referring to the previous problem, suppose that during a working day a
gang of counterfeiters put into circulation one million dollars in fake one-
dollar bills.
(a) If not detected, what will be the number in circulation as a function
of time?
(b) After 10 years, how many of these bills would still be in circulation?
336 Chapter 14 The Tanks-in-Series Model

14.5. Repeat Problem 13.13, but solve it using the tanks-in-series model instead
of the dispersion model.

14.6. A stream of fully suspended fine solids (v = 1 m3/min) passes through

two mixed flow reactors in series, each containing 1 m3 of slurry. As soon
as a particle enters the reactors, conversion to product begins and is
complete after two minutes in the reactors. When a particle leaves the
reactors, reaction stops. What fraction of particles is completely converted
to product in this system?

14.7. Fit the RTD of Fig. P14.7 with the tanks-in-series model.

Figure P14.7

14.8. From a pulse input into a vessel we obtain the following output signal

Time,min 1 3 5 7 9 11 13 15
Concentration 0 0 10 10 10 10 0 0
(arbitrary)

We want to represent the flow through the vessel with the tanks-in-series
model. Determine the number of tanks to use.

14.9. Strongly radioactive waste fluids are stored in "safe-tanks7' which are
simply long, small-diameter (e.g., 20 m by 10 cm) slightly sloping pipes.
To avoid sedimentation and development of "hot spots," and also to
insure uniformity before sampling the contents, fluid is recirculated in
these pipes.
To model the flow in these tanks, a pulse of tracer is introduced and
the curve of Fig. P14.9 is recorded. Develop a suitable model for this
system and evaluate the parameters.
 Problems 337

At injection

/- Pass

A Final value 7

Figure P14.9 RTD for a closed recirculating system.

14.10. A reactor with a number of dividing baffles is to be used to run the

reaction

A -+R with -rA = 0.05 CAmollliter. min

A pulse tracer test gives the following output curve:

Time,min I 0 10 20 30 40 50 60 70
Concentrationreading 1 35 38 40 40 39 37 36 35

(a) Find the area under the C versus t curve.

(b) Find the E versus t curve.
(c) Calculate the variance of the E curve.
(d) How many tanks in series is this vessel equivalent to?
(e) Calculate XAassuming plug flow.
(f) Calculate XAassuming mixed flow.
(g) Calculate XAassuming the tanks-in-series model.
(h) Calculate XAdirectly from the data.

14.11. A reactor has flow characteristics given by the nonnormalized C curve in

Table P14.11, and by the shape of this curve we feel that the dispersion
or tanks-in-series models should satisfactorily represent flow in the reactor.
(a) Find the conversion expected in this reactor, assuming that the disper-
sion model holds.
(b) Find the number of tanks in series which will represent the reactor
and the conversion expected, assuming that the tanks-in-series
model holds.
338 Chapter 14 The Tanks-in-Series Model

Table P14.11.
Time Tracer Concentration Time Tracer Concentration
0 0 10 67
1 9 15 47
2 57 20 32
3 81 30 15
4 90 41 7
5 90 52 3
6 86 67 1
8 77 70 0

(c) Find the conversion by direct use of the tracer curve.

(d) Comment on the difference in these results, and state which one you
think is the most reliable.

Data. The elementary liquid-phase reaction taking place is A + B -+

products, with a large enough excess of B so that the reaction is essentially
first order. In addition, if plug flow existed, conversion would be 99% in
the reactor.
Chapter 15
The Convection Model for
Laminar Flow

When a tube or pipe is long enough and the fluid is not very viscous, then the
dispersion or tanks-in-series model can be used to represent the flow in these
vessels. For a viscous fluid, one has laminar flow with its characteristic parabolic
velocity profile. Also, because of the high viscosity there is but slight radial
diffusion between faster and slower fluid elements. In the extreme we have the
pure convection model. This assumes that each element of fluid slides past its
neighbor with no interaction by molecular diffusion. Thus the spread in residence
times is caused only by velocity variations. This flow is shown in Fig. 15.1. This
chapter deals with this model.

15.1 THE CONVECTION MODEL AND ITS RTD

How to Tell from Theory Which Model to Use
The first question to ask is, "Which model should be used in a given situation?"
The following chart, adapted from Ananthakrishnan et al. (1965), tells what
regime you are in and which model to use. Just locate the point on Fig. 15.2
which corresponds to the fluid being used (Schmidt number), the flow conditions
(Reynolds number), and vessel geometry (LldJ. But be sure to check that your
system is not in turbulent flow. Remember that this chart only has meaning if
you have laminar flow. In this chart !3lud, is the reciprocal of the Bodenstein
number. It measures the flow contribution made by molecular diffusion. It is
NOT the axial dispersion number, Dlud, except in the pure diffusion regime.
The pure diffusion regime is not a very interesting regime because it represents
very very slow flow.
Gases are likely to be in the dispersion regime, not the pure convection regime.
Liquids can well be in one regime or another. Very viscous liquids such as
polymers are likely to be in the pure convection regime. If your system falls in
the no-man's-land between regimes, calculate the reactor behavior based on
340 Chapter 15 The Convection Model for Laminar Flow

Fluid close to the

wall moves slowly
I

Fastest flowing fluid

element is in the center

Figure 15.1 Flow of fluid according to the convec-

tion model.

the two bounding regimes and then try averaging. The numerical solution is
impractically complex to use.
Finally, it is very important to use the correct type of model because the RTD
curves are completely different for the different regimes. As an illustration, Fig.
15.3 shows RTD curves typical of these regimes.

Lld,

Figure 15.2 Map showing which flow models

should be used in any situation.
 15.1 The Convection Model and its RTD 341

The dispersion model

Pure convection glves gives a more or less
4 the same curve for all & distorted bell s h a ~ e d

Pure diffusion gives flow conditions

a pulse output at t = 0
(open-open vessel) \

0 0.5 1.0
f-Tracer in Mean for ail
at time zero three curves
Figure 15.3 Comparison of the RTD of the three models.

How to Tell from Experiment Which Model to Use

The sharpest way of experimentally distinguishing between models comes by
noting how a pulse or sloppy input pulse of tracer spreads as it moves downstream
in a flow channel. For example, consider the flow, as shown in Fig. 15.4. The
dispersion or tanks-in-series models are both stochastic models; thus, from Eq.
13.8 or Eq. 14.3 we see that the variance grows linearly with distance or

The convective model is a deterministic model; thus, the spread of tracer grows
linearly with distance, or

Whenever you have measurements of a at 3 points use this test to tell which
model to use. Just see if, in Fig. 15.4,

Figure 15.4 The changing spread of a tracer curve tells which model is the right
one to use.
342 Chapter 15 The Convection Model for Laminar Flow

Flux introduction Planar introduction

Proportional to velocity; Evenly distributed across pipe;

more tracer at centerline multiple injectors, a flash of
very little at the wall light on photosensitive fluid

Flux measurement Planar measurement

Mixing cup measurement; This could be a through-the-wall

catch all the exit fluid.
In essence this measures u C. measurement such as with a light
meter or radioactivity counter; also
a series of probes (for example
conductivity) across the tube.
This measures C a t an instant

Figure 15.5 Various ways of introducing and measuring tracer.

Pulse Response Experiment and the E Curve for Laminar Flow in Pipes
The shape of the response curve is strongly influenced by the way tracer is
introduced into the flowing fluid, and how it is measured. You may inject or
measure the tracer in two main ways, as shown in Fig. 15.5. We therefore have
four combinations of boundary conditions, as shown in Fig. 15.6, each with its
own particular E curve. These E curves are shown in Fig. 15.7.
As may be seen in Fig. 15.7, the E, E*, and E** curves are quite different,
one from the other.
E is the proper response curve for reactor purposes; it is the curve treated
in Chapter 11, and it represents the RTD in the vessel.
E* and *E are identical always, so we will call them E* from now on. One
correction for the planar boundary condition will transform this curve to
the proper RTD.
E** requires two corrections-one for entrance, one for exit-to transform
it to a proper RTD.
It may be simpler to determine E* or E** rather than E. This is perfectly all
right. However, remember to transform these measured tracer curves to the E
curve before calling it the RTD. Let us see how to make this transformation.

Flux flux Flux. planar Planar flux Planar. planar

Figure 15.6 Various combinations of input-output methods.

 15.1 The Convection Model and its R T D 343

Figure 15.7 Note how different are the output curves depending on how you introduce and
measure tracer.

For pipes and tubes with their parabolic velocity profile the various pulse
response curves are found to be as follows:
-
t2 t
E=- for t z -
2t3 2
and
1 1
E,=- for O r -
2 O3 2

- p* = w
E* = -t t
for tr -
2t2 2
and
1
E? = - 1
for 132-
2 O2 2

p** = CO

1
E**= - t
for t r -
2t 2
and
1
E$*= - 1
for O r -
2O 2

where p,p*,and p** are the mean values of the measured curves.
344 Chapter 15 The Convection Model for Laminar Flow

Flux flux Flux planar

Figure 15.8 Two different ways of measuring the output curves.

Note the simple relationship between E,E*,and E*".Thus at any time we

can write

E$*= 8 E$ = O2 EB
where 7 = -
v
v
t t2

Step Response Experiments and the F Curve for Laminar Flow in Pipes
When we do the step experiment by switching from one fluid to the other we
obtain the C,,,, curve (see Chapter ll),from which we should be able to find
the F curve. However, this input always represents the flux input, while the
output can be either planar or flux. Thus we only have two combinations, as
shown in Fig. [Link] these two combinations of boundary conditions their
equations and graphs are given in Eq. 6 and Fig. 15.9.

1
F=l-- for
4 O2
where 8= t
1
F*= I-- for 8 2 -
28

1
t Laminar flow in pipes
-------------....-.....---....-.--....

,g=&=L
t V/u

Figure 15.9 Different ways of measuring the out-

put gives different F curves.
 15.2 Chemical Conversion in Laminar Flow Reactors 345

Also each F curve is related to its corresponding E curve. Thus at any time t,
or 6,

F* = 1; E: dl = 101
0
* do and E:
E0 = or E: =@I
do 0,
(7)

The relationship is similar between E and F.

E Curves for Non-newtonians and for Non-circular Channels

Since plastics and nonnewtonians are often very viscous they usually should be
treated by the convective model of this chapter. The E, E*, and E** curves for
various situations besides newtonian fluids in circular pipes have been developed,
for example,
for power law fluids
for Bingham plastics
E curves have also been developed
for falling films
for flow between parallel plates
where line measurements are made rather than across the whole vessel
cross section.
These E equations and corresponding charts plus sources to various other analy-
ses can be found in Levenspiel, (1996).

15.2 CHEMICAL CONVERSION IN LAMINAR FLOW REACTORS

Single n-th Order Reactions
In the pure convection regime (negligible molecular diffusion) each element of
fluid follows its own streamline with no intermixing with neighboring elements.
In essence this gives macrofluid behavior. From Chapter 11the conversion expres-
sion is then

r 1
for zero reaction order

for first order reaction

for second order reaction

-CA

CA
-
CAO

C~~

c~
--
-
CAO
-I--
kt
CAO

= e-kt

1
1+ kCAOt
C ~ O
f o r t s --
k
346 Chapter 15 The Convection Model for Laminar Flow

For zero-order reaction of a newtonian in laminar flow in a pipe, integration of

Eq. 8 gives

For first-order reaction of a newtonian in laminar flow in a pipe,

where ei(y) is the exponential integral, see Chapter 16.

For second-order reaction of a newtonian in laminar flow in a pipe:

These performance expressions were first developed by Bosworth (1948) for

zero order, by Denbigh (1951) for second order, and by Cleland and Wilhelm
(1956) for first order reactions. For other kinetics, channel shapes, or types of
fluids insert the proper terms in the general performance expression and integrate.

Comments
(a) Test for the R T D curve. Proper RTD curves must satisfy the material
balance checks (calculated zero and first moments should agree with mea-
sured values)

0
E.~B=1 and 1' BE,dB
0
=1 (12)

The E curves of this chapter, for non-newtonians and all shapes of channels,
all meet this requirement. All the E* and E** curves of this chapter do
not; however, their transforms to E do.
(b) The variance and other R T D descriptors. The variance of all the E curves
of this chapter is finite; but it is infinite for all the E* and E** curves. So
be sure you know which curve you are dealing with.
In general the convection model E curve has a long tail. This makes
the measurement of its variance unreliable. Thus cr2 is not a useful parame-
ter for convection models and is not presented here.
The breakthrough time 8, is probably the most reliably measured and
most useful descriptive parameter for convection models, so it is widely
used.
(c) Comparison with plug flow for nth-order reaction is shown in Fig. 15.10.
This graph shows that even at high XA convective flow does not drastically lower
reactor performance. This result differs from the dispersion and tanks-in-series
models (see Chapters 13 and 14).
 15.2 Chemical Conversion in Laminar Flow Reactors 347

Convection versus -

1 I / I I /

Figure 15.10 Convective flow lowers conversion compared to plug flow.

Multiple Reaction in Laminar Flow

Consider a two-step first-order irreversible reactions in series

Because laminar flow represents a deviation from plug flow, the amount of
intermediate formed will be somewhat less than for plug flow. Let us examine
this situation.

PFR
LFR

PFR

LFR

M FR

Figure 15.11 Typical product distribution curves for laminar flow compared
with the curves for plug flow (Fig. 8.13) and mixed flow (Fig. 8.14).
348 Chapter 15 The Convection Model for Laminar Flow

The disappearance of A is given by the complicated Eq. 10, and the formation
and disappearance of R is given by an even more complicated equation. Devel-
oping the product distribution relationship, solving numerically, and comparing
the results with those for plug flow and for mixed flow gives Fig. 15.11; see
Johnson (1970) and Levien and Levenspiel (1998).
This graph shows that the LFR gives a little less intermediate than does the
PFR, about 20% of the way from the PFR to the MFR.
We should be able to generalize these findings to other more complex reaction
systems, such as for two component multistep reactions; to polymerizations; and
to nowNewtonian power law fluids.

REFERENCES

Ananthakrishnan, V., Gill, W. N., and Barduhn, A. J., AIChE J., 11,1063 (1965).
Bosworth, R. C. L., Phil. Mag., 39, 847 (1948).
Cleland, F. A., and Wilhelm, R. H., AIChE J., 2, 489 (1956).
Denbigh, K. G., J. Appl. Chem., 1,227 (1951).
Johnson, M. M., Ind. Eng. Chem. Fundamentals, 9, 681 (1970).
Levenspiel, O., Chemical Reactor Omnibook, Chap. 68, OSU Bookstores, Corvallis, OR
97339, 1996.
Levien, K. L., and Levenspiel, O., Chem. Eng. Sci., 54, 2453 (1999).

PROBLEMS

A viscous liquid is to react while passing through a tubular reactor in which flow
is expected to follow the convection model. What conversion can we expect in
this reactor if plug flow in the reactor will give 80% conversion?

15.1. Reaction follows zero-order kinetics.

15.2. Reaction is second order.

15.3. Assuming plug flow we calculate that a tubular reactor 12 m long would
give 96% conversion of A for the second-order reaction A --+R. However,
the fluid is very viscous, and flow will be strongly laminar, thus we expect
the convection model, not the plug flow model, to closely represent the
flow. How long should we make the reactor to insure 96% conversion of A?

15.4. Aqueous A (C,, = 1 mollliter) with physical properties close to water

(p = 1000 kg/m3, CB = m2/s) reacts by a first-order homogeneous
reaction (A -t R, k = 0.2 s-I) as it flows at 100 mm/s through a tubular
reactor (d, = 50 mm, L = 5 m). Find the conversion of A in the fluid
leaving this reactor.
 Problems 349

15.5. Aqueous A (CAO= 50 mol/m3) with physical properties close to water ( p =

1000 kg/m3, !3 = m2/s) reacts by a second-order reaction (k =
m3/mol.s) as it flows at 10 mmls through a tubular reactor (d, = 10 mm,
L = 20 m). Find the conversion of reactant A from this reactor.

15.6. We want to model the flow of fluid in a flow channel. For this we locate
three measuring points A, B, and C, 100 m apart along the flow channel.
We inject tracer upstream of point A, fluid flows past points A, B, and C
with the following results:
At A the tracer width is 2 m
At B the tracer width is 10 m
At C the tracer width is 14 m
What type of flow model would you try to use to represent this flow:
dispersion, convective, tanks-in-series, or none of these? Give a reason for
your answer.
Chapter 1
Earliness of Mixing,
Segregation, and RTD

The problem associated with the mixing of fluids during reaction is important for
extremely fast reactions in homogeneous systems, as well as for all heterogeneous
systems. This problem has two overlapping aspects: first, the degree of segregation
of the fluid, or whether mixing occurs on the microscopic level (mixing of individ-
ual molecules) or the macroscopic level (mixing of clumps, groups, or aggregates
of molecules); and second, the earliness of mixing, or whether fluid mixes early
or late as it flows through the vessel.
These two concepts are intertwined with the concept of RTD, so it becomes
rather difficult to understand their interaction. Please reread the first few pages
of Chapter 11 where these concepts are introduced and discussed.
In this chapter we first treat systems in which a single fluid is reacting. Then
we treat systems in which two fluids are contacted and reacted.

16.1 SELF-MIXING OF A SINGLE FLUID

Degree of Segregation
The normally accepted state of a liquid or gas is that of a microfluid, and
all previous discussions on homogeneous reactions have been based on the
assumption. Let us now consider a single reacting macrofluid being processed
in turn in batch, plug flow, and mixed flow reactors, and let us see how this state
of aggregation can result in behavior different from that of a microfluid.

Batch Reactor. Let the batch reactor be filled with a macrofluid containing
reactant A. Since each aggregate or packet of macrofluid acts as its own little
batch reactor, conversion is the same in all aggregates and is in fact identical to
what would be obtained with a microfluid. Thus for batch operations the degree
of segregation does not affect conversion or product distribution.

Plug Flow Reactor. Since plug flow can be visualized as a flow of small batch
reactors passing in succession through the vessel, macro- and microfluids act
 16.1 Self-Mixing of a Single Fluid 351

Individual molecules lose their Each aggregate retains its identity

identity, and reactant and acts as a batch reactor;
concentration in uniform reactant concentration varies from
throughout aggregate to aggregate
Figure 16.1 Difference in behavior i3f microfluids and macrofluids in mixed
flow reactors.

alike. Consequently the degree of segregation does not influence conversion or

product distribution.

Mixed Flow Reactor-Microfluid. When a microfluid containing reactant A is

treated as in Fig. 16.1, the reactant concentration everywhere drops to the low
value prevailing in the reactor. No clump of molecules retains its high initial
concentration of A. We may characterize this by saying that each molecule loses
its identity and has no determinable past history. In other words, by examining
its neighbors we cannot tell whether a molecule is a newcomer or an old-timer
in the reactor.
For this system the conversion of reactant is found by the usual methods for
homogeneous reactions, or

or, with no density changes,

where 7 is the mean residence time of fluid in the reactor.

Mixed Flow Reactor-Macrofluid. When a macrofluid enters a mixed flow reac-

tor, the reactant concentration in an aggregate does not drop immediately to a
low value but decreases in the same way as it would in a batch reactor. Thus a
molecule in a macrofluid does not lose its identity, its past history is not unknown,
and its age can be estimated by examining its neighboring molecules.
The performance equation for a macrofluid in a mixed flow reactor is given
by Eq. 11.13 as
-
--=I (-)
1 - X A - CA
CAO 0
CA
CAObatch
Edt
352 Chapter 16 Earliness of Mixing, Segregation, and RTD

where

V
E dt = -
v
-
e-"/V dt = e-t/i dt
t

Replacing Eq. 3 in Eq. 2 gives

This is the general equation for determining conversion of macrofluids in mixed

flow reactors, and it may be solved once the kinetics of the reaction is given.
Consider various reaction orders.
For a zero-order reaction in a batch reactor, Chapter 3 gives

Inserting into Eq. 4 and integrating gives

For afirst-order reaction in a batch, reactor Chapter 3 gives

On replacing into Eq. 4 we obtain

which on integration gives the expression for conversion of a macrofluid in a

mixed flow reactor

This equation is identical to that obtained for a microfluid; for example, see Eq.
5.14a. Thus we conclude that the degree of segregation has no effect on conversion
for first-order reactions.
 16.1 Self-Mixing of a Single Fluid 353
For a second-order reaction of a single reactant in a batch reactor Eq. 3.16 gives

On replacing into Eq. 4 we find

and by letting a = l/CAoktand converting into reduced time units 0 = tlt, this
expression becomes

This is the conversion expression for second-order reaction of a macrofluid in a

mixed flow reactor. The integral, represented by ei(a) is called an exponential
integral. It is a function alone of a , and its value is tabulated in a number of
tables of integrals. Table 16.1 presents a very abbreviated set of values for both
ei(x) and Ei(x). We will refer to this table later in the book.

Table 16.1 Two of the Family of Exponential Integrals

x2 x3
~i(x)= eYdu=0.57721+lnx+x+-+-+. .
Here are two useful 2-2! 3.3
exponential integrals x2 x3
-0.57721-lnx+x-- +--. . .
2.2! 3.3!

x Ei(x) ei(x) x Ei (x)

0 -to +m 0.2 -0.8218
0.01 -4.0179 4.0379 0.3 -0.3027
0.02 -3.3147 3.3547 0.5 0.4542
0.05 -2.3679 2.4679 1.0 1.8951
0.1 -1.6228 1.8229 1.4 3.0072

Reference: "Tables of Sines, Cosines and Exponential Integrals," Vols. I and 11, by WPA, for NBS (1940).
354 Chapter 16 Earliness of Mixing, Segregation, and RTD

Equation 8 may be compared with the corresponding expression for microflu-

ids, Eq. 5.14

For an nth-order reaction the conversion in a batch reactor can be found by

the methods of Chapter 3 to be

Insertion into Eq. 4 gives the conversion for an nth-order reaction of a macrofluid.

Difference in Performance: Early or Late Mixing, Macro- or Microfluids,

PFR or MFR
Figure 16.2 illustrates the difference in performance of macrofluids and microflu-
ids in mixed flow reactors, and they show clearly that a rise in segregation
improves reactor performance for reaction orders greater than unity but lowers
performance for reaction orders smaller than unity. Table 16.2 was used in
preparing these charts.

Early and Late Mixing of Fluids

Each flow pattern of fluid through a vessel has associated with it a definite clearly
defined residence time distribution (RTD), or exit age distribution function E.
The converse is not true, however. Each RTD does not define a specific flow
pattern; hence, a number of flow patterns-some with earlier mixing, others with
later mixing of fluids-may be able to give the same RTD.

Idealized Pulse RTD. Reflection shows that the only pattern of flow consistent
with this RTD is one with no intermixing of fluid of different ages, hence, that
of plug flow. Consequently it is immaterial whether we have a micro- or macro-
fluid. In addition the question of early or late mixing of fluid is of no concern
since there is no mixing of fluid of different ages.

Exponential Decay RTD. The mixed flow reactor can give this RTD. However,
other flow patterns can also give this RTD, for example, a set of parallel plug
flow reactors of proper length, a plug flow reactor with sidestreams, or a combina-
tion of these. Figure 16.3 shows a number of these patterns. Note that in patterns
a and b entering fluid elements mix immediately with material of different ages,
while in patterns c and d no such mixing occurs. Thus patterns a and b represent
the microfluid, while patterns c and d represent the macrofluid.
 16.1 Self-Mixing of a Single Fluid 355

Figure 16.2 Comparison of performance of mixed flow reactors treat-

ing micro- and macrofluids; for zero- and second-order reactions
with = 0.

Arbitrary RTD. In extending this argument we see that when the RTD is close
to that of plug flow, then the state of segregation of fluid as well as early or late
mixing of fluid has little effect on conversion. However, when the RTD ap-
proaches the exponential decay of mixed flow, then the state of segregation and
earliness of mixing become increasingly important.
For any RTD one extreme in behavior is represented by the macrofluid and
the latest mixing microfluid, and Eq. 2 gives the performance expression for this
case. The other extreme is represented by the earliest mixing microfluid. The
performance expression for this case has been developed by Zwietering (1959)
but is difficult to use. Although these extremes give the upper and lower bound
to the expected conversion for real vessels, it is usually simpler and preferable
to develop a model to reasonably approximate the real vessel and then calculate
conversions from this model. This is what actually was done in developing Fig.
13.20 for second-order reaction with axial dispersion. There we assumed that
the same extent of mixing occurs along the whole vessel.
%
Ch
Table 16.2 Conversion Equations for Macrofluids and Microfluids with E = 0 in Ideal Reactors
Plug Flow Mixed Flow
Microfluid or Macrofluid Microfluid Macrofluid

General kinetics c
CO
-
= ' (")
7
Jm

0 CO batch
e - l l ~ dt

nth-order reaction

( R = Ci;-'k~)

First-order reaction

( R = k7)

Second-order reaction c- 1 c --
- - 1 + m
co
( R = Cok7)
C,

R = -c
l+R
o- 1
2R -
C - ellR
- - -ei
co R
(k)
C
- -

R = C i - ' k ~reaction
, rate group for nth-order reaction, a time or capacity factor. T = i since E = 0 throughout.
 16.1 Self-Mixing of a Single Fluid 357

Mixedflow - /

<
, v ! ! l - * \
:---+
;C--T
;
\\ 1
- F d

Set of parallel
plug flow
(c)

,- Side leaving
plug flow

L S i d e feeding
plug flow
(6) (d)
Figure 16.3 Four contacting patterns which can all give the same expo-
nential decay RTD. Cases a and b represent the earliest possible mixing
while cases c and d represent the latest possible mixing of fluid elements
of different ages.

Summary of Findings for a Single Fluid

1. Factors affecting the performance of a reactor. In general we may write

kinetics, RTD, degree of segregation,

Performance: XAor cp
(g) ( = earliness of mixing

2. Effect of kinetics, or reaction order. Segregation and earliness of mixing

affect the conversion of reactant as follows

For n > 1 - . . Xmacro and Xmicro, late > Xmicro, early

For n < 1the inequality is reversed, and for n = 1conversion is unaffected

by these factors. This result shows that segregation and late mixing improves
conversion for n > 1, and decreases conversion for n < 1.
3. Effect of mixing factors for nonfirst-order reactions. Segregation plays no
role in plug flow; however, it increasingly affects the reactor performance
as the RTD shifts from plug to mixed flow.
4. Effect of conversion level. At low conversion levels X, is insensitive to
RTD, earliness of mixing, and segregation. At intermediate conversion
levels, the RTD begins to influence XA;however, earliness and segregation
still have little effect. This is the case of Example 16.1. Finally, at high
conversion levels all these factors may play important roles.
5. Effect on product distribution. Although segregation and earliness of mixing
can usually be ignored when treating single reactions, this often is not
so with multiple reactions where the effect of these factors on product
distribution can be of dominating importance, even at low conversion levels.
358 Chapter 16 Earliness of Mixing, Segregation, and RTD

As an example consider free-radical polymerization. When an occasional

free radical is formed here and there in the reactor it triggers an extremely
rapid chain of reactions, often thousands of steps in a fraction of a second.
The local reaction rate and conversion can thus be very high. In this situation
the immediate surroundings of the reacting and growing molecules-and
hence the state of segregation of the fluid-can greatly affect the type of
polymer formed.

EFFECT OF SEGREGATION AND EARLINESS OF MIXING

O N CONVERSION

A second-order reaction occurs in a reactor whose RTD is given in Fig. E16.1.

Calculate the conversion for the flow schemes shown in this figure. For simplicity
take C, = 1, k = 1, and r = 1 for each unit.

Microfluid Macrofluid
co
=-,
I
3
I ,:

To approx~matethe

(c)

Figure E16.1 ( a ) Microfluid, early mixing at molecular level; (b) Mi-

crofluid, fairly late mixing at molecular level; (c) Microfluid, late mixing
at molecular level; ( d ) Mac~ofluid,early mixing of elements; ( e ) Mac-
rofluid, late mixing of elements.
 16.1 Self-Mixing of a Single Fluid 359

I SOLUTION
Scheme A. Referring to Fig. E 1 6 . 1 ~we have for the mixed flow reactor

For the plug flow reactor

Scheme B. Referring to Fig. [Link], we have for the plug flow reactor

For the mixed reactor

Micro-fairly late: C; = 0.366

Scheme C, D, and E. From Fig. 12.1 the exit age distribution function for the
two equal-size plug-mixed flow reactor system is

2 el-2t1t -
E == t 1
, when T > -
t t 2
t 1
= 0, when Y < -
t 2
 360 Chapter 16 Earliness of Mixing, Segregation, and RTD

I Thus Eq. 3 becomes

With the mean residence time in the two-vessel system f = 2 min, this becomes

and replacing 1 + t by x we obtain the exponential integral

l
I From the table of integrals in Table 16.1 we find ei(2) =

Micro-late, and macro-late or early: C" = 0.362

0.048 90 from which

The results of this example confirm the statements made above: that macrofluids
and late mixing microfluids give higher conversions than early mixing microfluids
for reaction orders greater than unity. The difference is small here because the
conversion levels are low; however, this difference becomes more important as
conversion approaches unity.

Extensions for a Single Fluid

Partial Segregation. There are various ways of treating intermediate extents
of segregation, for example,
Intensity of segregation model-Danckwerts (1958)
Coalescence model-Curl (1963), Spielman and Levenspiel (1965)
Two environment and melting ice cube models-Ng and Rippin (1965) and
Suzuki (1970)
These approaches are discussed in Levenspiel (1972).

The Life of an Element of Fluid. Let us estimate how long a fluid element
retains its identity. First, all large elements are broken into smaller elements by
stretching or folding (laminar behavior) or by turbulence generated by baffles,
stirrers, etc., and mixing theory estimates the time needed for this breakup.
Small elements lose their identity by the action of molecular diffusion, and the
Einstein random walk analysis estimates this time as

(size of element)' -
--d2lement
t=
( diffusion \ !ZJ
{coefficient)
 16.2 Mixing of Two Miscible Fluids 361
Thus an element of water 1 micron in size would lose its identity in a very short
time, approximately

~ m ) ~
t= = sec
cm2/sec

while an element of viscous polymer 1.0 mm in size and 100 times as viscous as
water (10-30 W motor oil at room temperature) would retain its identity for a
long time, roughly

(10-I ~ m ) ~
t= = lo5sec .=; 30 hr
cm2/sec

In general, then, ordinary fluids behave as microfluids except for very viscous
materials and for systems in which very fast reactions are taking place.
The concept of micro- and macrofluids is of particular importance in heteroge-
neous systems because one of the two phases of such systems usually approxi-
mates a macrofluid. For example, the solid phase of fluid-solid systems can be
treated exactly as a macrofluid because each particle of solid is a distinct aggregate
of molecules. For such systems, then, Eq. 2 with the appropriate kinetic expression
is the starting point for design.
In the chapters to follow we apply these concepts of micro- and macrofluids
to heterogeneous systems of various kinds.

16.2 MIXING OF TWO MISCIBLE FLUIDS

Here we consider one topic, the role of the mixing process when two completely
miscible reactant fluids A and B are brought together. When two miscible fluids
A and B are mixed, we normally assume that they first form a homogeneous
mixture which then reacts. However, when the time required for A and B to
become homogeneous is not short with respect to the time for reaction to take
place, reaction occurs during the mixing process, and the problem of mixing
becomes important. Such is the case for very fast reactions or with very viscous
reacting fluids.
To help understand what occurs, imagine that we have A and B available,
each first as a microfluid, and then as a macrofluid. In one beaker mix micro A
with micro B, and in another beaker mix macro A with macro B and let them
react. What do we find? Micro A and B behave in the expected manner, and
reaction occurs. However, on mixing the macrofluids no reaction takes place
because molecules of A cannot contact molecules of B. These two situations are
illustrated in Fig. 16.4. So much for the treatment of the two extremes in behavior.
Now a real system acts as shown in Fig. 16.5 with regions of A-rich fluid and
regions of B-rich fluid.
Though partial segregation requires an increase in reactor size, this is not the
only consequence. For example, when reactants are viscous fluids, their mixing
in a stirred tank or batch reactor often places layers or "streaks" of one fluid
next to the other. As a result reaction occurs at different rates from point to
362 Chapter 16 Earliness of Mixing, Segregation, and RTD

A B A B
C-

reaction occurs no reaction occurs

Figure 16.4 Difference in behavior of microfluids and macrofluids

in the reaction of A and B.

point in the reactor, giving a nonuniform product which may be commercially

unacceptable. Such is the case in polymerization reactions in which monomer
must be intimately mixed with a catalyst. For reactions such as this, proper
mixing is of primary importance and often the rate of reaction and product
uniformity correlate well with the mixing energy input to the fluid.
For fast reactions the increase in reactor size needed because of segregation
is of secondary importance while other effects become important. For example,
if the product of reaction is a solid precipitate, the size of the precipitate particles
may be influenced by the rate of intermixing of reactants, a fact that is well
known from the analytical laboratory. As another example, hot gaseous reaction
mixtures may contain appreciable quantities of a desirable compound because
of favorable thermodynamic equilibrium at such temperatures. To reclaim this
component the gas may have to be cooled. But, as is often the case, a drop in
temperature causes an unfavorable shift in equilibrium with essentially complete
disappearance of desired material. To avoid this and to "freeze" the composition
of hot gases, cooling must be very rapid. When the method of quenching used
involves mixing the hot gases with an inert cold gas, the success of such a
procedure is primarily dependent on the rate at which segregation can be de-
stroyed. Finally the length, type, and temperature of a burning flame, the combus-
tion products obtained, the noise levels of jet engines, and the physical properties
of polymers as they are affected by the molecular weight distribution of the

A r A-rich region

L ~ - r i c h region B/ B-rich region

Figure 16.5 Partial segregation in the mixing of two miscible fluids in a reactor.
 16.2 Mixing of Two Miscible Fluids 363
material are some of the many phenomena that are closely influenced by the
rate and intimacy of fluid mixing.

Product Distribution in Multiple Reactions

When multiple reactions take place on mixing two reactant fluids and when
these reactions proceed to an appreciable extent before homogeneity is attained,
segregation is important and can affect product distribution.
Consider the homogeneous-phase competitive consecutive reactions

occurring when A and B are poured into a batch reactor. If the reactions are
slow enough so that the contents of the vessel are uniform before reaction takes
place, the maximum amount of R formed is governed by the k,lk, ratio. This
situation, treated in Chapter 8, is one in which we may assume microfluid behav-
ior. If, however, the fluids are very viscous or if the reactions are fast enough,
they will occur in the narrow zones between regions of high A concentration
and high B concentration. This is shown in Fig. 16.6. The zone of high reaction
rate will contain a higher concentration of R than the surrounding fluid. But
from the qualitative treatment of this reaction in Chapter 8 we know that any
nonhonrogeneity in A and R will depress formation of R. Thus partial segregation
of reaotants will depress the formation of intermediate.
For increased reaction rate, the zone of reaction narrows, and in the limit, for
an infinitely fast reaction, becomes a boundary surface between the A-rich and

L- High-reaction-rate zone
containing a high
concentration of R
Figure 16.6 When reaction rate is
very high, zones of nonhomogene-
ity exist in a reactor. This condition
is detrimental to obtaining high
yields of intermediate R from
the reactions
364 Chapter 16 Earliness of Mixing, Segregation, and RTD

Fast reaction Infinitely fast reaction

(some R formed) (no R formed)

A-r1ch
reglon 'a region A-rich
region
\ regi0n

Reaction zone

Reaction zone

J is a plane

I A-rich
region I
I
!
B-rich
region

Distance Distance
Figure 16.7 Concentration profiles of the components of the reactions

at a representative spot in the reactor between A-rich and B-rich fluid for a very
fast and for an infinitely fast reaction.

B-rich regions. Now R will only be formed at this plane. What will happen to
it? Consider a single molecule of R formed at the reaction plane. If it starts its
random wanderings (diffusion) into the A zone and never moves back into the
B zone, it will not react further and will be saved. However, if it starts off into
the B zone or if at any time during its wanderings it moves through the reaction
plane into the B zone, it will be attacked by B to form S. Interestingly enough,
from probabilities associated with a betting game treated by Feller (1957), we
can show that the odds in favor of a molecule of R never entering the B zone
become smaller and smaller as the number of diffusion steps taken by a molecule
gets larger and larger. This finding holds, no matter what pattern of wanderings
is chosen for the molecules of R. Thus we conclude that no R is formed. Looked
at from the point of view of Chapter 8, an infinitely fast reaction gives a maximum
nonhomogeneity of A and R in the mixture, resulting in no R being formed.
Figure 16. 7 shows the concentration of materials at a typical reaction interface
and illustrates these points.
 References 365

Figure 16.8 Stretching, folding, and

thinning of sheets of very viscous A
and B.

This behavior of multiple reaction could provide a powerful tool in the study
of partial segregation in homogeneous systems. It has been used by Paul and
Treybal(1971) who simply poured reactant B into a beaker of A and measured
the amount of R formed for a very fast reaction of Eq. 14.
Ottino (1989, 1994) discuss the whole problem of intermixing of fluids A and
B in terms of stretching, folding, thinning, and finally diffusional mixing of fluid
elements. Figure 16.8 tries to illustrate this mechanism; however, we must now
leave this fascinating subject.
These observations serve as a guide to the selection and design of equipment
favoring the formation of intermediate when reaction is very fast. The important
point is to achieve homogeneity in A and R throughout the reaction mixture
before reaction has proceeded to any significant extent. This is done by:
(a) making the reaction zone as large as possible by vigorous mixing.
(b) dispersing B in A in as fine a form as possible, rather than A in B.
(c) slowing the reaction.

REFERENCES

Curl, R. L., AZChE J . , 9, 175 (1963).

Danckwerts, P. V., Chem. Eng. Sci., 8, 93 (1958).
Feller, W., A n Introduction to Probability Theory and its Applications, Vol I , 2nd ed.,
John Wiley & Sons, New York, 1957, p. 254.
Levenspiel, O., Chemical Reaction Engineering, 2nd ed., Chap. 10, John Wiley & Sons,
New York, 1972.
Ng, D. Y. C., and Rippin, D. W. T., Third Symposium on Chemical Reaction Engineering,
Pergamon Press, Oxford, 1965.
Ottino, J. M., Scientific American, 56, Jan. 1989.
,- Chem. Eng. Sci., 49, 4005 (1994).
Paul, E. L., and Treybal, R. E., AZChE J., 17,718 (1971).
Spielman, L. A., and Levenspiel, O., Chem. Eng. Sci., 20,247 (1965).
Suzuki, M., personal communication, 1970.
Zweitering, Th. N., Chem. Eng. Sci., 11, 1 (1959).
Part I11

Reactions Catalyzed by Solids

Chapter 17 Heterogeneous Reactions-Introduction 1369

Chapter 18 Solid Catalyzed Reactions 1376
Chapter 19 The Packed Bed Catalytic Reactor 1427
Chapter 20 Reactors with Suspended Solid Catalyst, Fluidized Reactors of
Various Types 1447
Chapter 21 Deactivating Catalysts 1473
Chapter 22 GIL Reactions on Solid Catalyst: Trickle Beds, Slurry Reactors,
Three Phase Fluidized Beds 1500
Chapter 17
Heterogeneous
Reactions-Introduction

The second half of this book treats the kinetics and design of chemical reactors
for heterogeneous systems of various kinds, each chapter considering a different
system (see Chapter 1 for discussions of heterogeneous and homogeneous sys-
tems). For these systems there are two complicating factors that must be ac-
counted for beyond what is normally considered in homogeneous systems. First,
we have the complication of the rate expression, and second the complication
of the contacting patterns for two-phase systems. Let us briefly discuss these
in turn.

The Complications of the Rate Equation. Since more than one phase is present,
the movement of material from phase to phase must be considered in the rate
equation. Thus the rate expression in general will incorporate mass transfer
terms in addition to the usual chemical kinetics term. These mass transfer terms
are different in type and numbers in the different kinds of heterogeneous systems;
hence, no single rate expression has general application. Here are some sim-
ple examples.

THE BURNING OF A CARBON PARTICLE IN AIR

Tell how many rate steps are involved. The kinetics is given by

C + O,+CO,

and ignore the possible formation of CO.

From Fig. E17.1 we see that two steps in series are involved-mass transfer of
oxygen to the surface followed by reaction at the surface of the particle.
369
370 Chapter 17 Heterogeneous Reactions-Introduction

J'
r,sJ;,.

' 3 2
-
Model

Burning carbon particle

Gas film

Fig. E17.1

AEROBIC FERMENTATION

w many rate steps are involved when air bubbles through a tank of liquid

product material.

From Fig. E17.2 we see that there are up to seven possible resistance steps, only
one involving the reaction. How many you choose to consider depends on you
and on the situation.

To get an overall rate expression, write the individual rate steps on the same
basis (unit surface of burning particle, unit volume of fermenter, unit volume of
cells, etc.).

1 dNA - mol A reacted

-TA=
V dt volume of reactor fluid. time
 Chapter 17 Heterogeneous Reactions-Introduction 371

1 dNA - mol A reacted

-ri= W dt mass of solid. time

1d N A - mol A reacted
-rL= S dt interfacial surface. time

Now put all the mass transfer and reaction steps into the same rate form and
then combine. Thus

mol A reacted -
- (-rA)v = ( - r i ) W = (-rL)S
time

and if the steps are in series, as in Examples 17.1 and 17.2

roverall = r1 = r2 = r3

If they are in parallel

Consider steps in series. In general, if all the steps are linear in concentration,
then it is easy to combine them. However, if any of the steps are nonlinear, then
you will get a messy overall expression. Therefore, you may try to bypass this
nonlinear step in one of various ways. Approximating the r, versus C, curve by
a first-order expression is probably the most useful procedure.
Another point: in combining rates we normally do not know the concentration
of materials at intermediate conditions, so these are the concentrations that we
eliminate in combining rates. Example 17.3 shows this.
 372 Chapter 17 Heterogeneous Reactions-Introduction

0 VERALL RATE FOR A LINEAR PROCESS

Dilute A diffuses through a stagnant liquid film onto a plane surface consisting
of B, reacts there to produce R which diffuses back into the mainstream. Develop
the overall rate expression for the L/S reaction

A(1) + B(s) -+ R(1)

which takes place on this flat surface, see Fig. E17.3.

Figure E17.3.

SOLUTION

I By diffusion, the flux of A to the surface is

Reaction is first order with respect to A, so based on unit surface

At steady state the flow rate to the surface is equal to the reaction rate at the
surface (steps in series). So

and from Eqs. (i) and (ii)

 Chapter 17 Heterogeneous Reactions-Introduction 373

from which

(iii)

Replacing Eq. (iii) into either Eq. (i) or Eq. (ii) then eliminates C, which cannot
be measured, giving

Comment This result shows that llk, and llk" are additive resistances. It so
happens that the addition of resistances to obtain on overall resistance is permissi-
ble only when the rate is a linear function of the driving force and when the
processes occur in series.

OVERALL RATE FOR A NONLINEAR PROCESS

Repeat Example 17.3 with just one change: let the reaction step be second order
with respect to A, or

1 SOLUTION
Combining the reaction steps to eliminate C, as was done in Example 17.3, is
now not so simple, and gives

Contacting Patterns for Two-Phase Systems

There are many ways that two phases can be contacted, and for each the design
equation will be unique. Design equations for these ideal flow patterns may
be developed without too much difficulty. However, when real flow deviates
considerably from these, we can do one of two things: we may develop models
to mirror actual flow closely, or we may calculate performance with ideal patterns
which "bracket" actual flow. Fortunately, most real reactors for heterogeneous
systems can be satisfactorily approximated by one of the five ideal flow patterns
of Fig. 17.1. Notable exceptions are the reactions which take place in fluidized
beds. There special models must be developed.
374 Chapter 17 Heterogeneous Reactions-Introduction

Plug Alplug B Plug Alplug B Plug Alplug B

countercurrent cocurrent crosscurrent

Micro-micro

Reaction in either
A or B phase

Reaction in either

Plug Almixed B Mixed Alrnixed B

(two cases) (two cases)

Figure 17.1. Ideal contacting patterns for two flowing fluids.

Final Thoughts on Flow Modeling

In reactor design and scale-up, it is essential to select a flow model which reason-
ably represents our setup. Too often we put too little thought here, carelessly
picking a nonrepresentative model and then doing computer calculations to the
nth degree of accuracy. And then we are surprised when design and scale-up do
not agree with our predictions. A simple reasonable model is much better than
a precise and detailed model which does not represent the contacting. Often the
choice of a good flow model and the knowledge of how the flow pattern changes
with scale-up spells the difference between success and failure.
The preceding thoughts guide us in our approach to heterogeneous reaction,
which we will consider in the rest of the book.

PROBLEMS

17.1. Gaseous reactant A diffuses through a gas film and reacts on the surface
of a solid according to a reversible first-order rate,

where CA, is the concentration of A in equilibrium with the solid surface.

Develop an expression for the rate of reaction of A accounting for both
the mass transfer and reaction steps.
 Problems 375
17.2. Example 17.4 gives the final rate expression for film mass transfer followed
by a second-order rate expression for reaction on a plane surface. Please
derive this expression and show that it is correct.

17.3. In slurry reactors, pure reactant gas is bubbled through liquid containing
suspended catalyst particles. Let us view these kinetics in terms of the film
theory, as shown in Fig. P17.3. Thus, to reach the surface of the solid, the

Typical liquid-

Figure P17.3

reactant which enters the liquid must diffuse through the liquid film into
the main body of liquid, and then through the film surrounding the catalyst
particle. At the surface of the particle, reactant yields product according
to first-order kinetics. Derive an expression for the rate of reaction in terms
of these resistances.
Chapter 18
Solid Catalyzed Reactions

With many reactions, the rates are affected by materials which are neither re-
actants nor products. Such materials called catalysts can speed a reaction by a
factor of a million or much more, or they may slow a reaction (negative catalyst).
There are two broad classes of catalysts: those that operate at close to ambient
temperature with biochemical systems, and the man-made catalysts that operate
at high temperature.
The biochemical catalysts, called enzymes, are found everywhere in the bio-
chemical world and in living creatures, and without their action I doubt that life
could exist at all. In addition, in our bodies hundreds of different enzymes and
other catalysts are busily at work all the time, keeping us alive. We treat these
catalysts in Chapter 27.
The man-made catalysts, mostly solids, usually aim to cause the high-tempera-
ture rupture or synthesis of materials. These reactions play an important role in
many industrial processes, such as the production of methanol, sulfuric acid,
ammonia, and various petrochemicals, polymers, paints, and plastics. It is esti-
mated that well over 50% of all the chemical products produced today are made
with the use of catalysts. These materials, their reaction rates, and the reactors
that use them are the concern of this chapter and Chapters 19-22.
Consider petroleum. Since this consists of a mixture of many compounds,
primarily hydrocarbons, its treatment under extreme conditions will cause a
variety of changes to occur simultaneously, producing a spectrum of compounds,
some desirable, others undesirable. Although a catalyst can easily speed the rate
of reactions a thousandfold or a millionfold, still, when a variety of reactions
are encountered, the most important characteristic of a catalyst is its selectivity.
By this we mean that it only changes the rates of certain reactions, often a single
reaction, leaving the rest unaffected. Thus, in the presence of an appropriate
catalyst, products containing predominantly the materials desired can be obtained
from a given feed.
The following are some general observations.

1. The selection of a catalyst to promote a reaction is not well understood;

therefore, in practice extensive trial and error may be needed to produce
a satisfactory catalyst.
 18.1 The Rate Equation for Surface Kinetics 377
2. Duplication of the chemical constitution of a good catalyst is no guarantee
that the solid produced will have any catalytic activity. This observation
suggests that it is the physical or crystalline structure which somehow imparts
catalytic activity to a material. This view is strengthened by the fact that
heating a catalyst above a certain critical temperature may cause it to lose
its activity, often permanently. Thus present research on catalysts is strongly
centered on the surface structure of solids.
3. To explain the action of catalysts, it is thought that reactant molecules are
somehow changed, energized, or affected to form intermediates in the
regions close to the catalyst surface. Various theories have been proposed
to explain the details of this action. In one theory, the intermediate is viewed
as an association of a reactant molecule with a region of the surface; in
other words, the molecules are somehow attached to the surface. In another
theory, molecules are thought to move down into the atmosphere close to
the surface and be under the influence of surface forces. In this view the
molecules are still mobile but are nevertheless modified. In still a third
theory, it is thought that an active complex, a free radical, is formed at the
surface of the catalyst. This free radical then moves back into the main gas
stream, triggering a chain of reactions with fresh molecules before being
finally destroyed. In contrast with the first two theories, which consider the
reaction to occur in the vicinity of the surface, this theory views the catalyst
surface simply as a generator of free radicals, with the reaction occurring
in the main body of the gas.
4. In terms of the transition-state theory, the catalyst reduces the potential
energy barrier over which the reactants must pass to form products. This
lowering in energy barrier is shown in Fig. 18.1.
5. Though a catalyst may speed up a reaction, it never determines the equilib-
rium or endpoint of a reaction. This is governed by thermodynamics alone.
Thus with or without a catalyst the equilibrium constant for the reaction
is always the same.

Without catalyst the complex

has high potential energy
resulting in low rate of reaction

With catalyst the lower

energy barrier allows
higher rate of reaction

Final
- I state

Reactants Complex Products

*
Reaction path

Figure 18.1 Representation of the action of a catalyst.

378 Chapter 18 Solid Catalyzed Reactions

6. Since the solid surface is responsible for catalytic activity, a large readily
accessible surface in easily handled materials is desirable. By a variety of
methods, active surface areas the size of football fields can be obtained per
cubic centimeter of catalyst.
Though there are many problems related to solid catalysts, we consider only
those which are related to the development of kinetic rate equations needed in
design. We simply assume that we have a catalyst available to promote a specific
reaction. We wish to evaluate the kinetic behavior of reactants in the presence
of this material and then use this information for design.

The Spectrum of Kinetic Regimes

Consider a porous catalyst particle bathed by reactant A. The rate of reaction
of A for the particle as a whole may depend on:
O Surface kinetics, or what happens at the surfaces, interior or exterior of
the particle. This may involve the adsorption of reactant A onto the surface,
reaction on the surface, or desorption of product back into the gas stream.
O Pore diffusion resistance which may cause the interior of the particle to
be starved for reactant.
O Particle AT or temperature gradients within the particle. This is caused by
large heat release or absorption during reaction.
0 Film AT between the outer surface of the particle and the main gas stream.
For example, the particle may be uniform in temperature throughout but
hotter than the surrounding gas.
0 Film diffusion resistance or concentration gradients across the gas film
surrounding the particle.
For gaslporous catalyst systems slow reactions are influenced by O alone, in
faster reactions O intrudes to slow the rate, then O and/or 0enter the picture,
O unlikely limits the overall rate. In liquid systems the order in which these
effects intrude is O,0, 0 , and rarely O andlor 0.
In different areas of application (outside of catalytic kinetics too) different
combinations of these five factors enter the picture. Table 18.1 shows what we
normally encounter.

