0% encontró este documento útil (0 votos)
81 vistas117 páginas1 Modelos
SISTEMAS DINAMICOS DE PRIMER ORDEN
Cargado por
Javier ChaparroDerechos de autor
© © All Rights Reserved
Nos tomamos en serio los derechos de los contenidos. Si sospechas que se trata de tu contenido, reclámalo aquí.
Formatos disponibles
Descarga como PDF o lee en línea desde Scribd
0% encontró este documento útil (0 votos)
81 vistas117 páginas1 Modelos
SISTEMAS DINAMICOS DE PRIMER ORDEN
Cargado por
Javier ChaparroDerechos de autor
© © All Rights Reserved
Nos tomamos en serio los derechos de los contenidos. Si sospechas que se trata de tu contenido, reclámalo aquí.
Formatos disponibles
Descarga como PDF o lee en línea desde Scribd
Chapter 3
80
First-Order Dynamic
Systems
As briefly presented in Chapter 1, the dynamic response of processes is of prime con
sideration in the design, analysis, and implementation of process control systems. An
interesting and important characteristic of chemical processes is that their dynamies
change from one process to another, For instance, the response of temperature is dif:
ferent from the response of level. Further, the response of temperature in a heat
changer is different from the response of temperature in a fumace, The principal ob-
jective of this chapter is to show how to deseribe the dynamic response of simple
‘processes by using mathematical models, transfer fimetions, and block diagrams,
Though simple, these processes are taken ffom actual industrial applications. Chapter
4 presents more complex. processes.
‘The mathematical models will be developed starting from first principles. From the
models come the transfer fictions that reveal the terms that describe the process re
sponse: gain, time constant, and dead time (transportation delay or time delay). Most
of the mathematics reviewed in Chapter 2 is extensively used in this chapter,
The modeling of industrial processes usually stats with a balance of a conserved
quantity: mass or energy. The balance can be written as
Rate of masslenergy Rate of masslenergy Rate of accumulation
into contol = out of control of masslenergy
volume volume in control volume
In processes where chemical reactions are not present, the moles are also conserved.
Thus in these processes, we may substitute the term moles for mass in the balance
equation. Section 3-6 discusses processes where chemical reactions are present
In writing these balances, and all ther auxiliary equations, we call on our knowledge
of many areas of process engineering, such as thermodynamic, heat transfer, fluid flow,
mass transfer, and reaction engineering. This makes the modeling of industrial processes
mest interesting and_ challenging! 3-1 Processes and the Importance of Process Characteristics 81
3-1 PROCESSES AND THE IMPORTANCE OF
PROCESS CHARACTERISTICS
It is important to stat this chapter, and indeed to launch the entire subject of process
control, by explaining what a “process” is and describing its characteristics from a
process contro point of view. To do this, let us consider the heat exchanger of Chapter
1, shown again in Fig. [Link].
The controller's job is to control the process. In the example at hand, the controller
is to take action that keeps the outlet temperature, T(), ata specified value, its set pont.
However, the controller receives 2 signal from the transmit. It is through the trans-
‘itter Gat the controller “sees” the controlled variable. Thus, realistically, as far as the
controller is concemed, the controlled variable is the transmitter output (TO). The re-
lation between the tansmitter’s output and the physical variable to control, T(), is
teiven by the transmitter calibration as presented in Chapter 5.
In this example, the controller is to manipulate the steam valve position to maintain
the controlled variable at set point, Note, however, that the way the controller manip-
ulates the valve positon is by changing its ouput signal to the valve, The controller
does not manipulate the valve position directly; it manipulates only its output signal.