Table 18.1 Factors which Influence the Rate of Reaction of Particles

Porous Catalyst Burning of Cells and
Rate Influencing Catalyst Coated a Droplet Simple Living
Factor Particle Surface of Fuel Creatures
O Surface reaction Yes Yes No Yes
O Pore diffusion Yes No No Maybe
O Particle AT Not too likely No No No
0 Film AT Sometimes Rare All important No
O Film mass transfer No Yes All important Could be
 18.1 The Rate Equation for Surface Kinetics 379

Although here we introduce all the phenomena which affect the rate, the real
world is never so exciting that we have to concern ourselves with all five factors
at any one time. In fact, in the majority of situations with porous catalyst particles
we only have to consider factors @ and 0.
Here let us treat factors O and 0;then O and @ briefly.

18.1 THE RATE EQUATION FOR SURFACE KINETICS

Because of the great industrial importance of catalytic reactions, considerable
effort has been spent in developing theories from which kinetic equations can
rationally be developed. The most useful for our purposes supposes that the
reaction takes place on an active site on the surface of the catalyst. Thus three
steps are viewed to occur successively at the surface.

Step 1. A molecule is adsorbed onto the surface and is attached to an active site.

Step 2. It then reacts either with another molecule on an adjacent site (dual-
site mechanism), with one coming from the main gas stream (single-site mecha-
nism), or it simply decomposes while on the site (single-site mechanism).

Step 3. Products are desorbed from the surface, which then frees the site.

In addition, all species of molecules, free reactants, and free products as well
as site-attached reactants, intermediates, and products taking part in these three
processes are assumed to be in equilibrium.
Rate expressions derived from various postulated mechanisms are all of the
form

(kinetic term)(driving force or displacement from equilibrium)

rate of reaction =
(resistance term)

For example, for the reaction

occurring in the presence of inert carrier material U, the rate expression when
adsorption of A controls is

When reaction between adjacent site-attached molecules of A and B controls,

the rate expression is
380 Chapter 18 Solid Catalyzed Reactions

whereas for desorption of R, controlling it becomes

Each detailed mechanism of reaction with its controlling factor has its corre-
sponding rate equation, involving anywhere from three to seven arbitrary con-
stants, the K values. For reasons to be made clear, we do not intend to use
equations such as these. Consequently, we do not go into their derivations. These
are given by Hougen and Watson (1947), Corrigan (1954, 1955), Walas (1959),
and elsewhere.
Now, in terms of the contact time or space time, most catalytic conversion
data can be fitted adequately by relatively simple first- or nth-order rate expres-
sions (see Prater and Lago, 1956). Since this is so, why should we concern
ourselves with selecting one of a host of rather complicated rate expressions
which satisfactorily fits the data?
The following discussion summarizes the arguments for and against the use
of simple empirical kinetic equations.

Truth and Predictability. The strongest argument in favor of searching for the
actual mechanism is that if we find one which we think represents what truly
occurs, extrapolation to new and more favorable operating conditions is much
more safely done. This is a powerful argument. Other arguments, such as aug-
menting knowledge of the mechanism of catalysis with the final goal of producing
better catalysts in the future, do not concern a design engineer who has a specific
catalyst at hand.

Problems of Finding the Mechanism. To prove that we have such a mechanism

we must show that the family of curves representing the rate equation type of
the favored mechanism fits the data so much better than the other families that
all the others can be rejected. With the large number of parameters (three to
seven) that can be chosen arbitrarily for each rate-controlling mechanism, a very
extensive experimental program is required, using very precise and reproducible
data, which in itself is quite a problem. We should bear in mind that it is not
good enough to select the mechanism that well fits-or even best fits-the data.
Difference in fit may be explainable entirely in terms of experimental error. In
statistical terms these differences may not be "significant." Unfortunately, if a
number of alternative mechanisms fit the data equally well, we must recognize
that the equation selected can only be considered to be one of good fit, not one
that represents reality. With this admitted, there is no reason why we should not
use the simplest and easiest-to-handle equation of satisfactory fit. In fact, unless
there are good positive reasons for using the more complicated of two equations,
we should always select the simpler of the two if both fit the data equally well.
The statistical analyses and comments by Chou (1958) on the codimer example
in Hougen and Watson (1947) in which 18 mechanisms were examined illustrate
the difficulty in finding the correct mechanism from kinetic data, and show that
even in the most carefully conducted programs of experimentation the magnitude
of the experimental error will very likely mask the differences predicted by the
various mechanisms.
 18.2 Pore Diffusion Resistance Combined with Surface Kinetics 381

Thus it is hardly ever possible to determine with reasonable confidence which

is the correct mechanism.

Problems of Combining Resistances. Suppose that we have found the correct

mechanism and resultant rate equation for the surface phenomenon. Combining
this step with any of the other resistance steps, such as pore of film diffusion,
becomes rather impractical. When this has to be done, it is best to replace the
multiconstant rate equation by an equivalent first-order expression, which can
then be combined with other reaction steps to yield an overall rate expression.

Summary on Surface Kinetics. From this discussion we conclude that it is good

enough to use the simplest available correlating rate expression, hence first-order
or nth-order kinetics, to represent the surface reaction.
For additional comments questioning the validity of the active-site approach,
suggesting forms of kinetic equations to be used in reactor design, and suggesting
what is the real utility of the active site theory, see the opposing points of view
presented by Weller (1956) and Boudart (1956).

18.2 PORE DIFFUSION RESISTANCE COMBINED WITH SURFACE

KINETICS
Single Cylindrical Pore, First-Order Reaction
First consider a single cylindrical pore of length L, with reactant A diffusing into
the pore, and reacting on the surface by a first-order reaction

I dNA
A+product and -rlk,=---- - PC,
S dt

taking place at the walls of the pore, and product diffusing out of the pore, as
shown in Fig. 18.2. This simple model will later be extended.
The flow of materials into and out of any section of pore is shown in detail
in Fig. 18.3. At steady state a material balance for reactant A for this elementary
section gives

output - input + disappearance by reaction =0 (4.1)

or with the quantities shown in Fig. 18.3,

Rearranging gives
382 Chapter 18 Solid Catalyzed Reactions

Elementary section of catalyst

pore shown in Fig. 3

Distance along pore, x

Figure 18.2 Representation of a cylindrical catalyst

pore.

and taking the limit as Ax approaches zero (see the equation above Eq. 13.18a),
we obtain

Note that the first-order chemical reaction is expressed in terms of unit surface
area of the wall of the catalyst pore; hence k" has unit of length per time (see

Mantle area on which reaction

Cross-sectional area = Pr2 / is occurring, S = [Link].x

Flow out
*
(2IOut=
(%) -nrz 9 out

Lhl
Xin
I

xout

Disappearance of A = rate of disappearance (surface)

on surface by reaction (
unit surface

= (- b %) (surface) = X " C (~Z n r AX)

Figure 18.3 Setting up the material balance for the elementary slice
of catalyst pore.
 18.2 Pore Diffusion Resistance Combined with Surface Kinetics 383

Eq. 1.4). In general, the interrelation between rate constants on different bases
is given by

Hence for the cylindrical catalyst pore

Thus in terms of volumetric units Eq. 3 becomes

This is a frequently met linear differential equation whose general solution is

where

and where M I and M2 are constants. It is in the evaluation of these constants

that we restrict the solution to this system alone. We do this by specifying what
is particular about the model selected, a procedure which requires a clear picture
of what the model is supposed to represent. These specifications are called the
boundary conditions of the problem. Since two constants are to be evaluated,
we must find and specify two boundary conditions. Examining the physical limits
of the conceptual pore, we find that the following statements can always be
made. First, at the pore entrance

Second, because there is no flux or movement of material through the interior

end of the pore

With the appropriate mathematical manipulations of Eqs. 7 and 8 we then obtain

384 Chapter 18 Solid Catalyzed Reactions

Fractional distance into pore, xlL

Figure 18.4 Distribution and average value of reactant concentration within a catalyst
pore as a function of the parameter mL = L-

Hence the concentration of reactant within the pore is

--
+ e-m(L-x)
C A - em(L-x) -
-
cosh m ( L - x )
c* em^ +e-m~ cosh mL (10)

This progressive drop in concentration on moving into the pore is shown in Fig.
18.4, and this is seen to be dependent on the dimensionless quantity mL, or M,,
called the Thiele modulus.
To measure how much the reaction rate is lowered because of the resistance
to pore diffusion, define the quantity Bcalled the effectiveness factor as follows:

(actual mean reaction rate within pore)

Effectiveness factor, 6= (11)
(rate if not slowed by pore diffusion)
-
- r ~with, diffusion
r ~without
, diffusion resistance

In particular, for first-order reactions 6= CA/Ch because the rate is proportional

to the concentration. Evaluating the average rate in the pore from Eq. 10 gives
 18.3 Porous Catalyst Particles 385

with sealed ends

Volume change on reaction "\ \ \.\\

= 1 + E X= 1, no volume change
= 2, volume doubles
3,
= volume halves
\ \

Thiele modulus: mL = ~m
Figure 18.5 The effectiveness factor as a function of the parame-
ter mL or M,, called the Thiele modulus, prepared from Aris
(1957) and Thiele (1939).

the relationship

-
- CA - tanh mL
&first order - --
C, mL

which is shown by the solid line in Fig. 5. With this figure we can tell whether pore
diffusion modifies the rate of reaction, and inspection shows that this depends on
whether mL is large or small.
For small mL, or mL < 0.4, we see that & = 1, the concentration of reactant
does not drop appreciably within the pore; thus pore diffusion offers negligible
resistance. This can also be verified by noting that a small value for mL =
L means either a short pore, slow reaction, or rapid diffusion, all three
factors tending to lower the resistance to diffusion.
For large mL, or mL > 4, we find that G = llmL, the reactant concentration
drops rapidly to zero on moving into the pore, hence diffusion strongly influences
the rate of reaction. We call this the regime of strong pore resistance.

18.3 POROUS CATALYST PARTICLES

The results for a single pore can approximate the behavior or particles of various
shapes-spheres, cylinders, etc. For these systems the following apply.

1. Use of the proper diffusion cog$cient. Replace the molecular diffusion

coefficient CB by the effective diffusion coefficient of fluid in the porous
structure. Representative gkvalues for gases and liquids in porous solids
are given by Weisz (1959).
2. Proper measure of particle size. To find the effective distance penetrated
386 Chapter 18 Solid Catalyzed Reactions

by gas to get to all the interior surfaces we define a characteristic size

of particle

volume of particle
, any particle shape
I = lexterior surface available
{for reactant penetration /
thickness
,for flat plates
L

I = -,

=
for cylinders

f,for spheres
3. Measures of reactions rates. In catalytic systems the rate of reaction can
be expressed in one of many equivalent ways. For example, for first-or-
der kinetics

Based on volume of l d N A - k G , [ mols reacted

voids in the reactor -rA= V dt m3voids. s

Based on weight of
catalyst pellets
-ra 1 dNA -
= - --
[
mol reacted
W dt - ktCA, kgcat.s ] (15)

1-dNA - mol reacted

Based on catalyst surface - r l = -- - kUCA,
S dt -
m2cat. surf. s

Based on volume of 1 dNA mol reacted

catalyst pellets -rlkl= V, dt - P C A , [ m3solid. s ] (17)

Based on total reactor 1d N ~ mol reacted

volume -rT= V, dt m3reactor s

Use whichever definition is convenient. However, for porous catalyst parti-

cles rates based on unit mass and on unit volume of particles, r' and r"' are
the useful measures. Hence for nth order reactions

r
mol A (m3gas)"
(kg cat). s ] = k l C y w h e r e k ' = [(rnol A)"l (kg cat) s - 1 (19)
-I;
mol A
cat) .s] = k"'Cy where k" =
(m3gas)"
(mol A)"-l (m3cat) .s I
4. In a manner similar to what was done for a single cylindrical pore, Thiele
 18.3 Porous Catalyst Particles 387
(1939) and Aris (1957) related dwith MT for the various particle shapes as
follows:

-
--1 . tanh MT
MT

- -.1 Il(2MT)
MT z~(2MT)

- 1 1 1
MT ' (tanh 3MT - %)

where

These relationships are shown in Fig. 18.6. If you know ge,k"', and L you can
find the reaction rate from M T and Fig. 18.6. However, what if you want to
evaluate k from an experiment in which you measure a rate which could have
been slowed by diffusional resistance, but which you are unsure of?

Thiele modulus: MT = ~ d k " ' 1 9 ,

Figure 18.6 Effectiveness factor versus MT for porous particles of various shapes.
388 Chapter 18 Solid Catalyzed Reactions

5. Finding pore resistance effects from experiment. Here we have a simple

trick to help us. Define another modulus which only includes observable
and measurable quantities. This is known as the Wagner-Weisz-Wheeler
modulus M w (lucky for us that the three researchers who first dealt with
this problem had last names all starting with the same letter).

We shall call this the Wagner modulus.

6. Pore resistance limits. When reactant fully penetrates the particle and
bathes all its surfaces, then the particle is in the diffusion free regime. This
occurs when M T < 0.4 or M w < 0.15.
At the other extreme when the center of the particle is starved for reactant
and is unused then the particle is in the strong pore resistance regime. This
occurs when M T > 4 or M w > 4.
Figures 18.6 and 18.7 show these limits.
7. Particles of Different Sizes. Comparing the behavior of particles of size R,
and R2 we find in the diffusion-free regime

In the regime of strong diffusion resistance

Thus the rate varies inversely proportional to the particle size.

M~ M~
Figure 18.7 Shows the limits for negligible and for strong pore diffusion resistance.
 18.3 Porous Catalyst Particles 389
Extensions
There are many extensions to this basic treatment. Here we just mention a few.

Mixture of Particles of Various Shapes and Sizes. For a catalyst bed consisting
of a mixture of particles of various shapes and sizes Aris (1957) showed that the
correct mean effectiveness factor is

where f;,f ; , . . . are the volume fractions of particles of sizes 1, 2, . . . in

the mixture.

Molar Volume Change. With decrease in fluid density (expansion) during reac-
tion the increased outflow of molecules from the pores makes it harder for
reactants to diffuse into the pore, hence lowering d On the other hand, volumetric
contraction results in a net molar flow into the pore, hence increasing k For a
first-order reaction Thiele (1939) found that this flow simply shifted the d versus
M , curve as shown in Fig. 18.5.

Arbitrary Reaction Kinetics. If the Thiele modulus is generalized as follows

[see Froment and Bischoff (1962)l

then the &versus M , curves for all forms of rate equation closely follow the
curve for the first-order reaction. This generalized modulus becomes

for first-order reversible reactions

for nth-order irreversible reactions

nth-order reactions behave in an unexpected way in the region of strong pore

resistance. Combining the nth-order rate with the generalized modulus of Eq.
390 Chapter 18 Solid Catalyzed Reactions

28 we obtain

Thus, in the regime of strong pore diffusion, an nth-order reaction

behaves like a reaction of order (n + 1)/2 or
0 order becomes % order
1st order remains 1st order
2nd order becomes 1.5 order
3rd order becomes 2nd order

In addition the temperature dependency of reactions is affected by strong pore

resistance. From Eq. 29 the observed rate constant for nth-order reactions is

Taking logarithms and differentiating with respect to temperature and noting

that both the reaction rate and to a lesser extent the diffusional process are
temperature-dependent gives

With Arrhenius temperature dependencies for both reaction and diffusion we

have

and replacing in Eq. 31 gives

Since the activation energy for gas-phase reactions is normally rather high, say
-
80 240 kJ, while that for diffusion is small (about 5 kJ at room temperature
or 15 kJ at 1000"C), we can write approximately

=-Etrue
Eobs ,
2
 18.4 Heat Effects during Reaction 391
These results show that the observed activation energy for reactions influenced
by strong pore resistance is approximately one-half the true activation energy.

Summary-Diffusion Resistance in Pores

For first-order surface reaction we summarize our finding in compact form in
Eq. 34.

this flat plate

I;;/ rate without any

diffusion effects
= 1for no
diffusion
resistance
expression is a good
approximation for all
particle shapes

tanhM,
-ri
=kcAs& ... with ... &= ---
MT
effective diffusion
coefficient in porous
solids; m3 gaslm so1id.s

~~(-ri',,d
C~,obs
g e

&'= effectiveness factor, M T = Thiele modulus, Mw = Weisz modulus,

a fudge factor useful for predicting useful for interpreting
which varies betwen 0 reactor behavior from experiments since it
and 1,and which accounts known kinetic infor- only includes observables
for the resistance to pore mation, thus known k '
diffusion

To find how pore resistance influences the rate evaluate MTor Mw, then find
&from the above equations or figures, and insert &into the rate equation.
Desirable processing range: Fine solids are free of pore diffusion resistance
but are difficult to use (imagine the pressure drop of a packed bed of face
powder). On the other hand a bed of large particles have a small Ap but are
liable to be in the regime of strong pore diffusion where much of the pellets'
interior is unused.
For most effective operations what we want is to use the largest particle size
which is still free of diffusional resistance or

18.4 HEAT EFFECTS DURING REACTION

When reaction is so fast that the heat released (or absorbed) in the pellet cannot
be removed rapidly enough to keep the pellet close to the temperature of the
fluid, then nonisothermal effects intrude. In such a situation two different kinds
of temperature effects may be encountered:
Within-particle AT. There may be a temperature variation within the pellet.

Film AT. The pellet may be hotter (or colder) than the surrounding fluid.
392 Chapter 18 Solid Catalyzed Reactions

For exothermic reaction, heat is released and particles are hotter than the sur-
rounding fluid, hence the nonisothermal rate is always higher than the isothermal
rate as measured by the bulk stream conditions. However, for endothermic
reactions the nonisothermal rate is lower than the isothermal rate because the
particle is cooler than the surrounding fluid.
Thus our first conclusion: if the harmful effects of thermal shock, or sintering
of the catalyst surface, or drop in selectivity, do not occur with hot particles,
then we would encourage nonisothermal behavior in exothermic reactions. On
the other hand, we would like to depress such behavior for endothermic reactions.
We next ask which form of nonisothermal effect, if any, may be present. The
following simple calculations tell.
ForJilm AT we equate the rate of heat removal through the film with the rate
of heat generation by reaction within the pellet. Thus

and on combining we find

where L is the characteristic size of the pellet.

For within-particle AT the simple analysis by Prater (1958) for any particle
geometry and kinetics gives the desired expression. Since the temperature and
concentration within the particle are represented by the same form of differential
equation (Laplace equation) Prater showed that the T and CAdistributions must
have the same shape; thus at any point in the pellet x

and for the pellet as a whole

where ke, is the effective thermal conductivity within the pellet.

For temperature gradients within particles the corresponding nonisothermal
effective factor curves have been calculated by Carberry (1961), Weisz and Hicks
(1962), and others [see Bischoff (1967) for references]. Figure 18.8 illustrates
these curves in dimensionless form, and shows that the shape is very similar to
the isothermal curve of Fig. 18.6 with the following exception. For exothermic
reactions only, where pore resistance just begins to intrude, the effectiveness
factor can become greater than unity. This finding is not unexpected in light of
the above discussion.
 18.5 Performance Equations for Reactions Containing Porous Catalyst Particles 393

10
Generalized Thiele modulus of Eq. 26
Figure 18.8 Nonisothermal effectiveness factor curve for temper-
ature variation within the particle. Adapted from Bischoff (1967).

However, for gas-solid systems Hutchings and Carberry (1966) and McGreavy
and coworkers (1969, 1970) show that if reaction is fast enough to introduce
nonisothermal effects, then the temperature gradient occurs primarily across the
gas film, not within the particle. Thus we may expect to find a significant film
AT, before any within-particle AT becomes evident.
For detailed versions of Fig. 18.8 which show Bversus M , and 8versus Mw
plus discussion and problems dealing with nonisothermal reactors, see Chapter
22 in Levenspiel (1996).

18.5 PERFORMANCE EQUATIONS FOR REACTORS CONTAINING

POROUS CATALYST PARTICLES
For Plug Flow. Take a thin slice of the PFR. Then following the analysis of
Chapter 5 for homogeneous reactions we have the situation shown in Fig. 18.9.

([Link]) 9
[Link] = FAO
lj
Volume of catalyst, AV, Moles A reactedls,

Figure 18.9 Elementary slice of solid catalyzed plug flow re-

actor.
394 Chapter 18 Solid Catalyzed Reactions

At steady state a material balance for reactant A gives

input = output + accumulation . . . -

in symbols

F ~ O-F n - F ~ ~ X ~ o+u (t- r a )

~ ~ X ~= iF~~ AW

In differential form

FAodXA = ( - r a ) dW = ( - r z ) dVs

Integrating over the whole reactor gives

Note the similarity of this equation with Eq. 5.13 for homogeneous reactions.
To bring this analogy closer let

We have no name,for these two measures but if we wanted to we could call

them by the ugly terms weight-time, and volume-time,respectively. So for first-
order catalytic reactions Eq. 41 becomes

For Mixed Flow. Here we have, following the analysis of Chapter 5, for any
E A value

For first-order reactions with CAin= CAo,and E, Z0

18.5 Performance Equations for Reactions Containing Porous Catalyst Particles 395

Gas and
. . ,
solids

Solids

Gas Gas
Figure 18.10 Catalytic reactors where solid fraction f varies with
height.

For a Reactor Containing a Batch of Catalyst and a Batch of Gas

Extensions of the Simple Performance Equations. There are numerous applica-

tions of catalytic reactions where the fraction of solids f varies with height z in
the reactor (see Fig. 18.10).
For these situations the performance equations could more usefully be written
differently. With uo as the superficial gas velocity (velocity if solids are absent)
through the vertical reactor, Fig. 18.11shows what happens in a thin slice through
the reactor. A steady-state material balance gives

Input of A = output of A + disappearance of A

Cross-sectional FAout = ~ O A C A(1-XA

O out)
area, A Volume of solids in the
lernent, fAAz [m3solidsl

Reaction in the element

-rrf AAz [molls]

FAin = uOACAO(1-XA

Figure 18.11 Section of catalytic reactor which has a solid frac-

tion f.
396 Chapter 18 Solid Catalyzed Reactions

In symbols

In differential form

Integrating

For first-order reactions this expression reduces to

For the special case where E, = 0, f is constant, and the height of catalyst bed
is H, we have

The original derivation leading to Eqs. 40 to 47 is used in the next chapter on

packed beds. The extension leading to Eqs. 48 to 51 is used in Chapter 20, when
dealing with suspended solids reactors.

18.6 EXPERIMENTAL METHODS FOR FINDING RATES

Any type of reactor with known contacting pattern may be used to explore the
kinetics of catalytic reactions. Since only one fluid phase is present in these
reactions, the rates can be found as with homogeneous reactions. The only special
precaution to observe is to make sure that the performance equation used is
dimensionally correct and that its terms are carefully and precisely defined.
The experimental strategy in studying catalytic kinetics usually involves mea-
suring the extent of conversion of gas passing in steady flow through a batch of
solids. Any flow pattern can be used, as long as the pattern selected is known;
if it is not known then the kinetics cannot be found. A batch reactor can also
be used. In turn we discuss the following experimental devices:
Differential (flow) reactor
Integral (plug flow) reactor
Mixed flow reactor
Batch reactor for both gas and solid
 18.6 Experimental Methods for Finding Rates 397

Differential Reactor. We have a differential flow reactor when we choose to

consider the rate to be constant at all points within the reactor. Since rates are
concentration-dependent this assumption is usually reasonable only for small
conversions or for shallow small reactors. But this is not necessarily so, e.g., for
slow reactions where the reactor can be large, or for zero-order kinetics where
the composition change can be large.
For each run in a differential reactor the plug flow performance equation be-
comes

from which the average rate for each run is found. Thus each run gives directly
a value for the rate at the average concentration in the reactor, and a series of
runs gives a set of rate-concentration data which can then be analyzed for a
rate equation.
Example 18.2 illustrates the suggested procedure.

Integral Reactor. When the variation in reaction rate within a reactor is so

large that we choose to account for these variations in the method of analysis,
then we have an integral reactor. Since rates are concentration-dependent, such
large variations in rate may be expected to occur when the composition of
reactant fluid changes significantly in passing through the reactor. We may follow
one of two procedures in searching for a rate equation.

Integral Analysis. Here a specific mechanism with its corresponding rate equa-
tion is put to the test by integrating the basic performance equation to give,
similar to Eq. 5.17,

Equations 5.20 and 5.23 are the integrated forms of Eq. 5.17 for simple kinetic
equations, and Example 1 8 . 3 illustrates
~ this procedure.

Differential Analysis. Integral analysis provides a straightforward rapid proce-

dure for testing some of the simpler rate expressions. However, the integrated
forms of these expressions become unwieldy with more complicated rate expres-
sions. In these situations, the differential method of analysis becomes more
convenient. The procedure is closely analogous to the differential method de-
scribed in Chapter 3. So, by differentiating Eq. 53 we obtain

Example 18.3b illustrates this procedure.

398 Chapter 18 Solid Catalyzed Reactions

Inverted cup

Four rapidly spinning

wire baskets containing
catalyst pellets, W

F~Ov X ~ i n r C ~ i n

Spinning
shaft

Figure 18.12 Sketch of a Carberry basket-type experimental mixed

flow reactor.

Mixed Flow Reactor. A mixed flow reactor requires a uniform composition of

fluid throughout, and although it may seem difficult at first thought to approach
this ideal with gas-solid systems (except for differential contacting), such con-
tacting is in fact practical. One simple experimental device which closely ap-
proaches this ideal has been devised by Carberry (1964). It is called the basket-
type mixed pow reactor, and it is illustrated in Fig. 18.12. References to design
variations and uses of basket reactors are given by Carberry (1969). Another
device for approaching mixed flow is the design developed by Berty (1974), and
illustrated in Fig. 18.13. Still another design is that of a recycle reactor with
R = m. This is considered in the next section.
For the mixed flow reactor the performance equation becomes

Feed

Spinning
shaft
Figure 18.13 Principle of the Berty experimen-
tal mixed flow reactor.
 18.6 Experimental Methods for Finding Rates 399

from which the rate is

Thus each run gives directly a value for the rate at the composition of the exit fluid.
Examples 5.1, 5.2, and 5.3 and 18.6 show how to treat such data.

Recycle Reactor. As with integral analysis of an integral reactor, when we use

a recycle reactor we must put a specific kinetic equation to the test. The procedure
requires inserting the kinetic equation into the performance equation for recy-
cle reactors

and integrating. Then a plot of the left- versus right-hand side of the equation
tests for linearity. Figure 18.14 sketches an experimental recycle reactor.
Unfortunately such data would be difficult to interpret when done using a low
or intermediate recycle ratio. So we ignore this regime. But with a large enough
recycle ratio mixed flow is approached, in which case the methods of the mixed
flow reactor (direct evaluation of rate from each run) can be used. Thus a high
recycle ratio provides a way of approximating mixed flow with what is essentially
a plug flow device. But be warned, the problems of deciding how large a recycle
ratio is large enough can be serious. Wedel and Villadsen (1983) and Broucek
(1983) discuss the limitation of this reactor.

F I U in,
~
XAi"= 0

Figure 18.14 Experimental recycle re-

actor. When the recycle ratio is large
enough mixed flow is closely approxi-
mated.
400 Chapter 18 Solid Catalyzed Reactions

Follow change
of composition
with time

Catalyst W

Small composition
change across
reactor
Figure 18.15 Batch reactor (batch of catalyst and batch of fluid) for cata-
lytic reactions.

Batch Reactor. Figure 18.15 sketches the main features of an experimental

reactor which uses a batch of catalyst and a batch of fluid. In this system we
follow the changing composition with time and interpret the results with the
batch reactor performance equation.

The procedure is analogous with the homogeneous batch reactor. To ensure

meaningful results, the composition of fluid must be uniform throughout the
system at any instant. This requires that the conversion per pass across the
catalyst be small.
A recycle reactor without through-flow becomes a batch reactor. This type of
batch reactor was used by Butt et al. (1962).

Comparison of Experimental Reactors

1. The integral reactor can have significant temperature variations from point
to point, especially with gas-solid systems, even with cooling at the walls.
This could well make kinetic measurements from such a reactor completely
worthless when searching for rate expressions. The basket reactor is best
in this respect.
2. The integral reactor is useful for modeling the operations of larger packed
bed units with all their heat and mass transfer effects, particularly for systems
where the feed and product consist of a variety of materials.
3. Since the differential and mixed flow reactors give the rate directly they
are more useful in analyzing complex reacting systems. The test for anything
but a simple kinetic form can become awkward and impractical with the
integral reactor.
4. The small conversions needed in differential reactors require more accurate
measurements of composition than the other reactor types.
5. The recycle reactor with large recycle acts as a mixed flow reactor and
shares its advantages. Actually, to minimize heat effects the catalyst need
not be all at one location, but can be distributed throughout the recycle loop.
 18.6 Experimental Methods for Finding Rates 401

6. In exploring the physical factors of heat and mass transfer, the integral
reactor most closely models the larger fixed bed; however, the basket,
recycle, and batch GIS reactors are more suited for finding the limits for
such heat effects, for avoiding the regime where these effects intrude, and
for studying the kinetics of the reaction unhindered by these phenomena.
7. The batch GIS reactor, like the integral reactor, gives cumulative effects,
thus is useful for following the progress of multiple reactions. In these
reactors it is easier to study reactions free from heat and mass transfer
resistances (simply increase the circulation rate), and it is also simple to
slow down the progress of reactions (use a larger batch of fluid, or less
catalyst); however, direct modeling of the packed bed with all its complexi-
ties is best done with the integral flow reactor.
8. Because of the ease in interpreting its results the mixed flow reactor is
probably the most attractive device for studying the kinetics of solid cata-
lyzed reactions.

Determining Controlling Resistances and the Rate Equation

Interpretation of experiments becomes difficult when more than one resistance
affects the rate. To avoid this problem we should like, with preliminary runs, to
first find the limits of operations where the various resistances become important.
This will allow us to select conditions of operations in which the resistances can
be studied separately.

Film Resistance. First of all, it is best to see whether film resistance of any
kind (for mass or heat transfer) need be considered. This can be done in a
number of ways.
1. Experiments can be devised to see whether the conversion changes at
different gas velocities but at identical weight-time. This is done by using
different amounts of catalyst in integral or differential reactors for identical
values for weight-time, by changing the spinning rate in basket reactors, or
by changing the circulation rate in recycle or batch reactors.
2. If data are available we can calculate whether film resistance to heat transfer
is important by the estimate of Eq. 36., and whether film resistance to mass
transport is important by comparing the observed first-order rate constant
based on the volume of particle with the mass transfer coefficient for that
type of flow.
For fluid moving past a single particle at relative velocity u Froessling
(1938) gives

while for fluid passing through a packed bed of particles Ranz (1952) gives
402 Chapter 18 Solid Catalyzed Reactions

Thus we have roughly

-
kg -
1
for small dp and u
dP
-
kg -
dki2
u112
for large d, and u

Thus to see whether film mass transfer resistance is important compare

k rb,Vpversus kg&, (58)

If the two terms are of the same order of magnitude we may suspect that the
gas film resistance affects the rate. On the other hand, if k,,,V, is much smaller
than kg$,, we may ignore the resistance to mass transport through the film.
Example 18.1 illustrate this type of calculation. The results of that example
confirm our earlier statement that film mass transfer resistance is unlikely to
play a role with porous catalyst.

Nonisothermal Effects. We may expect temperature gradients to occur either

across the gas film or within the particle. However, the previous discussion
indicates that for gas-solid systems the most likely effect to intrude on the rate
will be the temperature gradient across the gas film. Consequently, if experiment
shows that gas film resistance is absent then we may expect the particle to be
at the temperature of its surrounding fluid; hence, isothermal conditions may be
assumed to prevail. Again see Example 18.1.

Pore Resistance. The effectiveness factor accounts for this resistance. Thus,
based on unit mass of catalyst we have

where

The existence of pore resistance can be determined by

1. Calculation if CZJe is known.
2. Comparing rates for different pellet sizes.
3. Noting the drop in activation energy of the reaction with rise in temperature,
coupled with a possible change in reaction order.

18.7 PRODUCT DISTRIBUTION IN MULTIPLE REACTIONS

More often than not, solid-catalyzed reactions are multiple reactions. Of the
variety of products formed, usually only one is desired, and it is the yield of this
material which is to be maximized. In cases such as these the question of product
distribution is of primary importance.
Here we examine how strong pore diffusion modifies the true instantaneous
fractional yield for various types of reactions; however, we leave to Chapter 7
 18.7 Product Distribution in Multiple Reactions 403

the calculation of the overall fractional yield in reactors with their particular
flow patterns of fluid. In addition, we will not consider film resistance to mass
transfer since this effect is unlikely to influence the rate.

Decomposition of a Single Reactant by Two Paths

N o resistance to pore diffusion. Consider the parallel-path decomposition

R (desired), r~ = k l C 2
f
A, \
S (unwanted), rs = k2Ci2

Here the instantaneous fractional yield at any element of catalyst surface is

given by

or for first-order reactions

1
=1 + (k21kl)
Strong resistance to pore diffusion. Under these conditions we have

and with Eq. 29

4ge
(a, + a, + 2)(k1 + k2)C2i1
Using a similar expression for r, and replacing both of these into the defining
equation for cp gives

+ -
- r~ 1
r
P O ~-S- - 1 + (k21k,)Cg-al)"
r -

and for equal-order or for first-order reactions

This result is expected since the rules in Chapter 7 suggest that the product
distribution for competing reactions of same order should be unaffected by
changing concentration of A in the pores or in the reactor.
404 Chapter 18 Solid Catalyzed Reactions

Reactions in Series
As characteristic of reactions in which the desired product can decompose further,
consider the successive first-order decompositions

When CA does not drop in the interior of catalyst particles, true rates are ob-
served. thus

-
(~obs- true or (5)
kl = (5)
kl
obs true

Strong resistance to pore diffusion. An analysis similar to that starting with

Eq. 2 using the appropriate kinetic rate expressions gives the concentration ratio
of materials in the main gas stream (or pore mouths) at any point in the reactor.
Thus the differential expression (see Wheeler, 1951 for details) is

For mixed flow with CAgoing from CAoto CA, Eq. 65 with CRo= 0 gives

For plug flow, integration with CRo= 0 gives

Comparing Eqs. 66 and 67 with the corresponding expressions for no resistance

in pores, Eqs. 8.41 and 8.37, shows that here the distributions of A and R are
given by a reaction having the square root of the true k ratio, with the added
modification that CRgis divided by 1 + y. The maximum yield of R is likewise
affected. Thus for plug flow Eq. 8.8 or 8.38 is modified to give

and for mixed flow Eq. 8.15 or 8.41 is modified to give

Table 18.2 shows that the yield of R is about halved in the presence of strong
resistance to diffusion in the pores.
 18.7 Product Distribution in Multiple Reactions 405

Table 18.2 The Role of Diffusion in Pores for First-Order Reactions in Series
CRg,,,,/CAofor Plug Flow CRg,,,ICAo for Mixed Flow
No Strong Percent No Strong Percent
k,/k, Resistance Resistance Decrease Resistance Resistance Decrease
1164 0.936 0.650 30.6 0.790 0.486 38.5
1/16 0.831 0.504 39.3 0.640 0.356 44.5
1I4 0.630 0.333 47.6 0.444 0.229 48.5
1 0.368 0.184 50.0 0.250 0.125 50.0
4 0.157 0.083 47.2 0.111 0.057 48.5
16 0.051 0.031 38.2 0.040 0.022 44.5

For more on the whole subject of the shift in product distribution caused by
diffusional effects, see Wheeler (1951).

Extensions to Real Catalysts

So far we have considered catalyst pellets having only one size of pore. Real
catalysts, however, have pores of various sizes. A good example of this are the
pellets prepared by compressing a porous powder. Here there are large openings
between the agglomerated particles and small pores within each particle. As a
first approximation we may represent this structure by two pore sizes as shown
in Fig. 18.16. If we define the degree of branching of a porous structure by a where

a = 0 represents a nonporous particle

a = 1represents a particle with one size of pore
a = 2 represents a particle with two pore sizes

then every real porous pellet can be characterized by some value of a.

Model of pore structure

porous powder

Figure 18.16 Porous structure with two sizes of pores as a model for a pellet of compressed
porous powder.
406 Chapter I 8 Solid Catalyzed Reactions

Now for strong pore diffusion in one size of pore we already know that the
observed order of reaction, activation energy, and k ratio for multiple reactions
will differ from the true value. Thus from Eqs. 30 and 32

1 1
Eobs= - Ediff+ - E
2 2
n-1
2

(5) = (?)In . - - for side-by-side reactions

kl obs

Carberry (1962a, b), Tartarelli (1968), and others have extended this type of
analysis to other values of a and to reversible reactions. Thus for two sizes of
pores, where reaction occurs primarily in the smaller pores (because of much
more area there), while both sizes of pores offer strong pore diffusional resistance,
we find

n-1
4

- - . for side-by-side reactions

More generally for an arbitrary porous structure

[(2),, = (2)'l2' - for side-by-side reactions

These findings show that for large a, diffusion plays an increasingly important
role, in that the observed activation energy decreases to that of diffusion, and the
reaction order approaches unity. So, for a given porous structure with unknown a,
the only reliable estimate of the true k ratio would be from experiments under
conditions where pore diffusion is unimportant. On the other hand, finding the
experimental ratio of k values under both strong and negligible pore resistance
should yield the value of a. This in turn should shed light on the pore structure
geometry of the catalyst.
 18.7 Product Distribution in Multiple Reactions 407

SEARCH OF THE RATE-CONTROLLING MECHANISM

An experimental rate measurement on the decomposition of A is made with a

particular catalyst (see pertinent data listed below).
(a) Is it likely that film resistance to mass transfer influences the rate?
(b) Could this run have been made in the regime of strong pore diffusion?
(c) Would you expect to have temperature variations within the pellet or
across the gas film?

Data
For the spherical particle:

dp = 2.4mmorL = R/3 =0.4mm = 4 X 10-4mcat

C2Je= 5 x m3/hr.m cat (effective mass conductivity)
k,, = 1.6 kJ/hr. m cat - K (effective thermal conductivity)

For the gas film surrounding the pellet (from correlations in the literature):

h = 160 kJ/hr. m2cat K (heat transfer coefficient)

kg = 300 m3/hr.m2cat (mass transfer coefficient)

I For the reaction:

AH, = -160 kJ/mol A (exothermic)

CAg= 20 mol/m3(at 1atm and 336°C)

-rT,obs = lo5mollhr .m3cat

Assume that the reaction is first order.

I SOLUTION

(a) Film mass transfer. From Eq. 58, and introducing numerical values, we
obtain

observed rate kr'r V (-r'rrA, obs/C~g)(nd;/6) obs . dp

---YZ,
-
- obs p -
rate if film resistance controls kg&, kg(ndi) c~gkg 6

-
- lo5mol/hr. m3cat 2.4 X m cat
(20 mol/m3)(300m3/hr.m2cat) ' 6

I The observed rate is very much lower than the limiting film mass transfer rate.
408 Chapter 18 Solid Catalyzed Reactions

Thus the resistance to film mass transfer certainly should not influence the
rate of reaction.
(b) Strong pore diffusion. Equations 24 and Fig. 18.7 test for strong pore
diffusion. Thus

This quantity M w is greater than 4, hence pore diffusion is influencing and

slowing the rate of reaction.
(c) Nonisothermaloperations. The estimate for the upper limit to temperature
variations is given by Eqs. 38 and 36. Thus within the pellet

-
- (5 X m3/hr.m cat)(20 mol/m3)(160kJ/mol)
(1.6 kJ/hr. m cat. K)

Across the gas film

These estimates show that the pellet is close to uniform in temperature, but
could well be hotter than the surrounding fluid.
The findings of this example use coefficients close to those observed in real
gas-solid systems (see the appendix), and the conclusions verify the discussion
of this chapter.

THE RATE EQUATION FROM A DIFFERENTIAL REACTOR

The catalytic reaction

is run at 3.2 atm and 117°C in a plug flow reactor which contains 0.01 kg of
catalyst and uses a feed consisting of the partially converted product of 20 liters/
hr of pure unreacted A. The results are as follows:
 18.7 Product Distribution in Multiple Reactions 409

Run 1 2 3 4
CAin, mollliter 0.100 0.080 0.060 0.040
CA,, mollliter 0.084 0.070 0.055 0.038

Find a rate equation to represent this reaction.

SOLUTION

Since the maximum variation about the mean concentration is 8% (run I), we
may consider this to be a differential reactor and we may apply Eq. 52 to find
the reaction rate.
Basing conversion for all runs on pure A at 3.2 atm and 117"C, we have

cAO-- N~~
v -PAO- 3.2 atm mol
RT (0.082 liter atmlmol .K)(390 K) = 0.1 -
liter

and

molA liters mol

FA o -- C A o v = ( ~ - l t e r ) ( 2 ~ l r ) = 2 G

Because the density changes during reaction, concentrations and conversions are
related by

CA
- - -1-XA 1 - CAICAo
or X, =
CAO 1 + EAXA 1 + EA(CAICAO)

where EA = 3 for the basis selected (pure A).

Table E18.2 shows the details of the calculations. Plotting - r i versus CA as
shown in Fig. E18.2 gives a straight line through the origin, indicating a first-

Table E18.2
X~in = x~out =

I--C ~ i n I-- C ~ o u t
c~~ c~~
-
C~in C~out
- c~av 1 + e A pC ~ i n c~out AxA = -r6 =-ax^
CAO CA, mollliter CAO X ~ ~ X ~~ i n t
- WIFAO
1-1- 1 - 0.84 0.0455 9.1
1 0.84 0.092 i--qj-
O 1 + 3(0.84)
= 0.0455 0.0455 -=
0.0112
0.8 0.70 0.075 0.0588 0.0968 0.0380 7.6
0.6 0.55 0.0575 0.1429 0.1698 0.0269 5.4
0.4 0.38 0.039 0.2727 0.2897 0.0171 3.4
410 Chapter 18 Solid Catalyzed Reactions

rnol
CA,
liter
Figure E18.2

order decomposition. The rate in terms of moles A reactedlhrekg cat is then

found from this figure to be

1 dNA liters )( m"l)

= W dt [Link] '*E

THE RATE EQUATION FROM AN INTEGRAL REACTOR

The catalytic reaction

is studied in a plug flow reactor using various amounts of catalyst and 20 liters1
hr of pure A feed at 3.2 atm and 117°C. The concentrations of A in the effluent
stream is recorded for the various runs as follows.

Run 1 2 3 4
Catalyst used, kg 0.020 0.040 0.080 0.160
CAout,mollliter 0.074 0.060 0.044 0.029

(a) Find the rate equation for this reaction, using the integral method of
analysis
(b) Repeat part (a), using the differential method of analysis.
 18.7 Product Distribution in Multiple Reactions 411

I SOLUTION
(a) Integral Analysis. From Example 18.2 we have for all experimental runs

Since the concentration varies significantly during the runs, the experimental
reactor should be considered to be an integral reactor.
As a first guess try a first-order rate expression. Then for plug flow Eq.
44 gives

c ~ =~ (1w+
kt - In- 1 - &AXA
FAO 1 - XA

and with cA, CAO,and FAo

replaced by numerical values this becomes

The two terms in parentheses should be proportional to each other with k t as

the constant of proportionality. Evaluating these terms in Table E18.3a for

Table E18.3a Calculations Needed to Test the Fit of Eq. (i); Integral Analysis

the data points and plotting as in Fig. 18.3a, we see that there is no reason to
suspect that we do not have a linear relationship. Hence we may conclude
that the first-order rate equation satisfactorily fits the data. With k t evaluated
from Fig. 18.3a, we then have

liters
= (95 h r . k g c a t ) ( c A ' m )

(b) Differential Analysis. Equation 54 shows that the rate of reaction is given
by the slope of the XAversus WIFAocurve. The tabulation (Table E18.3b)
based on the measured slopes in Fig. E18.3b shows how the rate of reaction
412 Chapter 18 Solid Catalyzed Reactions

CAOW w - .
hr kg cat
FAO - 20 ' liter

Figure E18.3a

Figure E18.3b
 18.7 Product Distribution in Multiple Reactions 413

I Table E18.3b Calculations Used for the Differential Analysis

(from Fig. E18.3b)

unused
5.62
4.13
2.715

rnol
CA, -
liter
Figure E18.3~

is found at various C,. The linear relation between - r i and CA in Fig.

E18.3~then gives for the rate equation:

PLUG FLOW REACTOR SIZE FROM A RATE EQUATION

Consider the catalytic reaction of Example 18.2. Using the rate equation found
for this reaction determine the amount of catalyst needed in a packed bed reactor
(assume plug flow) for 35% conversion of A to R for a feed of 2000 mollhr of
pure A at 3.2 atm and 117°C.
 414 Chapter 18 Solid Catalyzed Reactions

/ SOLUTION
The amount of catalyst needed is given by the first-order rate expression for
1 plug flow, Eq. 44. Thus

Replacing all the known values from Example 18.2 into this expression gives the
final result, or

mol A
2000 -
hr
= ( liter
96 hr .kg cat
)(md;)
0.1 -

= 140 kg catalyst

PLUG FLOW REACTOR SIZE FROM RATE

CONCENTRATION DATA

For the reaction of Example 18.2 suppose the following rate concentration data
are available:

CA,mollliter 0.039 0.0575 0.075 0.092

- r i , mol Alhr .kg cat 3.4 5.4 7.6 9.1

Directly from this data, and without using a rate equation, find the size of packed
bed needed to treat 2000 mollhr of pure A at 117°C (or ,C , = 0.1 mollliter,
E A = 3) to 35% conversion, all at 3.2 atm.

Note: Rate information such as this can be obtained from a differential reactor
(see Table E18.2), or from other types of experimental reactors.

SOLUTION

To find the amount of catalyst needed without using an analytic expression for
the rate concentration relationship requires graphical integration of the plug
flow performance equation, or
I
 18.7 Product Distribution in Multiple Reactions 415

Table E18.5

-r6, 1
- 1 - CA/O.l
X -
(given) (given) A -1 + 3CA/0.1

The needed 11-ra versus X, data are determined in Table E18.5 and are plotted
in Fig. E18.5. Integrating graphically then gives

Hence

W=
- (2000 F)
(0.0735
hr - kg cat
A mol = 147 kg cat

"0 0.1 0.2 0.3 0.35 0.4

XA,dimensionless

Figure E18.5
 416 Chapter I 8 Solid Catalyzed Reactions

i*
4*P I, i,t* s h a = .-
MIXED FLOW REACTOR SIZE

For the reaction of Example 18.2 determine the amount of catalyst needed in a
packed bed reactor with a very large recycle rate (assume mixed flow) for 35%
conversion of A to R for a feed rate of 2000 mollhr of pure A at 3.2 atm and
117°C. For the reaction at this temperature

A +4R, - r i = 96 C , , mollkg cat. hr

From Example 18.2, CAo= 0.1 mollliter and E, = 3.

SOLUTION

At 35% conversion the concentration of reactant is

1 - X, 1 - 0.35
= 'AO (1 + cAXA = ) + 3(0.35)) = 0'0317

Now for mixed flow Eq. 45 gives

W -x
-- ~ -~ x ~~ - ~x ~ o u t -
lt
- x~~~
F~~ -r i o u t Id C ~ ~ u t

(
0'35 -
W = 2000 96(0.0317)
-
) = 230 kg
=

Note: A mixed flow reactor requires more catalyst than does a plug flow reactor,
as expected.

1 EXAMPLE 18.7 MASS TRANSFER RESISTANCES

What is the most reasonable interpretation, in terms of controling resistances,

of the kinetic data of Table E18.7 obtained in a basket type mixed flow reactor
if we know that the catalyst is porous? Assume isothermal behavior.

Table E18.7
Leaving Spinning Measured
Pellet Concentration Rate of Reaction
Diameter of Reactant Baskets Rate, -rA
1 1 high 3
3 1 low 1
3 1 high 1
 References 417

Let us see whether film resistance or pore resistance could be slowing the rate
of reaction.
Runs 2 and 3 use different spinning rates but have the same reaction rate.
Therefore film diffusion is ruled out for the larger pellets. But Eq. 57 shows that
if film resistance is not important for large pellets it will not be important for
small pellets. Therefore film resistance does not influence the rate.
Next, comparing run 1 and either run 2 or 3 shows that

Eq. 25 then tells us that we are in the regime of strong pore resistance. Thus
our final conclusion is
negligible film resistance
strong pore diffusion resistance

REFERENCES

Aris, R., Chem. Eng. Sci., 6, 262 (1957).

Berty, J., Chem. Eng. Prog., 70 (S), 78 (1974).
Bischoff, K. B., Chem. Eng. Sci., 22, 525 (1967).
Boudart, M., AIChE J., 2, 62 (1956).
Broucek, R., Chem. Eng. Sci., 38, 1349 (1983).
Butt, J. B., Bliss, H., and Walker, C. A., AIChE J., 8, 42 (1962).
Carberry, J. J., AIChE J., 7, 350 (1961).
, AIChE J., 8, 557 (1962a).
, Chem. Eng. Sci., 17,675 (1926b).
, Ind. Eng. Chem., 56, 39 (Nov. 1964).
, Catalysis Reviews, 3, 61 (1969).
Chou, C. H., Znd. Eng. Chem., 50,799 (1958).
Corrigan, T. E., Chem. Eng., 61,236 (November 1954); 61,198 (December 1954); 62,199
(January 1955); 62,195 (February 1955); 62, 203 (May 1955); 62,227 (July 1955).
Froessling, N., Gerland Beitr. Geophys., 52, 170 (1938).
Froment, G. F., and Bischoff, K. B., Chemical Reactor Analysis and Design, p. 162, John
Wiley and Sons, New York, 1990.
Hougen, 0. A., and Watson, K. M., Chemical Process Principles, Part 111, John Wiley &
Sons, New York, 1947.
Hutchings, J., and Carberry, J. J., AIChE J., 12, 20 (1966).
Levenspiel, O., Chemical Reactor Omnibook, Chap. 22, OSU Bookstore, Corvallis,
OR, 1996.
McGreavy, C., and Cresswell, D. L., Can. J. Ch.E., 47, 583 (1969a).
, a n d , Chem. Eng. Sci., 24,608 (1969b).
McGreavy, C., and Thornton, J. M., Can. J. Ch. E., 48, 187 (1970a).
, Chem. Eng. Sci., 25,303 (1970b).
Prater, C. C., Chem. Eng. Sci., 8, 284 (1958).
, and Lago, R. M., Advances in Catalysis, 8, 293 (1956).
418 Chapter 18 Solid Catalyzed Reactions

Ranz, W. E., Chem. Eng. Prog., 48,247 (1952).

Satterfield, C. N., Mass Transfer in Heterogeneous Catalysis, M.I.T. Press, 1970.
Tartarelli, R., Chim. Ind. (Milan), 50, 556 (1968).
Thiele, E. W., Znd. Eng. Chem., 31, 916 (1939).
Walas, S., Reaction Kinetics for Chemical Engineers, McGraw-Hill, New York, 1959.
Wedel, S., and Villadsen, J., Chem. Eng. Sci., 38, 1346 (1983).
Weisz, P. B., Chem. Eng. Prog. Symp. Series, No. 25, 55, 29 (1959).
, and Hicks, J. S., Chem. Eng. Sci., 17, 265 (1962).
Weller, S., AIChE J., 2, 59 (1956).
Wheeler, A., Advances in Catalysis, 3, 250 (1951).

PROBLEMS

These problems are loosely grouped as follows:

Problems 1-19: Direct application of performance equations. These problems
should be tried first.
Problems 20-30: Pore diffusion kinetics.
Problems 31-40: Reactor size + pore diffusion

18.1. While being shown around Lumphead Laboratories, you stop to view a
reactor used to obtain kinetic data. It consists of a 5-cm ID glass column
packed with a 30-cm height of active catalyst. Is this a differential or
integral reactor?

18.2. A solid catalyzed first-order reaction, E = 0, takes place with 50% conver-
sion in a basket type mixed reactor. What will be the conversion if the
reactor size is trebled and all else-temperature, amount of catalyst, feed
composition, and flow rate-remains unchanged?

18.3. The following kinetic data on the reaction A -+ R are obtained in an

experimental packed bed reactor using various amounts of catalyst and
a fixed feed rate FA, = 10 kmollhr.

(a) Find the reaction rate at 40% conversion.

(b) In designing a large packed bed reactor with feed rate FA, = 400
kmollhr how much catalyst would be needed for 40% conversion.
(c) How much catalyst would be needed in part (b) if the reactor em-
ployed a very large recycle of product stream.

A gas containing A (2 mol/m3) is fed (1 m3/hr) to a plug flow reactor

with recycle loop (0.02 m3 loop volume, 3 kg of catalyst), and the output
composition from the reactor system is measured (0.5 mol A/m3).Find
the rate equation for the decomposition of A for the following cases. Be
sure to give the units of - r a , C
,, and k' in your final expression.
 Problems 419

18.4. Very large recycle, A - R, n = 112

18.5. Very large recycle, A -+ 3R, n = 1, 50% A-50% inerts in feed

18.6. No recycle, A -+ 3R, n = 2, 25% A-75% inerts in feed

Gaseous A reacts (A -t R) in an experimental reactor. From the following

conversion data at various conditions find a rate equation to represent
the reaction

u,, m3/hr 3 2 1.2 Mixed flow

CAo= 10 mol/m3
XA 0.2 0.3 0.5 W = 4 gm

18.8. W ,gm 0.5 1.0 2.5 Plug flow

CA, = 60 mol/m3
CA 30 20 10 u = 3 literlmin

The following kinetic data are obtained in an experimental Carberry type

basket reactor using 100 gm of catalyst in the paddles and different flow
rates from run to run:
A-R FAo,mol/min 0.14 0.42 1.67 2.5 1.25
CAo= 10 mol/m3 CA,mol/m3 8 6 4 2 1

18.9. Determine the amount of catalyst needed in a packed bed reactor for
75% conversion of 1000 mol Almin of a CAo= 8 mol/m3 feed.

18.10. Find W for mixed flow, XA = 0.90, CAo= 10 mol/m3,FAo= 1000 mollmin.

How much catalyst is needed in a packed bed reactor for 80% conversion
of 1000 m3/hr of pure gaseous A (CAo= 100 mol/m3) if the stoichiometry
and rate are given by

50CA mol
A+R, -a, =
1 + 0.02CA kg. hr

mol
A-R, -ri = 8Ci -
kg. hr

18.13. Gaseous feed with A and B (u, = 10 m3/hr)pass through an experimental

reactor packed with catalyst (W = 4 kg). Reaction occurs as follows:

mol
A+B+R+S, -,A=o.~c
A
[Link]
-
420 Chapter 18 Solid Catalyzed Reactions

Find the conversion of reactants if the feed contains CAo= 0.1 mol/m3
and C,, = 10 mol/m3.

18.14. A West Texas gas oil is cracked in a tubular reactor packed with silica-
alumina cracking catalyst. The liquid feed (mw = 0.255) is vaporized,
heated, enters the reactor at 630°C and 1atm, and with adequate tempera-
ture control stays close to this temperature within the reactor. The cracking
reaction follows first-order kinetics and gives a variety of products with
mean molecular weight mw = 0.070. Half the feed is cracked for a feed
rate of 60 m3 liquid/m3 reactor. hr. In the industry this measure of feed
rate is called the liquid hourly space velocity. Thus LHSV = 60 hr-l. Find
the first-order rate constants k' and kt" for this cracking reaction.
Data: Density of liquid feed: p, = 869 kg/m3
Bulk density of packed bed: p, = 700 kg/m3
Density of catalyst particles: p, = 950 kg/m3
This problem is prepared from Satterfield (1970).

18.15. Kinetic experiments on the solid catalyzed reaction A -3R are conducted
at 8 atm and 700°C in a basket type mixed reactor 960 cm3in volume and
containing 1 gm of catalyst of diameter d, = 3 mm. Feed consisting of
pure A is introduced at various rates into the reactor and the partial
pressure of A in the exit stream is measured for each feed rate. The results
are as follows:
Feed rate, literslhr
PAoutlp Ain 1 100 22
0.8 0.5
4
0.2
1
0.1
0.6
0.05
Find a rate equation to represent the rate of reaction on catalyst of this size.

18.16. "El jefe" (the chief) decided to do something to improve the low conver-
sion (XA = 0.80) of our first-order solid-catalyzed liquid-phase reaction.
Instead of ordering more of the expensive catalyst to fill the half-empty
reactor, he decided to save money by adding some tricky piping to the
present vertical packed bed reactor. When I saw what the mechanics were
putting together for him, (see Fig. P18.16), I told "el jefe" that it didn't
seem right. I could sense right away that my saying this annoyed him, but
all he said was, "All right, young lady. Why don't you tell me what
conversion to expect with this arrangement?" Please do this.

ug flow
,c, = 100
u, = 100

Figure P18.16
 Problems 421
18.17. The second-order reaction A -+R is studied in a recycle reactor with
very large recycle ratio, and the following data are recorded:
Void volume of reactor: 1 liter
Weight of catalyst used: 3 gm
Feed to the reactor: CAo= 2 mollliter
u0 = 1 literlhr
Exit stream condition: CAOut = 0.5 mollliter

(a) Find the rate constant for this reaction (give units).
(b) How much catalyst is needed in a packed bed reactor for 80% conver-
sion of 1000 literlhr of feed of concentration CAo= 1mollliter. No re-
cycle.
(c) Repeat part (b) if the reactor is packed with 1part catalyst to 4 parts
inert solid. This addition of inerts helps maintain isothermal conditions
and reduce possible hot spots.

Note: Assume isothermal conditions throughout.

18.18. A small experimental packed bed reactor (W = 1 kg) using very large
recycle of product stream gives the following kinetic data:
A+R, CA,mol/m3 1 2 3 6 9
CAo= 10 mol/m3 v,, literlhr
5 20 65 133 540

Find the amount of catalyst needed for 75% conversion for a flow rate of
1000 mol A/hr of a CA, = 8 mol/m3 feed stream
(a) in a packed bed reactor with no recycle of exit fluid
(b) in a packed bed reactor with very high recycle

18.19. A closed loop batch GIs reactor (see Fig. P18.19) is used for catalytic
rate studies. For this purpose feed gas with reactant is introduced into
the system and is rapidly circulated through the catalyst loop. From the
following composition-time data find a kinetic equation in units of moll
gm - min to represent this reaction.
t, min 0 4 8 16 36 PureAat609K
noatm 1 0.75 0.67 0.6 0.55 2A+R

Nonporous catalyst,
2 gin, 1 crn3 bulk,
void fraction = 0.5

Figure P18.19
422 Chapter 18 Solid Catalyzed Reactions

18.20. Our packed bed reactor runs the gas phase reaction A -+ R at 10 atm
and 336"C, and gives 90% conversion of a pure A feed. However, the
catalyst salesman guarantees that in the absence of any pore diffusion
resistance our reaction will proceed on his new improved porous catalyst
(ge= 2 X m3/m cat. s) with a rate given by

mol
-Y: = o.88cAm3 cat.

which is much better than what we now can do. The catalyst is rather
expensive, since it is formulated of compressed kookaburra droppings
and it is sold by weight. Still, we'll try it when we next replace our catalyst.
What diameter of catalyst balls should we order?

18.21. A reaction A + R is to take place on a porous catalyst pellet (d, = 6

mm, ge = m3/m cat. s). How much is the rate slowed by pore diffu-
sional resistance if the concentration of reactant bathing the particle is
100 mol/m3 and the diffusion-free kinetics are given by

mol
-rg = 0.1Ci-
m3cat. s

18.22. In the absence of pore diffusion resistance a particular first-order gas-

phase reaction proceeds as reported below.

-rl"
A --
mol/cm3cat. s
at
CA= mol/cm3, at 1atm and 400°C

What size of spherical catalyst pellets (ge= cm3/cm cat as) would
ensure that pore resistance effects do not intrude to slow the rate of re-
action?

The first-order decomposition of A is run in an experimental mixed flow

reactor. Find the role played by pore diffusion in these runs; in effect
determine whether the runs were made under diffusion-free, strong resis-
tance, or intermediate conditions.
 Problems 423

18.26. Find the activation energy of the first-order reaction from the following
data:

18.27. What can you tell about the influencing resistances for the porous catalyst
from the data of Table P18.27 obtained in a recycle type mixed flow
reactor. In all runs the leaving stream has the same composition, and
conditions are isothermal throughout.

Table P18.27
Measured
Quantity of Pellet Flow Rate of Recycle Reaction
Catalyst Diameter Given Feed Rate Rate, - r i
1 1 1 High 4
4 1 4 Higher still 4
1 2 1 Higher still 3
4 2 4 High 3

18.28. Experiments at 300°C in a packed bed reactor with very large recycle
stream give the results shown below for the first order catalytic decomposi-
tion A -+ R -,S. Under the best possible conditions (always at 300°C)
what CR,,axICAO may we expect, and how do you suggest we get this (what
flow pattern and particle size, large or small)?

4 1 0.5 No recycle
8 2 0.5

18.29. Experiments with a basket type mixed flow reactor on the solid catalyzed
decomposition A -+ R -+S give the results of Table P18.29. Under the
best possible reaction conditions (always at 300°C) what is the maximum
concentration of R we may expect? How do you suggest that this be ob-
tained?
424 Chapter 18 Solid Catalyzed Reactions

Table P18.29
Size of
Porous Pellets Temperature WIFAO CR,,X/~AO
6 mm 300°C 25 23%
12 mm 300°C 50 23%

18.30. Reactant A at CAo= 10 mol/m3 is to be passed through a packed bed

catalytic reactor where it will decompose into either R or S. To maximize
the formation of R and for 90% decomposition of A determine
whether to operate in the strong pore diffusion regime or in the diffu-
sion-free regime
whether to use plug flow or mixed flow (high recycle)
CR expected in the exit stream.
The kinetics of the decomposition, when free of pore diffusion resistance,
is given by

18.31. A packed bed reactor converts A to R by a first-order catalytic reaction,

A +R. With 9-mm pellets the reactor operates in the strong pore diffusion
resistance regime and gives 63.2% conversion. If these pellets were re-
placed by 18-mm pellets (to reduce pressure drop) how would this affect
the conversion?

18.32. We want to build a packed bed reactor filled with 1.2-cm porous catalyst
particles (p, = 2000 kg/m3,.CBc = 2 x m3/m cat. s) to treat 1 m3/s of
feed gas (113 A, 113 B, 113 inert) at 336OC and 1atm to 80% conversion of
A. Experiments with fine catalyst particles which are free from diffusional
resistance show that

How much catalyst must we use?

18.33. In an experimental mixed flow reactor with 10 gm of 1.2-mm catalyst

particles and a feed of 4 cm3/s of pure A at 1 atm and 336°C we get 80%
conversion for the first-order reaction

We want to design a commercial-sized reactor to treat large amounts of

feed to 80% conversion at the above temperature and pressure. Our choice
 Problems 425

is between a fluidized bed of 1-mm particles (assume mixed flow of gas)

and a packed bed of 1.5-cm particles. Which should we choose so as to
minimize the amount of catalyst needed? How much advantage is there
in this choice?
Additional Data: For the catalyst particles:

p, = 2000 kg/m3 gk= m3/m cat. s

In aqueous solution, and in contact with the right catalyst, reactant A is

converted to product R by the elementary reaction A 2R. Find the
-+

mass of catalyst needed in a packed bed reactor for 90% conversion of

lo4 mol A/hr of feed having C,, = lo3 mol/m3. For this reaction

18.34. k"' = 8x m3/m3bed. s

18.35. k"' = 2 m3/m3bed. s

Additional data:
Diameter of porous catalyst pellets = 6 mm
Effective diffusion coefficient of A in the pellet = 4 X m3/m cat. s
Voidage of packed bed = 0.5
Bulk density of packed bed = 2000 kg/m3 of bed

18.36. A first-order catalytic reaction A(1) R(1) is run in a long, narrow vertical
-+

reactor with upflow of liquid through a fluidized bed of catalyst particles.

Conversion is 95% at the start of operations when the catalyst particles
are 5 mm in diameter. The catalyst is friable and slowly wears away,
particles shrink and the fine powder produced washes out of the reactor.
After a few months each of the 5-mm spheres has shrunk to 3-mm spheres.
What should be the conversion at this time? Assume plug flow of liquid.
(a) Particles are porous and allow easy access for reactants (no resistance
to pore diffusion).
(b) Particles are porous and at all sizes provide a strong resistance to
pore diffusion.

18.37. At present we are running our catalytic first-order reaction in the strong
pore diffusion regime in a packed bed reactor filled with platinum impreg-
nated 6-mm particles of uniform size. A catalyst manufacturer suggests
that we replace our catalyst with 6-mm pellets consisting of fused 0.06-
mm grains. The voidage between grains in the pellet would be about 25%.
If these new pellets were free of pore diffusion resistance in their large
voids (between grains), but if the grains were still in the strong diffusional
resistance regime, how would this change affect the weight of catalyst
needed and the reactor volume?

18.38. Instead of impregnating the whole porous particle uniformly with platinum
(see Problem 18.37), suppose we only impregnate the outer layer of the
426 Chapter 18 Solid Catalyzed Reactions

spherical particle to a thickness of 0.3 mm. How much platinum do we

save by this change? Assume that we are in the strong pore diffusion
regime throughout.

18.39. Because the catalytic reaction A -+R is highly exothermic with rate highly
temperature-dependent, a long tubular flow reactor immersed in a trough
of water, as shown in Fig. P18.39, is used to obtain essentially isothermal
kinetic data. Pure A at O°C and 1 atrn flows through this tube at 10 cm3/
sec, and the stream composition is analyzed at various locations.

Distance from feed 0 12 24 36 48 60 72 84 (00)

input, meters
Partialpressureof 760 600 475 390 320 275 240 215 150
A, mm Hg

Determine what size of plug flow reactor operating at O°C and 1 atrn would
give 50% conversion of A to R for a feed rate of 100 kmollhr of pure A.

\\ \ Nonporous catalyst 2 gm,

1 cm3 bulk, fraction voids = 0.5
Figure P18.39

18.40. A closed-loop experimental G/S batch system as sketched in Fig. P18.19

is used to study the kinetics of a catalytic reaction A -+2R. Pure A is
introduced into the systems and is circulated at O°C and 1atrn at 10 cm3/sec.
The stream is analyzed from time to time with the following results:
Time, min 0 2 4 6 8 10 12 14 (00)

Partialpressure 760 600 475 290 320 275 240 215 150
of A, mm Hg
(a) Determine the size of a plug flow reactor operating at O°C and 1 atrn
required to effect a 50% conversion of A to R for a feed rate of 100
kmol Alhr.
(b) Repeat part a with the modification that an inert at a partial pressure
of 1 atrn is present in the closed loop so that the total pressure at the
start is 2 atm.
Chapter 19
The Packed Bed Catalytic
Reactor

Reactant gas can be made to contact solid catalyst in many ways, and each has
its specific advantages and disadvantages. Figure 19.1 illustrates a number of
these contacting patterns. These may be divided into two broad types, the fixed-
bed reactors of Fig. 1 9 . 1 b,
~ ~and c and the fluidized-bed reactors of Figs. [Link],
e, and f. The moving-bed reactor of Fig. [Link] is an intermediate case which
embodies some of the advantages and some of the disadvantages of fixed-bed
and fluidized-bed reactors. Let us compare the merits of these reactor types.
1. In passing through fixed beds, gases approximate plug flow. It is quite
different with bubbling fluidized beds where the flow is complex and not
well known, but certainly far from plug flow, and with considerable by-
passing. This behavior is unsatisfactory from the standpoint of effective
contacting and requires much more catalyst for high gas conversion, and
greatly depresses the amount of intermediate which can be formed in series
reactions. Hence, if efficient contacting in a reactor is of primary importance,
then the fixed bed is favored.
2. Effective temperature control of large fixed beds can be difficult because
such systems are characterized by a low heat conductivity. Thus in highly
exothermic reactions hot spots or moving hot fronts are likely to develop
which may ruin the catalyst. In contrast with this, the rapid mixing of solids
in fluidized beds allows easily and reliably controlled, practically isothermal,
operations. So if operations are to be restricted within a narrow temperature
range, either because of the explosive nature of the reaction or because of
product distribution considerations, then the fluidized bed is favored.
3. Fixed beds cannot use very small sizes of catalyst because of plugging and
high-pressure drop, whereas fluidized beds are well able to use small-size
particles. Thus for very fast reactions in which pore and film diffusion may
influence the rate, the fluidized bed with its vigorous gas-solid contacting
and small particles will allow a much more effective use of the catalyst.
4. If the catalyst has to be treated (regenerated) frequently because it deacti-
vates rapidly, then the liquid-like fluidized state allows it to be pumped easily
from unit to unit. This feature of fluidized contacting offers overwhelming
advantages over fixed bed operations for such solids.
427
428 Chapter 19 The Packed Bed Catalytic Reactor

Coolant-
in
in

t
Gas in
t
Reactant
gas in
(a) (b)

Packed bed, first stage,

small amount of catalyst
Reactant
Gas in gas in
(c)
t (d)

Regenerator (shown
as a f l u ~ d ~ z ebed)
d

Regenerated
catalyst returned to
the f l u ~ d ~ z ebed
d

Regenerator
gas
(e)
Figure 19.1 Various types of catalytic reactors.
 Chapter 19 The Packed Bed Catalytic Reactor 429

Gas out
Catalyst in Sidestream to
i.
regenerator

Fluidized bed Bucket elevator or

hydraulic lift to raise
catalyst to top of reactor
Mo\~ing bed -
of solids

Catalyst to
regenerator
@

fl '
~eactant
gas in
Reactant
gas in
f
- u

(f)
Figure 19.1 (Continued.)

With these points in mind, let us proceed to Fig. 19.1. Figure 1 9 . 1 is

~ a typical
packed bed reactor embodying all its advantages and disadvantages. Figure [Link]
shows how the problem of hot spots can be substantially reduced by increasing
the cooling surface. Figure 1 9 . 1 shows
~ how intercooling can still further control
the temperature. Note that in the first stage where reaction is fastest, conversion
is kept low by having less catalyst present than in the other stages. Units such
as these can all be incorporated in a single shell or can be kept separate with
heat exchanges between stages.
Figure [Link] shows a fluidized reactor for a stable catalyst which need not be
regenerated. The heat exchanger tubes are immersed in the bed to remove or
add heat and to control the temperature. Figure [Link] shows operations with a
deactivating catalyst which must be continually removed and regenerated. Figure
[Link] shows a three-stage countercurrent unit which is designed to overcome the
shortcomings of fluidized beds with regard to poor contacting. Figure [Link] shows
a moving-bed reactor. Such units share with fixed beds the advantages of plug
flow and disadvantages of large particle size, but they also share with fluidized
beds the advantages of low catalyst-handling costs.
Many factors must be weighed to obtain optimum design, and it may be that
the best design is one that uses two different reactor types in series. For example,
for high conversion of a very exothermic reaction we may well look into the use
of a fluidized bed followed by a fixed bed.
The main difficulties of design of catalytic reactors reduce to the following
two questions: (1) How do we account for the nonisothermal behavior of packed
beds? and (2) How do we account for the nonideal flow of gas in fluidized beds.
Consider a packed bed with heat exchange (Figs. 1 9 . 1 ~and [Link]). For an
exothermic reaction Fig. 19.2 shows the types of heat and mass movement that
will occur when the packed bed is cooled at the walls. The centerline will be
hotter than the walls, reaction will be faster, and reactants will be more rapidly
consumed there; hence, radial gradients of all sorts will be set up.
430 Chapter 19 The Packed Bed Catalytic Reactor

Walls cooled
Reactant Product
to 100"
I flows inward flows outward

Depletion of reactant Heat flows

at centerline outward
Figure 19.2 The temperature field in a packed bed reactor for an exother-
mic reaction creates radial movement of heat and matter.

The detailed analysis of this situation should include the simultaneous radial
dispersion of heat and matter, and maybe axial dispersion too. In setting up
the mathematical model, what simplifications are reasonable, would the results
properly model the real situation, would the solution indicate unstable behavior
and hot spots? These questions have been considered by scores of researchers,
numerous precise solutions have been claimed; however, from the point of view
of prediction and design the situation today still is not as we would wish. The
treatment of this problem is quite difficult, and we will not consider it here. A
good review of the state-of-the-art is given by Froment (1970), and Froment and
Bischoff (1990).
The staged adiabatic packed bed reactor of Fig. 1 9 . 1 ~presents a different
situation. Since there is no heat transfer in the zone of reaction the temperature
and conversion are related simply, hence the methods of Chapter 9 can be applied
directly. We will examine numerous variations of staging and heat transfer to
show that this is a versatile setup which can closely approximate the optimum.
The fluidized bed and other suspended solid reactor types are considered in
the next chapter.

Staged Adiabatic Packed Bed Reactors

With proper interchange of heat and proper gas flow, staged adiabatic packed
beds become a versatile system, which is able to approximate practically any
desired temperature progression. Calculation and design of such a system is
simple, and we can expect that real operations will closely follow these predic-
tions.
We illustrate the design procedure with the single reaction A + R with any
kinetics. This procedure can be extended to other reaction types without diffi-
culty. We first consider different ways of operating these reactors, and then
compare these and point out when one or other is favored.

Staged Packed Beds (Plug Flow) with Intercooling.' The reasoning in Chapter
9 shows that we would like the reacting conditions to follow the optimal tempera-

' This section follows directly from pp. 215-235 of Chapter 9. Hence it is suggested that the reader
familiarize himself with that section before proceeding here.
 Chapter 19 The Packed Bed Catalytic Reactor 431

I Exothermic reversible I I Exothermic irreversible I I Endothermic /

I Optimum 4 T,,, Optimum

Figure 19.3 Sketch showing how staged packed beds can closely approach the optimal tempera-
ture progression.

ture progression. With many stages available this can be closely approximated,
as shown in Fig. 19.3.
For any preset number of stages the optimization of operations reduces to
minimizing the total amount of catalyst needed to achieve a given conversion.
Let us illustrate the procedure for two-stage operations with reversible exothermic
reactions. The method of attack is shown in Fig. 19.4. In this figure we wish to
minimize the total area under the 11-6 versus X , curve in going from X , = 0
to X , = some fked or required conversion. In searching for this optimum we
have three variables which we can set at will: the incoming temperature (point
T,), the amount of catalyst used in the first stage (locates point b along the
adiabatic), and the amount of intercooling (locates point c along the bc line).
Fortunately, we are able to reduce this 3-dimensional search (5-dimensional for

Area measures

Figure 19.4 Optimum two-stage packed bed reactor.

432 Chapter 19 The Packed Bed Catalytic Reactor

three stages, etc.) to a one-dimensional search where Ta alone is guessed. The

procedure is as follows:
1. Guess T a .
2. Move along the adiabatic line until the following condition is satisfied:

This gives point b in Fig. 19.4, thus the amount of catalyst needed in the
first stage as well as the outlet temperature from that stage. Especially in
preliminary design it may not be convenient to use the criterion of Eq. 1.A
simple alternative is a trial-and-error search. Usually two or three carefully
chosen trials keeping away from low rate conditions will yield a good
design, close to the optimum.
3. Cool to point c which has the same rate of reaction as point b; thus

(-ra)leaving a reactor = ( -ra)entering the next reactor

4. Move along the adiabatic from point c until the criterion of Eq. 1 is
satisfied, giving point d.
5a. If point d is at the desired final conversion then we have guessed T, cor-
rectly.
5b. If point d is not at the desired final conversion try a different incoming
temperature T,. Usually three trials will very closely approach the op-
timum.
For three or more stages the procedure is a direct extension of that presented
here, and it still remains a one-dimensional search. This procedure was first
developed by Konoki (1956a) and later, independently, by Horn (1961a).
Overall cost considerations will determine the number of stages to be used,
so in practice we examine 1, then 2, etc., stages until a minimum cost is obtained.
Let us next consider the two other cases of Fig. 19.3. For irreversible exothermic
reactions the criterion for optimal operations has also been presented by Konoki
(1956b). For endothermic reactions the optimal criterion has yet to be developed.
In all these cases a trial-and-error search keeping far from the regions of low
rates is recommended.

Staged Mixed Flow Reactors. For very high recycle the staged recycle reactors
approach mixed flow. As shown in Fig. 19.5, in this case the reactors should
operate on the line of optimum temperature progression, the best distribution
of catalyst among the stages being found by the maximization of rectangles (see
Figs. 6.9-6.11). In effect we need to choose the distribution of catalyst so as to
maximize area KLMN which then minimizes the shaded area in Fig. 19.5.

Staged Packed Beds with Recycle. Here we have a flexible system which can
approach mixed flow and as such is able to avoid regions of low rates. Figure
 Chapter 19 The Packed Bed Catalytic Reactor 433

Figure 19.5 Optimum two-stage mixed flow reactor set up (infinite recycle
for staged packed beds).

19.6 illustrates two-stage operations with a recycle ratio R = 1, and a feed

temperature [Link] to three or more stages follows directly.
Konoki (1961) presents the criterion for optimal operations; however, in pre-
liminary design a few good guesses will suffice to closely approach optimal opera-
tions.
In recycle operations the heat exchangers can be located in a number of places
without affecting what goes on in the reactor. Figure 19.6 illustrates one of these;
other alternatives are shown in Fig. 19.7. The best location will depend on

Figure 19.6 Optimum two-stage packed bed reactor with recycle. The
conversions shown represent a recycle ratio R = 1 in both stages.
434 Chapter 19 The Packed Bed Catalytic Reactor

\I *T
Tf
Figure 19.7 Different location for the heat exchangers while keeping
the same reactor conditions as in Fig. 19.6.

convenience for startup, and on which location gives a higher heat transfer
coefficient (note that the exchanger arrangement of Fig. 19.7a has a higher
through-flow of fluid than the arrangement of Fig. 19.7b).

Cold Shot Cooling. One way of eliminating the interstage heat exchangers is
by properly adding cold feed directly into the second and succeeding stages of
the reactor. The procedure is shown in Fig. 19.8. The criterion for optimal
operations of such an arrangement is given by Konoki (1960), and, in somewhat
different form, by Horn (1961b). They found that the extent of interstage cooling
is given by Eq. 2, and this is shown in Fig. 19.8.
With cold shot cooling the calculation of reactor volumes by the 11-ri versus
X , curve beco~llesmore complicated because different amounts of feed are
involved in each stage. We can also cold shot cool with inert fluid. This will
affect both the 11-ri versus XAand T versus XAcurves.

Choice of Contacting System. With so many contacting alternatives let us sug-

gest when one or other is favored.
1. For endothermic reactions the rate always decreases with conversion; hence
we should always use plug flow with no recycle (see Chapter 9). For exother-
 Chapter 19 The Packed Bed Catalytic Reactor 435

Qin

Figure 19.8 Cold shot cooling eliminates interstage heat exchangers.

mic reactions the slope of the adiabatic line determines which contacting
scheme is best. The rest of these comments concern this case of exother-
mic reactions.
2. All else being equal, cold shot cooling has the advantage of lower cost
because interstage heat exchangers are not needed. However, cold shot
cooling is only practical when the feed temperature is very much below the
reaction temperature, and, in addition, when the temperature does not
change much during reaction. These conditions can be summarized as fol-
lows:
Cold shot cooling is practical when

Two situations, one when cold shot cooling is practical, the other when it
is not, are shown in Fig. 19.9.

Figure 19.9 Situations where cold shot cooling could be helpful and where it should not
be used.
436 Chapter 19 The Packed Bed Catalytic Reactor

3. For exothermic reactions if the slope of the adiabatic line is low (large
temperature rise during reaction) it is advantageous to avoid the low temper-
ature regime where the rate is very low. Thus use high recycle approaching
mixed flow. On the other hand, if the slope is high (small temperature rise
during reaction) the rate decreases with conversion and plug flow is to be
used. Typically, for pure gaseous reactant the slope of the adiabatic is small;
for a dilute gas or for a liquid it is large. As an example, consider a reactant
having C, = 40 Jlmol K and AH, = - 120 000 Jlmol and inerts with C, =
40 Jlmol .K:
For a pure reactant gas stream

For a dilute 1% reactant gas stream

For a 1-molar liquid solution

The adiabatic lines for these cases are sketched in Fig. 19.10 and illustrate
this point.

Use plug flow

Plug vs. mixed flow

1 5 0 0K
Extremely low
rates here
Figure 19.10 Sketch showing why plug flow is used for steep adiabatic
lines, and mixed flow (packed beds with large recycle) for lines with
small slope.
438 Chapter 19 The Packed Bed Catalytic Reactor

4. For exothermic reactions in staged reactors the above discussion can be

summarized as follows:

use plug flow.

For a dilute gas (or a solution) requiring large preheating to bring

Preliminaries for a Set of Problems Dealing with a Single Packed Bed Reactor
A single catalytic packed bed reactor is to be designed to treat 100 molls of
reactant A and produce product R. Feed gas enters at 2.49 MPa and 300 K, the
maximum allowable temperature is 900 K unless otherwise noted, the product
stream is wanted at 300 K, and the thermodynamics and kinetics of the exothermic
reaction are given to us in Fig. 19.11. Prepare a sketch showing the details of
the system you plan to use:

type of reactor: plug, recycle, or mixed ( m recycle)

amount of catalyst needed
heat duty ahead of the reactor, at the reactor itself, and after the reactor
the temperature of all flowing streams

Example 19.1 treats one case, Problems 19.13 to 19.16 treat four other cases.
In all these problems assume that

we are dealing with ideal gases.

Cp = 40 Jlmol .K for all materials, and at all temperatures. This means (from
Example 9.1) that AH, has the same value at all temperatures.

DESIGN OF A SINGLE ADIABATIC PACKED BED SYSTEM

Work out a good design for 80% conversion of a feed consisting of 1mol A and
7 mol inert.

First determine the slope of the adiabatic line. For this note that 8 moles enter1
mole of A. Thus

Cp = (40 Jlmol - K) (8) = 320 JI(mo1 of A + inerts) .K

Thus the slope of the adiabatic is

 Chapter 19 The Packed Bed Catalytic Reactor 439

Drawing various adiabatics in Fig. 19.11 it seems that the one shown on Fig.
[Link] looks best. So now tabulate XAversus ll(-ra) from Fig. 19.11. This gives

Plotting ll(-ra) versus XAgives Fig. [Link]. This tells right away that a recycle
reactor should be used. Thus

w - x~ 0.8(6 X 8) = 38.4 kg. slmol

F~~ -;a -

The recycle ratio, from Fig. [Link] is R = 1.

Figure [Link],b

The feed is available at 300 K, but enters the reactor at 600 K (from Fig.
[Link]), so it must be heated. Thus
 440 Chapter 19 The Packed Bed Catalytic Reactor

The product stream leaves the reactor at 800 K and must be cooled to 300 K, thus

We show our recommended design in Fig. E19.1~.

Figure E19.1~

Preliminaries for a Set of Problems Dealing with Two Packed Bed Reactors in
Series
Two catalyst filled packed bed reactors are to be designed to process 100
molls of reactant A so as to produce product R. Feed gas enters at 2.49 MPa
and 300 K, the allowable T,,, = 900 K, unless otherwise noted, T,, = 300 K,
the product stream is wanted at 300 K, and the thermodynamics and kinetics of
the exothermic reaction are given to us in Fig. 19.11. Prepare a sketch of your
recommended design and on it show

the flow arrangement selected: plug, recycle (give R value), or mixed (when-
ever R > 5). Do not consider injection of cold fluid between stages unless
the problem states that you are permitted to do so.
weight of catalyst needed in each stage.
location and duty of heat exchangers.
temperature of the flowing streams.

DESIGN OF A TWO ADIABATIC PACKED BED SYSTEM

Work out a good design for 85% conversion of a pure A feed to two packed beds.

I First determine the slope of the adiabatic line and draw it lightly on Fig. 18.11.
 Chapter 19 The Packed Bed Catalytic Reactor 441
A Optimum
1.0 / Equilibrium

0.66 ---------------<
---------
XA +
+ 4 ~ +

4++*- Adiab$iL +
+ ++-+ 1
+ Slope = -
2000
+4ce-

(a)
Figure E19.2a

This gives a very shallow adiabatic, as sketched in Fig. E19.2~.The rate continu-
ally increases as you move along this adiabatic, thus use a mixed flow reactor
operating at the optimum.
To minimize the amount of catalyst needed Chapter 6 says use the method
of maximization of rectangles, so tabulate XA versus ll(-r;),,,

And use the method of maximization of rectangles as shown in Fig. E19.2b.

Then from the performance equation

W shaded area
- = (XA)------ -
Fo (-r:)op, - (in Fig. E19.26

we have

W , = FA,(area), = 100 (2.376) = 237.6 kg

and

W2 = FA,(area), = 100 (3.819) = 381.9 kg

 442 Chapter 19 The Packed Bed Catalytic Reactor

Locus of
optima

-J-+
1 .o
XA
(6)
Figure E19.2b

Now to heat exchange:

For the first reactor. If we want to cool the feed before introducing it into the
first reactor we'd have to cool it to

which is well below absolute zero. This is impossible. So we have to cool it

somewhere inside the recycle reactor loop as shown in Fig. E19.2~.But wherever
you put the exchanger the amount of heating or cooling needed is the same.
So to go to 66% conversion at 820°C the amount of heat needed per mole of
A is

(820 - 300)40 + 0.66(-80000) = -32 000 Jlmol

heat in heat out

But for 100 molls of feed

Q , = (32 000 J/mo1)(100 molls) = -3.2 MW (cooling)

For the second reactor. To go from X , = 0.66 at 820 K to X , = 0.85 at 750

K requires, per mole

(750-820) 40 + (0.85 - 0.66) (-80 000) = -18 000 Jlmol

So for 100 molls

I Q2 = (- 18 000)(100) = - 1.8 MW (cooling)

 Problems 443

Similarly for the exchanger needed to cool the exit stream from 750 K to 300 K

Q3 = 100(40)(300 - 750) = -1.8 MW

So our recommended design is shown in Fig. E19.2~.

Figure E19.2~Note: exchanger Q, can be placed at A, B, or C and exchanger Q , can be placed

at D, E, or F.

REFERENCES

Froment, G .F., First Symp. on C.R.E., Washington, D.C., June 1970,Advances in Chemis-
try Series No. 109, A.C.S., 1972.
- and Bischoff, K. B., Chemical Reactor Analysis and Design, 2nd ed., John Wiley and
Sons, 1990.
Horn, F., 2. Electrochemie, 65,295 (1961a).
-, Chem. Eng. Sci., 14,20 (1961b).
Konoki, K. K., Chem. Eng. (Japan), 21,408 (1956a).
-, Chem. Eng. (Japan), 21,780 (1956b).
-, Chem. Eng. (Japan), 24,569 (1960).
-, Chem. Eng. (Japan), 25, 31 (1961).

PROBLEMS

19.1-19.8. Sketch the flow sheet for the two reactor system represented by the
X , versus T diagrams Figs. P.19.1 to Fig. P19.8, and on the sketch show
the flow rate of all streams for each 100 mol of entering fluid, and where
pertinent give the recycle ratios.
the location of the heat exchangers and indicate whether they cool
or heat.
444 Chapter 19 The Packed Bed Catalytic Reactor

Figures P19.1 to P19.8

 Problems 445

19.9-19.12. Sketch the XA versus T diagram for the two packed bed reactor
system shown in Figs. P19.9 to P19.12 for an exothermic reaction where
conversion: XAl = 0.6, XA2= 0.9.
recycle ratio: R,= 2, R, = 1.
heat exchangers all cool the reacting fluid.

12.

Figures P19.9 to p19.12

For the single reactor situation outlined in the text and followed by Exam-
ple 19.1 let us here consider four alternatives. We plan to use

19.13. . . . pure A feed, XA = 0.85, in a reactor which follows the optimal

temperature progression.

19.14. . . . a 50% A-50% inert feed, X, = 0.70, in an adiabatic reactor of

your choice.

19.15. . . . a 20% A-80% inert feed to an adiabatic reactor of your choice whose
outlet is at XA = 0.75 and T = 825 K.

19.16. . . . a 5% A-95% inert feed, XA = 0.5, in an adiabatic reactor of your

choice.
446 Chapter 19 The Packed Bed Catalytic Reactor

For the two-reactor situation outlined in the text and followed by Example 19.2
let us consider five alternatives, thus

19.17. . . . a pure A feed, XA = 0.85.

19.18. . . . a pure A feed, XA= 0.85 and T,,, = 550 K. In this problem, do not
worry about the possibility of an unstable operating point. But before
you build such a unit, you'd better check for this-otherwise you'll be
in trouble.

19.19. . . . a 20% A-80% inert feed, XA = 0.85

19.20. . . . a 40% A-60% inert feed, XA = 0.95

19.21. . . . a 5% A-95% inert feed, XA = 0.95 (try cold feed injection)

Chapter 20
Reactors with Suspended Solid
Catalyst, Fluidized Reactors of
Various Types

The formation of phthalic anhydride is highly exothermic, and even with the
most careful design the heat removal from packed bed reactors can become
uncheckable, leading to temperature runaways, meltdowns, and even explosions.
If the chief engineer of those reactors had been required to sit on the reactor
during start up, there would be fewer chief engineers ab0ut.l
The invention of the fluidized bed with its suspended and rapidly mixing solids
completely overcame this dangerous situation. This is because the rapid mixing
of solids, and the large heat sink (the solids) will only allow the bed temperature
to change very slowly, and it can be easily controlled.
Another problem-catalyst formulators (those magicians) have been very
successful in creating better and better catalysts, those that give higher and higher
rates of reaction. But to use the whole of the catalyst volume effectively we must
keep the Thiele modulus

This means using smaller and smaller particles as k ' is made larger and larger.
This leads us to use suspended solids. Also note that with these very effective
catalysts the required residence time of reactant gas becomes very small, say a
few seconds for a large 30-m high reactor.
Figure 20.1 shows the transition from fixed to BFB to TF to FF to PC reactors.

20.1 BACKGROUND INFORMATION ABOUT SUSPENDED

SOLIDS REACTORS
This is a vast subject, but we must be brief and can only touch on the highlights.
See Kunii and Levenspiel (1991) for a much more complete presentation.

"They ought to give medals for making phthalic," Chem. Week, 70, 40 (1952).
448 Chapter 20 Fluidized Reactors of Various Types

Fixed bed Bubbling Turbulent Fast Pneumatic

fluidized fluidized fluidized conveying
m m
Very
-Solids rDEl
solids - Solids

little solids -
throughflow

Streaks ' Solids

and Lean core
channels dense wall

No throughflow of solids Throughflow of solids, leading

to circulating fluidized beds

Figure 20.1 G/S contacting regimes, from low to very high gas velocity.

First, Geldart (1973) and Geldart and Abrahamson (1978) looked at how
different kinds of solids behaved when fluidized, and came up with the following
simple classification of solids which we now call the Geldart classification, thus
Geldart A, B, C, D. These are shown and described in Fig. 20.2.
Next consider the distribution of solids in the vertical vessel. Let f be the
volume fraction of solids at height z of the vessel. Then as shown in Fig. 20.3
when we go to higher and higher gas velocities the solids spread throughout
the vessel.

Bed expands, then

forms bubbles.
These split and *\ \ t
d
I I ! 1 1 1 1 1 1

-A*.,-
I

Bubbles grow large.

Bed internals help
keep bubble size

Very large exploding

bubbles. Low bed void-
age. Liable to spout

Cohesive '+o+,
Difficult to
fluidize. solids-
clump, and rise
as slugs. Gas O.l
channels 10 50 100 500 1000

Figure 20.2 Geldart classification of solids in BFB.

 20.1 Background Information about Suspended Solids Reactors 449

I I\ Pneumatic convevinn (PC)

11 \ \ Fast fluidization (FF)

low solid throughflow rate
high solid throughflow rate

9
a,
\ Turbulent fluidization (TF)
r:
Bubbling bed (BFB)

i
o,2-0,4 0.4-0.6
I I I
0 0.2 0.4 0.6

(Volume fraction of solids)./ (=)

m3 solids
-t

Figure 20.3 Distribution of solids in the various con-

tacting regimes.

GIS Contacting Regimes. To develop the language that tells what contacting
regime we have at hand consider solids of size d, in a bed of cross-sectional area
A which is fed gas at a superficial gas velocity u,, as shown in Fig. 20.4.
To simplify the equations let us start by defining two dimensionless quantities

(1)

Re, = u*d,*

(2)

Minimum Fluidizing Velocity. The solids will be suspended when the pressure
drop exceeds the weight of solids. This happens when the gas velocity exceeds
the minimum fluidizing velocity [Link] velocity is given by Ergun (1952), and
in dimensionless form is

Terminal Velocity, u,. Individual particles are blown out of the bed when the
gas velocity exceeds what is called the terminal velocity, u,. Haider and Levens-
450 Chapter 20 Fluidized Reactors of Various Types

f [-I -volume fract~onof sol~dsIn the bed

€[-I -volume fract~onof vo~ds~nthe bed

dJml -partrcle size

ps[kglm31 -part~cledens~ty

I
uo = superficial gas velocity
(velocity if vessel is empty of solids)
[m3 gas/m2 bed sl.
Figure 20.4 Notation for a bed of suspended solids.

pie1 (1989) give this velocity for spherical particles as

and for irregularly shaped particles of sphericity 4,

where the particle sphericity 4, is defined as

( surface of a sphere
4s= surface of a particle
volume

For fine particles we evaluate the size by screen analysis, which gives d,,,.
Unfortunately, there is no general relationship between dscrand d,. The best we
can say for pressure drop considerations is
dp = 4, d,, for irregular particles with no seeming longer or shorter di-
mension
* d, = ds, for irregular particles with one somewhat longer dimension
but with length ratio not greater than 2 : 1 (eggs, for example)
@ d 4 d for irregular particles with one shorter dimension but with
length ratio not less than 1:2 (pillows, for example)
Although a single particle will be entrained by a stream of gas flowing faster
than u,, this finding does not extend to a fluidized bed of particles. In the BFB
the gas velocity can be many times greater than u, with very little carryover of
solids. Thus the single particle terminal velocity is not very useful in estimating
when entrainment of solids will become appreciable.
 20.2 The Bubbling Fluidized Bed-BFB 451

dp* dp"
Figure 20.5 General flow regime diagram for the whole range of G / S contacting.

General Chart Showing GIs Contacting Regimes. Grace (1986) prepared a

graph to show the expected behavior of GIs systems all the way from BFB to
CFB. Figure 20.5 shows a somewhat modified version of this chart. In it you will
see Eqs. 3, 4, and 5, which tell when the bed will fluidize and when solids will
begin to entrain from the vessel.
We now look at the various contacting regimes in detail and see what predic-
tions are available for each regarding reactor behavior.

20.2 THE BUBBLING FLUIDIZED BED-BFB

Pass gas upward through a bed of fine particles. For superficial (or inlet) gas
velocities u, much in excess of this minimum the bed takes on the appearance
of a boiling liquid with large bubbles rising rapidly through the bed. In this state
we have the bubbling fluidized bed, BFB. Industrial reactors particularly for
solid catalyzed gas-phase reactions often operate as bubbling beds with gas
-
velocities u, = 5 30 u,,.
Calculations show that the conversion in bubbling beds may vary from plug
flow to well below mixed flow, see Fig. 20.6, and for many years the perplexing
and embarrassing thing about this was that often we could not reliably estimate
452 Chapter 20 Fluidized Reactors of Various Types

Results from 8 different studies

1.
I
Plug flow
I
I
I
0.01; I 20
I I I I
40 60
k"9T"' = k'7'

Figure 20.6 Conversion of reactant in BFB is usually poorer than for both
plug flow and mixed flow. Adapted from Kunii and Levenspiel (1991).

or guess what it would be for any new situation. Because of this, scale-up was
cautious and uncertain, and preferably left to others.
It was soon recognized that this difficulty stemmed from lack of knowledge
of the contacting and flow pattern in the bed: in effect, the bypassing of much
of the solids by the rising bubble gas. This led to the realization that adequate
prediction of bed behavior had to await a reasonable flow model for the bed.
Since the bubbling bed represents such severe deviations from ideal contacting,
not just minor ones as with other single-fluid reactors (packed beds, tubes, etc.),
it would be instructive to see how this problem of flow characterization has been
attacked. A wide variety of approaches have been tried. We consider these in turn.

Dispersion and Tanks in Series Model. The first attempts at modeling naturally
tried the simple one-parameter models; however, observed conversion well below
mixed flow cannot be accounted for by these models so this approach has been
dropped by most workers.

RTD Models. The next class of models relied on the RTD to calculate conver-
sions. But since the rate of catalytic reaction of an element of gas depends on
the amount of solid in its vicinity, the effective rate constant is low for bubble
gas, high for emulsion gas. Thus any model that simply tries to calculate conver-
 20.2 The Bubbling Fluidized Bed-BFB 453

sion from the RTD and the fixed rate constant in effect assumes that all elements
of gas, both slow and fast moving, spend the same fraction of time in each of
the phases. As we will show when we treat the details of gas contacting in
fluidized beds this assumption is a shaky one, hence the direct use of the RTD
to predict conversions, as developed for linear systems in Chapter 11, is quite inad-
equate.

Contact Time Distribution Models. To overcome this difficulty and still use
the information given by the RTD, models were proposed which assumed that
faster gas stayed mainly in the bubble phase, the slower in the emulsion. Gilliland
and Knudsen (1971) used this approach and proposed that the effective rate
constant depends on the length of stay of the element of gas in the bed, thus

short stay means small k

or k = k,tm
long stay means large k

where m is a fitted parameter. Thus combining with Eq. 11-13 we find for
the conversion

The problem with this approach involves obtaining a meaningful E function

to use in Eq. 7 from a measured C-curve. This approach has also been discarded.

Two-Region Models. Recognizing that the bubbling bed consists of two rather
distinct zones, the bubble phase and the emulsion phase, experimenters spent
much effort in developing models based on this fact. Since such models contain
six parameters, see Fig. 20.7, many simplifications and special cases have been
explored (eight by 1962,15 by 1972, and over two dozen to date), and even the
complete six-parameter model of Fig. 20.7 has been used. The users of this model,

..
.. . ..
.... ... . .

Bubble Emulsion
Fraction of
solids here, ml
Pul €! to Plup: to
rn ixed mixed

-\I-
Figure 20.7 Two-phase model to represent the bubbling fluidized bed, with
its six adjustable parameters, v,, V,, (DIuL),, (DIuL),, m , , K.
454 Chapter 20 Fluidized Reactors of Various Types

those dealing with FCC reactors, claim that this model fits their data beautifully.
However, they had to choose different sets of parameter values for each crude
oil feed, in each of their FCC reactors. Also some of the values for their parame-
ters made no physical sense, for example, a negative value for V , or v,.
With this as the situation we should also discard this type of model which
gives a perfect fit but predicts nothing, and brings no understanding with it. The
reason is that we have no idea how to assign values to the parameters for new
conditions. Thus this is just a curve-fitting model, and we should be able to
do better.

Hydrodynamic Flow Models. The discouraging result with the previous ap-
proaches lead us reluctantly to the conclusion that we must know more about
what goes on in the bed if we hope to develop a reasonable predictive flow
model. In particular we must learn more about the behavior of rising gas bubbles,
since they probably cause much of the difficulty.
Two developments are of particular importance in this regard. The first is
Davidson's remarkable theoretical development and experimental verification
[see Davidson and Harrison (1963) for details] of the flow in the vicinity of a
single rising bubble in a fluidized bed which is otherwise at minimum fluidizing
conditions. What he found was that the rise velocity of the bubble u,, depends
only on the bubble size, and that the gas behavior in the vicinity of the bubble
depends only on the relative velocity of rising bubble and of gas rising in the
emulsion u,. In the extremes he found completely different behavior, as shown
in Fig. 20.8. For catalytic reactions we are only interested in fine particle beds,
so let us ignore the large particle extreme from now on.
Now, for the fine particle bed gas circulates within the bubble plus a thin cloud
surrounding the bubble. Thus the bubble gas forms a vortex ring and stays
segregated from the rest of the gas in the bed. Theory says that

thickness of cloud
diameter of bubble

In fine particle beds In large particle beds

Ub + U, Ub < ue
Fast bubble Slow bubble
Clouded bubble Cloudless bubble

Figure 20.8 Extremes of gas flow in the vicinity of rising gas bubbles in BFBs.
 20.3 The K-L Model for BFB 455

As an example if the bubble rises 25 times as fast as the emulsion gas (not all
that uncommon because this ratio is 100 or more in some industrial operations),
then the cloud thickness is just 2% of the bubble diameter. This is the regime
which interests us.
The second finding on single bubbles is that every rising gas'bubble drags
behind it a wake of solids. We designate this wake by a, where

volume of wake a varies between 0.2 and 2.0

volume of bubble depending on the research study (9)

See Rowe and Partridge (1962,1965) for the original study which discovered this.

20.3 THE K-L MODEL FOR BFB

Hydrodynamic type flow models can be developed to represent the BFB, based
on the above two seemingly simple findings. Let us consider and develop the
simplest of these, the K-L BFB model.
Pass an excess of gas upward through a bed of fine particles. With a large
enough bed diameter we get a freely bubbling bed of fast bubbles. As simplifica-
tions, assume the following:

The bubbles are all spherical, all of the same size d,, and all follow the
Davidson model. Thus the bed contains bubbles surrounded by thin clouds
rising through an emulsion. We ignore the upflow of gas through the cloud
because the cloud volume is small compared to that of the bubble. This is
the regime where u, u, (see Fig. 20.8).
The emulsion stays at minimum fluidization conditions, thus the relative
G/S velocity stays constant in the emulsion.
* Each bubble drags up a wake of solids behind it. This generates a circulation
of solids in the bed, upflow behind the bubbles, and downflow everywhere
else in the bed. If this downflow of solids is rapid enough then gas upflow
in the emulsion is impeded, can actually stop, and even reverse itself. Such
downflow of gas has been observed and recorded, and occurs when

We ignore any upflow or downflow of gas in the emulsion. We show this model
in Fig. 20.9.

Let
u, = superficial gas velocity in the bed, m3 gas/m2bed -s
d = diameter, m
E = fraction of voids in the bed
subscripts b, c, e, w refer to bubble, cloud, emulsion, and wake, respectively.
subscripts m,mf,and f refer to packed bed, minimum fluidization, and bubbling
fluidized bed conditions, respectively.
456 Chapter 20 Fluidized Reactors of Various Types

Emulsion
For gas u,,
either up or-
down Cloud
US' - Bubble
down
Wake

uo
Cloud and wake

Figure 20.9 Model and symbols used to describe the K-L bubbling gas fluid-
ized bed.

In essence, given umf,emf,uo, a, and the effective bubble size in the bed d,, this
model tells you all the other properties of the bed-flows, region volumes,
interchange rates, and consequently reactor behavior.

Material Balance for Gas and for Solids

From Kunii and Levenspiel (1991) a material balance for the bed material gives

~ ... rise velocity of a single bubble in a

ubr = 0 . 7 1 1 ( ~ d ~ ) ~ 'm/s (10)
bed otherwise at umf

acceleration of gravity = 9.8mls2

u , = u , - umf+ ub1)mls rise velocity of bubbles in a bubbling bed (11)

m3bubbles
S = bed fraction in bubbles,
m3bed
 20.3 The K-L Model for BFB 457

Useful relationships:

H m ( l- Em)= Hmf(1 - Emf)= Hf (1 - Ef)

I -a=-= 1 - E %...H=height
1 - Emf Hf

a8 ub
us = . . . downflow of emulsion solids
i - ~ - ~ m/s
a'

ue = %f- us,mls . . . rise velocity of emulsion gas (can be + or -)

Emf

Using Davidson's theoretical expression for bubble-cloud circulation and the

Higbie theory for cloud-emulsion diffusion the interchange of gas between bubble
and cloud is then found to be

(interchange volume between

c,
Kbc= 4.50
(2) + 5.85 - -- b volume
)
and or c and b)ls
of bubble ,s-I

and between cloud-wake and emulsion

'I2 - interchange volumels -,

volume of bubble , S

volume of solids
fb = 0.001 - 0.01 =
in bubble
volume of bed
. . . rough estimate from experiment

fc = - Emf) [ 3umf/emf
Ubr - IEmf
+,I =
volume of solids in cloud and wake
volume of bed
(16)

fe =
-
(1 - ~ m ~ )-( l8) - fc - f b =
volume of solids in the rest of the emulsion
volume of bed
458 Chapter 20 Fluidized Reactors of Various Types

Application to Catalytic Reactions

In our development we make two questionable assumptions:
* We ignore the flow of gas through the cloud since the cloud volume is very
small for fast bubbles.
We ignore the flow of gas, either up or down, through the emulsion since
this flow is much smaller than the flow through the bubbles.
In effect, we consider the emulsion gas as stagnant. Of course more general
expressions can be developed for beds where bubbles have thick clouds (not too
large and fast bubbles), or where flow through the emulsion is significant (u,
close to umf,thus where u, = 1-2 umf).However, for fast bubble, vigorously
bubbling fine particle beds the above assumptions are reasonable.
We next see how to calculate performance in such a bed.

First-Order Reaction. Let the reaction be

mol
A+,, r A U r = k C A ,-
m3s.s
f

Then for any slice of bed we have

reactant rising 1-transfer

in bubble 6 Kbc,srl-
I reactant
in cloud
/-transfer
6 Kc,, srl-
I reactant
in emulsion

I
reaction: fb k"',s-l
I
reaction: f, k , s-l
I
reaction: fe k , s-I

bubble cloud emulsion

Accounting for these five resistances in series-parallel, eliminating cloud and

emulsion concentrations, and integrating from the bottom to the top of the
 20.3 The K-L Model for BFB 459

bed gives

We also find that the average gas composition seen by the solids is approximately
(from S. Kimura, personal communication)

of solids alone, WIp,

This quantity is important for noncatalytic G/S reactions because it is this CA

that the solids see and react with.
Let us now look at packed bed reactors. Assuming plug flow Kbc + w ,
Kc, + a, so Eq. 21 reduces to

- &lfrf'' k'
= k H p ( l - EP ) = ktT' = -
in (23)
u0 %A,
for plug flow CAP
- C ~ O- C ~ p
C ~ p= J ~ " ~ I I I (24)

Comparing Eq. 21 with 23, and Eq. 22 with 24 shows that a fluidized bed can
be treated as a plug flow reactor if
K is replaced by k'"

Comments. The five terms in brackets of the performance equation, Eq. 21,
represent the complex series-parallel resistances to mass transfer and reaction or

Reactant A transfer + in transfer - in

in bubble to cloud cloud to emulsion emulsion

reaction reaction reaction

in bubble in cloud in emulsion
460 Chapter 20 Fluidized Reactors of Various Types

+- Large bubble

+- Medium size bubble

0.06 -
-- Small bubble

0.04 I I I\ I I I I
2 4 6 8
f H k"'
BFB , dimensionless time
uo

Figure 20.10 Performance of a fluidized bed as a function of bubble size, as deter-

mined by Eq. 21. Compare with the plug flow and mixed flow predictions.

For very fast reaction (large k ' value) very little A gets as far as the emulsion
and the first two terms dominate. For slow reaction the latter terms become
increasingly important.
Since the bubble size is the one quantity which governs all the rate quantities
with the exception of k , we can plot the performance of a fluidized bed as a
function of d,, as shown in Fig. 20.10. Note that large d, gives poor performance
because of extensive bypassing of bubble gas, and that the performance of the
bed can drop considerably below mixed flow.
For multiple reactions the effect of this flow is much more serious still. Thus
for reactions in series the lowering in amount of intermediate formed can be
and usually is quite drastic.
Finally we have kept this presentation very brief. To help understand how to
use it please look at the following illustrative example.

:' First-Order Catalytic Reaction in a BFB

Reactant gas (u, = 0.3 m/s, v, = 0.3 .rr m3/spasses upward through a 2-m diameter
fluidized bed (umf= 0.03 m/s, emf= 0.5) containing 7 tons of catalyst (W = 7000
kg, p, = 2000 kg/m3). Reaction proceeds as follows:

m3
A + R, -rz = kU'C,. . .with k"' = 0.8 -
m3s.s
 20.3 The K-L Model for BFB 461

(a) Calculate the conversion of reactant.

(b) Find the proper mean concentration of A seen by the solids.
(c) If gas were made to flow downward through the solids we would have a
packed bed. Assuming plug flow of gas find the conversion of reactant for
this situation.

I Additional Data
C, = 100 mol/m3, CZj = 20 X m2/s, a = 0.33
Estimated bubble size in the bed: d, = 0.32 m
See Fig. E20.1 which represents this system.

I u,, = 0.3 rn/s

CAO= 1 0 0 rnol/m3

Figure E20.1

1 SOLUTION
Preliminary Determine the rise velocity of bubbles

Now see whether the fast bubble model of this chapter applies.
Check for slugging: The bubble size (32 cm) is small compared to the bed
size (200 cm), hence no slugging.
Check the fast bubble assumption: Take the velocity ratio

Since the bubble rises -25 times as fast as the emulsion gas we have a fast
bubble with thin cloud-less than 1 cm thick.
462 Chapter 20 Fluidized Reactors of Various Types

Thus we can safely use the bubbling bed model of this chapter.
(a) Calculate XA. Replacing numbers into the expressions for this model
gives, in turn,

fb = 0.001 - 0.01, choose it to be 0.001 1

Therefore

CA = 0.68 . . . o r X n = 32% (see Fig. 20.10)

A
'0

(b) Find cAseen by the solids. Since every particle samples all the gas in the
bed, 7'" = (C,, - CA)l-r*, or

(c) Calculate XAfor a fixed bed. From Eq. 11.44 or 11.51, for plug flow,
 20.3 The K-L Model for BFB 463

Therefore

This point is shown in Fig. 20.10.

Comments
The conversion in the fluidized bed is drastically lower than in the packed
bed (32% vs. 95%), and is even much below mixed flow (75%). This comes
from the severe bypassing of reactant gas in the large gas bubbles. Reduce
the size of bubbles in the bed and the conversion will rise spectacularly.
Gas enters at CAo= 100, leaves at C A = 68; however, the solids see a much
lower concentration of A, or C A= [Link] the solids, mostly in the emulsion,
are starved for gaseous reactant. This sort of finding, quite general for fast
bubble beds, is of great importance in GIS reactions. Thus in combustion
and roasting of fine solids these solids may be starved for 0, even though
the off gas from the bed may contain a significant amount of oxygen.
Example 20.1 and the solutions to Problems 20.1 and 20.2 at the end of this
chapter tell how conversion is affected by bubble size according to the
K-L model. This is shown in Fig. 20.11. In all four cases the rate constant
is the same and the catalyst is fixed at 7 tons, so k"' 7'" = f,,,,,HBFB
k l u , = 2.97.

Multiple Reactions and Product Distribution in Fluidized Beds

A derivation similar to that for single first-order reactions can be developed for
the Denbigh reaction system

1.:) t
.. ... .. .. .. .. .. ... .. .::
.....................
. .. .. .. ... .. .'.:. .:...
.:.:. Plug:.....:

. . ...:.:
..........:.......
. . . . . .
..
.;i:;:

. . . . . . .::;' :
:':._. .::.
. . . .;:i.,
_ _:.,:- _ .:' .:
.

t
Figure 20.11 Different bubble size gives different reactor perfor-
mance. Data from Example 20.1 and Problems 20.3 and 20.4.
464 Chapter 20 Fluidized Reactors of Various Types

and its special cases

These derivations are rather tedious, we will not give them here, but we refer
the reader to Levenspiel (1996) Chapter 25 for them. However, to illustrate the
general findings let us take one example:

k;' k;l' k;" = 0.8 m3/(m solid)3 . s

A-R-S
kj" = 0.025 m3/(m solid)3 . s

Let us suppose that these reactions take place in a BFB reactor having a gas
flow rate similar to that of Example 20.1.
Let us see how much catalyst is needed to achieve C,, ,,, for different sizes
of bubbles in the BFB. The results are shown in Fig. 20.12.

I
. . . .

. .

I+;i;
= 9.5 tons
= 15 tons

= 16 cm
.......

N = 32 tons
.&.
. :?,,:.
db = 32 crn
t ),

Figure 20.12 Different bubble sizes give different bed sizes for
maximum production of intermediate.
 20.4 The Circulating Fluidized Bed-CFB 465

The results of the calculations verify the general findings on bubbling fluid-
ized beds.

* The BFB always needs more catalyst than a fixed bed to achieve a given
conversion of reactant, or to reach C,, .,,,
For reactions in series the BFB always gives a lower yield of intermediate
compared to a fixed bed.
For reactions in parallel of the same reaction order the product distribution
is the same in fixed and BFBs.

Final Remarks About BFB

The expressions developed in this chapter show that if we know emf, estimate a,
and measure umfand u,, then all flow quantities and regional volumes can be
determined in terms of one parameter, the bubble size. Figure 20.9 then repre-
sents the model as visualized. The use of this model to calculate chemical reactor
behavior is straightforward and direct. The special feature of this model is that
its one parameter can be tested against what is measured and what is observed.
Various other hydrodynamic models have been proposed recently, using other
combinations of assumptions such as

Changing bubble size with height in the bed

Negligible bubble-cloud resistance
Negligible cloud-emulsion resistance
Nonspherical bubbles

In all cases the underlying rationale for these hydrodynamic models rests on the
observation that beds with identical solids and gas flow rates may develop either
large bubbles or small bubbles depending on bed diameter, distributor design,
baffle arrangement, etc.; thus, bubble size must enter as the primary parameter
in the model. A consequence of this argument is that models which do not
allow for different bubble sizes at given imposed bed conditions certainly cannot
be adequate.
The power of this class of model should be apparent. For example, even the
simplest of these models, the one considered here, gives unexpected predictions
(e.g., that most of the gas in the bed may be flowing downward) which are
subsequently verified. More important still, this type of model can be tested, it
can be shown to be wrong, and it can be rejected, because its one parameter,
the bubble size, can be compared with observation.

20.4 THE CIRCULATING FLUIDIZED BED-CFB

At gas velocities higher than those used for BFBs we successively enter the
turbulent (TB), fast fluidized (FF), and the pneumatic conveying (PC) regimes.
In these contacting regimes solids are entrained out of the bed and must be
replaced. Thus in continuous operations we have the CFB, shown in Fig. 20.1.
Flow models are very sketchy for these flow regimes. Let us see what is known.
466 Chapter 20 Fluidized Reactors of Various Types

- Gas and

-
solids

Lean
.. .. . . . . .
.. .,. .
:. .
. . . .. . . .

Solids

I
0.01-0.02 0.2-0.4
Gas
Figure 20.13 The TB and its solid distribution.

The Turbulent Bed (TB)

At higher gas velocities the BFB transforms into a TB-no distinct bubbles,
much churning, and violent solid movement. The surface of the dense bed fades
and solids are found increasingly in the lean region above the dense bed.
The concentration of solids in the upper lean region can be reasonable repre-
sented by an exponential decay function which starts from the value in the lower
region f, and falls to fik the limiting value in an infinitely high vessel. This is the
value for pneumatic conveying.
The flow of gas in the dense region is somewhere between the BFB and plug
flow. The TB is shown in Fig. 20.13.
Unfortunately no reasonable flow model has been developed for the dense
region of a TB. Research is needed here.
In beds of both coarse and fine solids one may observe a somewhat different
solid distribution with height-a distinct difference between dense and lean
regions and a sharp dense phase surface, as shown in Fig. 20.14. This behavior
is more typical of fluidized combustors, not catalytic reaction systems.

A
Gas and
solids

dense bed

Solids

t
Gas
Figure 20.14 Behavior of a TB of large and small particles.
 Lean

Dense
f--+ solids
Gas and

At a cross
section
Solids
Solid
fraction
20.4

Model

I
I
K;;
The Circulating Fluidized Bed-CFB

Upflow gas
velocity
Observed
467

0 R
Downf low
t
Gas
Wall

\ /
f Wall

At the cross section

Figure 20.15 Behavior of a FF bed.

The Fast Fluidized Bed (FF)

At even higher gas velocity (see Fig. 20.5) the bed enters the FF regime. One
characteristic of this transition is that the entrainment of solids rises dramatically
at this point. Bi et al. (1995) found this transition to occur at

In the FF regime the solid movement in the lower region of the vessel becomes
less chaotic and seems to settle to a lean core surrounded by a denser annulus
or wall zone.
The upper region retains its exponential decay behavior.
Figure 20.15 shows the distribution of solids in the whole FF bed and at a
cross section. The model which represents the FF bed is shown in Fig. 20.16.

Height

mas solids
and

fluid flow)

Solids

Gas
Figure 20.16 Model for a FF reactor.
468 Chapter 20 Fluidized Reactors of Various Types

The quantities that need to be known to predict the behavior of the FF reactor
are "a," Kc,, f X , and f,. Measurements show that

uoa = 2 - 4 s-I for Geldart A solids

= 5 s-l for Geldart AB solids
= 7 s-l for Geldart B solids

Measured values for fX and fd have been summarized by Kunii and Levenspiel
(1995). Values for Kc, have yet to be measured. The best we can do today is
estimate their order of magnitude from K,, and Kc, of the BFB. So at this time
we cannot predict reactor behavior. However, to see how to make the material
balance and conversion calculation see the numerical example given in Kunii
and Levenspiel (1997).

Pneumatic Conveying (PC)

Finally at the highest gas velocities we exceed what is called the choking velocity.
Above this the bed is in pneumatic transport. This transition velocity depends
on the solid flow rate, and according to Bi and Fan (1991) occurs at

In the PC regime particles are well distributed in the reactor, with no wall or
downflow zone, but with a slight decrease in solid fraction with height. So we
can assume plug flow of solids and of gas up the vessel.

The Downflow CFB

Fluidized catalytic cracking reactors, called "cat crackers" or FCC reactors, are

1 +
one of society's most important large-scale reactors. On an average, each such

RGas solids

and f = f* + (fd- f*kaz

Note: very little

. . solids in the vessel
..

Denser Solids

f
A 10.06
I 0.01-0.02
Gas
Figure 20.17 Model for a PC reactor.
 20.4 The Circulating Fluidized B e d - C F B 469

unit processes about 6000 m3/day (40 000 barrelslday) of crude oil to produce
gasoline, diesel fuel, and jet aircraft fuel to power the engines of today. There
are about 420 FCC reactors in the world today, running day and night to satisfy
the needs of our insatiable societies.
These reactors take long chain hydrocarbons and crack them to produce a
whole host of shorter chain hydrocarbons. To illustrate

we want
octane

The original cat cracker was invented in the early 1940s and was one of
chemical engineering's most important contributions to World War 11. However,
those early units did not use very selective catalysts and, with their large deviations
from plug flow, gave very little of whatever intermediate was desired.
In the 1960s better (more selective) and more active catalysts were created,
and the needed reaction time for the oil vapor was consequently reduced to
seconds, so the upflow FF reactors were invented. Approaching closer to plug
flow gave the designer better control of product distribution, and allowed produc-
tion of a larger fraction of desired product: for example, octane for automo-
bile fuel.
With this development, most of the BFBs in the world were cut up for scrap
to be replaced by upflow FF reactors.
Today, approaching year 2000, engineers want to do better still. Why? Because
increasing the production of desired product by 1% would increase the profit
per reactor by $1millon to $2 millionlyear. As Fig. 20.16 shows, the upflow FF
reactor has a practically stagnant or a downsliding wall zone of catalyst and gas,
and this results in deviation from plug flow.
How to avoid this? The obvious answer is to use a downflow FF reactor. These
can operate at very high gas velocities, and remain close to plug flow.
Today much exciting research is going on in this area, and we may some day,
not too far in the future, see many of the upflow short contact time FCC units
be replaced by downflow units. The payoff for this would be enormous.

Remarks About CFB

Here we have sketched the three regimes of CFB and their general behavior;
however, we have not presented their performance equations. The reason is that
the parameters for their reasonable models are uncertain today; hence, the
predictions based on these models will likewise be uncertain. Still the general
material balances and the form of the performance equations are available, see
Kunii and Levenspiel (1991, 1997).
470 Chapter 20 Fluidized Reactors of Various Types

20.5 THE JET IMPACT REACTOR

The idea here is to force two streams, one of reactant, the other of a very hot
heat carrier or catalyst, to collide at very high velocity and thereby mix intensely
and react at high temperature.
For an all-gas product, the product stream is rapidly quenched, while for a
gas-solid product, a cyclone separates the two phases, after which the gas is rapidly
cooled. By using the word "rapidly" we mean that the whole operation-mixing,
reacting, separating and quenching-is done in 0.1 to 0.3 seconds.
This type of reactor aims to challenge fast fluidization with its 1 to 10 second
gas residence time as the prime reactor for the catalytic cracking of petroleum.
The claim is that the higher cracking temperature and shorter residence time
will give a very different-and better-distribution of reaction products.
Another application is to ultrapyrolyze cellulose and other biomass wastes.
Commercial tests show that one can transform about 75% of wood into oil, and
about 70% of sawdust into oil-useful liquids having the consistency of light engine
oil. Much research is proceeding on this application (from Bergougnou, 1998).

REFERENCES

Bergougnou, M. A. private communication, 1998.

Bi, H. T., and Fan, L. S., Paper 101e, [Link].E. Annual Meeting, Los Angeles, 1991.
Bi, H. T., Grace, J. R., and Zhu, J. X., Trans. I. Chem. E., 73,154 (1995).
Davidson, J. F., and Harrison, D., Fluidized Particles, Cambridge Univ. Press, New
York, 1963.
Ergun, S., Chem. Eng. Prog., 48, 89 (1952).
Geldart, D., Powder Technol., 7, 285 (1973).
-, and Abrahamson, A. A., Powder Technol., 19,133 (1978).
Gilliland, E. R., and Knudsen, C. W., Chem. Eng. Prog. Symp. Ser. 67 (116), 168 (1971).
Grace, J. R., Can. J. Chem. Eng., 64, 353 (1986).
Haider, A., and Levenspiel, O., Powder Technol., 58, 63 (1989).
Kimura, S., private communication.
Kunii, D., and Levenspiel, O., Fluidization Engineering, 2nd ed., Butterworth-Heinemann,
Boston, MA, 1991.
-, in Fluidization VIZI, Tours, p. 17 (1995).
-, Chem. Eng. Sci., 52, 2471 (1997).
Levenspiel, O., Chemical Reactor Omnibook, OSU Bookstores, Corvallis, OR, 1996.
Rowe, P. N., and Partridge, B. A., Proc. Symp. on Interaction between Fluids and Particles,
p. 135, I. Chem. E. (1962); Trans. I. Chem. E., 43,157 (1965).

PROBLEMS

A suggestion for raising the conversion in Example 20.1 is to use more

solids. So for the same bed diameter (2 m) and same mean bubble size
(0.32 m) find XAand the mean concentration of gas seen by the solids CA
if we use
20.1. . . . 14 tons of catalyst
20.2. . . . 21 tons of catalyst
 Problems 471

In Example 20.1 the conversion is very low and unsatisfactory. One sugges-
tion for improvement is to insert baffles into the bed and thereby cut down
the effective bubble size. Find X, if

20.3. . . . d, = 16 cm

20.4. . . . d, = 8 cm

In Example 20.1 another suggestion for improving performance is to use a

shallower bed, keeping W unchanged, thereby decreasing the superficial
gas velocity. Find X, if we double the bed cross-sectional area (thus make
u, = 15 c d s ) and

20.5. . . . keep db unchanged at d, = 32 cm

20.6. . . . make d, = 8 cm by suitable baffles

20.7. In Example 20.1 still another suggestion is to use a narrower and taller
bed, keeping W unchanged. Following this suggestion find X , if we halve
the bed cross-sectional area (thus make u, = 60 cmls) and keep d, = 32 cm.

20.8. Example 20.1 showed that a 7-ton bed with 32-cm bubbles gave 32% conver-
sion. Suppose we diluted the catalyst 1: 1 with inert solids ending up with
a 16ton bed. How would this affect the conversion of reactant? Would it
be higher or lower than 32%?

20.9. Mathis and Watson in AIChE J., 2, 518 (1956) reported on the catalytic
conversion of cumene in both fluidized and packed beds of catalyst

C9Hl2+ 0 2 ~ C 6 H 5 0+H(CH,),O
catalyst

cumene phenol acetone

+C
In very dilute cumene-air mixtures the kinetics are essentially first-order
reversible with respect to cumene with an equilibrium conversion of 94%.
In packed bed experiments (H, = 76.2 mm) using downflow of gas
(u, = 64 m d s ) the conversion of cumene was found to be 60%. However,
with upflow of gas at the same flow rate in the same bed, the solids fluidize
(u,, = 6.1 mmls) and bubbles of gas were observed roughly 13.5 mm in
diameter. What conversion do you expect to find under these conditions?
Estimated values:
472 Chapter 20 Fluidized Reactors of Various Types

20.10. Calculate the conversion in a fluidized bed reactor for the catalytic hydro-
genation of nitrobenzene to analine.

catalyst
C6H5N02(g) + C6H~NH2(g)+ H2°(g)
(A) (R)
Data: H, = 1.4 m p, = 2.2 g/cm3 d, = 10 cm
d, = 3.55 m !Zeff= 0.9 cm2/s a = 0.33
T = 270°C umf= 2 cm/s E, = 0.4071
u, = 30 cm/s = 0.60

We plan to use an excess of hydrogen in which case we can ignore expan-

sion, and can assume simple first-order kinetics:

A +R -rx = k"' C , with kN'= 1.2 cm3/cm3cat . s

20.11. In a laboratory packed bed reactor (H, = 10 cm and u, = 2 cmls)

conversion is 97% for the first-order reaction A + R.
(a) Determine the rate constant k"' for this reaction.
(b) What would be the conversion in a larger fluidized bed pilot plant
( H , = 100 cm and u, = 20 cmls) in which the estimated bubble size
is 8 cm?
(c) What would be the conversion in a packed bed under the conditions
of part (b)?
Data: From experiment, umf= 3.2 cmls, emf= E, = 0.5

From the literature, !Z = Be= 3 X cm2/s, a = 0.34

Chapter 21
Deactivating Catalysts

The previous chapters assumed that the effectiveness of catalysts in promoting

reactions remains unchanged with time. Often this is not so, in which case the
activity usually decreases as the catalyst is used. Sometimes this drop is very
rapid, in the order of seconds; sometimes it is so slow that regeneration or
replacement is needed only after months of use. In any case, with deactivating
catalysts regeneration or replacement is necessary from time to time.
If deactivation is rapid and caused by a deposition and a physical blocking of
the surface this process is often termed fouling. Removal of this solid is termed
regeneration. Carbon deposition during catalytic cracking is a common example
of fouling

If the catalyst surface is slowly modified by chemisorption on the active sites

by materials which are not easily removed, then the process is frequently called
poisoning. Restoration of activity, where possible, is called reactivation. If the
adsorption is reversible then a change in operating conditions may be sufficient to
reactivate the catalyst. If the adsorption is not reversible, then we have permanent
poisoning. This may require a chemical retreatment of the surface or a complete
replacement of the spent catalyst.
Deactivation may also be uniform for all sites, or it may be selective, in which
case the more active sites, those which supply most of the catalyst activity, are
preferentially attacked and deactivated.
We will use the term deactivation for all types of catalyst decay, both fast and
slow; and we will call any material which deposits on the surface to lower its
activity a poison.
This chapter is a brief introduction to operations with deactivating catalysts.
We will consider in turn:
The mechanisms of catalyst decay
The form of rate equation for catalyst decay
How to develop a suitable rate equation from experiment
* How to discover the mechanism from experiment
Some consequences for design
474 Chapter 21 Deactivating Catalysts

Although this is basically a rather involved subject, still its very importance from
the practical standpoint requires at least an introductory treatment.

21.1 MECHANISMS OF CATALYST DEACTIVATION

The observed deactivation of a porous catalyst pellet depends on a number of
factors: the actual decay reactions, the presence or absence of pore diffusion
slowdown, the way poisons act on the surface, etc. We consider these in turn.

Decay Reactions. Broadly speaking, decay can occur in four ways. First, the
reactant may produce a side product which deposits on and deactivates the
surface. This is called parallel deactivation. Second, the reaction product may
decompose or react further to produce a material which then deposits on and
deactivates the surface. This is called series deactivation. Third, an impurity
in the feed may deposit on and deactivate the surface. This is called side-by-
side deactivation.
If we call P the material which deposits on and deactivates the surface, we
can represent these reactions as follows:
Parallel deactivation:

Series deactivation:

Side-by-side deactivation:

The key difference in these three forms of decay reactions is that the deposition
depends, respectively, on the concentration of reactant, product, and some other
substance in the feed. Since the distribution of these substances will vary with
position in the pellet, the location of deactivation will depend on which decay
reaction is occurring.
A fourth process for catalyst decay involves the structural modification or
sintering of the catalyst surface caused by exposure of the catalyst to extreme
conditions. This type of decay is dependent on the time that the catalyst spends
in the high temperature environment, and since it is unaffected by the materials
in the gas stream we call it independent deactivation.

Pore Diffusion. For a pellet, pore diffusion may strongly influence the progress
of catalyst decay. First consider parallel deactivation. From Chapter 18 we know
that reactant may either be evenly distributed throughout the pellet (M, < 0.4
 21.2 The Rate and Performance Equations 475

and 6 = 1) or may be found close to the exterior surface (M, > 4 and 8< 1).
Thus the poison will be deposited in a like manner-uniformly for no pore
resistance, and at the exterior for strong pore resistance. In the extreme of very
strong diffusional resistance a thin shell at the outside of the pellet becomes
poisoned. This shell thickens with time and the deactivation front moves inward.
We call this the shell model for poisoning.
On the other hand, consider series deactivation. In the regime of strong pore
resistance the concentration of product R is higher within the pellet than at
the exterior. Since R is the source of the poison, the latter deposits in higher
concentration within the pellet interior. Hence we can have poisoning from the
inside-out for series deactivation.
Finally, consider side-by-side deactivation. Whatever the concentration of re-
actants and products may be, the rate at which the poison from the feed reacts
with the surface determines where it deposits. For a small poison rate constant
the poison penetrates the pellet uniformly and deactivates all elements of the
catalyst surface in the same way. For a large rate constant poisoning occurs at
the pellet exterior, as soon as the poison reaches the surface.
The above discussion shows that the progress of deactivation may occur in
different ways depending on the type of decay reaction occurring and on the
value of the pore diffusion factor. For parallel and series poisoning, the Thiele
modulus for the main reaction is the pertinent pore diffusion parameter. For side-
by-side reactions, the Thiele modulus for the deactivation is the prime parameter.
Nonisothermal effects within pellets may also cause variations in deactivation
with location, especially when deactivation is caused by surface modifications
due to high temperatures.

Additional Factors Influencing Decay. Numerous other factors may influence

the observed change in activity of catalyst. These include pore mouth blocking
by deposited solid, equilibrium, or reversible poisoning where some activity
always remains, and the action of regeneration (this often leaves catalyst with
an active exterior but inactive core).
Most important of all, the observed deactivation may result from a number
of processes at work simultaneously; for example, the speedy immobilization of
the most active sites by poison PI, and then the slower attack of the remainder
of the sites by P,.
Although the possible influence of all these factors should be examined in the
real case, in this introductory treatment we will concentrate on the first two
factors: the decay reaction and pore diffusion. There are enough lessons here to
illustrate how to approach the more complete problem.

21.2 THE RATE AND PERFORMANCE EQUATIONS

The activity of a catalyst pellet at any time is defined as

rate at which the pellet converts reactant A - - T i

a= --
rate of reaction of A with a fresh pellet
476 Chapter 21 Deactivating Catalysts

and in terms of the fluid bathing the pellet the rate of reaction of A should be
of the following form:

main stream main stream

temperature concentration
(5)
present activity of
the catalyst pellet

Similarly, the rate at which the catalyst pellet deactivates may be written as

( d e a ~ t i o n )= ,(main stream
temperature
main stream
concentration
(6)
present state of
the catalyst pellet

In terms of nth-order kinetics, Arrhenius temperature dependency, and iso-

thermal conditions, Eq. 5 becomes, for the main reaction:

and for deactivation which in general is dependent on the concentration of gas-

phase species, Eq. 6 becomes

where d is called the order of deactivation, m measures the concentration depen-

dency and Ed is the activation energy or temperature dependency of the deacti-
vation.
For different decay reactions we may expect different forms for the above
equations. Thus

-ri = ktCia
For parallel deactivation
(A+R;A+PJ) da -
-- - k Cmad
dt d A

= k tC i a
For series deactivation
(A+R+PJ) da -
-- k cmad
dt d R
 21.2 The Rate and Performance Equations 477

-7; = k rC i a
For side-by-side deactivation
(A+R;P-Pi) da -
-- - k p a d
dt d P

-r; = k1C%a
For independent deactivation
(concentration independent) -4 = kdad
dt

In certain reactions, such as isomerizations and cracking, deactivation may be

caused both by reactant and product, or

A-R
da
AdPJ and ---- kd(CA
dt
+ CR)mad
R+Pi

Since CA + C, remains constant for a specific feed this type of deactivation

reduces to the simple-to-treat independent deactivation of Eq. 12.
Although the above nth-order expressions are quite simple they are general
enough to embrace many of the decay equations used to date [see Szepe and
Levenspiel (1968)l.

The Rate Equation from Experiment

Experimental devices for studying deactivating catalysts fall into two classes:
those which use a batch of solids, and those which use a through flow of solids.
Figure 21.1 shows some of these devices.

Batch solids-for slow deactivation

8 -+QCL
Batch for both
solid and fluid
Plug flow
for fluid
M~xedflow
for fluid
Recycle flow
for fluid

Flow of solids-for fast deactivation

Solids: mixed flow

Solids: - plug flow Fluid: questionable
flow pattern
Fluid: - plug flow Raining
solids

Figure 21.1 Slow deactivation may use a batch of solids in experimentation; fast
deactivation requires a flow of solids.
478 Chapter 21 Deactivating Catalysts

Because of the ease of experimentation batch-solids devices are much pre-

ferred; however, they can only be used when deactivation is slow enough (in the
order of minutes or longer) so that sufficient data on the changing fluid composi-
tion can be obtained before exhaustion of the catalyst. When deactivation is
very rapid (in the order of seconds or less) then a flowing-solids system must be
used. Cracking catalysts whose activity half-lives can be as short as 0.1 second
fall into this class.
The method of searching for a rate equation is analogous to that of homoge-
neous reactions: start with the simplest kinetic form and see if it fits the data. If
it doesn't, try another kinetic form and so on. The main complication here is
that we have an extra factor, the activity, to contend with. Nevertheless, the
strategy is the same; always start by trying to fit the simplest rate expression.
In the following sections we treat the batch-solids devices in some detail, and
then briefly consider the flowing-solids system.
Now the type of batch-solids flowing-fluid reactor that we find convenient
to use depends on whether the deactivation expression daldt is concentration
independent or not. When it is concentration independent, any type of batch-
solids system may be used and can be analyzed simply,but when it is concentration
dependent, then unless one particular type of reactor is used (that in which CA
is forced to stay unchanged with time) the analysis of the experimental results
becomes horrendously awkward and difficult.
We treat these two classes of experimental devices in turn.

Batch-Solids: Determining the Rate for Independent Deactivation

Let us illustrate how to interpret experiments from the various batch-solids
reactors of Fig. 21.1 and how to manipulate the basic performance equations for
these reactors by testing the fit for the simplest of equation forms for indepen-
dent deactivation.

with

This represents first-order reaction and first-order deactivation which, in addition,

is concentration independent.

Batch-Solids, Batch-Fluid. Here we need to develop an expression relating the

changing gas concentration with time. Using time as the one independent variable
throughout the run the kinetic expressions of Eq. 14 become
 21.2 The Rate and Performance Equations 479

Integrating Eq. 16 yields

and for unit initial activity, or a, = 1, this becomes

a = e-kdt

Replacing Eq. 18 in Eq. 15 we then find

and on separation and integration

c,
In-=- Wk' (1 - e - k d t )
CA Vkd

This expression shows that even at infinite time the concentration of reactant in
an irreversible reaction does not drop to zero but is governed by the rate of
reaction and of deactivation, or

Combining the above two expression and rearranging then yields

A plot as shown in Fig. 21.2 provides a test for this rate form.

Figure 21.2 Test of the kinetic expressions

of Eq. 14 using a batch-solids, batch-fluid re-
actor.
480 Chapter 21 Deactivating Catalysts

The batch-batch reactor becomes a practical device when the characteristic

times for reaction and deactivation are of the same order of magnitude. If they
are not and if deactivation is much slower, then CA, becomes very low and
difficult to measure accurately. Fortunately, this ratio can be controlled by the
experimenter by proper choice of WIV.

Batch-Solids, Mixed Constant Flow of Fluid. Inserting the rate of Eq. 14a iuto
the performance expression for mixed flow gives

and on rearrangement

In Eq. 24 the activity varies with the chronological time. To eliminate this quantity
integrate Eq. 14b (see Eq. 18) and insert in Eq. 24. Thus

Rearranging gives, in more useful form,

This expression shows how the reactant concentration at the outlet rises with
time, while the plot of Fig. 21.3 provides the test of this kinetic equation. If the

Figure 21.3 Test of the kinetic expressions

of Eq. 14 using a batch of solids and steady-
state mixed flow of fluids.
 21.2 The Rate and Performance Equations 481
data fall on a straight line then the slope and intercept yield the two rate constants
of Eq. 14.
We should mention that this and the succeeding batch-solids derivations are
based on the pseudo steady-state assumption. This assumes that conditions
change slowly enough with time for the system to be at steady state at any
instant. Since a batch of solids can only be used if deactivation is not too rapid
this assumption is reasonable.

Batch-Solids, Mixed Changing Flow of Fluid (to keep C, fixed). For steady
flow in a mixed reactor we have found

To keep CAconstant the flow rate must be slowly changed with time. In fact, it
must be lowered because the catalyst is deactivating. Hence the variables in this
situation are r 1 and t. So, on rearranging we have

Figure 21.4 shows how to test for the kinetic expressions of Eq. 14 by this pro-
cedure.
Actually there is no particular advantage in using varying flow over constant
flow when testing for the kinetics of Eq. 14 or any other independent deactivation.
However, for other deactivation kinetics this reactor system is by far the most
useful because it allows us to decouple the three factors C, T, and a and study
them a pair at a time.

Ir
lr kcreax
in flow
w+
-., ~

. -
2 7

c~~- c~
Intercept = In -
k'CA

Figure 21.4 Test of the kinetic expressions

of Eq. 14 using a batch of solids and chang-
ing flow of fluid in a mixed reactor so as to
keep C, constant.
482 Chapter 21 Deactivating Catalysts

Figure 21.5 Test of the kinetic expressions

of Eq. 14 using a batch of solids and steady-
state plug flow of fluid.

Batch-Solids, Plug Constant Flow of Fluid. For plug flow the performance
equation combined with the rate of Eq. 14a becomes

Integrating and replacing a by the expression of Eq. 18 gives

which becomes, on rearrangement,

Figure 21.5 shows how to test for the kinetics of Eq. 14, and how to evaluate
the rate constants with data from this type of reactor.

Batch-Solids, Plug Changing Flow of Fluid (to keep CA,out fixed). At any instant
in the plug flow reactor Eq. 29 applies. Thus noting that r 1 and t are the two
variables we obtain, on suitable rearrangement,

Figure 21.4, with one modification (the intercept given by the last term of Eq.
31), shows how to test for the kinetic expressions of Eq. 14 with this device.
So far we have illustrated how to use a batch, plug flow or mixed flow of fluid
to search for the rate constants of a particular rate form, Eq. 14.
 21.2 The Rate and Performance Equations 483

As long as the deactivation is concentration independent, the activity and

concentration effects can be decoupled and any of the above experimental devices
will give simple easy-to-interpret results. Thus the above analyses can be extended
with no difficulty to any reaction order n and any deactivation order d, as long
as m = 0, or for

On the other hand, if the deactivation is concentration dependent, or m # 0,

then the concentration and activity effects do not decouple and analysis becomes
difficult unless the proper experimental device is used, one deliberately chosen
so as to decouple these factors. Reactors which keep C, constant by slowing the
flow rate of feed, as treated above, are such devices. They can be used to find
the rate constants for the reactions

as well as for parallel, series, and side-by-side deactivation [see Levenspiel

(1972, 1996)l.

How Pore Diffusion Resistance Distorts the Kinetics of Reactions with

Deactivating Catalysts
Consider the following rate scheme for reactions on spherical particles

-
da = kdad
-
dt
1where k'p, = k

For n o deactivation but with or without pore diffusion resistance these rate
expressions become

where

& = 1 for no diffusional slowdown

1
&=-, 1 M,=L k' for strong diffusional
MT
484 Chapter 21 Deactivating Catalysts

With deactivation, Eq. 34 above becomes

where

L = 1 for no diffusional slowdown

1=
1
for strong diffusional
& M -L
Td -
MT; resistance

Also as time passes, from Eq. 37,

a=exp(-k,t) ford=l
a = [I + (d - l)k,t]ll(l-d) for d # 1

These expressions show that in the regime of strong pore diffusion resistance a
decreases (see Eq. 39), causing MTdto also decrease. This means that 8 rises
with time, as shown in Fig. 21.6. However a decreases faster than &rises so the
reaction rate decreases with time.

here

Figure 21.6 Effectiveness factor increases with time as catalyst

deactivates: batch Slplug flow G.
 21.2 The Rate and Performance Equations 485

Table 21.1 Performance equations for plug flow Glbatch S, with the simple first
order system of Eq. 14
No Resistance to Pore Diffusion Strong Resistance to Pore Diffusion
C ~ O CAO kt7'
In-=-
Without In - = k'7'
deactivation
(40) CA MT . (41)
(Chapters 5, 18) (Chapter 18)

c ~ O
In - = (k'a)~'
CAO (kla)+
In-=-
With cA . (42) CA MTd
. . (43)
deactivation = k17' ' exp(-kdt)
k17'
--
MT
. -p(- y)
Performance Equations in the Regime of Strong Diffusional Resistance
There are so many different forms of deactivation rate, it is not worthwhile
trying to present their performance equations. Let us illustrate what happens
with just one rate form, the simplest,

For batch Slplug flow G integration of the various diffusion deactivation regimes
gives the values in Table 21.1.
For given k ' ~ ' in
, other words for given treatment rate, Fig. 21.7 shows that CA
at the exit rises with time, or that conversion decreases with time.

h
Eq. 27
k'~'

Eq. 42, slope = -kd ... no pore diffusion

resistance
kd
Eq. 43, slope = --
* ... tostrong resistance
pore diffusion

*t

Figure 21.7 With deactivation and pore diffusional resistance

reactor performance decreases with time.
 486 Chapter 21 Deactivating Catalysts

Comments

(a) For mixed flow we get equations and charts identical to the above except
for the following change

(b) For other reaction orders and deactivation orders similar equations can
be developed for the above four kinetic regimes. For example, for second-
order reaction batch Slmixed flow G we have, with Eq. 18.28

(c) Figure 21.7 for first-order systems shows that the conversion decreases
more slowly in the presence of strong pore diffusion resistance than in the
diffusion-free regime. This result should apply generally to all reactor types
and reaction kinetics.

INTERPRETING KINETIC DATA IN THE PRESENCE OF

PORE DIFFUSION RESISTANCE AND DEACTIVATION
I
The catalytic decomposition of reactant (A+R) is studied in a packed bed
reactor filled with 2.4-mm pellets and using a very high recycle rate of product
gases (assume mixed flow). The results of a long-time run and additional data
are given below.

Be= 5 X 10-lom3/m cat. s

t, hr O
ps = 1500 kg/m3cat
XA 0.75 0.64 0.52 0.39
+ = 4000 kg s,m3

Find the kinetics of reaction and deactivation, both in the diffusion-free and in
the strong pore diffusion resistance regime.

I SOLUTION
First of all, note that deactivation occurs during the run, so guess or try to fit
the data with the simplest rate form for such situations, or
 21.2 The Rate and Performance Equations 487

The performance equations for this rate form and mixed flow are analogous to
those for plug flow, shown in Table 21.1,

In diffusion-free = k'T1a = ktT1e-kdl

regime

In strong diffusion a=
regime
- (ii)

see Eq. 38

If this rate form is correct then for both strong or no diffusion resistance effects
a plot of in (c
c*, - 1) vs t should give a straight line. Making this tabulation
and plot, as shown in Fig. E21.1 shows that the above rate form fits the data.

1.2 k Intercept = I . 1

0 0.75 3 1.10
2 0.64 1.78 0.58
Slope = -0.26 4 0.52 1.08 0.08
6 0.39 0.64 -0.45
0.4

-0.4 1 \ /This is a good straight line; hence,

the guessed rate form fits the data.

Figure E21.1

Let us next see whether the reactor is operating in the diffusion-free or strong
diffusion regime.

Guess No Intrusion of Diffusional Resistances. Then from Eq. (i) we have

(2 1
In -- 1 = In ( k ' d ) - kdt

From Fig. E21.1

the intercept = 1.1

the slope = -0.26
} thus [ kt = 7.5 x

kd = 0.26 hr-l
m3
kgas
488 Chapter 21 Deactivating Catalysts

Now check the Thiele modulus to verify that we really are in the diffusion
resistance free regime. Thus at t = 0

Unfortunately, this value indicates strong resistance to pore diffusion. This con-
tradicts our starting assumption, so our original guess was wrong.

Guess That the Runs Were Made in the Regime of Strong Resistance to Pore
Diffusion. Then Eq. (ii) becomes

From Fig. E21.1 we have

the intercept = 1.1

the slope = -0.26
kd = 0.52 hr-l

From the value for the intercept

we find

Check the Thiele modulus at t = 0

This value of Thiele modulus represents strong pore diffusional resistance. This
is consistent with our original guess.
Hence the final rate equations:
(a) In the diffusion-free regime (for very small d,) with deactivation

mol .
- da
. with . - . - -
-rk = 0.27 C,a, = 0.52 a, hr-I
kg-s dt
 21.3 Design 489

(b) In the strong pore digusion resistance regime (for large d,) with deacti-
vation

-rA = 0.27 C,a& ... with ... da - 0.52 a

--
dt
f

combining all gives

3 lo-' CAal",
-,-i= - mol . . with . da
L
-
kgas
. - - = 0.52 a, hr-l
dt

Note: In the strong pore diffusion regime the rate is lower but the catalyst
deactivates more slowly. Actually, for the catalyst used here if we could have
been free of diffusional resistances reaction rates would have been 360 times
as fast as those measured.

Reflection on this example leaves us with the impression that even with the
simplest of rate forms the analysis is quite involved. It suggests that it is not
worthwhile in this general text to try to treat more complicated rate forms.

21.3 DESIGN
When reacting fluid flows through a batch of deactivating catalyst the conversion
drops progressively during the run, and steady state conditions cannot be main-
tained. If conditions change slowly with time, then the average conversion during
a run can be found by calculating the steady-state conversion at various times
and summing over time. In symbols, then,

When conversion drops too low the run is terminated, the catalyst is either
discarded or regenerated, and the cycle is repeated. Example 21.2 illustrates this
type of calculation.
There are two important and real problems with deactivating catalysts.
The operational problem: how to best operate a reactor during a run. Since
temperature is the most important variable affecting reaction and deactivation
this problem reduces to finding the best temperature progression during the run.
The regeneration problem: when to stop a run and either discard or regenerate
the catalyst. This problem is easy to treat once the first problem has been solved
for a range of run times and final catalyst activities. (Note: each pair of values
for time and final activity yields the corresponding mean conversion.)
490 Chapter 21 Deactivating Catalysts

XA decreases with time as

catalyst deactivates.
Constant flow
Thus FR decreases with time
Isothermal
XA and composition stay
Flow decreases with constant. However, F R
time so as to keep decreases with time.
XA constant.

XA and FR are kept

Constant molar constant.
flow 4 @
Temperature is boosted
to keep XA constant.

Figure 21.8 Three possible reactor policies for a batch of deactivat-

ing catalyst.

The operational problem has been solved analytically for a rather general
family of kinetic equations:

Note the restriction here, that deactivation is concentration independent.

Consider the operational problem. For rather slow deactivation we can use a
batch of catalyst with a choice of running the reactor in one of three ways, as
shown in Fig. 21.8. Whatever policy is chosen the performance equations for the
rates of Eqs. 41 and 42 are obtained by solving the following expressions:

with

The integrated performance equations take on a whole host of forms

for plug or mixed flow
for n = 1,2, . . .
for d = 0, 1, 2, 3
Some of these are given in Chapter 32 of Levenspiel (1996). We do not present
them here.
In general we can evaluate zA
b solving Eq. 40 with Eqs. 43 and 44. Referring
to Fig. 21.8 we see that for case 6
a decreases with time, and so does [Link]
 21.3 Design 491

case aa decreases with time so the feed rate must also be lowered so that at
any time

.
a , ,
AO, inihal

Nearly always we find that case @ is optimum. Here Eq. 43 becomes

constant constant

so klaCA, should remain constant as the temperature is raised. This means that
in the regime of n o pore diffusion resistance we must change T with time such that

ideal gases with

nth-order reactions

In the regime of strong resistance to pore diffusion we must change T with time
such that

klaCBe= constant . - . for liquids

ideal gases and
nth-order reactions

Discussion
The overall problem is finding which policy @ a @ is best. We have the
following findings:
( a ) For the longest run time for given FAand for any a,,,, if the deactivation
is very temperature sensitive compared to reaction, then the optimal policy
is a rising temperature with time so as to keep the conversion from the
reactor unchanged, thus use policy @, ending the run at the maximum
allowable temperature P,as shown in Fig. 21.9, see Szepe (1966, 1968,
1971).
(b) If deactivation is less temperature sensitive than is the reaction use policy
a
@ or whichever is more convenient, but make the run at the maximum
allowable temperature T*.
(c) For deactivations which are concentration dependent the catalyst activity
will vary with position in the reactor. The optimum requires that the
temperature vary with position along the reactor as well as with time. This
is a difficult situation to analyze.
 492 Chapter 21 Deactivating Catalysts

Maximum allowable temperature

/-
Constant XA

I
1
trun trun
Starting T Correct Starting T
is too low starting T is too high
(a) (b) (4
Figure 21.9 These sketches show the correct and incorrect starting temperatures for policy @.

(d) For catalysts which deactivate very rapidly packed bed reactors are imprac-
tical, and one has to use solid-flow systems. We do not treat these here.
Such systems are treated in Chapter 33 of Levenspiel (1996).

DEACTIVATION IN A PACKED BED REACTOR

We plan to run an isomerization of A to R in a packed bed reactor (pure A

feed, FA,= 5 kmollhr, W = 1-ton catalyst, T = 3 atm, T = 730 K). The catalyst
deactivates so we plan to make 120 day runs, then regenerate the catalyst.
(a) Plot conversion and activity versus time for the run
(b) find & for the 120 day run
The rate of reaction with CAin mol/m3 is described by

mol A
= 0.2 C i a
kg cat. hr

I and the rate of deactivation is given by

I (a) . . . - -da
-- 8.3125 X lo-',
dt
day-'
This expression represents poisoning by a feed impurity.
da
(b) . . . - - = 10-3(CA+ C,)a, day-l
dt
This represents poisoning by both reactant and by product, thus pore
diffusion resistance does not influence the deactivation rate.
(c). . . - - da
- - 3.325 a',
dt
day-l
This represents fairly strong pore diffusion resistance.
da
( d ) . . . - - - - 666.5 a3, day-'
dt
This represents very strong pore diffusion resistance.
 21.3 Design 493

SOLUTION

In general for parts (a), (b), (c), and (d) we can write

mol A
-ri = 0.2 C i a
kg cat. hr
PAO-
CAo=-- 3 atm = 50-
mol
[Link] m3
RT (82.06 x -
mol .K

(1000 kg) (so 2)

(5000 e) = 10-
kg. hr
m3

or on rearranging

Let us now replace the activity term in Eq. (i).

Part (a)

or a = 1 - 8.3125 X t (ii)

Replacing Eq. (ii) in Eq. (i) and evaluating XA at various values of t gives the
topmost curve shown in Fig. E21.2.

da
Part (b) Separating and integrating - - = 10-3(CA C,)a gives
dt
+
 494 Chapter 21 Deactivating Catalysts

I 4

Time, days
Figure E21.2 Decrease in conversion as a function of time for various
deactivation orders.

Replacing Eq. (iii) in Eq. (i) and evaluating X, at various t values gives the
corresponding curve in Fig. E21.2.

da
Part (c) Separating and integrating - - = 3.325 a2 gives
dt

Again, replacing Eq. (iv) in (i) gives the corresponding curve in Fig. E21.2.

da
Part (d) Separating and integrating -- = 666.5 a3 gives
dt

1 Combining Eq. (v) in (i) gives the lowest curve shown in Fig. E21.2.
 Problems 495

From Fig. E21.2 we find by graphical integration that for the 120-day period,
XA,,,,,, = 0.99, X, ,, = 0.20 for

d = 0 the mean conversion FA= 0.96

d = 1the mean conversion X, = 0.73
d = 2 the mean conversion FA= 0.40
d = 3 the mean conversion XA= 0.30

I It also clearly shows how different is the progress of the reaction with the different
orders of deactivation.

REFERENCES

Levenspiel, O., J. Catal., 25, 265 (1972).

, Chemical Reactor Omnibook, OSU Bookstores, Corvallis, OR, 1996.
Szepe, S., Ph.D. Thesis, Illinois Institute of Technology, 1966; also see Szepe, S., and 0 .
Levenspiel, Chem. Eng. Sci., 23, 881 (1968); "Catalyst Deactivation," p. 265, Fourth
European Symposium on Chemical Reaction Engineering, Brussels, September 1968,
Pergamon, London, 1971.

PROBLEMS

The kinetics of a particular catalytic reaction A +R are studied at temperature

T in a basket reactor (batch-solids and mixed flow of gas) in which the gas
composition is kept unchanged, despite deactivation of the catalyst. What can
you say about the rates of reaction and deactivation from the results of the
following runs? Note, to keep the gas concentration in the reactor unchanged
the flow rate of reactant had to be lowered to about 5% of the initial value.

21.1. . . . Run 1
CAo= 1mollliter
XA= 0.5
t, time from start of run, hr
rl, gm cat. minlliter 1 0 1 2 3
1 e e2 e3

Run 2
CAo= 2 mollliter t, hr
XA= 0.667 rl, gm cat. minlliter l 2 o 2e l 2e2z 2e3
3

21.2. . . . Run 1
CAo= 2 mollliter t, hr
XA= 0.5 rl, [Link]/liter
496 Chapter 21 Deactivating Catalysts

Run 2
C, = 20 mollliter t, hr
XA = 0.8 TI,[Link]/liter

21.3. In an automobile's catalytic converter, CO and hydrocarbons present in

the exhaust gases are oxidized. Unfortunately the effectiveness of these
units decreases with use. The phenomenon was studied by Summers and
Hegedus in J. Catalysis, 51, 185 (1978) by means of an accelerated aging
test on a palladium impregnated porous pellet packed bed converter.
From the reported data on hydrocarbon conversion shown below, develop
an expression to represent the deactivation rate of this catalyst.

This problem was prepared by Dennis Timberlake.

21.4. A recycle reactor with very high recycle ratio is used to study the kinetics
of a particular irreversible catalytic reaction, A --+ R. For a constant flow
rate of feed (7' = 2 kg. seclliter) the following data are obtained:
Time after start of operation, hr
XA 1 1
0.889
2
0.865
4
0.804
The progressive drop in conversion suggests that the catalyst deactivates
with use. Find rate equations for the reaction and for the deactivation
which fit these data.

21.5. The reversible catalytic reaction

proceeds with decaying catalyst in a batch reactor (batch-solids, batch-

fluid). What can you say of the kinetics of reaction and deactivation from
the following data:

21.6. The following data on an irreversible reaction are obtained with decaying
catalyst in a batch reactor (batch-solids, batch-fluid) What can you say
about the kinetics

21.7. With fresh catalyst the packed bed reactor is run at 600K. Four weeks
later when the temperature reaches 800K the reactor is shut down to
reactivate the catalyst. In addition, at any instant the reactor is isothermal.
 Problems 497

Assuming optimal operations what is the activity of the catalyst at the

time of reactivation.
Data: The rate of reaction with fresh catalyst is

The rate of deactivation is unknown.

Our reaction A - t R proceeds isothermally in a packed bed of large,

slowly deactivating catalyst particles and is performing well in the strong
pore diffusion regime. With fresh pellets conversion is 88%; however,
after 250 days conversion drops to 64%. How long can we run the reactor
before conversion drops to

It has been suggested that we replace these large particles with very small
particles so as to operate wholly in the diffusion-free regime and thus use
less catalyst for the same conversions. How long a run time can we expect
before the conversion drops from 88% to 64% if the catalyst is used in

21.9. . . . a packed bed reactor?

21.10. . . . a fluidized solids reactor (assume mixed flow of fluid)?

21.11. Under conditions of strong pore diffusion the reaction A --+ R proceeds
at 700°C on a slowly deactivating catalyst by a first-order rate

-ra = 0.030 CAa, [mollgm min]

Deactivation is caused by strong absorption of unavoidable and irremov-

able trace impurities in the feed, giving third-order deactivation kinetics,
or

We plan to feed a packed bed reactor (W = 10 kg) with u = 100 liters1

min of fresh A at 8 atm and 700°C until the catalyst activity drops to 10%
of the fresh catalyst, then regenerate the catalyst and repeat the cycle,
(a) What is the run time for this operation?
(b) What is the mean conversion for the run?

21.12. In catalytic dehydrogenation of hydrocarbons the catalyst activity decays

with use because of carbon deposition on the active surfaces. Let us study
this process in a specific system.
498 Chapter 21 Deactivating Catalysts

A gaseous feed (10% C, HI, - 90% inerts, ?T = 1 atm, T = 555°C) flows

(7'= 1.1 kg. hr/m3) through a packed bed of alumina-chromia catalyst.
The butane decomposes by a first-order reaction

C,HIo4 C,H,-carbon
\
other gases

and the behavior with time is as follows:

Examination of the 0.55-mm pellets shows the same extent of carbon

deposition at the entrance and the exit of the reactor, suggesting concentra-
tion independent deactivation. Develop rate equations for reaction and
deactivation.
This problem was devised from the information given by Kunugita et
al., J. Chem. Eng. (Japan), 2, 75 (1969).

The enzyme catalase effectively decomposes hydrogen peroxide

H202- catalase 1
H 2 0 + - 0,
2

and the kinetics of this reaction are to be evaluated from an experiment

in which dilute H202flows through a packed bed of kieselguhr particles
impregnated with immobilized enzyme.
From the following data, reported by Krishnaswamy and Kitterell,
AZChE J., 28, 273 (1982), develop rate expressions to represent this
decomposition, both in the diffusion free regime and in the strong pore
diffusion regime, for the catalyst at hand. Note that the conversion de-
creases with time in all runs showing that the catalyst deactivates with use.

21.13. . . . run E (modified)

Elapsed Time, hr XA
0
1.25
r1 = 4100 kg cat. s/m3 2.0
zp = 72 2 3.0
p, = 630 kg/m3 cat 4.25
Se= 5 X 10-lo m3/m cat. s 5.0
6.0
7.0
 Problems 499

21.14. . . . run B

Elapsed Time, hr XA
0.25 0.57
1.O 0.475
T' =
-
4560 kg cat. s/m3 2.0 0.39
d, 1.45 X 10-3m
= 3.0 0.30
p, = 630 kg/m3 cat 4.0 0.23
9, = 5 X 10-lo m3/m cat. s 5.0 0.186
6.0 0.14
7.0 0.115

21.15. At 730K the isomerization of A to R (rearrangement of atoms in the

molecule) proceeds on a slowly deactivating catalyst with a second-order
rate

- r i = k'Cia = 200 Cia, [mol Alhr .gm cat]

Since reactant and product molecules are similar in structure, deactivation

is caused by both A and R. With diffusional effects absent, the rate of
deactivation is found to be

We plan to operate a packed bed reactor containing W = 1 metric ton

of catalyst for 12 days using a steady feed of pure A, FA,= 5 kmollhr at
730 K and 3 atm (CAo= 0.05 mollliter).
(a) First evaluate -da/dt, d , and then the general expression for 1 - XA.
(b) What is the conversion at the start of the run?
(c) What is the conversion at the end of the run?
(d) What is the average conversion over the 12-day run?
Chapter 22
G/L Reactions on Solid
Catalysts: Trickle Beds, Slurry
Reactors, and Three-Phase
Fluidized Beds

These multiphase reactions are of the type

dissolve on solid

and they can be run in a number of ways, as shown in Fig. 22.1. The packed bed
contactors use large solid particles, the slurry reactors very fine suspended solids,
while the fluidized bed can use either depending on the flow rates.
Flow pattern: Overall, all other things equal (which they are not) countercur-
rent plug flow is the most desirable flow pattern while the worst of all in terms
of driving forces is mixedlmixed flow.

22.1 THE GENERAL RATE EQUATION

Consider the following reaction and stoichiometry

on catalyst
A(g -+ 1) + bB(1) =products .

,ye c
-Y utA - - A A B -r; = -~;;'/b. . mol A/m3cat. s
where
-r;= klCACB k g = k;;'lb. . .mf/molB.m3cat.s
 22.1 The General Rate Equation 501

Trickle High velocity

Packed or bed flow. L is
blown through
fixed bed:
by the gas.

Agitated slurry
Suspended reactor / Sd:91$
iT
particles: (suspended through L
fine solids) containing fine
suspended
'
L Rising solids).
bubbles

@ Three-phase fluidized bed. these vary widely in G/L

ratio from G bubbling through an L slurry to a
G fluidized bed of damp particles.

Figure 22.1 Various ways of running GIL reactions catalyzed by solids.

Gas reactant must first dissolve in the L, then both reactants must diffuse or
move to the catalyst surface for reaction to occur. Thus the resistance to transfer
across the GIL interface and then to the surface of solid both enter the general
rate expression.
To develop the rate equation, let us draw on the two-film theory, and let us
use the following nomenclature:

PA, Henry's law constant [Pa m311moll

Typical element gas-liquid interfacial area

a. =
of volume contain volume of reactor
G,L, and S.
, a, = external surface of particles for 6fs
volume of reactor spheres d,

= volume
volume Ofof particles,
reactor called solid loading

.' volume Of liquid

= volume of reactor'
called liquid holdup
502 Chapter 22 G/L Reactions o n Solid Catalysts

Gas fllm coeff~cient Liquid fllm coef

coefflc~entsL?OAe, EDs, [m31/m cat . sl

Centerl~neof pellet

Figure 22.2 Sketch showing the resistances involved in the GIL reaction on a catalyst
surface.

Graphically we show the resistances as in Fig. 22.2. We can then write the
following general rate equations:

For A: 1

these rates

with rate constant (kAmCB)

Henry's law constant
HA = PA/ CA

For B: k-Y; = 1
+ 1 CB~

for the first order reaction of B

with rate constant (kgcA)

Now either Eq. 2 or Eq. 3 should give the rate of reaction. Unhappily, even
with all the system parameters known ( k , a, f, etc.) we still cannot solve these
expressions without trial and error because rB
is not known in Eq. 2 and FAis
not known in Eq. 3. However, we usually encounter either of two simplifying
extremes, which are very useful.

Extreme 1: CBlB CAP In systems with pure liquid B and slightly soluble gas
A we can take
-
CBs= CB,within
pellet = CBr - . . same value everywhere
 22.2 Performance Equations for an Excess of B 503

With C, constant the reaction becomes first order with respect to A overall and
the above rate expressions with their required trial and error all reduce to one
directly solvable expression

Lfirst-order rate
constant for A

Extreme 2: CBlG CAI. In systems with dilute liquid reactant B, highly soluble
A, and high pressure, we can take

CAI= X, - . throughout the reactor

The rate then becomes first order with respect to B and reduces to

-rg = 1
+ 1 CBl
mol ~ / r reactom
n ~

& first-order rate constant which

is used to calculate gB

How to Test Whether these Extremes Apply, and Other Comments

(a) By the unequal signs or G we mean two or three times as large.

(b) More generally compare the rates calculated from Eqs. 4 and 5 and use
the smaller one. Thus,
if r & , 4 r&, ,, then C,, is in excess and extreme 1 applies.
if r &,,a r&, 5, then Eq. 5 gives the rate of reaction.
(c) Nearly always does one or other of the extremes apply.

22.2 PERFORMANCE EQUATIONS FOR AN EXCESS OF B

All types of contactors-trickle beds, slurry reactors, and fluidized beds-can
be treated at the same time. What is important is to recognize the flow patterns
of the contacting phases and which component, A or B, is in excess. First consider
an excess of B. Here the flow pattern of liquid is not important. We only have
to consider the flow pattern of the gas phase. So we have the following cases.
 22.2 Performance Equations for an Excess of B 505

22.4, we can write

where

pA(rir,- pAO) dilute pArO dilute - pA

1-XA= --
P A O ( ~- P A ) only P A Oa~= const. PAO
and
n(rO- pAO)dpA dilute rOdpA dilute dpA
-dXA = --
-P
PAO(~ A)~ pAOr a = const. pAo

Overall, around the whole reactor

Mixed Flow GIBatch L (excess of B)

With throughflow of L its composition stays roughly constant when B is in excess.
However, with a batch of L its composition slowly changes with time as B is
being used up, but B is roughly constant in the reactor at any time, as shown in
Fig. 22.5. Here the material balance at any time t becomes

l ncreases
with time
XA. exit! PA
xA,exit,~att=~}/ ' '

XA, exit, f at t = tf Batch volume V,

CEOat t = 0
CBf at t = tf

Figure 22.5 Gas bubbles through a batch of liquid, B in

excess.
506 Chapter 22 G/L Reactions o n Solid Catalysts

1
-
(-r;) ith Eq. 12 the area
or
1
-
XA,exit

Figure 22.6 Evaluation of the reaction

time for a batch of liquid.

The general procedure for finding the processing time is as follows:

Pick a number From I11 From I and I1

of CBvalues calculate calculate
c~~
-
(-TA)O
-
XA,exit, o
-
- - -

CB~ (-rAIf XA,exit, f

Then from I1 and 111. or I and I1 solve for t

as shown in Fig. 22.6.

Plug Flow GIBatch L (excess of B)

As with the previous case, CB changes slowly with time; however, any element
of gas sees the same CB as it flows through the reactor, as shown in Fig. 22.7.

Increases
[with time

with time
-Same CB everywhere.
It can be circulated if
wanted. However CB
Consider a slice decreases with time
as B is used up

Figure 22.7 Plug flow of gas through a batch of

liquid.
 22.2 Performance Equations for an Excess of B 507

Consider a slice of contactor in a short time interval in which CBis practically

unchanged. Then a material balance gives

dXA = (-r'f)dVr
FAO (13)

Integrating gives the exit conversion of A

Considering B we may now write

and on integrating we find the processing time to be

The procedure is as follows:

Solve Eq. 14
Choose CB to Give XA,
CBO XA,exit ( t = 0)
- -
- -

c~f XA,exit (at t )

Then solve Eq. 15 graphically to find the time as shown in Fig. 22.8.

C ~fixed
, C ~ O

Figure 22.8 Evaluation of the reac-

tion time in a batch L reactor.
508 Chapter 22 G/L Reactions o n Solid Catalysts

Batch
L

Figure 22.9 Pure gaseous A can be recompressed and recycled.

Special Case of Pure Gaseous A (excess of B)

One often encounters this situation, especially in hydrogenations. Here one
usually recycles the gas, in which case pA and CA stay unchanged; hence, the
preceding equations for both batch and flow systems simplify enormously, see
Fig. 22.9. When solving problems, it is suggested that one write down the basic
material balances and then carefully see what simplifications apply . . . is pA
constant? . . . and so on.

Comments The rate expressions used so far have been first order with respect
to A and first order with respect to B. But how do we deal with more general
rate forms, for example:

To be able to combine the chemical step with the mass transfer steps in simple
fashion, we must replace the above awkward rate equation with a first-order
approximation, as follows:

-ri
= kic: C? +-'/ = [(k'"Cm)C2-I] cA (17)

"I;
mean values at
locations where
reaction occurs

This approach is not completely satisfactory, but it is the best we can do. Thus,
instead of Eq. 4, the rate form to be used in all the performance expressions
will be
 22.4 Which Kind of Contactor to Use 509

22.3 PERFORMANCE EQUATIONS FOR AN EXCESS OF A

Here the flow pattern of gas is of no concern. All we need to worry about is the
flow pattern of liquid.

Plug Flow LIAny Flow G (tower and packed bed operations)

Making the material balances gives on integration

L.-
given by Eq. s

Mixed Flow LIAny Flow G (tank operations of all types)

Here the performance equation is simply

(given by Eq. 5

Batch LIAny Flow G

Noting that C, = constant throughout time (because it is in excess), the perfor-
mance equation for B becomes

22.4 WHICH KIND OF CONTACTOR TO USE

The selection of a good contactor depends on

* where the controlling resistance lies in the rate expression,

the advantages of one contacting pattern over another, and
the difference in auxiliary equipment needed.

The overall economics which accounts for these three factors will determine
which set up and reactor type is best. Let us briefly look at these factors in turn.
510 Chapter 22 G/L Reactions on Solid Catalysts

The Rate: We should favor the contactor which favors the weakest step in the
rate. For example,
if the main resistance lies in the GIL film, use a contactor with large interfacial
surface area.
if the resistance lies at the LIS boundary, use a large exterior surface of
solid, thus large f, or small particles.
if the resistance to pore diffusion intrudes, use tiny particles.
From prediction we can find the weakest step by inserting all transfer coeffi-
cients (kg, k,, . . .) and system parameters (a,, a,, . . .) into the rate equation
and then see which resistance term dominates. Unfortunately, the values for
these quantities usually are not well known.
From experiment, we can change one or another factor in the rate expression;
for example,
solid loading (this changes f, alone, hence, only changes the last resistance
term in the rate expression)
size of catalyst particle (affects both Band a,)
intensity of agitation of the liquid (affects the mass transfer terms in the rate)
TIT, CB,PA.

This should tell which factor strongly affects the rate and which does not.
Boosting the weakest step in the rate by a proper choice of particle size, solid
loading and reactor type can strongly affect the overall economics of the process.

Contacting: Plug flow of the limiting component, the one which is not in excess,
is certainly better than mixed flow. However, except for very high conversions
this factor is of minor importance.

Supporting equipment: Slurry reactors can use very fine catalyst particles, and
this can lead to problems of separating catalyst from liquid. Trickle beds don't
have this problem, and this is the big advantage of trickle beds. Unfortunately,
these large particles in trickle beds mean much lower reaction rates. With regard
to rate, the trickle bed can only hold its own
for very slow reactions on porous solids where pore diffusion limitations do
not appear, even for large particles and
for very fast reactions on nonporous catalyst-coated particles.
Overall, the trickle bed is simpler, the slurry reactor usually has a higher rate,
and the fluidized bed is somewhere in between.

22.5 APPLICATIONS
Here is a short list of applications of these reactors.
* The catalytic hydrogenation of petroleum fractions to remove sulfur impuri-
ties. Hydrogen is very soluble in the liquid; high pressure is used while the
impurity is present in the liquid in low concentration. All these factors tend
to lead to extreme 2 (excess of A).
 22.5 Applications 511

The catalytic oxidation of liquid hydrocarbons with air or oxygen. Since

oxygen is not very soluble in the liquid, while the hydrocarbon could well
be present in high concentration we could end up in extreme 1(excess of B).
The removal of dissolved organics from industrial waste water by catalytic
oxidation as an alternative to biooxidation. Here oxygen is not very soluble
in water, but the organic waste is also present in low concentration. It is
therefore not clear in what regime the kinetics lie. The catalytic oxidation
of phenol is an example of such an operation.
0 The removal of airborne pollutants by adsorption and/or reaction. These
operations usually lead to extreme 1 (excess of B).
The illustrative examples and many of the problems which follow are adapted
from or are extensions of problems prepared by Ramachandran and Choud-
hary (1980).

HYDROGENATION OF ACETONE IN A PACKED

BUBBLE COLUMN

Aqueous acetone (C,, = 1000 mol/m3 1, v, = m3 11s) and hydrogen (1 atm,

v, = 0.04 m3 gls, HA = 36845 Pa.m3 Ilmol) are fed to the bottom of a long,
slender column (5-m high, 0.1-m2cross section) packed with porous Raney nickel
catalyst (d, = 5 x m cat, p, = 4500 kg/m3 cat, f , = 0.6, ge = 8 X 10-lo m3
Ilm cat. s) and kept at 14"C, as shown in Fig. E22.1. At these conditions acetone
is hydrogenated to propanol according to the reaction

on
H, (g -1) + CH3COCH3(I)cataly;t CH3CHOHCH3
-- ----------
(A) (B)

I with rate given by

What will be the conversion of acetone in this unit?

Additional Data

The mass transfer rate constants are estimated to be

the sum of the gas + liquid film conductances

 5u Chapter 22 G/L Reactions o n Solid Catalysts

I
I 9,= 8 x 1 0 - l o m31/m cat. s
5m (of l ~ q u ~indporus catalyst)

-ri = rk = k'~;'*, mollkg cat s

k' = 2.35 X
.. .
(rn31/kgcat s) ( m o l 1 m ~ 1 ) ~ ' ~

.....
=
pAO 1 atm CBO= 1 0 0 rnol/rn31
Hydrogen
v, = 0.04 m3&
Figure E22.1

SOLUTION

Before rushing to our equations to do the appropriate integrations needed for

plug flow, let us consider the situation

CBo= 1000 while CA is given by Henry's law as

Comparing values shows that C, 9 CA;hence, we are in extreme 1 (excess

of B).
Next, we are dealing with pure hydrogen; thus, pA is constant throughout
the packed column. And since the rate is only dependent on CA and not
on CB, this means that the rate of reaction is constant throughout the
column.
 22.5 Applications 513
Let us next look at the rate. From Eq. 18.28 we have for a half order reaction

Replacing all known values into Eq. 18 gives

= 0.0317 mol/m3 r s .

Next, to the material balance of Eq. 11. With constant rate this becomes

(not particularly use this term in which

useful Fso = v,CBo= (1000) = 0.1 molls

Thus on rearranging

= 0.158, or 16% conversion

HYDROGENATION OF A BATCH OF BUTYNEDIOL IN A

SLURRY REACTOR
I
Hydrogen gas is bubbled into an agitated tank (V, = 2 m3r) containing liquid
butynediol (CBQ= 2500 mol/m3 1) plus a dilute suspension of palladium-impreg-
nated porous catalyst pellets (d, = 5 X m cat, p, = 1450 kg/m3 cat, ge =
5 x 10-lo m3 Urn cat .s, f , = 0.0055). Hydrogen dissolves in the liquid ( H A =
 514 Chapter 22 G / L Reactions on Solid Catalysts

Unused H2 is
recompressed and

Butynediol --t butenediol

Porous catalyst
d, = 50 prn
p, = 1450 kg/rn3
9,= 5 x 10-lo m31/m cat. s
f, = 0.0055

Find t for 90% conversion to butenediol

Figure E22.2~

148 000 Pa.m3 llmol) and reacts with the butynediol on the catalyst surface as
follows (see Fig. E22.2~):

H, (g +1) + butynediol(1)
-- ---------
- on
butynediol

(A) (B)

and at 35°C

-a, = klCACB and k' = 5 x m6llkg am01 cat s

Unused hydrogen is recompressed and recirculated, and the whole operation

takes place at 1.46 atm and 35°C.
Find out how long it will take for 90% conversion of reactant.

I Additional Data

I The mass transfer rates are given as

I the sum of the gas and liquid films

I SOLUTION

First compare CA and CB:

cA= = -PA
=
HA
1.46(101 325)
148 000
CBo= 2500 and CBf= 250 mol/m3I
=

1
10 mol/m3 1 Both at the beginning and
at the end of the batch run
CB+ CA;therefore, the
system is in extreme 1.
 22.5 Applications 515
Although C, stays constant throughout the batch run, CB does not, so we will
have a changing rate with time and with [Link], let us evaluate the rate
at any particular value of CB.

therefore,

and

Replacing in the rate expression of Eq. 18 gives

(ii)

Choose a
Number of MT from 8, from -r,"'
from
CBValues Eq. 6) Eq. 18.6 Eq. (ii) I/(-r,"')
2500 5 0.19 0.8105 1.23
1000 3.16 0.29 0.6367 1.57
250 1.58 0.5 0.3687 2.71

Now from Eq. 12 the reaction time is given by

With b = 1 and Vl = V,, making the plot of Fig. E22.2b, and taking the area
under the curve gives

t = 3460 s, or 58 min
516 Chapter 22 G/L Reactions on Solid Catalysts

Data points from the table

td

Figure E22.2b

REFERENCE

Ramachandran, P. A., and Choudry, V. R., Chem. Eng., p. 74 (December 1, 1980).

PROBLEMS

22.1. Trickle bed oxidation. Dilute aqueous ethanol (about 2-3%) is oxidized
to acetic acid by the action of pure oxygen at 10 atm in a trickle bed reactor
packed with palladium-alumina catalyst pellets and kept at 30°C. According
to Sato et al., Proc. First Pacific Chem. Eng. Congress, Kyoto, p. 197,1972,
the reaction proceeds as follows:

02(g- + I ) + CH3CH20H(I)
------- ----------
-on
CH3COOH(I) + H 2 0

(A) (B)
with rate

Find the fractional conversion of ethanol to acetic acid if gas and liquid
are fed to the top of a reactor in the following system:
Gas stream: vg = 0.01 m3/s, HA = 86 000 Pa-m3/mol
Liquid stream: v, = 2 X m3/s, CBo= 400 mol/m3
Reactor: 5 m high, 0.1 m2 cross section, f, = 0.58
Catalyst: d, = 5 mm, p, = 1800 kg/m3
Be= 4.16 x 10-lo m3/m cat. s
Kinetics: kAga,= 3 x mol/[Link]. s, kA,ai = 0.02 s-l
k,, = 3.86 x m/s
 Problems 517
22.2. Slurry column oxidation. Instead of using a trickle bed reactor for ethanol
oxidation (see previous problem), let us consider using a slurry reactor.
For this type of unit

Take all flows and other values from the previous problem, and then find
the expected fractional conversion of ethanol in this reactor.

22.3. Slurry tank hydrogenation. Predict the conversion of glucose to sorbitol

in a stirred slurry reactor using pure hydrogen gas at 200 atm and 150°C.
The catalyst used is porous Raney nickel, and under these conditions
Brahme and Doraiswamy, IEC/PDD, 15,130 (1976) report that the reaction
proceeds as follows:

solid
H2(g 1) + glucose, C ~ H I ~ O ,c;d;llg;t
+ ( ~ ) sorbitol, C6H1406(1)
(A) (B)
with

Data:
Gas stream: u, = 0.2 m3/s, HA = 277 600 Pa. m3/mol
Liquid stream: v, = 0.01 m3/s, C,, = 2000 mol/m3
Reactor: V, = 2 m3, f, = 0.056
Catalyst: dp = 10 pm, ps = 8900 kg/m3, gk= 2 X m3/m cat.s
Kinetics: (kA,ai),+, = 0.05 s-', kAc= rnls

22.4. Multistage bubble column hydrogenation. In the previous problem, conver-

sion to sorbitol is not as high as desired so let us consider an alternative
design, one which uses an upflow of gas and liquid through a long, narrow
multistage column 0.25 m2 in cross section and 8 m high containing sus-
pended solids (d, = m and f, = 0.4). What will be the conversion with
this arrangement?

Data: (k,,~,),,~ = 0.025 s-', k,, = m/s

All other values are unchanged from the previous problem.

22.5. Three-phasefluidized bed hydrogenation. Aniline is to be hydrogenated

in a three-phase fluidized bed of porous clay particles impregnated with
nickel catalyst. The well-agitated batch of liquid is kept at 130°C by heat
exchanger tubes passing through the fluidized bed, and by bubbling hydro-
518 Chapter 22 G/L Reactions o n Solid Catalysts

gen vigorously and at a high rate through the bed at 1 atm. According to
Govindarao and Murthy, J. Appl. Chem. Biotechnol. 25, 196 (19751, at
these conditions reaction proceeds as follows:

Ni
3H2(g -1) f 2 C&5NH2(1) ..[Link];tC6HllNHC&5 + NH;
------- --------

(A) (B) cyclohexyl-aniline

with rate

-rL = krCA, k' = 0.05 m3/kgcat. s

Find the time needed for 90% conversion of this batch of aniline.

Data:
Gas stream: pure H2 at 1 atm, HA = 28 500 Pa. m3/mol
Batch of liquid: C,, = 1097 mol/m3
Reactor: f, = 0.10, = 0.65, f, = 0.25
Catalyst: d, = 300 pm, p, = 750 kg/m3
Be= 8.35 X 10-lo m3/m cat as
Kinetics: (kA,ai),+, = 0.04 s-l, k,, = m/s
Assume that the fraction of NH, in the gas stream is very small at any time.

22.6. Bubble column hydrogenation. Consider a different design to effect the

hydrogenation of the previous problem, one which uses a long, narrow
bubble column of semisuspended 3-mm catalyst particles (f, = 0.4, f, =
0.5, fg = 0.1). The batch of liquid aniline is circulated through an external
heat exchanger (volume of liquid in the loop outside the reactor equals
the total volume of the reactor), and hydrogen is bubbled through the
column. Find the time needed for 90% conversion of aniline in this unit.

All other values not mentioned here remain unchanged from the previ-
ous problem.

22.7. Trickle bed gas absorber-reactor. Sulfur dioxide is to be removed from a

gas by passing the gas and water through a bed of highly porous activated
carbon kept at 25°C. In this system sulfur dioxide and oxygen dissolve in
water and react on the solid to give sulfur trioxide, as follows:
 Problems 519
where

Find the fraction of sulfur dioxide removed from a gas stream under the
following conditions:
Gas stream: v, = 0.01 m3/s, n = 101 325 Pa
entering SO, = 0.2%, H = 380 000 Pa. m3/mol
entering 0, = 21%, H = 87 000 Pa. m3/mol
Liquid stream: v l = 2 X m3/s
Reactor: 2 m high, 0.1 m2 cross section, f, = 0.6
Catalyst: d, = 5 mm, p, = 850 kg/m3
CBe = 5.35 x 10-lo m3/m s - s
Kinetics: = 0.01 s-',
(kiai),+, kc = m/s

22.8. Hydrogenation in a slurry reactor. The batch hydrogenation of Example

22.2 takes just about an hour to run. Let us suppose that in practical
operations we can run eight batches of fluid per day in this unit. Thus, in
the long run a batch of fluid is processed every three hours.
Another way of running this reaction is to feed the agitated reactor
continuously at such a rate that we get 90% conversion of butynediol. How
do these two processing rates compare in the long term? Give your answer
as FBO,COntinUOUS/FBO,
batch. Assume that the liquid feed composition, gas compo-
sition and pressure, mass transfer and chemical rates are the same in both
batch and continuous operations.
Part IV

Non-Catalytic Systems

Chapter 23 Fluid-Fluid Reactions: Kinetics 1523

Chapter 24 Fluid-Fluid Reactions: Design 1540
Chapter 25 Fluid-Particle Reactions: Kinetics 1566
Chapter 26 Fluid-Particle Reactions: Design I589
Chapter 23
Fluid-Fluid Reactions: Kinetics

Heterogeneous fluid-fluid reactions are made to take place for one of three
reasons. First, the product of reaction may be a desired material. Such reactions
are numerous and can be found in practically all areas of the chemical industry
where organic syntheses are employed. An example of liquid-liquid reactions is
the nitration of organics with a mixture of nitric and sulfuric acids to form
materials such as nitroglycerin. The chlorination of liquid benzene and other
hydrocarbons with gaseous chlorine is an example of gas-liquid reactions. In the
inorganic field we have the manufacture of sodium amide, a solid, from gaseous
ammonia and liquid sodium:

Fluid-fluid reactions may also be made to take place to facilitate the removal
of an unwanted component from a fluid. Thus, the absorption of a solute gas by
water may be accelerated by adding a suitable material to the water which will
react with the solute being absorbed. Table 23.1 shows the reagents used for
various solute gases.
The third reason for using fluid-fluid systems is to obtain a vastly improved
product distribution for homogeneous multiple reactions than is possible by using
the single phase alone. Let us turn to the first two reasons, both of which concern
the reaction of materials originally present in different phases.
The following factors will determine how we approach this process.

The Overall Rate Expression. Since materials in the two separate phases must
contact each other before reaction can occur, both the mass transfer and the
chemical rates will enter the overall rate expression.

Equilibrium Solubility. The solubility of the reacting components will limit

their movement from phase to phase. This factor will certainly influence the
form of the rate equation since it will determine whether the reaction takes place
in one or both phases.

The Contacting Scheme. In gas-liquid systems semibatch and countercurrent

contacting schemes predominate. In liquid-liquid systems mixed flow (mixer-
523
524 Chapter 23 Fluid-Fluid Reactions: Kinetics

Table 23.1 Absorption Systems with

Chemical Reactionu
Solute Gas Reagent
co2 Carbonates
co2 Hydroxides
co2 Ethanolamines
CO Cuprous amine complexes
co Cuprous ammonium chloride
SO2 Ca(OH12
SO2 Ozone-H20
SO2 HCr04
SO2 KOH
c12 Hz0
c12 FeC1,
H2S Ethanolamines
H2S Fe(OH13
SO3
C2H4 KOH
C2H4 Trialkyl phosphates
Olefins Cuprous ammonium complexes
NO FeSO,
NO Ca(OH12
NO H2SO4

" Adapted from Teller (1960).

settlers) and batch contacting are used in addition to counter and concurrent con-
tacting.
Many possible permutations of rate, equilibrium, and contacting pattern can
be imagined; however, only some of these are important in the sense that they
are widely used on the technical scale.

23.1 THE RATE EQUATION

For convenience in notation let use talk of GIL reactions, even though what we
say holds equally for LIL reactions. Further, let us assume that gaseous A is
soluble in the liquid but that B does not enter the gas. Thus A must enter
and move into the liquid phase before it can react, and reaction occurs in this
phase alone.
Now the overall rate expression for the reaction will have to account for the
mass transfer resistance (to bring reactants together) and the resistance of the
chemical reactions step. Since the relative magnitude of these resistances can
vary greatly we have a whole spectrum of possibilities to consider.
 23.1 The Rate Equation 525

Our analysis considers the following second-order reaction

;\(g-I) + b?(l) -
4
R(s or l o r g), -r, = kCACB

\present in gas, but 'present in liquid 'reaction occurs in liquid only,

soluble in liquid with and unable to maybe close to the interface
solubility given by enter the gas (in the liquid film), maybe
PA, = HACA~ phase in the main body of liquid

For notation consider a unit volume of contactor Vr with its gas, liquid, and solid

,- G L in contact
in this volume

Solid may be
V, = volume of
present
liquid

The rate of reaction is usefully written in a number of ways, as follows:

These rates are related by

526 Chapter 23 Fluid-Fluid Reactions: Kinetics

Since reactant A must move from gas to liquid for reaction to occur, diffusional
resistances enter the rate. Here we will develop everything in terms of the two-
film theory. Other theories can and have been used; however, they give essentially
the same result, but with more impressive mathematics.

The Rate Equation for Straight Mass Transfer (Absorption) of A

Here we have two resistances in series, of the gas film and of the liquid film.
Thus, as shown in Fig. 23.1, the rate of transfer of A from gas to liquid is given
by the rate expressions, for the gas film

mol
m3 contactor. P a . s

and for the liquid film

C
njliquid ( rn31iauid
---L- - -~
\ - ---
m2 surface . s

Gas side Liquid side

r Interface

Assume equilibrium at interface, or

pAi = H ~ Ci ~ i
and take HAi = constant, thus

Figure 23.1 Setting up the rate equation for straight mass transfer based
on the two film theory.
 23.1 The Rate Equation 527

,- Interface

Reaction can take place in both the

liquid film and in the main body of
liquid. Fast reactions occur in a
narrow zone within the film, slow
reactions spread through the film
and main body of liquid.

7 " 7

Gas Liquid Main body

film film ofliquid

Figure 23.2 Setting up the rate equation for absorption of A in the liquid,
and reaction in the liquid, based on the two-film theory.

Combining Eqs. 2 and 3 with Henry's law pAi= HACAito eliminate the unknown
interface conditions pAiand CAiwe obtain the final rate expression for straight
mass transfer at any point in the absorber.

+-
Pa . m3 liquid
mol

The Rate Equation for Mass Transfer and Reaction

Here we have three factors to consider: what happens in the gas film; in the
liquid film; and in the main body of the liquid, as shown in Fig. 23.2.
All sorts of special forms of the rate equation can result depending on the
relative values of the rate constants k, kg, and k,, the concentration ratio of
reactants pAICB,and Henry's law constant [Link] turns out that there are eight
cases to consider, going from the extreme of infinitely fast reaction rate (mass
transfer control) to the other extreme of very slow reaction rate (no mass transfer
resistance need be considered).
The eight special cases, each with its particular rate equation, are, from infi-
nitely fast to very slow reaction, as follows:

Case A: Instantaneous reaction with low C,

Case B: Instantaneous reaction with high CB
Case C: Fast reaction in liquid film, with low CB
Case D: Fast reaction in liquid film, with high C,
528 Chapter 23 Fluid-FluidReactions: Kinetics

A
PA

c
.-
0 I PA
* I
m
E'
G9 @ Reaction only
.w in film
m
+
-
X Reaction
EE'
+
W

PA
.. .. v

PA In film and
main body

0 CA

Phase
interface\

Reaction only
in main body
of liquid

- Gas Liquid
Figure 23.3 Interface behavior for the liquid-phase reaction
A (from gas) + bB (liquid) +products (liquid)
for the complete range of rates of the reaction and of the mass transfer.

Case E and F: Intermediate rate with reaction in the film and in the main body
of the liquid
Case G: Slow reaction in main body but with film resistance
Case W: Slow reaction, no mass transfer resistance
We show these eight cases in Fig. 23.3.
 23.1 The Rate Equation 529

We discuss these special cases and present their particular rate equations later,
after we present the general rate equation.

+AN' =

------ -----
gas film liquid film liquid bulk
resistance resistance resistance

The absorption of A from gas is larger when reaction occurs within the liquid
film than for straight mass transfer. Thus for the same concentrations at the two
boundaries of the liquid film we have

rate of take up of A
Liquid film
when reaction occurs
rate of take up of A for same
c, c ~ , c~ (6)
in the two cases
\ straight mass transfer /

The value of E is always greater or equal to one. The only problem now is to
evaluate E, the enhancement factor. Figure 23.4 shows that E is dependent on
two quantities:

the enhancement factor for

an infinitely fast reaction (7)

maximum possible conversion in the

film compared with maximum transport (8)
through the film

M , stands for the Hatta modulus, in recognition of the scientist who first dealt
with this problem, Hatta (1932).
Let us now examine the eight special cases.

Case A: Instantaneous Reaction with Respect to Mass Transfer. Since an ele-

ment of liquid can contain either A or B, but not both, reaction will occur at a
plane between A-containing and B-containing liquid. Also, since reactants must
diffuse to this reaction plane the rate of diffusion of A and B will determine the
rate, so that a change in p, or C, will move the plane one way or the other (see
Fig. 23.5). At steady state the flow rate of B toward the reaction zone will be b
530 Chapter 23 Fluid-Fluid Reactions: Kinetics

If Ei > 5 M, we then have

pseudo first order reaction by
the G/L interface in which case
E=MH
More precisely

All reaction occurs

M,=- 'ie,tc, If Ei < MX then we have

~ A I 5
instantaneous reaction at a plane
on the L film in which case
E Ei"
More precisely:
E~~( E ~ -1)
E r Ei- --- + ...
M,:

Figure 23.4 The enhancement factor for fluid-fluid reactions as a function of MH and Ei,
modified from the numerical solution of van Krevelens and Hoftijzer (1954).

times the flow rate of A toward the reaction zone. Thus,

---------- ----------- -------------

A in gas film A in liquid film B in liquid film

where k,, and k,,, k,, are the mass transfer coefficients in gas and liquid phases.
The liquid side coefficients are for straight mass transfer without chemical reac-
tion and are therefore based on flow through the whole film of thickness x,.
At the interface the relationship betweenp, and C, is given by the distribution
coefficient, called Henry's law constant for gas-liquid systems. Thus,
 23.1 The Rate Equation 531

is at the interfa

PA

Case A Case B
Eq. 13 Eq. 16
Figure 23.5 Concentration of reactants as visualized by the two-film theory for an infinitely
fast irreversible reactions of any order, A + bB4products. Case A-low C,, Case B-
high C, (see Eq. 17).

In addition, since the movement of material within the film is visualized to occur
by diffusion alone, the transfer coefficients for A and B are related by1

Eliminating the unmeasured intermediates x, x,, p,,, CAiby Eqs. 9, 10, and 12,
we obtain

for Case A: ( kA ~ P A> y)

For the special case of negligible gas-phase resistance, for example, if you used
pure reactant A in the gas phase, then p, = pAi or kg + w , in which case Eq.

Alternatives to the film theory are also in use. These models [Higbie (1935); Danckwerts (1950,
1955)l view that the liquid at the interface is continually washed away and replaced by fresh fluid
from the main body of the liquid, and that this is the means of mass transport. These unsteady-state
surface renewal theories all predict

as opposed to Eq. 12, for the film theory.

With the exception of this one difference, these models, so completely different from a physical
standpoint, give essentially identical predictions of steady-state behavior. Because of this, and because
the film theory is so much easier to develop and use than the other theories, we deal with it exclusively.
532 Chapter 23 Fluid-Fluid Reactions: Kinetics

13 reduces to

Case B: Instantaneous Reaction; High CB. Returning to the general situation

shown in Fig. 23.5, if the concentration of B is raised, or more precisely, if

then this condition, combined with Eq. 5, requires that the reaction zone move
to and stay at the interface rather than remain in the liquid film. This is shown
in Fig. 23.5. When this happens, the resistance of the gas-phase controls, and
the rate is not affected by any further increase in concentration of B. In addition,
Eq. 9 simplifies to

for Case B: (kAgpA 4)

" kBCB (16)

Equation 17 tells whether Case A or Case B applies in any situation. Thus,

Let us now look at the other cases.

Case C: Fast Reaction; Low CW The plane of reaction for case A now spreads
into a zone of reaction in which A and B are both present. However, reaction
is fast enough so that this reaction zone remains totally within the liquid film.
Thus, no A enters the main body of liquid to react there.
Since the last resistance term in the general rate equation, Eq. 5, is negligible
(large k ) , the rate form for this case is

Case D: Fast Reaction; High CB,Hence Pseudo First-Order Rate with Respect
to A. For the special case where CBdoes not drop appreciably within the film,
it can be taken to be constant throughout, and the second-order reaction rate
(Case C) simplifies to the more easily solved first-order rate expression. Thus,
 23.1 The Rate Equation 533

Case C Case D
(fast second order) (pseudo-first order)
Eq. 18 Eq. 19
Figure 23.6 Location of reaction in the liquid film for a fast (but not infinitely fast) second-
order reaction. Case C-low C,, Case D-high C,.

the general rate expression, Eq. 5 , reduces to

Figure 23.6 sketches Cases C and D.

Cases E and F:Intermediate Rate with Respect to Mass Transfer. Here reaction
is slow enough for some A to diffuse through the film into the main body of the
fluid. Consequently, A reacts both within the film and in the main body of the
fluid. Here we have to use the general rate expression with its three resistances,
Eq. 5.

Case G: Slow Reaction with Respect to Mass Transfer. This represents the
somewhat curious case where all reaction occurs in the main body of the liquid;
however, the film still provides a resistance to the transfer of A into the main
body of liquid. Thus, three resistances enter into the rate expression, and Eq. 5
reduces to

Case li:Infinitely Slow Reaction. Here the mass transfer resistance is negligible,
the compositions of A and B are uniform in the liquid, and the rate is determined
by chemical kinetics alone.

Figure 23.7 shows cases G and H.

534 Chapter 23 Fluid-Fluid Reactions: Kinetics

G film L film L main body

m- *

Case G Case H
Eq. 20 Eq. 21

Figure 23.7 Slow reactions Case G still shows film resistance. Case H shows no film resis-
tance.

Review of the Role of the Hatta Number, MH

To tell whether reaction is fast or slow, we focus on unit surface of gas-liquid
interface, we assume that gas-phase resistance is negligible, and we define a film
conversion parameter

2 - maximum possible conversion in the film

M H - maximum diffusional transport through the film
(22)

If MH 9 1, all reaction occurs in the film, and surface area is the controlling rate
factor. On the other hand, if M, G 1 no reaction occurs in the film, and bulk
volume becomes the controlling rate factor. More precisely, it has been found
that:
1. If MH > 2, reaction occurs in the film and we have Cases A, B, C, D.
2. If 0.02 < MH < 2, we then have the intermediate Cases E, F, G.
3. If MH < 0.02, we have the infinitely slow reaction of Case H.
When MH is large, we should pick a contacting device which develops or
creates large interfacial areas; energy for agitation is usually an important consid-
eration in these contacting schemes. On the other hand, if MH is very small, all
we need is a large volume of liquid. Agitation to create large interfacial areas
is of no benefit here.
Table 24.1 of the next chapter presents typical data for various contacting
devices, and from this we see that spray or plate columns should be efficient
devices for systems with fast reaction (or large MH), while bubble contactors
should be more efficient for slow reactions (or small MH).

Clues to the Kinetic Regime from Solubility Data

For reactions which occur in the film, the phase distribution coefficient H can
suggest whether the gas-phase resistance is likely to be important or not. To
 23.1 The Rate Equation 535

show this we write the expression for straight mass transfer of A across the gas
and liquid films

(gas film liquid film

Now for slightly soluble gases HAis large; hence, with all other factors unchanged
the above rate equation shows that the liquid film resistance term is large. The
reverse holds for highly soluble gases. Thus, we see that:
Gas film resistance controls for highly soluble gases.
Liquid film resistance controls for slightly soluble gases.

Table 23.2 Typical Values of HA = pAiICAi,Pa . m3/mol, for Common Gases in

Water

slightly soluble gas 4 -highly soluble gas

Note that gases are more soluble at lower temperatures. Additional values for
many different gases can be extracted, with difficulty, from Perry and Green
(1984) and from references given by Danckwerts (1970).
Since a highly soluble gas is easy to absorb and has its main resistance in the
gas phase, we would not need to add a liquid-phase reactant B to promote the
absorption. On the other hand, a sparingly soluble gas is both difficult to absorb
and has its main resistance in the liquid phase; hence it is this system which
would benefit greatly by a reaction in the liquid phase.

Final Comments
To find the size of process unit needed for a given job (this is discussed in the
next chapter), we need to know the overall rate of reaction. This chapter explains
how to evaluate the overall rate of reaction.
Many sources report on the physical and chemical constants used in this
chapter. I recommend the following ones:
Doraiswamy and Sharma (1984): an extensive treatment of this whole subject
Shah (1979): evaluation of the mass transfer coefficients in various types of
equipment. Also experimental devices for finding these coefficients.
Danckwerts (1970): easy to follow discussion, sources for Henry's law con-
stants, equations to use for finding mass transfer coefficients.
 536 Chapter 23 Fluid-Fluid Reactions: Kinetics

FINDING THE RATE OF A G / L REACTION

Air with gaseous A bubbles through a tank containing aqueous B. Reaction

occurs as follows:

For this system

k A p = 0.01 mol/hr - [Link] fi = 0.98

k A p = 20 hr-l HA = 105 Pa. m3/mol,very low solubility
g',, = CBB,= m2/hr a = 20 m2/m3

I For a point in the absorber-reactor where

I pA = 5 X lo3Pa and CB= 100 mol/m3

(a) locate the resistance to reaction (what % is in the gas film, in the liquid
film, in the main body of liquid)
(b) locate the reaction zone
(c) determine the behavior in the liquid film (whether pseudo first-order reac-
tion, instantaneous, physical transport, etc.)
(d) calculate the rate of reaction (mol/m3 hr)

I SOLUTION
This chapter has only analyzed second-order reactions, however, this problem
deals with a third-order reaction. Since no analysis is available for other than
second-order reactions, let us replace our third-order reaction with a second-
order approximation. Thus,

To find the rate from the general expression (Eq. 5), we need to first evaluate
E, and M,. Let us do this:

~ B C B H=A + 100 x lo5

(Ei)first guess = + = lo3
bgd~i 2(5 x 10')
1(

> 5 MH, then for any other smaller guess for pAi we will still
Since (Ei)firstguess
have E, > 5 M,. Therefore, from Fig. 23.4 we have pseudo first-order reaction
 Problems 537
in the film with

E = MH = 100

Now to the rate expression, Eq. 5 ,

-rA =
NII PA
1 H+
A H A+ -
k~,a APE kc;fl
- 5 x 103 = 33 mollhr . m3 reactor
1. -lo5 lo5
tol +

20$100) (lo6) (loo2) (.098)

I Thus,

II (a) 213 of the resistance is in the gas film, 113 is in the liquid film +--
(b) the reaction zone is in the liquid film +
( c ) reaction proceeds by a pseudd first-order reaction of A, at the interface +
(d) the rate Is - r r = 33 mollhr -m3+-

REFERENCES

Danckwerts, P. V., Trans. Faraday Soc., 46, 300 (1950).

, AIChE J., 1, 456 (1955).
, Gas-Liquid Reactions, McGraw-Hill, New York, 1970.
Doraiswamy, L. K., and Sharma, M. M., Heterogeneous Reactions, Vol. 2, John Wiley
and Sons, New York, 1984.
Hatta, S., Technological Reports, TBhoku University, 10, 119 (1932); from Sherwood,
T. K., and Pigford, R. L., Absorption and Extraction, McGraw-Hill, New York, 1952.
Higbie, R., Trans. [Link].E., 31, 365 (1935).
Perry, R. H., and Green, D. W., Chemical Engineers' Handbook, 6th ed., Section 3,
McGraw-Hill, New York, 1984.
Shah, Y. T., Gas-Liquid-Solid Reactor Design, McGraw-Hill, New York, 1979.
Teller, A. J., Chem. Eng., 67, 111 (July 11, 1960).
van Krevelens, D. W., and Hoftijzer, P. J., Rec. Trau. Chim., 67, 563 (1948); Trans. I.
Chem. E., 32, 5360 (1954).

PROBLEMS

Gaseous A absorbs and reacts with B in liquid according to

538 Chapter 23 Fluid-Fluid Reactions: Kinetics

in a packed bed under conditions where

k A p = 0.1 mollhr .m2of reactor. Pa fi = 0.01 m3liquid/m3reactor

kAla= 100 m3liquid/m3reactor. hr gAl= gBl= m2/hr

a = 100 m2/m3reactor

At a point in the reactor where pA = 100 Pa and CB = 100 mol/m3 liquid

(a) calculate the rate of reaction in mol/hr .m3 of reactor.

(b) describe the following characteristics of the kinetics:
location of the major resistance (gas film, liquid film, main body
of liquid)
behavior in the liquid film (pseudo first-order reaction, instantaneous,
second-order reaction, physical transport)
for the following values of reaction rate and Henry's law constant.

k , m3 liquidlmol .hr H A , Pa. m3 liquidlmol

23.1. 10 lo5
23.2. lo6 lo4
23.3. 10 lo3
23.4. lo-4 1
23.5. lo-z 1
23.6. lo8 1

23.7. Redo Example 23.1 with just one change. Let us suppose that C, is very
low, or C, = 1.

23.8. At high pressure CO, is absorbed into a solution of NaOH in a packed

column. The reaction is as follows:

CO, + 2NaOH -+Na2C03 + H 2 0 with -rA, = kCACB

---- -----
(A) (B)
Find the rate of absorption, the controlling resistance, and what is happen-
ing in the liquid film, at a point in the column where pA = lo5 Pa and
C, = 500 mol/m3.

Data: kAga= mol/m2.s .Pa HA = 25 000 Pa. m3/mol

This problem was adapted from Danckwerts (1970).

 Problems 539

23.9. Hydrogen sulfide is absorbed by a solution of methanolamine (MEA) in

a packed column. At the top of the column, gas is at 20 atm and it contains
0.1% of H,S, while the absorbent contains 250 mol/m3 of free MEA. The
diffusivity of MEA in solution is 0.64 times that of H2S. The reaction is
normally regarded as irreversible and instantaneous.

H,S + RNH,
---- -----
HS-
-+ + RNH;
(A) (B)
For the flow rates and packing used

k~ = 60 mol/m3 - s atm
HA = 1 x m3- atmlmol, Henry's law constant for H2Sin water.

(a) Find the rate of absorption of H,S in MEA solution.

(b) To find out whether it is worthwhile using MEA absorbent, determine
how much faster is absorption with MEA compared to absorption in
pure water.
This problem was adapted from Danckwerts (1970).
 Fluid-Fluid Reactors: Design

We must first choose the right kind of contactor, then find the size needed. There
are two kinds of contactor-towers and tanks, and Fig. 24.1 shows some examples.
As may be expected, these contactors have widely different GIL volume ratios,
interfacial areas, kg and k,, and concentration driving forces. The particular
properties of the system you are dealing with, the solubility of gaseous reactant,
the concentration of reactants, etc.-in effect the location of the main resistance
in the rate equation-will suggest that you use one class of contactor and not
the other.
Table 24.1 shows some of the characteristics of these contactors.

Factors to Consider in Selecting a Contactor

(a) Contacting pattern. We idealize these as shown in Fig. 24.2.

Towers approximate plug Glplug L.
Bubble tanks approximate plug Glmixed L.
Agitated tanks approximate mixed Glmixed L.
As we shall see, towers have the largest mass transfer driving force and
in this respect have an advantage over tanks. Agitated tanks have the
smallest driving force.
(b) kg and k,. For liquid droplets in gas kg is high, k, is low. For gas bubbles
rising in liquid kg is low, k, is high.
(c) Flow rates. Packed beds work best with relative flow rates of about F,IF, 7
10 at 1 bar. Other contactors are more flexible in that they work well in
a wider range of F,IFg values.
(d) If the resistance is in the gas andlor liquid films you want a large interfacial
area "a," thus most agitated contactors and most columns. If the L film
dominates, stay away from spray contactors. If the G film dominates stay
away from bubble contactors.
(e) If the resistance is in the main body of the L you want large f, = V,/V,.
Stay away from towers. Use tank contactors.
(f) Solubility. For very soluble gases, those with a small value of Henry's law
constant H (ammonia, for example), gas film controls, thus you should
 Selection 541

Irrigated Staged bubble

Spray tower packed bed Plate tower tower

Can have perforated plates,

Static mixer and pulsing action

Bubble tank Agitated tank

G

\ Ether with or
without flow of L
/
Figure 24.1 Tower and tank contactors for GIL reactions.

avoid bubble contactors. For gases of low solubility in the liquid, thus high
H value (O,, N,, as examples) liquid film controls, so avoid spray towers.
(g) Reaction lowers the resistance of the liquid film, so
For absorption of highly soluble gases, chemical reaction is not helpful.
For absorption of slightly soluble gases, chemical reaction is helpful and
does speed up the rate.

Nomenclature. We use the following symbols in our development.

A, = cross-sectional area of column.
a = interfacial contact area per unit volume of reactor (m2/m3).
f , = volume fraction of liquid (-).
i = any participant, reactant or product, in the reaction.
A, B, R, S = participants in the reaction.
U = carrier or inert component in a phase, hence neither reactant
nor product.
T = total moles in the reacting (or liquid) phase.
Y, = pAlpU,moles Almole inert in the gas (-).
X, = CA/Cu,moles Almole inert in the liquid (-).
F i , F; = molar flow rate of ail the gas and the liquid (molls).
F, = F i p , l ~ , upward molar flow rate of inerts in the gas (molls).
 24.1 Straight Mass Transfer 543

--- Steady
state
operations

Plug Glplug L Plug Glplug L Plug Glmixed L Mixed Glmixed L

Batch
Unsteady-state operations
of L
(semibatch) where the liquid
composition changes with time

Mixed Glbatch uniform L Plug Glbatch uniform L

Figure 24.2 Contacting patterns for GIL contactors.

F,
F ; C,ICT, downward molar flow rate of inerts in the liquid phase
=
(molls).
With this nomenclature, we have the following relationships among the various
concentration measures.

The performance equations which are written in terms of F, and Fl are useful
when the flowing streams have inert carrier material. The equation written in
terms of FB and F ; are useful when the flowing streams only contain reactive
materials and no inerts.

24.1 STRAIGHT MASS TRANSFER

Since the approach for reacting systems is a straightforward extension of straight
mass transfer, let us first develop equations for absorption alone of A by liquid

A (gas) + A (liquid)
544 Chapter 24 Fluid-Fluid Reactors: Design

then go to reacting systems.

Note the similarity in the performance equations.

Plug Flow GIPlug Flow L-Countercurrent Flow in a Tower

To develop the performance equation, we combine the rate equation with the
material balance. Thus for steady-state countercurrent operations we have for
a differential element of volume

(A lost by gas) = (A gained by liquid) = (-rr)dV,

Integrating for the whole tower gives

In brief, the design procedure is summarized in Fig. 24.3. For dilute systems
CA< C , and pA G n, so Fb = F, and F ; = F,. In this situation the differential
material balance becomes

and for any two points in the absorber

 24.1 Straight Mass Transfer 545

+ Gas -&Liquid --f

Interface at any point

in tower
High CA

The operating line is slightly curved here.

The shaded area

YA
Figure 24.3 Illustration of the design procedure for straight mass transfer in countercur-
rent towers.

The rate expression reduces to

= K A , ~ ( P-
A ~ 2=)K A ~ -CCAI
~

Thus the general integrated rate expression of Eq. 2 becomes, with Eq. 3,

coefficient on gas with gas P A , or

1
- -1- +-- 1 HA p i = HACA -1
- -- 1 + - CL=pAIHA
K ~ gk ~ g k ~ l K ~ l k ~ l
546 Chapter 24 Fluid-Fluid Reactors: Design

For the other contacting patterns of Fig. 24.2 plug Glplug L cocurrent, mixed
Glmixed L, plug Glmixed L, mixed Glplug L, mixed Glbatch L see Levenspiel
(1996) Chapter 42, or recall the equations and methods from your lessons in
mass transfer and unit operations.

24.2 MASS TRANSFER PLUS NOT VERY SLOW REACTION

Here we only treat the reaction A (g -+1) + bB(1) products (1). We assume
-+

that the rate is fast enough so that no unreacted A enters the main body of the
liquid. This assumes that the Hatta modulus is not very much smaller than unity.

Plug Flow GIPlug Flow L-Mass Transfer + Reaction in a Countercurrent

Tower

For a differential slice of absorber-reactor we write

(A lost)
by gas
= (
b by liquid
) (
B lost = disappearance
of A by reaction
----- ------- ------------
I I1 I11

For Dilute Systems. p, = n and C , = C , in which case the above expressions

simplify to
 24.2 Mass Transfer Plus Not Very Slow Reaction 547

Rearranging and integrating I and 11, I1 and 111, I and I11 gives the following:

In general

-- ...with Fi (XBl - xB2)

Fg(Y~2-Y ~ l=)-
b

For dilute systems from point 1 to and point i in the tower

F
d C ~ ...with g'IT( ~ & ? - ~ A=l )F I ( c ~ ~ -B
' 1)
b CT

To Solve for Vr
pick a fewp, values, usually pAl ,p,, and one intermediate value are enough,
and for each p, find the corresponding CB.
evaluate the rate for each point from

integrate the performance equation graphically

Plug Flow GIPlug Flow L-Mass Transfer + Reaction in a Cocurrent Tower

Here simply change Fl to -Fl (for upflow of both streams) or F, to -F, (for
downflow of both streams) in the equations for countercurrent flow. Be sure to
find the proper C, value for eachp,. The rest of the procedure remains the same.
548 Chapter 24 Fluid-Fluid Reactors: Design

Mixed Flow GIMixed Flow L-Mass Transfer + Reaction in an Agitated

Tank Contactor
Since the composition is the same everywhere in the vessel make an accounting
about the vessel as a whole. Thus,

disappearance of
A by reaction

Whole vessel is well

mixed and at exit
I
G PAI~
conditions

In symbols these equalities become

4
Fg(YAin - Y ~ o u t ) = - (XBin - X ~ o u t ) = (-'?)/at exit conditions
b for both G and L

and for dilute systems

F Fl
(PAin - PAout ) = -(CBin - CBout) = (-'?)]at exit Vr
bC,

Tojind V, the solution is direct; evaluate -rT from known stream compositions
and solve Eq. 11 or 12.
To jind C,,,, and p,,, given Vr guess pAout , evaluate CBout,then -rT, then Vr.
Compare the calculated Vr value with the true value. If different, guess an-
other PAOUt .

Plug Flow GIMixed Flow L-Mass Transfer + Reaction in Bubble

Tank Contactors
Here we must make two accountings, a differential balance for the loss of A
from the gas because G is in plug flow, and an overall balance for B because L
is in mixed flow.
Focusing on a bit of rising gas, we have

(ti::)( =
disappearance of
A by reaction
... o r . . . FgdYA=(-rT)ILatexit
conditions
dVr (13)
 24.2 Mass Transfer Plus Not Very Slow Reaction 549

Fg I pAin
All liquid
is at cBOut

For the liquid as a whole and for the gas as a whole, a balance about the whole
reactor gives

Integrating Eq. 13 along the path of the bubble and also using Eq. 14 gives

In general

...with... Fg(YAi,- YAou;) =

( X ~ i n- X ~ ~ u t )

for liquid at C B ~ ~ ~
For dilute systems
F
...with ... g(~Ain-pAout)

=a(cBi.- CB0.t)
~ C T

If Vr is to be found and the exit conditions are known, then the procedure is
direct. Pick a number of p, values and integrate graphically.
If p,,,, and CBou,are to be found in a reactor of known volume Vr then we
require a trial and error solution. Simply guess C,,,, and then see if Vc,,c,,,,,d -
-
v,,,, .

Mixed Flow GIBatch Uniform L-Absorption + Reaction in a Batch Agitated

Tank Contactor
Since this is not a steady-state operation, composition and rates all change with
time, as shown in Fig. 24.4. At any instant the material balance equates the three
550 Chapter 24 Fluid-Fluid Reactors: Design

Rises with time as B is consumed

At any instant CB is the same

V, volume of L -
everywhere in the tank. However
CB decreases with time because
of reaction with A.
At the start CB = CBO
V , volume of . At the end CB = C B ~
G/L emulsion
f
Constant -/
Figure 24.4 History of a batch of reacting liquid.

quantities shown below and thus in general

Vl ~ C B
Fg(YAin- YAoui)
=
b---dt--= ( -
-rr)Vr
-----
-------------
I I1 I11
loss of A from gas decrease of disappearance of A or B
B with time by reaction. In the rate
in L expression use p~~~~since
G i s in mixed flow.

For dilute systems

F (p
A
a Ain
-
Aout
)= v,~ C =B( - r r ) V r
b dt

To Find the Time Needed for a Given Operation

Choose a number of CBvalues, say C,,, CBfand an intermediate CBvalue.
For each C, value guess pA0,, .
Next calculate M H , E,, and then E and - r r . This may require trial and
error, but not often.
See if terms I and I11 are equal to each other

and keep adjusting pAOu,until they do.

As a shortcut: if pA < a and if E = MH then E is independent of pA in which
case I and I11 combine to give
 24.2 Mass Transfer Plus Not Very Slow Reaction 551

Next combine terms I1 and I11 to find the processing time

t=-1
h CBOdC,
. . . solve graphically
b cBf- r r

This time can be compared with the minimum needed if all A reacts and
none escapes the vessel. This situation is represented by p,,,, = 0 at all
times. Thus

1 1 amount of B reacted
b(
- 0-f
-
b (away in the vessel
t-:- =
,Lull . ,
amount of A entering
the vessel in unit time

Combining t and tmingives the efficiency of utilization of A. Thus

percent of entering A
which reacts with B

Example 24.6 illustrates this procedure for batch absorber-reactors.

TOWERS FOR STRAIGHT ABSORPTION

The concentration of undesirable impurity in air (at 1 bar = lo5 Pa) is to be

reduced from 0.1% (or 100 Pa) to 0.02% (or 20 Pa) by absorption in pure water.
Find the height of tower required for countercurrent operations.

Data

For consistency let us use SI units throughout.

For the packing
552 Chapter 24 Fluid-Fluid Reactors: Design

The solubility of A in water is given by Henry's law constant

HA = p,,/C,, = 12.5 Pa. m3/mol

The flow rates per meter squared cross section of tower are

F,/A, = 1 x 105mol/hr m2

F,/A, =7 X lo5mollhr .m2

The molar density of liquid under all conditions is

C, = 56 000 mol/m3

Figure E24.1 shows the quantities known at this point.

point in the
tower

Figure E24.1

Our strategy is to first solve the material balance, then determine the tower
height. Since we are dealing with dilute solutions we may use the simplified form
of the material balance. So for any point in the tower p , and CA are related by
Eq. 4.
 24.2 Mass Transfer Plus Not Very Slow Reaction 553

from which the concentration of A in the liquid leaving the tower is

(ii)

The expression for tower height is, from Eq. 6,

(iii)

Now evaluate terms

This expression shows that

I G film resistance = 3.1251128.125 = 0.024, or 2.4%

L film resistance = 1251128.125 = 0.976, or 97.6%

and

(KAga)= 11128.125 = 0.0078 mollhr .[Link]

Next evaluate p, - p z . Thus with Eq. (i) we get

Inserting Eqs. (iv) and (v) in (iii) gives

(1 x lo5mollhr - m2)
h=
(lo5Pa)(0.0078 mollhr .[Link])

= (128.125) ( loo
2 -0 20) =
- 512.5 rn

Comment. Here the tower is very high, unacceptably high. Also note that most
of the resistance (over 97%) lies in the liquid film, making this a liquid-film-
controlling process. However, if we added component B to the liquid which
reacts with A, we should be able to speed things-up. Let's see if this-is so.
 554 Chapter 24 Fluid-Fluid Reactors: Design

TOWERS FOR HIGH CONCENTRATION OF LIQUID

REACTANT

To the water of Example 24.1 add a high concentration of reactant B , C,, =

800 mol/m3 or approximately 0.8 N. Material B reacts with A extremely rapidly

I A(g -,1) + B(1) +product (1), k =w

Assume that the diffusivities of A and B in water are the same, thus

k,, = k,/ = k,

Figure E24.2 shows what is known at this point.

Figure E24.2

The strategy in solving the problem is as follows

Step 1. Express the material balance and find CB2in the exit stream.
Step 2. Find which of the many forms of rate equation should be used.
Step 3. Determine the tower height.

Step 1. Material balance. For dilute solutions with rapid reaction Eq. 6 gives
for any point in the tower, p,,, CB3
 24.2 Mass Transfer Plus Not Very Slow Reaction 555

At the bottom of the tower p,, = pA2,so

1
C - -(10 020 - 100) = 793.6 mol/m3
B2 - 12.5

Step 2. Form of rate equation to use. Check both ends of the tower:

kAgapA= (0.32)(20) = 6.4 mol/hr .m3

at top
kpCB = (0.1)(800) = 80 mol/hr .m3

kAgapA= (0.32)(100) = 32
at bottom
k,aC, = (0.1)(793.6) = 79.36

At both ends of the tower kAgpA< klCB;therefore, gas-phase resistance controls

and we have a pseudo first-order reaction as given by Eq. 16 of Chapter 23

Step 3. Height of tower. From Eq. 10

Comment. Even though the liquid phase controls in physical absorption (see
Example 24.1), it does not necessarily follow that it should still control when
reaction occurs. In fact, we see here in Example 24.2 that it is the gas phase
alone which influences the rate of the overall process. Reaction serves merely
to eliminate the resistance of the liquid film. Also note the remarkable improve-
ment in performance; 5 versus 500 m.

TOWERS FOR LOW CONCENTRATION OF LIQUID

REACTANT; CASE A
I
I Repeat Example 24.2 using a feed with CB, = 32 mol/m3, instead of 800 moll
m3, see Fig. E24.3.
556 Chapter 24 Fluid-Fluid Reactors: Design

Figure E24.3

As in the previous examples, solve by making a material balance, check the

form of rate equation to use, then apply the performance equation to find the
tower height.

Step 1. Material Balance. As in Example 24.2, for any point in the tower,

and for the bottom of the tower where p, = 100 Pa,

Step 2. Which Rate firm to Use. Again check both ends of the tower to see
which rate form applies

kA,apA = 0.32(20) = 6.4

at top
klaCB= 0.1(32) = 3.2

kQapA = 0.32(100) = 32
at bottom
klaCB = 0.1(35.6) = 2.56

At both ends of the tower kAgapA> k,aCB, therefore, the reaction takes place
within the liquid film and Eq. 13 in Chapter 23 should be used,

reactor. hr
 24.2 Mass Transfer Plus Not Very Slow Reaction 557

I Step 3. Height of Tower. From Eq. 6,

TOWERS FOR INTERMEDIATE CONCENTRATIONS OF

LIQUID REACTANT

Repeat Example 24.2 using a feed in which CB = 128 mol/m3.

SOLUTION

Refer to Fig. E24.4, and solve as with the previous examples.

--- 0
1 Reaction at
interface

Reaction in
liquid film
--- @
PA2=100~a> b c B 2 = ?

Figure E24.4

Step 1. Material balance. As with Examples 24.2 and 24.3 we have at any point
in the tower,

and at the bottom of the tower,

Step 2. Form of rate equation to use. Check both ends of the tower:

kAgapA= 6.4 rnollhr m3

at top
klaCB = 12.8 mol .hr .m3

kAgapA= 32
at bottom
k,aC, = 12.16
558 Chapter 24 Fluid-Fluid Reactors: Design

At the top kAgpA< klCB;hence, Eq. 16 in Chapter 23 must be used. At the

bottom kAgpA> k,CB;hence, Eq. 13 in Chapter 23 must be used.
Let us now find the condition at which the reaction zone just reaches the
interface and where the form of rate equation changes. This occurs where

Solving with the material balance we find that the change occurs at pA = 39.5 Pa.

Step 3. Height of Tower. Writing the performance equation we have, from Eq. 6,

(iii)

Noting that two different rate forms must be used, we have

Comment. In this example we see that two distinct zones are present. Situations
may be encountered where even another zone may be present. For example, if
the entering liquid contains insufficient reactant, a point is reached in the tower
where all this reactant is consumed. Below this point physical absorption alone
takes place in reactant-free liquid. The methods of these examples, when used
together, deal in a straightforward manner with this three-zone situation and
van Krevelens and Hoftijzer (1948) discuss actual situations where these three
distinct zones are present.
Comparing solutions for the four examples shows how reaction increases the
effectiveness of the absorption process.

REDO EXAMPLE 24.2 BY THE GENERAL METHOD

In Example 24.2 we found which of the eight special cases (see Fig. 23.3) applied
and then used its corresponding rate equation (it was Eq. 23.16). Alternatively
we could have used the general rate expression (Eq. 23.5). This is what we will
show here.
 24.2 Mass Transfer Plus Not Very Slow Reaction 559

1 SOLUTION

From Example 24.2 a material balance gives the tower end conditions, as shown
in Fig. E24.5. Now the rate of reaction at any point in the tower is, from Eq. 23.5,

Figure E24.5

Evaluate E at various points in the tower. For this we need to first evaluate MH
and E,.

A t the Top of the Tower. From Fig. 23.4

MH = = W, because k =w
k,,

(ii)

We have to guess the value of p,,. It can be anywhere between 0 Pa (gas film
controls) up to 20 Pa (liquid film controls). Let us guess no gas-phase resistance.
Then pAi = p,, in which case

and from Fig. 23.4, for MH = a, Ei = 500, we see that

 560 Chapter 24 Fluid-Fluid Reactors: Design

Replacing in Eq. (i) we find

<93% b%
...resistance

Our guess was wrong, so let's try again. Let us guess the other extreme, pAi =
0, meaning that the total resistance is in the gas film. Then from Eq. (ii) we see
that Ei= w, E = co and the rate equation becomes

(iii)

Thus our guess was correct.

At the Bottom of the Tower. We follow the same procedure and find the same
result. Thus the rate at all points in the tower is given by Eq. (iii). The height
of the tower, from Eq. 10, is then (see step 3 of Example 24.2)

Suggestion. Whenever MH > Eiwe end up having to guess pAi,and that is

tedious. In those cases try to use the special case expressions.
In other cases (and this is what we usually find) the general rate equation is
easier to use.

/ EXAMPLE 24.6 REACTION OF A BATCH OF LIQUID

We wish to lower the concentration of B in the liquid (V, = 1.62 m3, C, =

55555.6 mol/m3) of an agitated tank reactor by bubbling gas (F, = 9000 mollhr,
n = lo5 Pa) containing A (pAh = 1000 Pa) through it. A and B react as follows:

(a) How long must we bubble gas through the vessel to lower the concentration
from C,, = 555.6 to CBf= 55.6 mol/m3?
(b) What percent of entering A passes through the vessel unreacted?

I Additional Data

k,,a = 0.72 mollhr .[Link] f , = 0.9 m3liquid/m3total

kA,a = 144 hr-I gA= qB= 3.6 X m2/hr, a = 100 m2/m3
H A = lo3Pa. m3/mol k = 2.6 x los m3/[Link]
 24.2 Mass Transfer Plus Not Very Slow Reaction 561

Figure E24.6

I SOLUTION

Let us sketch what is known in Fig. E24.6.

E i = l + --
p Ai
51000
55 555.6, or higher

Since pA 4 n and E = MH the text says we can use the shortcut outlined above
Eq. 19. Let us do it.

At the end, following a similar treatment, we find

M H z5
E = M H = 5.0
Ei= 55.6, or higher
562 Chapter 24 Fluid-Fluid Reactors: Design

The rate of reaction at the beginning and at the end of the run is just about the
same, so

Thus the run time needed is

The minimum time required is

tmin. = -c~f) - -
1.62(555.6 - 55.6)
vdcBO = 8.91 hr
Fg(pAi,l(a - pAin)) 9000(1000/(105- 100)) =

Thus the fraction of reactant which passes through the tank untreated is

Fraction = 9'13 - 8'91 = 0.025 =

-
8.19

REFERENCES

van Krevelens, D. W., and Hoftijzer, P., Rec. Trau. Chim., 67, 563 (1948).
Kramers, H., and Westerterp, K. R., Elements of Chemical Reactor Design and Operation,

PROBLEMS

24.1. The four pA versus CA sketches of Fig. 24.2 represent various possible
ideal contacting schemes of gas with liquid. Sketch the contacting scheme
for straight physical absorption corresponding to the p, versus CA op-
erating lines X Y shown in Fig. P24.1.

(a) (6) (c)

Figure P24.1
 Problems 563

We plan to remove about 90% of the A present in a gas stream by

absorption in water which contains reactant B. Chemicals A and B react
in the liquid as follows:

B has a negligible vapor pressure, hence does not go into the gas phase.
We plan to do this absorption in either a packed bed column, or an
agitated tank contactor.
(a) What volume of contactor is needed?
(b) Where does the resistance of absorption reaction lie?

Data

For the gas stream:

Fg= 90 000 mollhr at n = lo5Pa
p,, = 1000 Pa
pAout= 100 Pa

Physical data
CB = 3.6 X m2/hr
C, = 55 556 mol H20/m3liquid, at all CB

For the packed bed

Fl = 900 000 mol/hr kAla= 72 hr-l
CBi, = 55.56 mol/m3 a = 100 m2/m3
kka = 0.36 mollhr .[Link] fi = Vl/V = 0.08

For the agitated tank

Fl = 9000 mol/hr kAla= 144 hr-l
CBin= 5556 mol/m3 (about 10%B) a = 200 m2/m3
kAga= 0.72 mol/hr .[Link] fi = Vl/V = 0.9

Note that Fland CBinare very different in packed beds and tank contactors,
and here is the reason why. Packed columns need Fl/Fg= 10 for satisfactory
operations. This means large F,, and so as not to waste reactant B, it is
introduced in low concentration. On the other hand, tank contactors do
not have this flow restriction. Thus we can use low Fl and high CBin,as
long as we introduce sufficient B to react with A.
564 Chapter 24 Fluid-Fluid Reactors: Design

Type of
Contactor
T = Tower,
Henry's Law For Countercurrent
Constant: H A , Reaction: k, A = Agitated
Pa. m3/mol m3/[Link] Tank
In these problems A
of straight mass
transfer assume T
that no B is
present in the T
system.

24.11. Danckwerts and Gillham, in Trans. I. Chem. E., 44, 42, March 1966,
studied the rate of CO, absorption into an alkaline buffered solution of
K2C03and KHCO,. The resulting reaction can be represented as

CO,(g -1) + OH-(1) -+HCO; with -r, = kCACB

(A) (B)

In the experiment pure CO, at 1 atm was bubbled into a packed column
irrigated by rapidly recirculating solution kept at 20°C and close to con-
stant C,. Find the fraction of entering CO, absorbed.

Data
Column: V , = 0.6041 m3 f, = 0.08 a = 120 m2/m3
Gas: n = 101 325 Pa H A = 3500 Pa. m3/mol v, = 0.0363 m3/s
Liquid: CB= 300 mol/m3 = 1.4 x
gA,= gB1 m2/s
Rates: k = 0.433 m3/mol. s kAla = 0.025 s-I
This problem is by Barry Kelly.

24.12. A column packed with 5-cm polypropylene saddles (a = 55 m2/m3) is

being designed for the removal of chlorine from a gas stream (G = 100
molls .m2, 2.36% Cl,) by countercurrent contact with an NaOH solution
 Problems 565

(L = 250 mol/s.m2, 10% NaOH, C, = 2736 mol/m3) at about 40-45°C

and 1 atm.
How high should the tower be for 99% removal of chlorine? Double
the calculated height to take care of deviations from plug flow.

Data
The reaction C1, + 2NaOH -+ product is very very fast and irreversible.
For these very high flow rates (close to the limits allowed) an extrapolation
of the correlations in Perry 6th ed., section 14, gives

k,a = 133 mollhr . [Link] HA = 125 X 106atm .m3/mol

k,a = 45 hr-l = 1.5 X m2/s

Repeat Example 24.6 with the following two changes

Henry's Law Constant Second-Order Reaction Rate, Constant

H A , Pa. m3/mol k, m3/[Link]
Chapter 25
Fluid-Particle Reactions:
Kinetics

This chapter treats the class of heterogeneous reactions in which a gas or liquid
contacts a solid, reacts with it, and transforms it into product. Such reactions
may be represented by

A(fluid) + bB(so1id) 4 fluid products (1)

+solid products (2)

+fluid and solid products (3)

As shown in Fig. 25.1, solid particles remain unchanged in size during reaction
when they contain large amounts of impurities which remain as a nonflaking ash
or if they form a firm product material by the reactions of Eq. 2 or Eq. 3. Particles
shrink in size during reaction when a flaking ash or product material is formed
or when pure B is used in the reaction of Eq. 1.
Fluid-solid reactions are numerous and of great industrial importance. Those
in which the solid does not appreciably change in size during reaction are as
follows.
1. The roasting (or oxidation) of sulfide ores to yield the metal oxides. For
example, in the preparation of zinc oxide the sulfide ore is mined, crushed,
separated from the gangue by flotation, and then roasted in a reactor to
form hard white zinc oxide particles according to the reaction

Similarly, iron pyrites react as follows:

2. The preparation of metals from their oxides by reaction in reducing atmo-

spheres. For example, iron is prepared from crushed and sized magnetite
 Particle Size 567

Initial Partly Completely

unreacted reacted reacted
particle particle particle

-
Time
-
Time Final particle is
hard, firm, and
unchanged in size

Initial
unreacted
particle

-
Time Time
@ -
Time
o
Particle shrinks
with time, finally
disappearing.

Flaking ash or gaseous

products cause
shrinkage in size
Figure 25.1 Different sorts of behavior of reacting solid particles.

ore in continuous-countercurrent, three-stage, fluidized-bed reactors ac-

cording to the reaction

3. The nitrogenation of calcium carbide to produce cyanamide

4. The protective surface treatment of solids such as the plating of metals.

The most common examples of fluid-solid reactions in which the size of solid
changes are the reactions of carbonaceous materials such as coal briquettes,
wood, etc. with low ash content to produce heat or heating fuels. For example,
with an insufficient amount of air, producer gas is formed by the reactions

With steam, water gas is obtained by the reactions

568 Chapter 25 Fluid-Particle Reactions: Kinetio

Other examples of reactions in which solids change in size are as follows.

1. The manufacture of carbon disulfide from the elements

2. The manufacture of sodium cyanide from sodium amide

3. The manufacture of sodium thiosulfate from sulfur and sodium sulfite

Still other examples are the dissolution reactions, the attack of metal chips by
acids, and the rusting of iron.
In Chapter 17 we pointed out that the treatment of heterogeneous reaction
required the consideration of two factors in addition to those normally encoun-
tered in homogeneous reactions: the modification of the kinetic expressions
resulting from the mass transfer between phases and the contacting patterns of
the reacting phases.
In this chapter we develop the rate expressions for fluid-solid reactions. The
next chapter will then use this information in design.

25.1 SELECTION OF A MODEL

We should clearly understand that every conceptual picture or model for the
progress of reaction comes with its mathematical representation, its rate equation.
Consequently, if we choose a model we must accept its rate equation, and vice
versa. If a model corresponds closely to what really takes place, then its rate
expression will closely predict and describe the actual kinetics; if a model differs
widely from reality, then its kinetic expressions will be useless. We must remem-
ber that the most elegant and high-powered mathematical analysis based on a
model which does not match reality is worthless for the engineer who must make
design predictions. What we say here about a model holds not only in deriving
kinetic expressions but in all areas of engineering.
The requirement for a good engineering model is that it be the closest represen-
tation of reality which can be treated without too many mathematical complexi-
ties. It is of little use to select a model which very closely mirrors reality but
which is so complicated that we cannot do anything with it. Unfortunately, in
today's age of computers, this all too often happens.
For the noncatalytic reaction of particles with surrounding fluid, we consider
two simple idealized models, the progressive-conversion model and the shrinking
unreacted-core model.

Progressive-Conversion Model (PCM). Here we visualize that reactant gas

enters and reacts throughout the particle at all times, most likely at different
rates at different locations within the particle. Thus, solid reactant is converted
continuously and progressively throughout the particle as shown in Fig. 25.2.
 25.1 Selection 0.f a Model 569
Low High
conversion conversion

Radial position

Figure 25.2 According to the progressive-conversion model, reaction pro-

ceeds continuously throughout the solid particle.

Shrinking-Core Model (SCM). Here we visualize that reaction occurs first at

the outer skin of the particle. The zone of reaction then moves into the solid,
leaving behind completely converted material and inert solid. We refer to these
as "ash." Thus, at any time there exists an unreacted core of material which
shrinks in size during reaction, as shown in Fig. 25.3.

Comparison of Models with Real Situations. In slicing and examining the cross
section of partly reacted solid particles, we usually find unreacted solid material
surrounded by a layer of ash. The boundary of this unreacted core may not

Low
Ash

1, , I * k*
Time
*
I I
I I
I I
I I I I zone I I

+
0,
c
.2 E
4- 0
, I I I
I
I
m m I
$ 9 1 I I
8.2 1 I I
SSI I I I I I
0 I I I I

R O R R O R R O R
Radial position

Figure 25.3 According to the shrinking-core model, reaction pro-

ceeds at a narrow front which moves into the solid particle. Reactant
is completely converted as the front passes by.
570 Chapter 25 Fluid-Particle Reactions: Kinetics

always be as sharply defined as the model pictures it; nevertheless, evidence

from a wide variety of situations indicates that in most cases the shrinking-core
model (SCM) approximates real particles more closely than does the progressive-
conversion model (PCM). Observations with burning coal, wood, briquettes, and
tightly wrapped newspapers also favor the shrinking-core model. For further
discussion on the many other models used (at least ten), see Chapter 55 in
Levenspiel (1996).
Since the SCM seems to reasonably represent reality in a wide variety of
situations, we develop its kinetic equations in the following section. In doing this
we consider the surrounding fluid to be a gas. However, this is done only for
convenience since the analysis applies equally well to liquids.

25.2 SHRINKING-CORE MODEL FOR SPHERICAL PARTICLES

OF UNCHANGING SIZE
This model was first developed by Yagi and Kunii (1955, 1961), who visualized
five steps occurring in succession during reaction (see Fig. 25.4).
Step 1. Diffusion of gaseous reactant A through the film surrounding the
particle to the surface of the solid.
Step 2. Penetration and diffusion of A through the blanket of ash to the surface
of the unreacted core.
Step 3. Reaction of gaseous A with solid at this reaction surface.

I I I I I l l I t
R rc 0 r c r R
Radial position
Figure 25.4 Representation of concentrations of reactants and prod-
ucts for the reaction A(g) + bB(s) -+ solid product for a particle
of unchanging size.
 25.2 Shrinking-Core Model for Spherical Particles of Unchanging Size 571

Step 4. Diffusion of gaseous products through the ash back to the exterior
surface of the solid.
Step 5. Diffusion of gaseous products through the gas film back into the main
body of fluid.
In some situations some of these steps do not exist. For example, if no gaseous
products are formed, steps 4 and 5 do not contribute directly to the resistance
to reaction. Also, the resistances of the different steps usually vary greatly one
from the other. In such cases we may consider that step with the highest resistance
to be rate-controlling.
In this treatment we develop the conversion equations for spherical particles
in which steps 1, 2, and 3, in turn, are rate-controlling. We then extend the
analysis to nonspherical particles and to situations where the combined effect
of these three resistances must be considered.

Diffusion Through Gas Film Controls

Whenever the resistance of the gas film controls, the concentration profile for
gaseous reactant A will be shown as in Fig. 25.5. From this figure we see that
no gaseous reactant is present at the particle surface; hence, the concentration
driving force, CAg- C, becomes C,, and is constant at all times during reaction
of the particle. Now since it is convenient to derive the kinetic equations based
on available surface, we focus attention on the unchanging exterior surface of a
particle S,. Noting from the stoichiometry of Eqs. 1,2, and 3 that dNB = bdNA,
we write

1 dNB - 1 dNB - b dNA

= bkg(CAg- C,) = bkgCAg= constant (4)
S, dt 4nR2 dt 4n-R2 dt

.--- 1- /Surface of shrinking

unreacted core
,/as film '\,

Surface of particle

I I I I I I Concentration

*
R r 0 r R
Radial position
Figure 25.5 Representation o f a reacting particle when diffusion
through the gas film is the controlling resistance.
572 Chapter 25 Fluid-Particle Reactions: Kinetics

If we let p, be the molar density of B in the solid and V be the volume of a

particle, the amount of B present in a particle is

(~2:)
N B = e V = --- (m3solid)

The decrease in volume or radius of unreacted core accompanying the disappear-

ance of dNB moles of solid reactant is then given by

Replacing Eq. 6 in 4 gives the rate of reaction in terms of the shrinking radlus
of unreacted core, or

where kg is the mass transfer coefficient between fluid and particle; see the
discussion leading up to Eq. 24. Rearranging and integrating, we find how the
unreacted core shrinks with time. Thus,

Let the time for complete conversion of a particle be [Link] by taking r, = 0

in Eq. 8, we find

The radius of unreacted core in terms of fractional time for complete conversion
is obtained by combining Eqs. 8 and 9, or

This can be written in terms of fractional conversion by noting that

 25.2 Shrinking-Core Model for Spherical Particles of Unchanging Size 573
Therefore

Thus we obtain the relationship of time with radius and with conversion, which
is shown graphically in Figs. 25.9 and 25.10, pp. 582 and 583.

Diffusion through Ash Layer Controls

Figure 25.6 illustrates the situation in which the resistance to diffusion through
the ash controls the rate of reaction. To develop an expression between time
and radius, such as Eq. 8 for film resistance, requires a two-step analysis. First
examine a typical partially reacted particle, writing the flux relationships for this
condition. Then apply this relationship for all values of r,; in other words, inte-
grate r, between R and 0.
Consider a partially reacted particle as shown in Fig. 25.6. Both reactant A
and the boundary of the unreacted core move inward toward the center of the
particle. But for GIS systems the shrinkage of the unreacted core is slower than
the flow rate of A toward the unreacted core by a factor of about 1000, which
is roughly the ratio of densities of solid to gas. Because of this it is reasonable
for us to assume, in considering the concentration gradient of A in the ash layer
at any time, that the unreacted core is stationary.

QAs =flux of A through exterior surface

of particle (inward +, outward -)
QA = flux of A through surface
of any radius r
QA, =flux of A to the reaction surface

Typical position in diffusion region

I
1
'.7----
I
TI Y
, I II
I I I I I
4 I I I I I

Radial position
Figure 25.6 Representation of a reacting particle when diffusion through the ash
layer is the controlling resistance.
574 Chapter 25 Fluid-Particle Reactions: Kinetics

With L I S systems we have a problem because the velocity ratio is closer to unity
than to 1000. Yoshida et al. (1975) consider a relaxation of the above assumption.
For G I S systems the use of the steady-state assumption allows great simplifica-
tion in the mathematics which follows. Thus the rate of reaction of A at any
instant is given by its rate of diffusion to the reaction surface, or

dNA - 4nr2QA = 4rR2Q, = 4rr:QA, = constant

-- (12)
dt

For convenience, let the flux of A within the ash layer be expressed by Fick's
law for equimolar counterdiffusion, though other forms of this diffusion equation
will give the same result. Then, noting that both QA and dCA/dr are positive,
we have

where CZje is the effective diffusion coefficient of gaseous reactant in the ash layer.
Often it is difficult to assign a value beforehand to this quantity because the
property of the ash (its sintering qualities, for example) can be very sensitive to
small amounts of impurities in the solid and to small variations in the particle's
environment. Combining Eqs. 12 and 13, we obtain for any r

--dNA-
~ C A
- 4rr2ge-= constant
dt dr

Integrating across the ash layer form R to r,, we obtain

This expression represents the conditions of a reacting particle at any time.

In the second part of the analysis we let the size of unreacted core change
with time. For a given size of unreacted core, dNA/dt is constant; however, as
the core shrinks the ash layer becomes thicker, lowering the rate of diffusion of
A. Consequently, integration of Eq. 15 with respect to time and other variables
should yield the required relationship. But we note that this equation contains
three variables, t, NA, and r,, one of which must be eliminated or written in
terms of the other variables before integration can be performed. As with film
diffusion, let us eliminate NA by writing it in terms of r,. This relationship is
given by Eq. 6; hence, replacing in Eq. 15, separating variables, and integrating,
 25.2 Shrinking-Core Model for Spherical Particles of Unchanging Size 575

we obtain

For the complete conversion of a particle, r, = 0, and the time required is

The progression of reaction in terms of the time required for complete conversion
is found by dividing Eq. 16 by Eq. 17, or

3
t
-T= 1 - 3 ( 2 ) 2 + 2 ( $ )

which in terms of fractional conversion, as given in Eq. 10, becomes

t
- = 1 - 3(1 - XB)'l3
7
+ 2(1 - xB)

These results are presented graphically in Figs. 25.9 and 25.10, pp. 582 and 583.

Chemical Reaction Controls

Figure 25.7 illustrates concentration gradients within a particle when chemical
reaction controls. Since the progress of the reaction is unaffected by the presence
of any ash layer, the rate is proportional to the available surface of unreacted
core. Thus, based on unit surface of unreacted core, r,, the rate of reaction for
the stoichiometry of Eqs. 1, 2, and 3 is

where k is the first-order rate constant for the surface reaction. Writing NB in
terms of the shrinking radius, as given in Eq. 6, we obtain
576 Chapter 25 Fluid-Particle Reactions: Kinetics

,//
,----
Gas film
.\,

R r, 0 r, R
Radial position
Figure 25.7 Representation of a reacting particle when chem-
ical reaction is the controlling resistance, the reaction being
A(g) + bB(s) --, products.

which on integration becomes

The time r required for complete conversion is given when rc = 0, or

The decrease in radius or increase in fractional conversion of the particle in

terms of r is found by combining Eqs. 21 and 22. Thus,

This result is plotted in Figs. 25.9 and 25.10, pp. 582 and 583.
 25.3 Rate of Reaction for Shrinking Spherical Particles 577

25.3 RATE OF REACTION FOR SHRINKING SPHERICAL PARTICLES

When no ash forms, as in the burning of pure carbon in air, the reacting particle
shrinks during reaction, finally disappearing. This process is illustrated in Fig.
25.8. For a reaction of this kind we visualize the following three steps occurring
in succession.
Step 1. Diffusion of reactant A from the main body of gas through the gas
film to the surface of the solid.
Step 2. Reaction on the surface between reactant A and solid.
Step 3. Diffusion of reaction products from the surface of the solid through
the gas film back into the main body of gas. Note that the ash layer
is absent and does not contribute any resistance.
As with particles of constant size, let us see what rate expressions result when
one or the other of the resistances controls.

Chemical Reaction Controls

When chemical reaction controls, the behavior is identical to that of particles
of unchanging size; therefore, Fig. 25.7 and Eq. 21 or 23 will represent the
conversion-time behavior of single particles, both shrinking and of constant size.

Gas Film Diffusion Controls

Film resistance at the surface of a particle is dependent on numerous factors,
such as the relative velocity between particle and fluid, size of particle, and fluid
properties. These have been correlated for various ways of contacting fluid with

R O R
Radial position
Figure 25.8 Representation o f concentration of reactants and
products for the reaction A(g) + bB(s) -+rR(g) between a
shrinking solid particle and gas.
578 Chapter 25 Fluid-Particle Reactions: Kinetics

solid, such as packed beds, fluidized beds, and solids in free fall. As an example,
for mass transfer of a component of mole fraction y in a fluid to free-falling
solids Froessling (1938) gives

During reaction a particle changes in size; hence kg also varies. In general kg

rises for an increase in gas velocity and for smaller particles. As an example,
Fig. 12 and Eq. 24 show that

-
kg -
dP
for small d, and u

-
kg -
d;I2
u1/2
for large dp and u (26)

Equation 25 represents particles in the Stokes law regime. Let us develop conver-
sion-time expressions for such particles.

Stokes Regime (Small Particles). At the time when a particle, originally of size
R,, has shrunk to size R, we may write

Thus, analogous to Eq. 7, we have

Since in the Stokes regime Eq. 24 reduces to

we have on combining and integrating

 25.4 Extensions 579

The time for complete disappearance of a particle is thus

and on combining we obtain

This relationship of size versus time for shrinking particles in the Stokes regime
is shown in Figs. 25.9 and 25.10, pp. 582 and 583, and it well represents small
burning solid particles and small burning liquid droplets.

25.4 EXTENSIONS
Particles of Different Shape. Conversion-time equations similar to those devel-
oped above can be obtained for various-shaped particles, and Table 25.1 summa-
rizes these expressions.

Combination of Resistances. The above conversion-time expressions assume

that a single resistance controls throughout reaction of the particle. However,
the relative importance of the gas film, ash layer, and reaction steps will vary as
particle conversion progresses. For example, for a constant size particle the gas
film resistance remains unchanged, the resistance to reaction increases as the
surface of unreacted core decreases, while the ash layer resistance is nonexistent
at the start because no ash is present, but becomes progressively more and more
important as the ash layer builds up. In general, then, it may not be reasonable
to consider that just one step controls throughout reaction.
To account for the simultaneous action of these resistances is straightforward
since they act in series and are all linear in concentration. Thus on combining
Eqs. 7,15, and 20 with their individual driving forces and eliminating intermediate
concentrations we can show that the time to reach any stage of conversion is
the sum of the times needed if each resistance acted alone, or

ttotal = &lrn alone + tash alone + treaction alone

Similarly, for complete conversion

I 7total = 7film alone + 7ash alone + 7reaction alone

In an alternative approach, the individual resistances can be combined directly

to give, at any particular stage of conversion,
 25.4 Extensions 581

---- ------- --
filn ash reaction

As may be seen, the relative importance of the three individual resistances varies
as conversion progresses, or as r, decreases.
On considering the whole progression from fresh to completely converted
constant size particle, we find on the average that the relative roles of these
three resistances is given by

For ash-free particles which shrink with reaction, only two resistances, gas film
and surface reaction, need to be considered. Because these are both based on
the changing exterior surface of particles, we may combine them to give at
any instant

Various forms of these expressions have been derived by Yagi and Kunii (1955),
Shen and Smith (1965), and White and Carberry (1965).

Limitations of the Shrinking Core Model. The assumptions of this model may
not match reality precisely. For example, reaction may occur along a diffuse
front rather than along a sharp interface between ash and fresh solid, thus giving
behavior intermediate between the shrinking core and the continuous reaction
models. This problem is considered by Wen (1968), and Ishida and Wen (1971).
Also, for fast reaction the rate of heat release may be high enough to cause
significant temperature gradients within the particles or between particle and
the bulk fluid. This problem is treated in detail by Wen and Wang (1970).
Despite these complications Wen (1968) and Ishida et al. (1971), on the basis
of studies of numerous systems, conclude that the shrinking core model is the
best simple representation for the majority of reacting gas-solid systems.
There are, however, two broad classes of exceptions to this conclusion. The
first comes with the slow reaction of a gas with a very porous solid. Here reaction
can occur throughout the solid, in which situation the continuous reaction model
may be expected to better fit reality. An example of this is the slow poisoning
of a catalyst pellet, a situation treated in Chapter 21.
The second exception occurs when solid is converted by the action of heat,
and without needing contact with gas. Baking bread, boiling missionaries, and
582 Chapter 25 Fluid-Particle Reactions: Kinetics

roasting puppies are mouthwatering examples of such reactions. Here again the
continuous reaction model is a better representation of reality. Wen (1968) and
Kunii and Levenspiel (1991) treat these kinetics.

25.5 DETERMINATION OF THE RATE-CONTROLLING STEP

The kinetics and rate-controlling steps of a fluid-solid reaction are deduced by
noting how the progressive conversion of particles is influenced by particle size
and operating temperature. This information can be obtained in various ways,
depending on the facilities available and the materials at hand. The following
observations are a guide to experimentation and to the interpretation of experi-
mental data.

Temperature. The chemical step is usually much more temperature-sensitive

than the physical steps; hence, experiments at different temperatures should
easily distinguish between ash or film diffusion on the one hand and chemical
reaction on the other hand as the controlling step.

Time. Figures 25.9 and 25.10 show the progressive conversion of spherical solids
when chemical reaction, film diffusion, and ash diffusion in turn control. Results
of kinetic runs compared with these predicted curves should indicate the rate-
controlling step. Unfortunately, the difference between ash diffusion and chemi-
cal reaction as controlling steps is not great and may be masked by the scatter
in experimental data.

t/7

Figure 25.9 Progress of reaction of a single spherical particle

with surrounding fluid measured in terms of time for com-
plete reaction.
 25.5 Determination of the Rate-Controlling Step 583

Figure 25.10 Progress of reaction of a single spherical particle

with surrounding fluid measured in terms of time for com-
plete conversion.

Conversion-time curves analogous to those in Figs. 25.9 and 25.10 can be

prepared for other solid shapes by using the equations of Table 25.1.

Particle Size. Equations 16, 21, and 8 with Eq. 24 or 25 show that the time
needed to achieve the same fractional conversion for particles of different but
unchanging sizes is given by
t cc R1.5to 2.0 for film diffusion controlling (the exponent drops as
Reynolds number rises) (36)
t cc R2 for ash diffusion controlling (37)
tccR for chemical reaction controlling (38)
Thus kinetic runs with different sizes of particles can distinguish between reac-
tions in which the chemical and physical steps control.

Ash Versus Film Resistance. When a hard solid ash forms during reaction, the
resistance of gas-phase reactant through this ash is usually much greater than
through the gas film surrounding the particle. Hence in the presence of a nonflak-
ing ash layer, film resistance can safely be ignored. In addition, ash resistance is
unaffected by changes in gas velocity.

Predictability of Film Resistance. The magnitude of film resistance can be

estimated from dimensionless correlations such as Eq. 24. Thus an observed rate
approximately equal to the calculated rate suggests that film resistance controls.
584 Chapter 25 Fluid-Particle Reactions: Kinetics

t Chemical reaction
step controls

Temperature (n
Figure 25.11 Because of the series relationship
among the resistances to reaction, the net or ob-
served rate is never higher than for any of the
individual steps acting alone.

Overall Versus Individual Resistance. If a plot of individual rate coefficients

is made as a function of temperature, as shown in Fig. 25.11, the overall coefficient
given by Eq. 34 or 35 cannot be higher than any of the individual coefficients.
With these observations we can usually discover with a small, carefully planned
experimental program which is the controlling mechanism.
Let us illustrate the interplay of resistances with the well-studied gas-solid
reaction of pure carbon particles with oxygen:

C + 0, +co,

[B(s) + A(g) +gaseous products]

with rate equation

Since no ash is formed at any time during reaction, we have here a case of
kinetics of shrinking particles for which two resistances at most, surface reaction
and gas film, may play a role. In terms of these, the overall rate constant at any
instant from Eq. 35 is
 25.5 Determination of the Rate-Controlling Step 585

Figure 25.12 Rate of combustion of pure carbon particles; this figure

is adapted from Yagi and Kunii (1955).

kg is given by Eq. 24, while k" is given by the following expression of Parker
and Hottel (1936):

where R is in J/mol.K, T is in kelvin, and C, is in gram moles per liter.

Figure 25.12 shows all this information in convenient graphical form and allows
determination of a" for different values of the system variables. Note that when
film resistance controls, the reaction is rather temperature insensitive but is
dependent on particle size and relative velocity between solid and gas. This is
shown by the family of lines, close to parallel and practically horizontal.
586 Chapter 25 Fluid-Particle Reactions: Kinetics

In extrapolating to new untried operating conditions, we must know when to

be prepared for a change in controlling step and when we may reasonably expect
the rate-controlling step not to change. For example, for particles with nonflaking
ash a rise in temperature and to a lesser extent an increase in particle size may
cause the rate to switch from reaction to ash diffusion controlling. For reactions
in which ash is not present, a rise in temperature will cause a shift from reaction
to film resistance controlling.
On the other hand, if ash diffusion already controls, then a rise in temperature
should not cause it to shift to reaction control or film diffusion control.

REFERENCES

Froessling, N., Gerland Beitr. Geophys., 52, 170 (1938).

Ishida, M., and Wen, C. Y., Chem. Eng. Sci., 26, 1031 (1971).
Ishida, M., Wen, C. Y., and Shirai, T., Chem. Eng. Sci., 26, 1043 (1971).
Kunii, D., and Levenspiel, O., Fluidization Engineering,2nd edition, Butterworth, Boston,
MA, 1991.
Levenspiel, O., Chemical Reactor Omnibook, OSU Bookstores, Corvallis, OR, 1996.
Parker, A. L., and Hottel, H. C., Znd. Eng. Chem., 28, 1334 (1936).
Shen, J., and Smith, J. M., Znd. Eng. Chem. Fund., 4,293 (1965).
Wen, C. Y., Znd. Eng. Chem., 60 (9), 34 (1968).
Wen, C. Y., and Wang, S. C., Znd. Eng. Chem., 62 (8), 30 (1970).
White, D. E., and Carberry, J. J., Can. J. Chem. Eng., 43, 334 (1965).
Yagi, S., and Kunii, D., 5th Symposium (International) on Combustion, Reinhold, New
York, 1955, p. 231; Chem. Eng. (Japan), 19,500 (1955).
Yagi, S., and Kunii, D., Chem. Eng. Sci., 16, 364, 372, 380 (1961).
Yoshida, K., Kunii, D., and Shimizu, F. J., Chem. Eng. (Japan), 8,417 (1975).

PROBLEMS

25.1. A batch of solids of uniform size is treated by gas in a uniform environ-

ment. Solid is converted to give a nonflaking product according to the
shrinking-core model. Conversion is about $ for a reaction time of 1 h,
conversion is complete in two hours. What mechanism is rate controlling?

25.2. In a shady spot at the end of Brown Street in Lewisburg, Pennsylvania,

stands a Civil War memorial-a brass general, a brass cannon which
persistent undergraduate legend insists may still fire some day, and a stack
of iron cannonballs. At the time this memorial was set up, 1868, the
cannonballs were 30 inches in circumference. Today due to weathering,
rusting, and the once-a-decade steel wire scrubbing by the DCW, the
cannonballs are only 29.75 in. in circumference. Approximately, when will
they disappear completely?

25.3. Calculate the time needed to burn to completion particles of graphite

(R, = 5 mm, p, = 2.2 gm/cm3, k" = 20 cmlsec) in an 8% oxygen stream.
 Problems 587

For the high gas velocity used assume that film diffusion does not offer
any resistance to transfer and reaction. Reaction temperature = 900°C.

25.4. Spherical particles of zinc blende of size R = 1 mm are roasted in an 8%

oxygen stream at 900°C and 1 atm. The stoichiometry of the reaction is

Assuming that reaction proceeds by the shrinking-core model calculate

the time needed for complete conversion of a particle and the relative
resistance of ash layer diffusion during this operation.

Data
Density of solid, p, = 4.13 gm/cm3 = 0.0425 mol/cm3
Reaction rate constant, k" = 2 cmlsec
For gases in the ZnO layer, 9 j e = 0.08 cm2/sec
Note that film resistance can safely be neglected as long as a growing ash
layer is present.

On doubling the particle size from R to 2R the time for complete conver-
sion triples. What is the contribution of ash diffusion to the overall resis-
tance for particles of size

Spherical solid particles containing B are roasted isothermally in an oven

with gas of constant composition. Solids are converted to a firm nonflaking
product according to the SCM as follows:

From the following conversion data (by chemical analysis) or core size
data (by slicing and measuring) determine the rate controlling mechanism
for the transformation of solid.

25.7. dp, mm XB t, min 25.8. d, XB t, sec

1 1 4 1 0.3 2
1.5 1 6 1 0.75 5

25.9. dp, mm XB t, sec 25.10. d,, XB t, sec

1 1 200 2 0.875 1
1.5 1 450 1 1 1
588 Chapter 25 Fluid-Particle Reactions: Kinetics

25.11. Uniform-sized spherical particles UO, are reduced to UO, in a uniform

environment with the following results:

If reaction follows the SCM, find the controlling mechanism and a rate
equation to represent this reduction.

25.12. A large stockpile of coal is burning. Every part of its surface is in flames.
In a 24-hr period the linear size of the pile, as measured by its silhouette
against the horizon, seems to decrease by about 5%.
(a) How should the burning mass decrease in size?
(b) When should the fire burn itself out?
( c ) State the assumptions on which your estimation is based.
Chapter 26
Fluid-Particle Reactors: Design

Three factors control the design of a fluid-solid reactor; the reaction kinetics
for single particles, the size distribution of solids being treated, and the flow
patterns of solids and fluid in the reactor. Where the kinetics are complex and
not well known, where the products of reaction form a blanketing fluid phase,
where temperature within the system varies greatly from position to position,
analysis of the situation becomes difficult and present design is based largely on
the experiences gained by many years of operations, innovation, and small
changes made on existing reactors. The blast furnace for producing iron is proba-
bly the most important industrial example of such a system.
Though some real industrial reactions may never yield to simple analysis, this
should not deter us from studying idealized [Link] satisfactorily represent
many real systems and in addition may be taken as the starting point for more
involved analyses. Here we consider only the greatly simplified idealized systems
in which the reaction kinetics, flow characteristics, and size distribution of solids
are known.
Referring to Fig. 26.1, let us discuss briefly the various types of contacting in
gas-solid operations.

Solids and Gas Both in Plug Flow. When solids and gas pass through the
reactor in plug flow, their compositions change during passage. In addition, such
operations are usually nonisothermal.
The plug flow contacting of phases may be accomplished in many ways: by
countercurrent flow as in blast furnaces and cement kilns [Fig. 26.l(a)], by cross-
flow as in moving belt feeders for furnaces [Fig. 26.l(b)], or by cocurrent flow
as in polymer driers [Fig. 26.l(c)].

Solids in Mixed Flow. The fluidized bed [Fig. 26.l(d)] is the best example of
a reactor with mixed flow of solids. The gas flow in such reactors is difficult to
characterize and often is worse than mixed flow. Because of the high heat capacity
of the solids, isothermal conditions can frequently be assumed in such operations.

Semibatch Operations. The ion exchange column of Fig. 26.l(e) is an example

of the batch treatment of solids in which the flow of fluid closely approximates
590 Chapter 26 Fluid-Particle Reactors: Design

a-
Air Air
(a) Blast furnace (b) Moving feeder for
coal furnaces

Hot gas

t
Product
( c ) Rotary dryer for heat-sensitive materials

Fluid in

Solids in

Bed of
solids

Fluid out
(d) Fluidized-bed reactor (e) Ion exchange bed

Figure 26.1 Various contacting patterns in fluid-solid reactors: (a-d)

countercurrent, crosscurrent, and cocurrent plug flow; ( d ) intermediate
gas flow, mixed solid flow; (e) semibatch operations.

the ideal of plug flow. On the contrary, an ordinary home fireplace, another
semibatch operation, has a flow which is difficult to characterize.

Batch Operations. The reaction and dissolution of a batch of solid in a batch of

fluid, such as the acid attack of a solid, is a common example of batch operations.
Analysis and design of fluid-solid systems are greatly simplified if the composi-
 Fluid-Particle Reactors: Design 591

tion of the fluid can be considered to be uniform throughout the reactor. Since
this is a reasonable approximation where fractional conversion of fluid-phase
reactants is not too great, where fluid backmixing is considerable, or where
solids wander about the reactor, sampling all the fluid as in fluidized beds, this
assumption frequently can be used without deviating too greatly from reality.
We use this assumption in the analyses that follow.
We then conclude this chapter with a brief treatment of extremely fast reactions
which are representative of some combustions. Here the analysis simplifies con-
siderably, since the kinetics do not enter the picture.
Let us now turn to a number of frequently met contacting patterns, and let
us develop their performance equations, employing in every case the assumptions
of uniform gas composition within the reactor.

Particles of a Single Size, Plug Flow of Solids, Uniform Gas Composition

The contact time or reaction time needed for any specific conversion of solid is
found directly from the equations of Table 25.1.

Mixture of Particles of Different but Unchanging Sizes, Plug Flow of Solids,

Uniform Gas Composition
Consider a solid feed consisting of a mixture of different-size particles. The size
distribution of this feed can be represented either as a continuous distribution
or as a discrete distribution. We use the latter representation because screen
analysis, our way of measuring size distributions, gives discrete measurements.
Let F be the quantity of solid being treated in unit time. Since the density of
solid may change during reaction, F is defined as the volumetric feed rate of
solid in the general case. Where density change of the solid is negligible, F can
represent the mass feed rate of solid as well. In addition, let F(R,) be the quantity
of material of size about Ri fed to the reactor. If R, is the largest particle size
in the feed, we have for particles of unchanging size

Figure 26.2 shows the general characteristics of a discrete size distribution.

When in plug flow all solids stay in the reactor for the same length of time t,.
From this and the kinetics for whatever resistance controls, the conversion XB(Ri)
for any size of particle Ri can be found. Then the mean conversion xB
of the
solids leaving the reactor can be obtained by properly summing to find the overall
contribution to conversion of all sizes of particles. Thus,

mean value for

(the fraction of)
I3 unconverted
= 2.
allsizes
(
fraction of reactant
B unconverted in
particles of size Ri )(
fraction of
feed which is
of size Ri ) (1)
 592 Chapter 26 Fluid-Particle Reactors: Design

&- 4 Rm Shaded area is feed

Total feed: size

(any size)
Size of particles in the feed

Figure 26.2 Representation of the feed rate of a mixture

of particles.

or in symbols

where R(t, = T) is the radius of the largest particle completely converted in

the reactor.
Equation 2 requires some discussion. First of all, we know that a smaller
particle requires a shorter time for complete conversion. Hence some of our
feed particles, those smaller than R(tp = T), will be completely reacted. But if
we automatically apply our conversion-time equations to these particles we
can come up with X, values greater than unity, which makes no sense physi-
cally. Thus the lower limit of the summation indicates that particles smaller
than R(tp = T) are completely converted and do not contribute to the fraction
unconverted, 1 - zB.

CONVERSION OF A SIZE MIXTURE IN PLUG FLOW

I A feed consisting

30% of 50-pm-radius particles

40% of 100-pm-radius particles
30% of 200-pm-radius particles

is to be fed continuously in a thin layer onto a moving grate crosscurrent to a

flow of reactant gas. For the planned operating conditions the time required for
complete conversion is 5, 10, and 20 min for the three sizes of particles. Find
 Fluid-Particle Reactors: Desiga 593

Solid B
( 100 >m - 40%
200 p m - 30%
I
t,, = 8 min t

Figure E26.1

the conversion of solids on the grate for a residence time of 8 min in the reactor
(see Fig. E26.1).

From the statement of the problem we may consider the solids to be in plug
flow with t, = 8 min and the gas to be uniform in composition. Hence for a
mixed feed Eq. 2 is applicable, or

1 - XB= [I - XB(50pm)] F(50pm)+[1-XB(100pm)]

F F(100 ~ m +) ...

and ~ ( 5 pm)
0 = 5 rnin

F(lOO pm) = 0.40 and ~ ( 1 0 pm)

0 = 10 min
F

F(200 = 0.30 and ~ ( 2 0 pm)

0 = 20 rnin
F

Because for the three sizes of particles

R l : R,: R, = T ~T: ~T: ~ ,

we see from Eq. 25.38 that chemical reaction controls and the conversion-time
characteristics for each size is given by Eq. 25.23 or
 594 Chapter 26 Fluid-Particle Reactors: Design

I Replacing in Eq. (i) we obtain for unconverted reactant

------------- -----------
for R = 100 pm for R = 200 pm

II Hence the fraction of solid converted equals 93.2%.

Note that the smallest size of particles is completely converted and does not
contribute to the summation of Eq. (i).

Mixed Flow of Particles of a Single Unchanging Size, Uniform Gas Composition

Consider the reactor of Fig. 26.l(d) with constant flow rates of both solids and
gas into and out of the reactor. With the assumption of uniform gas concentration
and mixed flow of solids, this model represents a fluidized-bed reactor in which
there is no elutriation of fine particles.
The conversion of reactant in a single particle depends on its length of stay
in the bed, and its appropriate controlling resistance is given by Eq. 25.11,25.18,
or 25.23. However, the length of stay is not the same for all the particles in the
reactor; hence we must calculate a mean conversion zB
of material. Recognizing
that the solid behaves as a macrofluid, this can be done by the methods leading

[ ][
to Eq. 11.13. Thus, for the solids leaving the reactor

fraction of reactant fraction of exit

= i, unconvertedfor str;~zl:II(
particles staying in stayed in the reactor (3)
B unconverted particles the reactor for time for a time between
ages between t and t + dt

or in symbols

1 - X B = Jm ( 1 - X B ) E dt, XB5 1
0
or (4)
1 - XB= J T ( 1 - X B ) E dt
0

where E is the exit age distribution of the solids in the reactor (see Chapter 11).
For mixed flow of solids with mean residence time 2 in the reactor, see Fig.
26.3, we find from Fig. 11.14 or Eq. 14.1 that
 Fluid-Particle Reactors: Design 595

CA,out

t lo
as seen by
the solids

Solids in
mixed flow I
(fluidized)
I t
Figure 26.3 Conversion of one size of solids in
mixed flow.

Thus, for mixed flow of the single size of solid which is completely converted in
time T,we obtain

This expression may be integrated for the various controlling resistances.

For film resistance controlling, Eq. 25.11 with Eq. 6 yields

which on integration by parts gives

-
-
XB= -t (1 - e~li)
7

or in equivalent expanded form, useful for large ?IT,thus for

very high conversion

(8b)

For chemical reaction controlling, Eq. 25.23 replaced in Eq. 6 gives

596 Chapter 26 Fluid-Particle Reactors: Design

Integrating by parts using the recursion formula, found in any table of integrals,
we obtain

B-
- t 6
-
7 (3' (3'
- +6 -

or in equivalent form, useful for large 717, or for very high

conversion

For ash resistance controlling replacement of Eq. 25.18 in Eq. 6 followed by

integration leads to a cumbersome expression which on expansion yields [see
Yagi and Kunii (1961)l

Figure 26.4 Mean conversion versus mean residence time

in mixed flow reactors, single size of solid.
 Fluid-Particle Reactors: Design 597

50

Gas f ~ l mres~stancecontrols, Eqs 25 11 and 8

React~oncontrols, Eqs. 25.23 and 10

P g 10
$/a
II
'"I3 5

1 I I I I . . /
0.01 0.05 01 0.5 1.O
1 -x,
Figure 26.5 Comparison of holding times needed to effect a given conver-
sion for mixed flow and plug flow of a single size of solid.

-
1-x II
B-5t 420 t
_ 4iio(;r (;r
l9 (IT +- - 0.00149 + . a a

Figures 26.4 and 26.5 present these results for solids in mixed flow in convenient
graphical form. Figure 26.5 shows clearly that at high conversion the mixed
flow reactor requires a much larger holding time for solids than does a plug
flow reactor.
Extension to multistage operations is not difficult; see Levenspiel (1996, p.
52.14) or Kunii and Levenspiel (1991).

CONVERSION OF A SINGLE-SIZED FEED IN A MIXED

FLOW REACTOR
I
Yagi et al. (1951) roasted pyrrhotite (iron sulfide) particles dispersed in asbestos
fibers and found that the time for complete conversion was related to particle
size as follows:

Particles remained as hard solids of unchanging size during reaction.

A fluidized-bed reactor is planned to convert pyrrhotite ore to the correspond-
ing oxide. The feed is to be uniform in size, I = 20 min, with mean residence
 598 Chapter 26 Fluid-Particle Reactors: Design

time 7 = 60 min in the reactor. What fraction of original sulfide ore remains un-
converted?

Since a hard product material is formed during reaction, film diffusion can be
ruled out as the controlling resistance. For chemical reaction controlling Eq.
25.38 shows that

I whereas for ash layer diffusion controlling Eq. 25.37 shows that

As the experimentally found diameter dependency lies between these two values,
it is reasonable to expect that both these mechanisms offer resistance to conver-
sion. Using in turn ash diffusion and chemical reaction as the controlling resis-
tance should then give the upper and lower bound to the conversion expected.
The solids in a fluidized bed approximate mixed flow; hence, for chemical
reaction controlling, Eq. 10, with 717 = 20 min160 min = i, gives

For ash layer diffusion controlling Eq. 11 gives

Hence the fraction of sulfide remaining is between 6.2% and 7.8%,or on averaging

Mixed Flow of a Size Mixture of Particles of Unchanging Size,

Uniform Gas Composition
Often a range of particle sizes is used as feed to a mixed flow reactor. For such
a feed and a single-exit stream (no elutriation of the fines) the methods leading
to Eqs. 2 and 6, when combined, should yield the required conversion.
Consider the reactor shown in Fig. 26.6. Since the exit stream is representative
of the bed conditions, the size distributions of the bed as well as the feed and
exit streams are all alike, or
 Fluid-Particle Reactors: Design 599

Fluidized bed: W, yB

,- ' ' ? -.. -,

S O ~ I ~ nSF,
: XB = 0 ,? 1-,-. *- , Solids out: F ,

R Gas R

Figure 26.6 Fluidized bed with single exit stream treating a size
mixture of solids. Note that the size distribution of the flow streams
and the bed are all the same.

where Wis the quantity of material in the reactor and where W(R,) is the quantity
of material of size Ri in the reactor. In addition, for this flow the mean residence
time 7(Ri)of material of any size Ri is equal to the mean residence time of solid
in the bed, or

-
t = t -( R . ) = -W
= (
weight of all solids in the reactor)
(13)
9 (feed rate of all solids to the reactor)

Letting X,(R,) be the mean conversion of particles of size Ri in the bed, we have
from Eq. 6

However, the feed consists of particles of different sizes; hence the overall mean
of B unconverted in all these sizes is

mean value for fraction of exit or

fraction unconverted
entering stream consisting
in particles of size Ri
unconverted of particles of size Ri

or in symbols
 600 Chapter 26 Fluid-Particle Reactors: Design

Combining Eqs. 14 and 15 and replacing the first term expression with Eqs.
8,10, or 11for each size of particle, we obtain in turn, forfilm diffusion controlling,

for chemical reaction controlling,

for ash diffusion controlling,

where 7(Ri)is the time for complete reaction of particles of size Ri. The following
example illustrates the use of these expressions.

CONVERSION OF A FEED MIXTURE IN A MIXED

FLOW REACTOR
I
A feed consisting
30% of 50-pm-radius particles
40% of 100-pm-radius particles
30% of 200-pm-radius particles
is to be reacted in a fluidized-bed steady-state flow reactor constructed from a
vertical 2-m long 20-cm ID pipe. The fluidizing gas is the gas-phase reactant, and
at the planned operating conditions the time required for complete conversion is
5,10, and 20 min for the three sizes of feed. Find the conversion of solids in the
reactor for a feed rate of 1 kg solidslmin if the bed contains 10 kg solids.

Additional Information:

The solids are hard and unchanged in size and weight during reaction.
A cyclone separator is used to separate and return to the bed any solids that
may be entrained by the gas.
The change in gas-phase composition in the bed is small.

From the statement of the problem we may consider the solids to be in mixed
flow. For a feed mixture Eq. 15 is applicable, and since chemical reaction controls
(see Example 26.1), this equation reduces to Eq. 17, where from the problem
 Fluid-Particle Reactors: Design 601

statement

F = 1000 gmlmin -t = -W
= 10 000 gm = min
W = 10 000 gm F 1000 gmlmin

F(50 pm) = 300 gmlmin and 0 = 5 min

~ ( 5 pm)
F(100 pm) = 400 gmlmin and ~(100 pm) = 10 min
F(200 pm) = 300 gmlmin and ~ ( 2 0 pm)
0 = 20 min

Replacing in Eq. 17 we obtain

= - [I
1-x
B-
5min
----
4lOmin
1 (L)~+
20 10
. . 300 gmlmin.]
1000 gmlmin
for R = 50 pm

+ [- o m i n ) - -1 (lo)'+.
~ ( l-
4 10min 20 10 1000
for R = 100 pm

+ [I- (20
- min) - -1 ( 2 0 ) ~
4 10min 20 10
+ .. .]joo
1000
for R = 200 pm

I The mean conversion of solids is then

FINDING THE SIZE OF A FLUIDIZED BED

In a gas-phase environment, particles of B are converted to solid product as

follows:

Reaction proceeds according to the shrinking core model with reaction control
and with time for complete conversion of particles of 1 hr.
A fluidized bed is to be designed to treat 1 tonlhr of solids to 90% conversion
using a stoichiometric feed rate of A, fed at C,,. Find the weight of solids in
602 Chapter 26 Fluid-Particle Reactors: Design

Note: The gas in the bed

is not at CAO

x '
' Gas at CAO

Figure E26.4

the reactor if gas is assumed to be in mixed flow. Note that the gas in the reactor
is not at C,,. Figure E26.4 sketches this problem.

SOLUTION

In a CA, environment with reaction controlling

1
and in any other environment T cc -
c*o
Now for equal stoichiometric feed XA = F B .Thus, the leaving gas is at 0.1 CAo.
Since the gas is in mixed flow, the solids see this exit gas, or T = 10 hr.
From Eq. 10 let us find TI? which gives = 0.9. Thus, solve

Solving by trial and error we find

Thus, the needed weight of bed is

W = iFB,= 23(1) = 23 tons

 Fluid-Particle Reactors: Design 603

Instantaneous Reaction
When reaction between gas and solid is fast enough so that any volume element
of reactor contains only one or other of the two reactants, but not both, then
we may consider reaction to be instantaneous. This extreme is approached in
the high temperature combustion of finely divided solids.
In this situation prediction of the performance of the reactor is straightforward
and is dependent only on the stoichiometry of the reaction. The kinetics do
not enter the picture. Let us illustrate this behavior with the following ideal
contacting patterns.

Batch Solids. Figure 26.7 shows two situations, one which represents a packed
bed, the other a fluidized bed with no bypassing of gas in the form of large gas
bubbles. In both cases the leaving gas is completely converted and remains that
way so long as solid reactant is still present in the bed. As soon as the solids are
all consumed, and this occurs the instant the stoichiometric quantity of gas has
been added, then the conversion of gas drops to zero.

Countercurrent Plug Flow of Gas and Solids. Since only one or other reactant
can be present at any level in the bed, there will be a sharp reaction plane where
the reactants meet. This will occur either at one end or the other of the reactor
depending on which feed stream is in excess of stoichiometric. Assuming that
each 100 moles of solid combine with 100 moles of gas, Figs. 2 6 . 8 ~and b show
what happens when we feed a little less gas than stoichiometric and a little more
than stoichiometric.
We may wish reaction to occur in the center of the bed so that both ends can
be used as heat exchange regions to heat up reactants. This can be done by
matching the gas and solids flow rates; however, this is inherently an unstable
system and requires proper control. A second alternative, shown in Fig. 26.8c,

Completely
converted gas

Completely
Unconverted converted gas
solids

Progressive
conversion of
solids with tlrne

Cornpletley Moving Suspended

converted Sollds are
solids
reaction solids , ,. un~forrnly
front
rnlxed

I Gas
Gas
Figure 26.7 A batch of solids contacted with gas; instantaneous reaction.
604 Chapter 26 Fluid-Particle Reactors: Design

Solids in: FBo= 1 0 0 Solids in: FBo= 1 0 0 Solids in: FBo= 1 0 0

I
- Reaction
zone

Completely
converted
solids

Gas in: FAO

= 99 Gas in: FAO
= 101 Gas in: FAo= 1 0 1

(4 (6) (c)

Figure 26.8 In countercurrent flow the location of the reaction zone depends on
which component is in excess of stoichiometric.

introduces a slight excess of gas at the bottom of the bed, and then removes a
bit more than this excess at the point where reaction is to occur.
Moving bed reactors for oil recovery from shale is one example of this kind of
operation. Another somewhat analogous operation is the multistage counterflow
reactor, and the four- or five-stage fluidized calciner is a good example of this.
In all these operations the efficiency of heat utilization is the main concern.

Cocurrent and Crosscurrent Plug Flow of Gas and Solids. In cocurrent flow,
shown in Fig. 26.9a, all reaction occurs at the feed end, and this represents a
poor method of contacting with regard to efficiency of heat utilization and pre-
heating of entering materials.
For crosscurrent flow, shown in Fig. 26.9b, there will be a definite reaction
plane in the solids whose angle depends solely on the stoichiometry and the
relative feed rate of reactants. In practice, heat transfer characteristics may
somewhat modify the angle of this plane.

Mixed Flow of Solids and Gas. Again in the ideal situation either gas or solid will
be completely converted in the reactor depending on which stream is in excess.

Extensions
Modifications and extensions of the methods presented here, for example,
to more complicated particle kinetics
to growing and shrinking particles in single reactors and in solid circula-
tion systems
to changing gas composition in a single reactor and from stage to stage in
multistage operations
to deviations from ideal plug and mixed flow
to elutriation of fines from a reactor
are treated elsewhere: see Kunii and Levenspiel (l991), and Levenspiel (1996).
 Problems 605

Solids,~ Gas

Completely Unreacted
converted gas gas n; ; ;R
?

t
Gas in
t t
Gas in

Figure 26.9 Cocurrent and crosscurrent contacting of gas-solids; instantaneous re-

action.

Arbitrary Flow of Solids. Since particles flow as macrofluids,their mean conver-

sion is given by Eq. 11.13. Thus with any RTD and with known gas composition
we have

1- B = J (1 - s i n e s o
particle stream
dt

where 1 - X , is given by Eq. 25.23 for SCMIreaction control, and by Eq. 25.18
for SCMIash diffusion control.

REFERENCES

Kunii, D., and Levenspiel, O., Fluidization Engineering, Second edition, Butterworth,
Boston, MA, 1991.
Levenspiel, O., Chemical Reactor Omnibook, OSU Bookstores, Corvallis, O R 97339,1996.
Yagi, S., and Kunii, D., Chem. Eng. Sci., 16, 364, 372, 380 (1961).
, Takagi, K., and Shimoyama, S., J. Chem. Soc. (Japan), Znd. Chem. Sec.,
54, 1 (1951).

PROBLEMS

A stream of particles of one size are 80% converted (SCMIash diffusion control,
uniform gas environment) on passing through a reactor. If the reactor is made
twice the size but with the same gas environment, same feed rate, and same flow
pattern of solids, what would be the conversion of solids? The solids are in

26.1. plug flow. 26.2. mixed flow.

606 Chapter 26 Fluid-Particle Reactors: Design

A solid feed consisting of

20 wt% of 1-mm particles and smaller
30 wt% of 2-mm particles
50 wt% of 4-mm particles
passes through a rotating tubular reactor somewhat like a cement kiln
where it reacts with gas to give a hard nonfriable solid product (SCMI
reaction control, T = 4 h for 4-mm particles).

26.3. Find the residence time needed for 100% conversion of solids.

26.4. Find the mean conversion of the solids for a residence time of 15 min.

26.5. Particles of uniform size are 60% converted on the average (shrinking
core model with reaction controlling) when flowing through a single fluid-
ized bed. If the reactor is made twice as large but contains the same
amount of solids and with the same gas environment what would be the
conversion of solids?

26.6. Solids of unchanging size, R = 0.3 mm, are reacted with gas in a steady
flow bench scale fluidized reactor with the following result.

F, = 10 gmlsec, W = 1000 gm, X, = 0.75

Also, the conversion is strongly temperature-sensitive suggesting that the
reaction step is rate-controlling. Design a commercial sized fluidized bed
reactor (find W) to treat 4 metric tonslhr of solid feed of size R = 0.3
mm to 98% conversion.

26.7. Solve Example 26.3 with the following modification: the kinetics of the
reaction is ash diffusion controlled with T(R = 100 ~ m =) 10 min.

26.8. Repeat Example 26.4 if twice the stoichiometric ratio of gas to solid, still
at C,,, is fed to the reactor.

26.9. Repeat Example 26.4 if the gas is assumed to pass in plug flow through
the reactor.

26.10. Consider the following process for converting waste shredded fibers into
a useful product. Fibers and fluid are fed continuously into a mixed flow
reactor where they react according to the shrinking core model with the
reaction step as rate controlling. Develop the performance expression for
this operation as a function of the pertinent parameters and ignore elutri-
ation.
 Problems 607

Hydrogen sulfide is removed from coal gas by passing the gas through a
moving bed or iron oxide particles. In the coal gas environment (consider
uniform) the solids are converted from Fe203to FeS by the SCMIreaction
control, 7 = 1 hr. Find the fractional conversion of oxide to iron sulfide
if the RTD of solids in the reactor is approximated by the E curves of
Figs. P26.11-P26.14.

E E E

t t t
0 90 rnin 15 45 rnin 30 90 rnin

Figures P26.11-P26.14
Biochemical Reaction Systems

Chapter 27 Enzyme Fermentation 1611

Chapter 28 Microbial Fermentation-Introduction and Overall Picture 1623
Chapter 29 Substrate Limiting Microbial Fermentation 1630
Chapter 30 Product Limiting Microbial Fermentation 1645
Chapter 27
Enzyme Fermentation

The term "fermentation" can be used in its original strict meaning (to produce
alcohol from sugar-nothing else) or it can be used more or less broadly. We
will use the modern broad definition:

From the simplest to the most complex, biological processes may be classed
as fermentations, elementary physiological processes, and the action of living
entities. Further, fermentations can be divided into two broad groups: those
promoted and catalyzed by microorganisms or microbes (yeasts, bacteria,
algae, molds, protozoa) and those promoted by enzymes (chemicals pro-
duced by microorganisms). In general, fermentations are reactions wherein
a raw organic feed is converted into product by the action of microbes or
by the action of enzymes.

This whole classification is shown in Fig. 27.1.

Enzyme fermentations can be represented by

(organic feed, A) - enzyme E

(product chemical, R)

Microbial fermentations can be represented by

microbe C
(organic feed, A) which acb as catalys: (product, R) + (more cells, C) (2)

The key distinction between these two types of fermentation is that in enzyme
fermentation the catalytic agent, the enzyme, does not reproduce itself, but acts
as an ordinary chemical, while in microbial fermentation the catalytic agent, the
611
612 Chapter 27 Enzyme Fermentation

Classification:
I processes I- Action of living
entities

Fermentation Elementary
.. physiological
.
processes
J \ Y
Catalvsed bv ENZYMES Catalvsed by microorgansrns,
-high molecular weight (MICROBESor CELLS)
compounds, whichBre -yeasts - molds
very specific, and usually - bacteria - protozoa
only act on one compound - algae
Figure 27.1 Classification of biological processes.

cell or microbe, reproduces itself. Within the cells it is the enzyme which catalyses
the reaction, just as in enzyme fermentation; however, in reproducing itself the
cell manufactures its own enzyme.
In this chapter we introduce enzyme fermentations, in the following chapters
we take up microbial fermentations.

27.1 MICHAELIS-MENTEN KINETICS (M-M KINETICS)

In a sympathetic environment, with just the right enzyme for catalyst, organic
A will react to produce R. Observations show the behavior of Fig. 27.2.
A simple expression which accounts for this behavior is

a constant. called the Michaelis constant

AAR~F

CA
+
High CEO

Low CEO
At high CA:The rate is independent
of CA
At low CA: Rate CA, hence the
rate is 1st order with
respect to CA.
At all CA:The rate is proportional
to CEO,the enzyme
concentration

Figure 27.2 Typical rate-concentration curves for enzyme catalyzed reactions.

 27.1 Michaelis-Menten Kinetics (M-M Kinetics) 613
Rate: - r ~= r~ [First order behavior
A when CA<< CM Zero order behavior

rmax= k3CE0
/
-------L ------------
/I
J
------ '"
Rate-concentration curve
for Michaelis-Menten
At CA= CM half the kinetics

* CA
k2 + k3
CA=CM=-
kl

Figure 27.3 Special features of the M-M equation, Eq. 3.

In searching for the simplest mechanism to explain these observations and this
rate form, Michaelis and Menten (1913) came up with the two-step elementary
reaction mechanism

rfree enzyme

3
X--R + E ... and ...\
CEO= CE + CX
A+E+/ Lintermediate /" I
\total enzyme cenzyme attached
to reactant

Example 2.2 develops and explains the relationship between the above mecha-
nism and rate equation. Let us look at some of the special features of the
Michaelis-Menten equation.
when CA = C, half the enzyme is in free form, the other half combined.
when CA 9 C, most of the enzyme is tied up as complex X.
when CA < C, most of the enzyme is in free form.
Graphically we show this equation in Fig. 27.3.
Next let us see how to evaluate the two rate constants of this important enzyme
fermentation equation.

Batch or Plug Flow Fermentor

For this system, integration of the M-M equation gives [see Eq. 3.57 or see
Michaelis and Menten (1913)l

first-order term zero-order term

This concentration-time behavior is shown in Fig. 27.4.

614 Chapter 27 Enzyme Fermentation

CA
Zero-order is approached

0- *
0

Figure 27.4 Concentration-time behavior of the

M-M equation.

Unfortunately this equation cannot be plotted directly to find the values of

the constants k3 and C,. However, by manipulation we find the following form
which can be plotted, as shown in Fig. 27.5, to give the rate constants

Mixed Flow Fermentor

Inserting the M-M equation into the mixed flow performance expression gives

Unfortunately we cannot devise a plot of this equation to give k3 and C,.

However, on rearrangement we find an equation form which does allow a direct
evaluation of k3 and CM,or

In graphical form this gives (see Fig. 27.6).

Alternative Methods for Evaluating k and C,

Biologists and life scientists have developed a tradition of fitting and extracting
the rate constants of the M-M equation with a multistep procedure, as follows
 27.1 Michaelis-Menten Kinetics (M-M Kinetics) 615
A

c ~ O- c~

I
I , CEO~
?

CM+ CAo In CAolCA

k3

- C ~ This plot gives one

line for all CEO

I for each CEO

Figure 27.5 Either plot can be used to test and fit the M-M equation (Eq. 6) from batch reactor data.

first measure CAOutversus r for data taken in any one of the three ideal
reactors-batch, mixed, or plug flow.
then evaluate -rA at various C A , either directly from mixed flow data
( - r , = (C,, - CAout)/r)
or by taking slopes for the plug flow or batch data,
as shown in Chapter 3.
plot CA versus (-r,) in one of two ways

(-r,) versus ( - r A ) / C A . . .the Eadie plot

l / ( - r A ) versus l / C A . . .the Lineweaver plot
and from these plots extract C , and k.

The CRE method which leads to Eq. 6 or Eq. 8 fits the measured CA versus
r data from any of the three ideal reactor types, is direct, is less prone to fiddling,
and is more reliable.
616 Chapter 27 Enzyme Fermentation

One line for all CEO

Separate line
for each CEO

Figure 27.6 Either plot can be used to test and fit the M-M equation (Eq. 8) using data
taken in a mixed-flow reactor.

27.2INHIBITION BY A FOREIGN SUBSTANCE-COMPETITIVE AND

NONCOMPETITIVE INHIBITION
When the presence of substance B causes the slowdown of the enzyme-substrate
reaction of A to R, then B is called an inhibitor. We have various kinds of
inhibitor action, the simplest models being called competitive and noncompetitive.
We have competitive inhibition when A and B attack the same site on the
enzyme. We have noncompetitive inhibition when B attacks a different site on
the enzyme, but in doing so stops the action of A. In simple pictures this action
is shown in Fig. 27.7.
Pharmacological significance: the study of enzymes and inhibition is one of
the major methods in determining the action of existing drugs and in developing
new drugs. This approach has changed the whole direction of pharmacological
research in recent years. The main thrust today is
to study the disease biochemically, then
synthesize a chemical to block the action of a crucial enzyme.
Let us develop kinetic expressions for these two types of inhibition.
 27.2 Inhibition by a Foreign Substance 617

Note: when B 1s on
the surface ~tw ~ l l
not let A enter
or leave

Compet~t~ve Noncompef~t~ve 1
lnh~b~t~on ~nh~bltlon

Figure 27.7 A simple representation of the action of two types of inhibitors.

Kinetics of Competitive Inhibition

With A and B competing for the same site on the enzyme we have the follow-
ing mechanism:

Cnote that B only attacks the free enzyme

With a procedure as shown in Chapter 2 (see Example 2.2) we end up with the
following rate equation:
-

r, = k3 c ~ o c ~ -
- k3 C~
cM+ cA+ NCBoCM CM(l+ NCBo)+ CA
k2 + k3 --
CM= ---- mol
kl ' m3
where
N=- -
k5' mol

Compared to systems without inhibition (Eq. 3) we see that all we need do here
is replace CMby CM(l + NCBo).

Kinetics of Noncompetitive Inhibition

Here A attacks one site on the enzyme, B attacks a different site, but in doing
so it stops the action of A. The mechanism of this action is represented by
618 Chapter 27 Enzyme Fermentation

Note that B attacks the enzyme irrespective of whether A is attached to it or

not. The overall rate is then

Compared to enzyme reactions without inhibition we see that both k3 and CM

are modified here. Thus, for noncompetitive inhibition,

replace k b k3
1 + LCBa

replace C, by C, (11++ NCBo

LCBo)

How to Tell between Competitive and Noncompetitive

Inhibition from Experiment
With C versus t data obtained from either batch, mixed flow or plug flow runs
make one of the recommended plots for inhibition-free systems (see Fig. 27.5
or Fig. 27.6). With inhibition these plots are modified as shown in Figs. 27.8
and 27.9.

Competitive Noncompetitive
c,, =0 Low High CBo 4 CEO= 0 Low High CEO

Figure 27.8 Effect of inhibition on batch or plug flow data.

 Problems 619

A Competitive h Noncompetitive
CBO= 0 LOW High CBO CBo= 0 Low High CBo

0
C M ( l+ NCBo) CM C M ( l+ NCBo)
/ k3 k3 ;k3 k3
/ /
-CM L -CM L-

Figure 27.9 Effect of inhibition on mixed flow reactor data.

Comments
The M-M equation is the simplest expression to represent enzyme catalyzed
reactions. It has been modified and extended in many ways. The two modes of
inhibition introduced here are the simplest imaginable. Others are more complex
to represent mathematically, so always try these first.

REFERENCE

Michaelis, L., and Menten, M. L., Biochem. Z., 49, 333 (1913).

PROBLEMS

27.1. Substrate A and enzyme E flow through a mixed flow reactor (V = 6 liter).
From the entering and leaving concentrations and flow rate find a rate
equation to represent the action of enzyme on substrate.

CEO,mollliter C,,, mollliter C,, mollliter v, literlhr

27.2. At room temperature sucrose is hydrolyzed by the enzyme sucrase as

follows:

sucrase
sucrose -products
620 Chapter 27 Enzyme Fermentation

Starting with sucrose (C,, = 1 mol/m3) and sucrase (CEO= 0.01 mol/m3)
the following data are obtained in a batch reactor (concentrations are
calculated from optical rotation measurements)

C,, mol/m3 0.68 0.16 0.006

t, hr 2 6 10

Find a rate equation to represent the kinetics of this reaction.

27.3. In a number of separate runs different concentrations of substrate and

enzyme are introduced into a batch reactor and allowed to react. After a
certain time the reaction is quenched and the vessel contents analyzed.
From the results found below find a rate equation to represent the action
of enzyme on substrate.

Run c,,, m01/m3 CAo m01/m3 c,, m01/m3 t, hr

27.4. Carbohydrate A decomposes in the presence of enzyme E. We also suspect

that carbohydrate B in some way influences this decomposition. To study
this phenomenon various concentrations of A, B, and E flow into and out
of a mixed flow reactor (V = 240 cm3).
(a) From the following data find a rate equation for the decomposition.
(b) What can you say about the role of B in the decomposition?
(c) Can you suggest a mechanism for this reaction?

27.5. Enzyme E catalyzes the decomposition of substrate A. To see whether

substance B acts as inhibitor we make two kinetic runs in a batch reactor,
one with B present, the other without B. From the data recorded below
(a) Find a rate equation to represent the decomposition of A.
(b) What is the role of B in this decomposition?
(c) Suggest a mechanism for the reaction.
 Problems 621
Run I . CAo = 600 mol/m3, CEO= 8 gm/m3, no B present

Run 2. CAo = 800 mol/m3, CEO= 8 gm/m3, CB = CBo= 100 mol/m3

Cellulose can be converted to sugar by the following enzymatic attack

cellulase -
cellulase
sugar

and both celluboise and glucose act to inhibit the breakdown. To study
the kinetics of this reaction a number of runs are made in a mixed flow
reactor kept at 50°C and using a feed of finely shredded cellulose (CAo=
25 kg/m3), enzyme (CEO= 0.01 kg/m3, same for all runs), and various
inhibitors. The results are as follows:

Series 2 Series 3
Series 1 with Cellobiose with Glucose
Exit Stream no Inhibitor CBO= 5 kg/m3 CGo= 10 kg/m3
Run C A , kg/m3 T, min 7,min T, min

27.6. Find a rate equation to represent the breakdown of cellulose by cellulase

in the absence of inhibitor.

27.7. What is the role of cellubiose in the breakdown of cellulose (find the type
of inhibition, and rate equation).

27.8. What is the role of glucose in the breakdown of cellulose (find the type of
inhibition, and the rate equation).
The rate data for these problems are modified from Ghose and Das,
Advances in Biochemical Engineering, 1, 66 (1971).

27.9. Given the Michaelis-Menten rate form to represent enzyme-substrate reac-

tions (or catalyst-reactant reaction)

A-R,
enzyme
-rA= ~CACEO
CA + (constant)
622 Chapter 27 Enzyme Fermentation

which of the following contacting patterns when operated properly gives

good reactor behavior (close to minimum reactor size) and which will
not, if the reactor is to be fed two feed streams, one containing C
,,, the
other CEO?
Chapter 28
Microbial Fermentation-
Introduction and Overall Picture

Natural fermentation is a complex situation with a hodge-podge of foods and

cells all busily reacting. In this introduction let us only consider the very simplest
of situations:
one type of microbe C acting. We sometimes call this the cell or bug.
one type of needed food A. This is called the substrate by life science workers.
If the food is right the bugs eat it, multiply, and in the process produce waste
material R. In symbols

In some cases the presence of product R inhibits the action of the cells, no matter
how much food is available and we have what is called poisoning by product,
or product poisoning. Wine making is an example of this

squished grapes, fruit, buss_ more

cereals, potatoes, etc.) (bugs) +

As the concentration of alcohol rises the cells multiply more slowly, and at about
12% alcohol the bugs quit. Alcohol is the poison here.
Activated sludge treatment of waste water is an example of a fermentation
which is free of product poisoning

organic waste) (more) + (breakdown products

material bugs CO,H,O, ...

Sometimes we are interested in the breakdown of A, as in waste water treat-

ment. In other situations we are interested in producing cells C, as in the growing
of yeast or single-cell protein for food. In still others we want the cell's waste
material R, as in the production of penicillin and other antibiotics.
Let us see what typically happens with a single type of bug and a single food.
623
624 Chapter 28 Microbial Fermentation

Slope of
growth
curve
Maximum slope,
:. maximum growth High CA,
rate of cells, .: cells have
k more than
cco enough food
In -
cc

-4 Time lag k
LOWcA,
thus cells are
- * CA
Here the cells
have plenty
starved for food of food
Figure 28.1 Cell growth in a uniform friendly environment.

Constant Environment Fermentation, Qualitative

What happens when we introduce a batch of microbes into a friendly constant
composition medium having food of concentration C,? First the microbes take
some time to adapt to their new environment, then they grow exponentially.
Thus we have the behavior shown in Fig. 28.1. Roughly-the time lag is a result
of the "shock" to the cells in finding themselves in these new surroundings.
the growth rate of cells (after the time lag) is given by Monod as

the concentration o f A where the cells

reproduce at 112 their m a x i m u m rate

Batch Fermentor, Qualitative

Here cells reproduce, the composition of the substrate changes, and product
which can be toxic to the cells forms. Typically we see
an induction period (time lag)
a growth period
a stationary period and
a dying of cells

Stationary

Cells start
dying

Exponential

Time
 Chapter 28 Microbial Fermentation 625
Other forms of lag curves are
Usual form of lag curve sometimes observed, for example
t t
Mature cells have the
least trouble adapting

Age of cells introduced into Age of cells introduced into

fresh medium fresh medium
Figure 28.2 The time lag depends on the age of cells.

A few words follow about these regimes.

(a) Lag. As the cells in a container use up their food supply they stop multi-
plying, their enzyme activity decreases, low molecular weight chemicals
diffuse out, and the cells change character. They age. So when they are
introduced into a new environment a time lag is observed as the cells
remanufacture the chemicals needed for growth and reproduction. In gen-
eral any change in environment results in an induction period as the cells
adjust. We find, as shown in Fig. 28.2.
(b) Growth and stationary phase. Cells grow exponentially in a uniform envi-
ronment, but in a batch system the medium changes so the growth rate
changes. The eventual drop in cell growth is governed either by
depletion of food or
accumulation of toxic materials (toxic to the cell).
Graphically we summarize this in Fig. 28.3.

Mixed Flow Fermentor

Here the cells are in a uniform environment. No adaptation is needed and cell
multiplication proceeds at a constant rate determined by the composition of the
fluid in the vessel. This is frequently represented by a Monod-type equation

The k value depends on all sorts of things: temperature, presence of trace ele-
ments, vitamins, toxic substances, light intensity, etc.
626 Chapter 28 Microbial Fermentation

Substrate limiting Poison limiting

(final Cc depends on the initial (Cc never exceeds a certain value.
amount of food. Product does Product slows and then stops
not affect the rate) the reaction)

t
I
Start with

much
Rapid
x- high CAo

I growth
/

In general

Cc, ,,, is independent of CAo

For an initial limits cell growth

concentration CAO
in the batch here
is the maximum of food is limiting
cell production *
possible

Figure 28.3 In a batch reactor the maximum cell production depends on the limiting mechanism.

Product Distribution and Fractional Yields

For the stoichiometric equation

A&C + r~
let us use the following shorthand notation for the instantaneous fractional yields

d(C formed)

d(R formed)
@ = '(RIA) = d(A used)

d(R formed)
@ = '(R'C) = d(C used)
 Chapter 28 Microbial Fermentation 627

Then we have the following relationships

and

In general the stoichiometry can be messy with fractional yields changing with
composition. Treatment of that case can be difficult. We'd like therefore to make
the simplification that all cp values remain constant at all compositions. This
assumption may be reasonable for mixed flow, or for the exponential growth
period of batch reactors, otherwise it is questionable.
Let us make this assumption anyway-all cp values stay constant. In this case
for any change we can write

C - C C A 0 - C ) . or... Cc=Cco+ @ ( C A O - c A q

CR-CRo=@(CAo-CA) . or... CR=CRO+@(CAO-~A)) (4)

CR-CRO= @ (CC-CCO) ...or... C R = C R O + @ (CC-CCOd

Kinetic Expressions
The rate of cell multiplication depends in general on the availability of food and
on the build up of wastes which interfere with cell multiplication. The simplest
reasonable rate forms are shown below. We use these in the following chapters.
Availability of Food. For a reasonable quantitative expression make the anal-
ogy with enzyme kinetics.
For enzymes:

A+E=X
7 at high CA ... rR = k c E O
7
and CEO= CE + CX
J at all CA ... r - kCE0 CA
/- R - C A + C fM
Michaelis-Menten M-M constant
equation
628 Chapter 28 Microbial Fermentation

For microbes:

Cpregnant +
+
+ Crestingc Cpregnant

+R
2Cresting
1) at high CA

at low CA
... r ~ ' k c C
7
.-. r ~= k c C c A / c M ) (6)

Many other kinetic forms have been proposed and have been used in the past;
however, they have all been forgotten since Monod came out with his expression.
Its simplicity won the day. So we will use this type of expression throughout to
relate the rate of cell growth to substrate concentration.

Effect of Harmful Wastes. As harmful wastes R build up, they interfere with
cell multiplication. Thus, the observed Monod rate constant kob,decreases with
rise in C,. A simple form of this relationship is

\( ,
kobs= k 1-
:p order of product poisoning

Lrate constant in the absence

of harmful waste material

where Ca is that concentration of R where all cell activity stops, in which case
kob,becomes zero. This expression is shown in Fig. 28.4.

k
Linear decrease in rate
constant as R builds up

0
Concentration of waste where -/
all cell activity stops
Figure 28.4 The observed k decreases as poi-
sonous product R builds up (see Eq. 7).
 Chapter 28 Microbial Fermentation 629

General Kinetic Expression. The simplest expression of the Monod type which
can account for both factors in microbial fermentation is

r c = - r ~ @ = ~ ~ @ = k ObsCA + CM

generalized
Monod equation as CRrises

...where... kobs=

concentration where
all reaction stops

In general, then, reaction and cell multiplication will slow down either by deple-
tion of A (famine) or by build up of R (environmental pollution).

Planned Treatment of the Subject

The next two chapters treat in turn the performance expressions and design
consequences for
poison-free Monod kinetics. Here food limitation alone affects the growth
rate of cells.
product poisoning kinetics. Here some product formed during fermentation
slows the rate.
We also assume in these chapters a constant fractional yield throughout, and since
everything is liquid we take E, = 0, and we will use concentrations throughout.
More generally, an excess of either substrate or cells in the broth can also
slow the rate of fermentation. In these situations the Monod equation should
be suitably modified [see Han and Levenspiel, Biotech. and Bioeng., 32, 430
(1988), for a discussion and general treatment for all these forms of inhibition].
Chapter 29
Substrate-Limiting
Microbial Fermentation

If we assume a constant fractional yield and no slowing of the rate as a result

of product poisoning or increase in cell crowding in the broth, then the general
rate equation of the previous chapter, Eq. 28.8, reduces to the well-known Mo-
nod equation

rc = @ (-'A) = k C C~
~ . where cc- C , = @ (cA0
- cA) (1)
C A CM +

(Monad constant

and where C,, and C , are the feed or starting compositions.

Let us examine ideal reactors processing feed which is reacting away according
to these kinetics.

29.1 BATCH (OR PLUG FLOW) FERMENTORS

Consider the progress of this reaction. At the start CAOis high, Cc, is low; at the
end, C A+ 0 while C , is high. Thus, the rate at the beginning and at the end of
a run is low, but it will be high at some intermediate composition. By putting
drcldt = 0, we find that the maximum rate occurs at

C ~rnax, rate = 4 C
C& + C ~ ( + C ~ ~ @ ~- C~
~ )

What this means is that with a feed C,,, Cco to any system, the key to proper
,,, all in one step, then use plug flow
design is to use mixed flow to reach C,, ,
beyond this point. Thus, it is always important to know CA,,, ,, .
630
 29.1 Batch (or Plug Flow) Fermentors 631

Now let us return to our batch (or plug flow) reactor. Applying its performance
equation we find

Integration then gives

... with Cc - Cco = @(cAo- c A )

If a time lag is involved just add t,,, to the above times to find t,,,,,. If we wish
we can write the performance equation in terms of C, instead of C, and C,.
Just remember that

The sketches of Fig. 29.1 show the main properties of this performance equation.

f"; Run out

I of food

the exponential
if both CAand Cc
change appreciably
during the run

Figure 29.1 Behavior of batch or plug flow reactors for Monod type microbial fermentation.
632 Chapter 29 Substrate-Limiting Microbial Fermentation

Figure 29.2 Evaluation of the Monod equation constants from

batch data, method (a).

How to Find the Monod Constants from the Batch Experiment

Method (a). Rearrange Eq. 3 to give

Then plot the data as shown in Fig. 29.2

Method (b). First find rc by taking dCc/dt data, then rearrange the Monod
equation to give

Then plot as shown in Fig. 29.3 to find CMand k.

Figure 29.3 Evaluation of the constants of the Monod

equation from batch reactor data, method (b).
 29.2 Mixed Flow Ferrnentors 633

Comments on Batch Operations

Method (a) uses all the data directly and presents it as a linear plot. This
method is probably better, all round.
Method (b) requires taking derivatives or slopes from experimental data, is
more tedious, and is probably less reliable.
For high CA,thus C , P C M ,put CM= 0 in the Monod equation. This gives
rc = k c c , in which case the performance expression, Eq. 3, simplifies to

cc
k~~= In - ... an exponential growth curve
cco

For low C,, thus CA < C,, the Monod equation becomes simple autocata-
lytic, and the performance expression, Eq. 3, reduces to

k~~ = c~ c ~ ~ C...cS-shaped growth curve. See the

In ------
@ Cco C ~ C c o autocatalytic equation in Chapter 3
CAO+

For very high Cc the poison-free Monod equation just can't apply, for even
if there is plenty enough food the cells will crowd out each other, and growth
will slow down and will eventually stop. So, for very high cell concentration,
we must go to product poison kinetics.
It is awkward to try to evaluate the rate constants of the Monod equation
from batch or plug flow data. Mixed flow data are so much simpler to
interpret, as we shall see.

29.2 MIXED FLOW FERMENTORS

No Cells in Feed Stream, C,, = 0
Assume Monod kinetics (no product poisoning), constant fractional yields 9,
and no cells entering in the feed stream. Then the mixed flow performance
equation becomes

A ci
T,,,= - where i = A, C, or R
Ti

Figure 29a
634 Chapter 29 Substrate-Limiting Microbial Fermentation

Replacing r, from Eq. 1 into Eq. 6 then gives

In terms of CA

... or CA = CM CM
... for kz, > 1 + -
kzm = c~ c~
+
cA m c ~ o
In terms of Cc

kz, =
@(c~O+ c ~ -) c~
@cA0-cC

CAO
In terms of CR

kz, =
@~ A -OCR

no solution possible
if k z m < l + - CM
'A0

This intriguing expression was developed by Monod (1949), and independently,

at about the same time, by Novick and Szilard (1950).
To evaluate the kinetic constants from a set of mixed flow runs, rearrange Eq.
7 to give

and the plot as shown in Fig. 29.4

From the performance equation we can show that everything-washout, opti-
mum processing time, maximum production rate-all depend on C, and c,,,
combined as follows:
 29.2 Mixed Flow Fermentors 635

Figure 29.4 Evaluating the Monod constants

from mixed flow reactor data.

A
, . @A$I
I \ CC, optimum = ----
I \ N+l
I \ -
CC,optlmum -- N
C ~ma~.
, possible + '

CC,opt --------

* k7m

I I Washout occurs here. Maximum production rate of cells

I I (maximum rate of everything)
I I occurs here.
I I
I I
II
I '
Figure 29.5 Summary of the mixed flow behavior of reactions which follow Monod kinetics.
636 Chapter 29 Substrate-Limiting Microbial Fermentation

Thus, optimum operations of a single mixed flow fermentor occurs when

(10)

and washout occurs at

All this is shown in Fig. 29.5.

Feed Stream Contains Cells, C,, Z 0

With both feed and cells entering the fermentor, the Monod expression (Eq. 1)
in the performance expression for the MFR (Eq. 6) gives

Examples [Link], [Link] and [Link] use this expression.

29.3 OPTIMUM OPERATION OF FERMENTORS

With poison-free Monod kinetics and a given feed, we have a U-shaped l l r
versus C curve, as shown in Fig. 29.6.
From Chapter 5 we have learned that with this form of rate-concentration
curve we should operate as follows:
To reach any point between A and B run part of the feed at A in mixed
flow and mix with the rest of the feed.
To reach any point between A and 0 go directly to point A in mixed flow,
then use plug flow beyond A.
These two rules represent the key to optimum reactor behavior.

Figure 29.6 Rate-concentration behav-

ior of Monod kinetics.
 29.3 Optimum Operation of Fermentors 637

MIXED REACTORS FOR MONOD KINETICS

Let us illustrate the above important optimization principle by suggesting how

to best operate various combinations of mixed flow reactors for a fermentation
which follows Monod kinetics, and where

k = 2, CAO= 3, CM= 1. @= 1 and V = 1 for each reactor

and all quantities are expressed in consistent units. Consider in turn the follow-
ing setups

(a) A single MFR, with a feed rate v = 3

(b) A single MFR, with v = 1
( c ) A single MFR, with v = 113
(d) Two MFR, v = 3
( e ) Two MFR, v = 1.5
(f) Two MFR, v = 0.5

Do not calculate the outlet concentration of A or C.

SOLUTION

Preliminary. From the optimization rules for mixed flow reactions (see Eqs. 9
and l o ) , we have

- N2
rm, washout - (N2 - l ) k

(a) High Feed Rate, Single Reactor, v = 3. If all the feed goes through the
MFR the mean residence time r = V l v = 113 is too short and the cells don't
stay in the reactor long enough and will wash out. So bypass some of the feed
and let the reactor operate at its optimum, as shown in Fig. [Link].

(b) Intermediate Feed Rate, Single MFR, v = 1. Here if all the feed passes
through the reactor r = V l v = 1, which is optimum, so operate that way, see
Fig. [Link].
 638 Chapter 29 Substrate-Limiting Microbial Fermentation

Figure E29.1

(c) Low Feed Rate, Single Reactor, v = 1/3. Here r = Vlv = ll(113) = 3,
which is longer than the optimum, so pass everything through the reactor, as
shown in Fig. E29.1~.

(d) High Feed Rate, 2 MFR, v = 3. Noting that for each MFR we want r =
1, which is optimum, we end up with the scheme shown in Fig. [Link].

(e) Intermediate Feed Rate, 2 MFR, v = 1.5. Here the best we can do is keep
the first reactor at the best conditions, as shown in Fig. [Link].

Cf) Low Feed Rate, 2 MFR, v = 0.5. Here the feed rate is too low to keep
either reactor at the optimum so we end up with the scheme shown in Fig. E29.1f.

g PLUG FLOW REACTOR FOR MONOD KINETICS

Let us extend Example 29.1 to the situation where we use a plug flow reactor
(PFR) with or without recycle in place of mixed flow reactors (MFR). Find the
optimum in the following situations.

I (a) The volume of PFR Vp = 3, v

(b) Vp = 2, v = 3
=2
 29.3 Optimum Operation of Fermentors 639

First find the optimum conditions

2
N=2, r,,,=l and 7,, =-
3

( a ) Low feed rate, T = Vp/v = 3 / 2 = 1.5. Here two optimal arrangements

are shown in Fig. E29.2a.
(b) High feed rate, T = Vp/v = 2/3. If all the feed flows through the reactor
we will have washout. So the optimum is as shown in Fig. E29.2b.

Approach mixed flow

(a)

(b)
Figure E29.2

GLUCOSE FOR E. COLI BACTERIA

The E. coli microbe grows contentedly on glucose according to Monod kinetics

as follows:

40 CACc kg cells
r -
C- CA+0.4 m 3 . h r with kg glucose
\c*= m3

What feed rate of glucose solution (C,, = 6 kg/m3)to a mixed flow reactor (V =
1 m3) would give the maximum consumption rate of glucose, and maximum
production rate of E. coli cells?
 640 Chapter 29 Substrate-Lirniting Microbial Fermentation

To find the maximum consumption rate of glucose, first find Nand then use the
information given in Fig. 29.5.

I The feed rate of glucose is

The maximum consumption rate of glucose is

I The maximum production rate of E. coli is

Comments
The extension of the performance equations developed in this chapter to
plug flow with recycle of exit fluid
mixed flow with a feed which contains C,, Z 0
cell separation, concentration, and recycle
and additional discussion are found in Levenspiel (1996) Chapter 83.
The problems which follow quantitatively verify the findings of the first
two examples.
 Problems 641
In the literature on continuous fermentation much effort is spent calculating
what happens in this or that set up. Most of the schemes are nowhere near
optimum so we do not consider them.
When the cp values are constant throughout, at high and low conversions,
then we have just one independent variable to deal with as composition
changes with time or position. Thus, we can use any one concentration, C,,
C,, or C, in the performance expression. We can compare performance of
various reactor types, plug flow, mixed flow, etc. with no difficulty. This is
what we have done here.
In general, however, cp = f (C,, C,, C,). And when cp varies with composi-
tion, then things get more difficult and we cannot directly compare reac-
tor types.
In 1939, as part of his thesis, Jacques Monod proposed the equation which
we use here. The thesis was published as a book in 1948 and was later
condensed and translated into English in 1949 (see end-of-chapter reference).
In microbiology they use the terms
substrate for the feed.
dilution rate for 117. In chemical engineering, we call this the space
velocity. Here we use neither term. We use space time 7.
chemostat, turbidostat for mixed flow reactor.
We should be aware of the difference in language.

REFERENCES

Levenspiel, O., Chemical Reactor Omnibook, OSU Bookstores, Corvallis, O R (1996).

Monod J., Ann. Reu. Microbiology, 3, 371 (1949)
, Annales de l'lnstitut Pasteur, 79, 390 (1950). Recherches sur la Croissance des
Cultures Bacteriennes, Second ed., Herman, Paris, 1958.
Novick, A., and Szilard, L., Proc. N.A.S., Washington, 36, 708 (1950).

PROBLEMS

29.1. A culture of E. coli was grown on lactose in a mixed flow reactor (V =

1 liter) using various flow rates of a CAo= 160 mg lactoselliter feed. The
following results were obtained:

---

v, literlhr CA,mglliter Cell Concentration, Arbitrary

0.2 4 15.6
0.4 10 15
0.8 40 12
1.0 100 6

Find a rate equation to represent this growth.

642 Chapter 29 Substrate-Limiting Microbial Fermentation

E. coli lives and grows on mannitol with the following kinetics

cA= gm mannitolim3, @ = 0.1 gm cells/gm mannitol

'-c A'ACc,
r -
+2

Find the outlet concentration of cells from the reactor when 1 m3/hr of
mannitol solution (CAo= 6 gm/m3)is fed directly to a mixed flow reactor
of volume

Can you do better and produce more cells (if so, find Cc) by proper bypass or
recycle of fluid from the reactor for the system of

29.4. problem 2? 29.5. problem 3?

29.6. How curious-two different flow rates of CAo= 500 mol/m3 feed to our
1 m3 mixed flow reactor produces the same 100 gmlhr of yeast cells in
the exit stream, namely
at 0.5 m3/hr of feed for which we find CA = 100 mol/m3
at 1 m3/hr of feed for which we find CA = 300 mol/m3
Substrate limiting Monod kinetics should well represent yeast formation.
From this information, find
(a) the fractional yield of yeast,
(b) the kinetic equation for yeast formation,
(c) the flow rate for maximum yeast production,
(d) the maximum production rate of yeast.

A stream of reactant A (CAo= 3, CRO= 0, CCO= 0) is to be decomposed

by the following microbial fermentation

In the following problems sketch your recommended reactor setup with recycle,
bypass, etc., and on each sketch indicate pertinent quantities.
What is the lowest C, which can be obtained in a single mixed flow reactor
of size Vm= 1 for a feed rate of
 Problems 643

What is the lowest CAwhich can be obtained with two properly connected
mixed flow reactors, each of volume V = 1, for a feed rate

What is the lowest C, which can be obtained with three wisely connected
mixed flow reactors, each of volume V, = 1, for a feed rate

For a feed rate v = 3 what is the smallest size of plug flow reactor with
appropriate piping (bypass or recycle or slide taps on the reactor) which
will give

29.13. C, = 0.5, side tap allowed

29.14. Cc = 1.25, side tap not allowed

29.15. Cc = 1.44, side tap allowed

29.16. Find the lowest CA obtainable from a plug flow reactor of volume V, =
4 (bypass, recycle and/or side taps are all allowed) for a feed rate v = 6

29.17. Given the Monod equation to represent a microbial fermentation

cells C
A-R+C -rA= ~CACC
CA + constant

Which of the contacting patterns of Fig. P29.17 could be optimum, and

which can never be? By optimum we mean minimizing reactor volume
for a feed consisting only of CAo.

(4
Figure P29.17

29.18. In his thesis, which was published as a book in 1948, Monod first proposed
the celebrated equation that bears his name. As experimental support for
this equation he presented results from four batch reactor runs on the
644 Chapter 29 Substrate-Limiting Microbial Fermentation

growth of a pure bacterial culture in a lactose solution (see Monod, 1958,

p. 74). Here are the reported data for one of his runs.

- - -

Time Interval -
Number At, hr CA cc
0.54 137 15.5 to 23.0
0.36 114 23.0 to 30.0
0.33 90 30.0 to 38.8
0.35 43 38.8 to 48.5
0.37 29 48.5 to 58.3
0.38 9 58.3 to 61.3
0.37 2 61.3 to 62.5

Fit the Monod equation to this data.

29.19. In Example [Link] we could have used any one of the three contacting
schemes shown in Fig. P29.19. We stated without proof that the bypass
scheme was best. Show that this is so by calculating C,,,, for the three
contacting schemes of Fig. P29.19.

(a)

Figure P29.19
Chapter 30
Product-Limiting Microbial
Fermentation

With sufficient food and harmonious environment, cells multiply freely. However,
no matter how much food is available, eventually either cells crowd each other
out or their waste products inhibit their growth. We call this product poisoning.
Hence, Monod kinetics always is a special case of a more general rate form
which includes product poisoning. A simple equation of the general rate form
for this situation is

order of product poisoning

f

rate constant in a ) t k o b sdecreases as

poison-free environment product builds up

In the special case of sufficient food, or C, + C,, and n = 1,the above equation

C!
reduces to the simplest expression for product poison control

reaction stops when CR reaches C

:

We start with the rate form of Eq. 2, then we extend the treatment to systems
where n Z 1, or Eq. 1. Let us also develop everything in terms of C, in which
case Eq. 2 becomes
646 Chapter 30 Product-Limiting Microbial Ferrnentatioi.l

dr, -
The maximum rate then occurs where - - 0. Solving gives
~ C R

1
'R, ma, rate = (c, + cg - @c,)
Thus, there always is a composition where the rate is optimum.

30.1 BATCH OR PLUG FLOW FERMENTORS FOR n = 1

CA CAo
cc
CR

Figure 30a

To relate time to concentration, we integrate the performance expression of Eq. 3

or in terms of product R,

Graphically this performance equation is shown in Fig. 30.1.

Since the llr, versus CR curve is U-shaped, the optimum way of running the
plug flow reactor is shown in Fig. 30.2.

1
-
rA
Maximum rate

r t b or 7p

Figure 30.1 Graphical representation of Eq. 5.

 30.2 Mired Flow Ferrnentors for n = 1 647

At-high
- - - -conversion
-----

At-low
- - -conversion
------

1
-
R'

here the rate is maximum 0 CR*

2
Figure 30.2 Optimum operation of a plug flow fermentor for poison limiting
kinetics.

30.2 MIXED FLOW FERMENTORS FOR n = 1

For the case of negligible time lag for the feed cells which enter their new
environment, and for high C,, we have

For the special case where C, = 0 and CRo= 0 the above general expression
simplifies to

for kr, > 1 (10)

To evaluate the kinetic constants from mixed flow experiments rearrange Eq.
10 as in Eq. 11 and plot as shown in Fig. 30.3.

The properties of Eq. 10 are displayed in Fig. 30.4.

648 Chapter 30 Product-Limiting Microbial Fermentation

Figure 30.3 Evaluation of the rate constants of

Eq. 2 from data taken in a mixed flow reactor.

Comments
For mixed flow with C, = 0, C,, = 0, and any high C,,.
washout occurs at kr, = 1 for any feed
maximum production rate of cells and product R is obtained at

kr, =2. . . and. . . C, = C$/2

c ~ O ------_
CA is high and close to constant

CA
This graph only for
c, = 0
I
I
I
----7---
Limiting CA
c, = 0 I
I
I
I Maximum possible CR

Largest slope (the tangent to

the cuwe) gives the highest
production rate of R

C~ opt

0 1 2
Washout r' Optimum r is just double rwhwt

Figure 30.4 Properties of the mixed flow equation for the kinetics of Eq. 2.
 30.2 Mixed Flow Ferrnentors for n = 1 649
the maximum production rate of cells and product is found to be

the C, curve is similar in shape to the CRcurve and is proportional to it. Thus,
it rises from 0 to @c;.

Optimum operations for multistage systems follow the same pattern as for
poison-free systems. The general rule is to use mixed flow to reach the
maximum rate in one step. Proceed beyond this with plug flow.
Note that the maximum rate occurs at Cg/2 when C, = 0 and CRo= 0.

Fermentation with n # 1 Poison Limiting Kinetics

For n-th order product poisoning kinetics the rate equation is

These kinetics are displayed in Fig. 30.5. The performance equation for plug
flow is quite messy; however, for mixed flow the performance equation can be
obtained directly. Thus in general, for Cco # 0 and CRo# 0 we have

Figure 30.5 Reaction slowdown is strongly depen-

dent on the order of poisoning, n.
Chapter 30 Product-Limiting Microbial Fermentation

4
CR

Conditions of maximum production rate

CR,opt =

kVC& n n
F R ,max = ( C ~ v ) o=~ -
t --
(n + I)"+ I

t k7m

Figure 30.6 Behavior of a mixed flow reactor for the poison influenced kinetics of Eq. 12.

and for the special case where Ca = 0 and ,C

, =0

The properties of this equation, washout, maximum production, etc. are displayed
in Fig. 30.6. To find the kinetic constants Cg, k and n from experiment first
evaluate Cg in a batch run using an excess of reactant A and letting t -+w . Then
rearrange the mixed flow performance equation to give

log 7, = -log k + n log

(ci"",j

and plot as shown in Fig. 30.7. This will give the kinetic constants k and n.

O F i log (-) CF?

1
log -
k

Figure 30.7 Finding the order of product poisoning and the

rate constant of Eq. 12 from mixed flow reactor data.
 30.2 Mixed Flow Ferrnentors f a r r t = 1 651

Discussion
The similarity in shape of the mixed flow graphs for product-limiting and sub-
strate-limiting (Monod) kinetics (Fig. 29.5 and Fig. 30.4), has led many investiga-
tors to fit product poisoning systems with the simple Monod equation. The fit
will be good, but don't try to use the fitted equation for different feed conditions.
Your predictions are likely to be way off because the logic of the extension
is wrong.
You must first find which of these two factors is rate limiting. This is easy to
do, so there is no excuse for using the wrong expression. In one case the final
extent of reaction is dependent on C, and not on CRo;in the other case just
the opposite holds. The discussion and sketches of Chapter 28 show this.
Expressions for related situations such as plug flow with recycle, and where
both substrate and product both affect the rate are developed by Levenspiel
(1996) Chapter 84.

FRUIT FLY COCKTAIL

Crushed fruit flies (A) ferment to produce an aromatic alcoholic drink (R) with
product-limiting kinetics estimated as follows:

n = 1in a springtime operation

Cg = 0.12 kg alclkg solution
p = 1000 kg/m3

What is the most alcohol you can produce (kglhr) in a commercial-sized mixed
flow reactor (V, = 30 m3)? Also calculate the concentration of alcohol in the
cocktail, and the feed rate needed of fragrant fresh frappCed fluit fries.

From the given data and from Fig. 30.4, we find the conditions which yield the
optimum. Thus

cg 120 kg kg alc -
C ~opt, =
-=(
2 m3 )2
= 60 -- 6% alcohol
m3

Again, from Fig. 30.4 we find

652 Chapter 30 Product-Limiting Microbial Fermentation

and

But rapt =
v
- SO the optimum feed rate is
Vopt

"opt = <
I/ -
--30v3 -
2 -
25.98 m3/hr

The production rate of alcohol, again from Fig. 30.4, is

REFERENCES

Levenspiel, O., Chemical Reactor Omnibook, Chapter 84, OSU Bookstores, Corvallis,
OR, 1996.

PROBLEMS

Material R is to be produced from the following microbial fermentation,

using a feed stream C,, = lo6, CRo= 0, Cco = 0. All quantities are given
in consistent SI units.

C, = 50

A%R+C with cg = 12

In each of the following problems sketch your recommended reactor setup and
on the sketch indicate pertinent quantities.
What CRis obtainable in a single mixed flow reactor of size Vm= 1for a feed rate
30.1. . . . v = 1 30.2. . . . v = 4
What CR is obtainable using two mixed flow reactors, each of volume
Vm = 1, for a feed rate
30.3. . . .v = 1 30.4. . . . v = 3
 Problems 653
For a feed rate v = 3 what size of plug flow reactor with appropriate
piping, recycle, bypass or anything you want to use is needed to give

30.8. The microbial fermentation of A produces R as follows

10A + Cell + 18R + 2 Cells

and experiments in a mixed flow reactor with CAo= 250 mol/m3show that

CR= 24 mol/m3when r = 1.5 hr

CR = 30 mol/m3when r = 3.0 hr

In addition, there seems to be a limiting upper value for CR at 36 moll

m3 for any r,,, , CA,or CC.
From this information determine how to maximize the fractional yield
of R, or @I, from a feed stream of 10m3/hr of CAo= 350 mol/rn3. Cell
or product separation and recycle are not practical in this system, so only
consider a once-through system. Present your answer as a sketch showing
reactor type, reactor volume, CRin the exit stream, and the moles of R
producedlhr.

In spring the fermentation of fruit flies proceeds with n = 1 kinetics, as

shown in Example 30.1. However, in winter or summer, maybe because
of a difference in temperature, the poisoning proceeds differently. So
repeat Example 30.1 with one change:

1
30.9. in winter, n =- 30.10. in summer, n = 2
2
Note: To tell how the value of n affects the reactor performance, compare
your answers here with the answer in Example 30.1.

30.11. Shredded chemical reactor textbooks are degraded to glucose in a pilot

plant well stirred fermenter (V, = 50 lit) under the action of a word
gobbling bug. With a large excess of these shredded incomprehensible
words, the presence of glucose becomes the rate limiting factor. We sum-
marize the findings as follows
when v = 16 bookslhr CR= 54 pmol/liter
when v = 4 bookslhr CR= 75 pmolllit
when v + 0 CR+ 90 pmolllit
Determine the flow rate of books which maximizes glucose production
per word gobbled up, and find this production rate.
654 Chapter 30 Product-Limiting Microbial Fermentation

30.12. Professor Microbe has submitted a paper for publication in which he

studied the growth of a new strain of bug in a mixed flow fermenter
(V, = 46.4) using a pure substrate feed (C,, = 150, CR0= Cco = 0). His
raw data is as follows

20.0 125 with @ = 0.5

22.0 150 (washout)

He asserts, without giving details, that this data clearly represents poison-
limiting kinetics with rate constants

The reviewer of the paper, Dr. Ferment, counters that Microbe is quite
wrong, that the data in fact represents substrate limiting Monod kinetics
with

But, out of orneriness, he didn't present the details of his calculations

either. The editor can't determine who is right (this is not his field), so
he sends the paper and the review to duWayne Zuelhsdorff. What is
duwayne's answer? Is Microbe, or Ferment, or both, or neither, right?
 Appendix-Miscellany

A. Newton's Law

at earth's surface a = g = 9.806 m/s2

f

conversion I kg.m
factor: g, = -
s2.N

B. Length

1010 lo6 39.37 3.280 83 1 0.000 6214

angstrom micron inch foot ml mile

C. Volume

I
in3
I
fluid oz liter
I I
US gal Imp. gal
I
it3
I
(hhl I
(oil)
I I

42 US gal 55 US gal

D. Mass

lb short ton metric ton

avoirdupois C ,000 ,b C or "tonne" ?yy2:bn

656 Appendix-Miscellany

E. Pressure

ThePasca1:lPa = I -N
= I-=kg 10-
dyne
m2 m.s2 cm2
lb
1atm = 760 mm Hg = 14.696f= 29.92 in Hg = 33.93 ft H 2 0 = 101 325 Pa
in2
1 bar = lo5 Pa . . . close to 1 atm, sometimes called a technical atmosphere
1 inch H 2 0 = 248.86 Pa = 250 Pa
F. Work, Energy, and Heat
kg.m2
Thejou1e:l J = 1 N . m = 1-
s2

I I I I I I I I I I
erg ft . lbf cal kgf. m lit. atm Btu kcal Hp . hr kW . hr
f

G. Molecular Weight

In SI units: ~ -
( m ~ )=, 0.032 kg
rnol

(mw),, kg
= 0.0289 -... etc.
rnol

H. Ideal

the gas/ \ kgIm3

contstant

cal - ft3. atm

R= = 1.987 -- 0.7302 lb mol.
rnol .K

liter. atm - [Link]

= 0.082 06 Btu = 82.06 X -
m01 . K - 1.987 lb rnol. OR m01 . K

- Pa liter
8.314 - = 8314 ------
rnol .K rnol . K
 Appendix-Miscellany 657

I. Viscosity ( p )
kg
The poiseuille: 1 P1 = -
m.s

for water: p,,., = P1

for gases: p = P1
for air: p 2 ~= -1.8
~X P1

J. Density p = [s]
kg
for water: p = 1000-
m3

p(mw) & (101 325)(0.0289) = kg

for ideal gas: p = --
R T 2wc (8.314)(293) m3

K. Diffusivity CB and !3, =

[?I
in liquids gas in porous media in in gas
Knudsen regime, or small +
pore diameter commercial
catalyst
(gas at 1atm)<
gas in porous media for large
pore diameter, or bulk diffusion

independent of n ... for liquids

1
G27 cc T3I2,!3 cc - ... for bulk diffusion of gases
n
!3 O= TIi2, independent of n ... for Knudsen diffusion of gases
In a tube of diameter d Knudsen diffusion occurs when nd < 0.01, Pa. m.
In this situation.
658 Appendix-Miscellany

In gas or any fluid: !2 =

Dimensions:

I
In a porous structure: !2 =
m solid. s

L. Concentration CA =
[
m~n":id ]
0.1 1 10 102 103 104 105
I I I I I I

1% in gas at
t
pure gas at 1N
t
pure gas at pure
I
1atm, 1000K 1atm, 273K aqueous 1000 atm, water
273K
mol
I-=
m3
6.24 X lo-'- lbft3
mol

M. Thermal Conductivity k and ken =

[s]
I I I I I I I
A
4 -C------) 4 *
gases liquids and water metals
H2 gas

insuiators C nonmetals; porous structures; alumina,

silica, activated carbon, etc.

W cal - Btu
1-= 0.239 -- 0.578 hr. f t .
m.K m.K.s

k and keffare independent of n

Dimensions:
I
In gas or liquid: k =
[ W
rn fluid K]

In porous structures: k,, =

" 1
m structure. K
 Appendix-Miscellany 659

N. Heat Transfer Coefficient h = -

[mFK]
W cal - Btu
1-= 0.239
m2.K rn2.K.s- 0.176 hr. ft2.
Gas to particle: h = 8 1200 -
Gas to fine entrained particles (fast fluidized systems, FCC, etc.) h = 1000-1200
Liquid to particle: h = 80 1200 -
- 1 for gases
10 for liquids
hd
In packed beds: Nu = = 2 + 1.8(~e,)l'~
(pr)lI3 ... Rep > 100
k

0. Mass Transfer Coefficient kg =

m2surface s
Gas to particle: kg = 0.02 -2
Liquid to particle: kg = 2 x -2 x
- 1 for gases
= 1000 for liquids

In packed beds: Sh = ~
f'
= 2 + 1 . 8 ( ~ e , ) l(sc)"~ ... Rep > 80

For gases: kg llr

P. Rate of Reaction -fi = moles of A disappearing

m3of thing s 1
I I I I I I ...
4 t
t ... 4 ...
cellular rxs., working hard gases in porous
industrial water catalyst particles
treatment plants 4 +
I coal furnaces
human being
at rest

jet engines rocket engines bimolecular reaction in

which every collision
counts, -1 atm, and 400°C
660 Appendix-Miscellany

Q. Dimensionless Groups
iviscous effects

P- molecular momentum transfer ... Schmidt number

molecular mass transfer
a
\diffusion effects

-
- lo-' = 10' . f o r liquids
(10~)(10-~)
/viscous effects

= % = molecular momentumt transfer . . Prandtl number

k molecular heat transfer
heat conduction

= 0.66 - 0.75 for air, A, COz, CH4,CO, H2, He, N2, and other common gases
r 1.06 for steam
-
= 10 1000 for most liquids
-
= 0.006 0.03 for most liquid metals

(inertial effects

t
Re = dup = total momentum transfer ... Reynolds number
P- molecular momentum transfer
L
\ viscous effects
turbulent and laminar, conduction and convection
f
t
hd total heat transfer ... Nusselt number
= k = molecular heat transfer
t
heat conduction alone

k d total mass transfer

Sh = g =
molecular mass transfer
... Sherwood number

@=k- d u p C ~ - (Re)(pr) = total momentum transfer

molecular heat transfer
... Peclet number

= $ = (Re)(Sc) =
total momentum transfer
molecular mass transfer
... Bodenstein number

fluid overtaking caused by molecular diffusion, velocity differences,

( turbulent eddier etc.
-
- movement by longitudinal dispersion ... dispersion groups
movement by bulk flow
This is a new and different type of dimensionless group introduced by workers
in chemical reaction engineering. Unfortunately someone started calling the
reciprocal of this group the Peclet number-this is wrong. It is neither the Peclet
number nor its mass transfer analog, which is widely called the Bodenstein
number in Europe. The difference rests in the use of D in place of 93;hence,
these groups have completely different meanings.

A name is needed for this group. Until one is chosen let us use:

- intensity of axial dispersion

ud "'

-... vessel dispersion number.

uL
 Name Index

Abrahamson, A.A., 448,470 Corcoran, W.H., 110,115 Ghose, T.K., 621

Adams, J., 252 Corrigan, T.E., 204,380,417 Gill, W.N. 339,348
Ananthakrishnan, V., 339, Cresswell, D.L., 393, 417 Gillham, A.J., 564
348 Curl, R.L., 360, 365 Gilliland, E.R., 453, 470
Aris, R., 299, 311, 317, 385, Govindarao, V.M.H., 518
387, 389, 417 Danckwerts, P.V., 258, 277, Grace, J.R., 451,467,470
Arrhenius, A., 27, 72 360, 365, 531, 535, 537, Green, D.W., 535, 537
538, 539,564
Barduhn, A.J., 339, 348 Das, K., 621 Haider, A., 449, 470
Bashby, B., 78 Davidson, J.F., 454, 455, 457, Haldane, J.B .S., 37
Bennackers, A.A.C.M., 161, 470 Han, K., 629
164, 317 den Hartog, H.W., 325, 335 Harrell, J.E., 325, 335
Bergougnou, M.A., 470 Denbigh, K.G., 158, 164, 193, Harrison, D., 454, 470
Berty, J., 398, 417 194, 196, 197, 200, 251, Hatta, S., 529, 534, 537
Bi, H.T., 467, 468,470 346, 348 Hegedus, L., 496
Binns, D.T., 245, 246 Dolbear, A.E., 35 Hellin, M., 80
Bischoff, K.B., 314, 315, 317, Doraiswamy, L.K., 517, 535, Hicks, J.S., 392, 418
389, 392, 417,430, 443 537 Higbie, R., 457, 531, 537
Bliss, H., 417
Hoftijzer, P.J., 530, 537, 558,
Bodenstein, M., 82 Einstein, A., 360 562
Bosworth, R.C.L., 346, 348 Ergun, S., 449, 470 Holmes, D.B., 325, 335
Boudart, M., 381, 417
Holmes, S., 117
Brahme, P.H., 517
Fabre, H., 34 Horn, F., 432, 434, 443
Briggs, G.E., 37
Fan, L.S., 468, 470 Hottel, H.C., 585, 586
Broucek, R., 399
Feller, W., 364, 365 Hougen, O.A., 380,417
Butt, J.B., 400, 417
Fillesi, P., 140 Hull, D.E., 318
Butt, W.M., 81
Fitzgerald, T.J., 140 Husain, A., 245,246
Carberry, J.J., 392, 393, 398, Froessling, N., 401, 417, 578, Hutchings, J., 393, 417
406,417, 581, 586 586
Cates, D.L., 200 Froment, G.F., 389,417, 430,
Ishida, M., 581, 586
Catipovic, N., 193, 200 443
Chandan, B., 206 Frost, A.A., 77
Chou, C.H., 417 Jackson, R., 237
Choudhary, V.R., 511,516 Gangiah, K., 245, 246 Jagadeesh, V., 318
Cleland, F.A., 346,348 Geldart, D., 448, 470 Johnson, M.M., 348
664 Name Index

Jones, R.W., 131, 147 Obando, R., 237 Thornton, J.M., 417
Jungers, J.C., 33, 80, 182, 183, Ogg, R., 36 Trambouze, P.J., 243, 246
200 Ottino, J.M., 365 Treybal, R.E., 365

Kantyka, T.A., 245,246 Parker, A.L., 585, 586

van der Laan, E.T., 300, 317
Kelly, B., 564 Partridge, B.A., 455, 470
van der Vusse, J.G., 246, 253,
Kent, J.W., 318 Paul, E.L., 365
325, 335
Kimura, S., 470 Pearson, R.G., 77
van Heerden, C., 226,228,
Kitterell, J.R., 498 Pease, R.N., 110
237
Knudsen, C.W., 453, 470 Perona, J.J., 325, 335
van Krevelens, D.W., 530,
Konoki, K.K., 432, 434, 443 Perry, R.H., 535, 537
537, 558, 562
Krishnaswamy, S., 498 Piret, E.L., 243, 246
van Swaaij, W.P.M., 161, 164,
Kunii, D., 447, 452, 456, 468, Polthier, 318
317
469, 470, 570, 574, 581, Prater, C.C., 380, 392, 417
Villadsen, J., 399, 418
582, 585, 586, 596, 597,
Villeneuve, Admiral, 168, 169
604, 605 Ramachandran, P.A., 511, von Rosenberg, D.U., 306,
Kunugita, E., 498 516 317
Ranz, W.E., 401,418 Voncken, R.M., 325, 335
Lacey, W.N., 110, 115 Rippin, D.W.T., 360, 365
Lago, R.M., 380,417 Rowe, P.N., 455,470
Laidler, K.J., 33, 77 Wagner-Weisz-Wheeler, 388
Levenspiel, O., 134, 147, 193, Sandy, R., 205 Walas, S., 380, 418
200, 301, 302, 304, 310, Satterfield, C.N., 418, 420 Walker, C.A., 417
314, 315, 317, 345, 348, Satyanarayana, M., 318 Wang, S.C., 581,586
360, 365, 393, 417, 447, Senior, M.G., 237 Watson, C.C., 471
449, 452, 456, 464, 468, Shah, Y.T., 535,537 Watson, Dr., 118
469, 470, 477, 490, 492, Sharma, M.M., 535, 537 Watson, K.M., 380, 417
495, 546, 570, 582, 586, Shen, J., 581, 586 Weber, M., 323, 335
597, 604, 605, 629, 640, Shimizu, F.J., 574, 586 Wedel, S., 399, 418
641, 651,652 Shimoyama, S., 597, 605 Wehner, J.F., 313, 317
Levien, K.L., 348 Shirai, T., 586 Weisz, P.B., 385, 392, 418
Lindemann, F.A., 21, 33 Sir Boss, 117 Welland, R.C., 245, 246
Sjenitzer, 319 Weller, S., 381, 418
MacMullin, R.B., 193, 200, Smith, J.M., 581, 586 Wen, C.Y., 581, 582, 586
323, 335 Smith, W.K., 301, 317 Westerterp, K.R., 161, 164,
Magoo, S.E., 77 Spielman, L.A., 360, 365 317
Mathis, J.F., 471 Standish, N., 318 Wheeler, A., 405, 418
McGreavy, C., 393,417 Summers, 496 White, D.E., 581, 586
Menten, M.L., 21, 26, 33, 37, Suzuki, M., 360, 365 Wilhelm, R.H., 313, 317, 346,
79, 613, 619 Szepe, S., 134, 147, 477, 491, 348
Michaelis, L., 21, 26, 33, 37, 495
79, 613, 619 Szilard, L., 634, 641
Yagi, S. 570, 581, 585, 586,
Monod, J., 634, 641
596, 597, 605
Moore, W.J., 33 Takagi, K., 597, 605 Yoshida, K., 574, 586
Murthy, K.V.R., 518 Tartarelli, R., 406, 418
Taylor, G.I., 311, 317
Nelson, Lord, 169 Teller, A.J., 524, 537 Zhu, J.X., 467, 470
Ng, D.Y.C., 360, 365 Thiele, E.W., 384, 385, 387, Zuelhsdorff, duWayne, 654
Novick, A., 634, 641 389,418 Zwietering, Th.N., 355, 365
 Subject Index

Absorber-reactors reactions in series, 53 porous particle, 385

choice of reactor, 540-543 reactions of shifting order, rate of reaction, 386
design chart, 530 59 surface reaction, 379
examples and problems, reversible reactions, 56 Catalytic reactors
551-565 search for a rate equation, comparison, 400
performance equations, 75 experimental, 396-401
546-551 second order kinetics, 42 performance equations,
rate equation, 527-534 third order kinetics, 45 393-396
Absorbers, 524-527 varying volume, 67-72 CFB, 465-470
performance equations, zero order kinetics, 47 contacting regimes, 448
453-456 Battle of Trafalgar, 168 downflow CFB, 468
Acrolein production, 252 Benzene chlorination prob- fast fluidized, FF, 467
Activation energy lem, 252 jet impact reactor, 470
in strong diffusion regime, Bodenstein number, 339, 660 pneumatic conveying, PC,
390 BR. see Batch reactor 468
of reactions, 27 turbulent bed, TB, 466
Aggregation, 258, 350 Chemostat, 641
Catalyst
Antifreeze, production, 248 Circulating fluidized bed, see
homogeneous, 50
Autocatalytic reactions, 140 CFB
solid, 376
Axial dispersion Classification of reactions, 2
Catalytic reactions
definition, 293 Closed vessels, 300
general rate equation, 391
in packed beds, 311 mean and variance 300
heat effects during, 391
in pipe flow, 310 Coca-Cola problem, 250
in real catalysts, 405
intensity, 309 Cold shot cooling, 434
influencing factors, 378
model, 295 Compartment models,
kinetic equations, 379
Batch reactor, 38- kinetic regimes, 378 284-287
basic performance equa- multiple reactions, 403 diagnosing reactor ills, 287
tion, 91 pore diffusion effects, 381, Convection model, 339
constant volume, 39 403 E curves, 342
first order kinetics, 41 Catalytic kinetics F curves, 344
half life method, 48 controlling resistance, for general kinetics, 345
homogeneous catalyzed 401-402 for n = 0,1, 2 reactions,
reactions, 50 in a single pore, 381 346
n-th order kinetics, 46 pore diffusion intrusion, for reactions-in-series, 347
reactions in parallel, 49 381 when to be used, 339, 341
666 Subject Index

Conversion-concentration Enhancement factor, 529 circulating solids, 465-470

relationship, 86-88 Enzyme fermentation, 611 flow diagram, 451
Convolution example, 272 Equilibrium from thermo, 210 flow models, 452-455
Convolution integral, 270 examples, 213 flow regimes, 448
CSTR, see Mixed flow reactor Expansion factor, 86 Geldart classification, 448
Exponential integral K-L model, 455-463
table. 353 minimum velocity, 449
Davidson bubble, 454 terminal velocity, 449
Deactivating catalysts, 473 Fluidized reactors, 458
design of reactors, 489 conversion equation, 459
F curve, 264
mechanism, 474 example, 460
for convection model, 344
optimum reactor policy, Fractional yield, 242
for dispersion model,
491 in fermentors, 626
302-304
order of deactivation, 476 for tanks-in-series model,
pore diffusion effects, 483
327 Gas-solid systems, see Fluid-
rate equations, 475
Fast fluidization, FF, 467 solid
rate from experiment,
Fermentation, 611 Geldart classification, 448
477-483
Fermentation by enzymes, GILIS reactions, 500-511
Denbigh reactions, 194-198
611-622 application, 510
problem, 251
Fermentor, microbial, 623 choice of reactor, 509
Differential method of analy-
batch, 624, 630 examples and problems,
sis, 39
fractional yield, 626 511-519
batch data, 63 kinetics, 627 rate equation, 500-503
catalytic, 397
mixed flow, 625 reactor performance,
example, 65 Monod equation, 628
Dilution rate, 641 503-509
product distribution, 626
Dirac delta function, 275
product limiting, 645 Hatta number, 529
Dispersion, see axial dis-
substrate limiting, 630 role in reactors, 534
persion
Fermentor performance Heat effects in catalytic reac-
Dispersion model, 293
for mixed flow, 633,647 tions, 391
for large DIuL, 299
for plug flow, 630, 646 Heat of reaction, 208
for n = 1 reactors, optimum operations, 636
312-315 Heterogeneous reactions,
Flow models definition, 2
for n = 2 reactors, 315 see axial dispersion, 293
for open vessels, 300 Holding time, 109
see compartment, 283
for reactors, 312 Homogeneous reactions,
see convection, 339 definition, 2
for small DluL, 296
see tanks-in-series, 321
Dispersion number, 294, 296
which to use, 339, 341
Downflow CFB, 468 Ideal reactors, 90-112
Fluid-fluid systems, see Ab-
sorber-reactors table of performance
Fluid-solid reactions equations, 111, 112
E curve, 262
for convection model, 342 models, 568-582 Inhibition
for dispersion model, 297, Fluid-solid reactors, 589 competitive, 617
299 for a size mixture, 591 noncompetitive, 617
for tanks-in-series model, instantaneous reactions, Integral method of analysis,
323 603 38
mixed flow of solids, batch data, 41
Eadie plot, 615
Earliness of mixing, 259, 350, 594-600 catalytic, 397
354 Fluidized beds, 447 example, 60
conversion for, 273, 351 3 phase, 500
Effectiveness factor, 384, 391 BFB, 451 Jet impact reactor, 470
 Subject Index 667

K-L model Models, see Flow models Plug flow reactor,

assumptions, 455 Models for fluid-solid reac- basic performance equa-
examples, 460 tions tion, 94
for multiple reactions, 463 finding the rate controlling comparison with mixed
for single reactions, 458 step, 582 flow, 122
material balance, 456 for constant size particles, Pneumatic conveying (PC),
Kinetics of deactivating cata- 570-579 468
lysts for shrinking particles, Pore diffusion
distortion by pore diffu- 577-579 effect on deactivating cata-
sion, 483 progressive conversion, lysts, 483-486
from experiment, 477-483 568 PRF, see Plug flow reactor
shrinking core, 569 Problems
table of kinetics, 580 acrolein production, 252
Laminar flow model, see con-
Modulus Battle of Trafalgar, 168
vection model
Hatta, 529, 530 benzene chlorination, 252
Lineweaver plot, 615
Thiele, 391, 394 chirping crickets, 35
Liquid-solid systems, see
Wagner, 388, 391 Coca-Cola problem, 250
Fluid-solid
Monod equation, 628, 641 electric power stations, 8
Multiple reactors, 124 FCC reactors, reaction
Macro fluids, 351 best arrangement, 135 rate, 9
conversion equations, 352, comparison with plug flow,
flow in blast furnaces, 317
353 128
gambling Bashby, 78
lifetime of an element, 360
gambling Magoo, 77
reactor tables, 356
Open vessels, 300 garbage collection, 205
Maximization of rectangles,
mean and variance, 301 grinding of paint pigment,
133
Optimum reactor policy 202
Maximizing operations
with deactivating catalyst, making dollar bills, 335
definitions, 242, 243
491 phthalic anhydride produc-
Mean of a tracer curve, 294,
Order of deactivation, 476 tion, 254
300, 301
Order of product poisoning, pollution of the Ohio
for the dispersion model,
645 River, 319
294,300, 301
for the tanks-in-series popcorn popping popcorn
model, 323 Packed bed catalytic reactor, popper, 89
MFR, see Mixed flow reactor 427 rock and gravel, 205
Michaelis-Menten kinetics, choice of reactor type, 435 running speed of ants, 34
612 staged plug flow, 430 Sherlock Holmes mystery,
in mixed flow, 614 staged mixed flow 432 117
in plug flow, 613 staged reactors, 430 Slobbovian wars, 205
Microbial fermentation, 611, staged recycle, 432 storage of radioactive
623 Parallel reactions, 153 wastes, 336
Minimum fluidizing velocity, examples, 159, 161, 163 Trambouze reactions, 248
449 in mixed flow, 157 van der Vusse reactions,
Mixed flow reactor, in plug flow, 157 253
basic performance equa- Particle size water treatment plants, 8
tion, 96 effective, 386 Product distribution
comparison with plug flow, Particles effect of activation energy,
122 size, 450 236
Mixing of two fluids, 361 shape, 450 effect of temperature, 235
effect on product distribu- Peclet number, 660 Proper choice of reactors,
tion, 363 Performance equation, 2 240-242
668 Subject Index

Rate constant, k, 75 RTD, 257 adiabatic operations, 220

Reaction rate, experimental methods, 261 design chart, 219
Arrhenius law, 27 definition of the E curve, effect on reactions, 215
change with temperature, 260 equilibrium constant, 210
28,72 F, C, E curves, 264,266 multiple reactions, 235
order of reaction, 16 pulse experiment, 262 non-adiabatic operations,
predictions, 32 step experiment, 263 223
rate constant, 27 reactor examples, 229-235
temperature dependent Selectivity, definition, 158 single reaction, 207
term, 27 Segregation, 258, 350 AH,, 207
Reaction rate, definition, 3, Semibatch reactor, 83 Terminal velocity, 449
13,14 Series-parallel reactions Thiele modulus, 391
how fast or slow, 5 examples, 192 definition, 384
Reactions Denbigh network, 194 for different kinetics, 389
elementary, 15 general rules, 187 generalized, 389
kinetic models, 18 graphical representation, Three-phase reactors, see GI
parallel, 15 190, 191 LIS reactors
Michaelis-Menten, 25 irreversible, 181 Trafalgar, 168
molecularity, 16 plug flow or batch, 188 Trickle bed reactor, see G/L/S
non-elementary, 15 Sherlock Holmes mystery, reactors
searching for a mecha- 117 Turbidostat, 641
nism, 29 Slurry reactors, see GILIS Turbulent fluidized bed, TB,
series, 15 reactors 466
shift in mechanism, 31 Solubility and rate, 534
Reactions in parallel, see Par- Space-time, 93, 109 Unsteady-state flow reactor,
allel reactions Space-velocity, 93 83
Reactions in series, 170 Speed of reactions, 5
first order irreversible, 170 Sphericity of particles, 450 Variance of a tracer curve
in laminar flow, 347 Steady-state flow reactor, 83 definition, 294
in mixed flow, 175 Substrate, 641 for the dispersion model,
in plug flow or batch, 173 294, 300, 301
reversible, 181 Tanks-in-series model, 321 for the tanks-in-series
Reactor design, closed recirculation sys- model, 323
introduction, 83 tem, 325
Reactor ills, 288 E curve, 323
Reactor performance equa- Wagner-Weisz-Wheeler mod-
F curve, 327
tion, 2 for general kinetics, 328 ulus, 388, 391
Reactor types, 14 for n = 1 reactors, 328
Recycle reactor, 136 for n = 2 reactors, 328 Xylene oxidation problem,
comparison with plug flow, for reactors, 328 247
139 Temperature and pressure,
optimum recycle, 142 207- Yield, fractional, 242