‘Thus, as far as the controller is concemed, the manipulated variable is its own out
put (CO),
We can now define the process as anything between the controller output (CO) and
the controller input. Most often, the controller input is provided by the transmiter
Coulput, There are some instances, however, in wihich this may not be the case, such as
‘when a mathematical manipulation, such as filering, is done on the signal from the
transmitter before itis received by the controller. Because the transmitter usually pro-
vides the input to the controller directly, we can say that coniroller input is equal to
the transmitter output (TO). We will point out those examples in which this isnot the
case, In Fig. [Link], the process is anything within the area delineated by the solid
curve, The process includes the VP transducer, the valve, the heat exchanger with
associated piping, the sensor, and the transmitter
0.92 |
Tac |
\
ie ca
Figure [Link] Heat exchanger emol gstem 82 Chapter 3 FistOrder Dynamic Systems
ther inputs
wiv] one | sf} 2 Ltt
Pig Seam
Figure 3.1 th Heat exchanger temperature contol
For firther understanding of what we have just discussed, consider Fig. 3-1, Tb, The
diagram shows all tie parts of the process and how they are related. It shows that the
‘output signal from the controller (CO) enters the VP transducer, producing @ pneumatic
signal. This signal then goes to the valve (V), producing a steam flow. This flow enters
the heat exchanger (HLE) and, along with other process inputs, produces an output
temperature T(Q). This temperature is measured by a sensor (S), and the output signal
from the sensor, maybe in millivolts, is received by the transmitter (I), which produces
a signal (TO) to the controler. Thus the diagram clearly shows that as far as the
controller is concemed, the controlled variable is the transmitter’s output (TO) and the
‘manipulated variable is the controller's output (CO).
Why is it important to understand the characteristics of the process to be controled?
[As we noted in Chapter 1, the control performance provided by the controller depends
fon the adjustment or specification of differeat terms in the controller. Setting these
terms is refered to as tuning the controller. The optimum controller tuning depends on
the process to be contolled and on the tuning criterion. Every contoller must be tuned
specifically for the process it contols. Consequently, to tune a contoller, we must fist
understand the characteristics, or bebavior, of the process to be contolled.
‘Another way to explain the need to understand the characteristics of the process is
to realize that in tuning the controller, what we are doing is “adapting” the contoller
to the process, Thus it makes sense first to obtain the process characteristics and then
to tune the controller, or adapt the “ontoller characteristics,” to that of the process. If
this is done correctly, the complete closed-loop control system, process plus controller,
will perform as require.
‘The present chapter and Chapter 4 discuss pracesses and their characteristics. Chapter
5 briefly presents some terms related to transmitters and also discusses control valves
and controllers and their characteristics. Finally, Chapter 6 puts everything together, it
“closes the loop.” Chapter 7 shows how to tune the feedback controller once the process
characteristics are known, Herein, then, ies the importance of knowing, understanding,
and oblaining the process characteristics. We cam tune the controller only after the
steady state and dynamic characteristics of the process are known,
32 THERMAL PROCESS EXAMPLE
Consider the well-stined tank shown in Fig. 32.1. In this process, constant and equal
inlet and outlet volumetric flows, liquid densities, and heat capacities are assumed; all
of these properties are known. The liquid in the tank is assumed to be well mixed, and 3-2 Thermal Process Example 83
Figure 321 Thermal proces.
the tank is well insulated-that is, there are negligible heat losses to the suroundings
Finally, the energy input by the stiner is assumed negligible
We are interested in developing the mathematical model and tansfer function that
deseribe how the outlet temperature, T(), responds to changes in inlet temperature,
1,(0) An unsteady-stte energy balance on the contents of the tank, the contral volume,
sives us the desired relation between the inlet and outlet temperatures. That is
Rate of energy _, Rate of energy _ Rate of accumulation
into tank ‘out of tank of energy’ in tank
or, in terms of an equation,
Johto ~ fone) = MAO
whee
f= volumenie flow, n/t
py p= itkt and outlet liquid densities, respectively, kgm?
= volume of liquid in tank, m?
‘u(0) bx) = inlet and outet liquid enthalpies, respectively, kg
u(t) = intemal energy of guid in tank, hg
In terms of temperatures, using as reference state for u(t) and b(t) the pure component
in the liquid state at O°F and the pressure of the system, the foregoing equation can be
written as
— AWC,
dt
Foyle, - foc ye, BL}
where
Cy» C, ~ inlet and outlet liquid heat capacities at constant pressure,
respectively, J/kg“C
C, = liquid heat copacity at constant volume, J/kg-"C
1(0), TE) ~ inlet and outlet temperatures, respectively, °C 84 Chapter 3 First-Order Dynamic Systems
Because the densities and the heat capacities are assumed constant over the operating
temperature range, the last equation can be written as
a)
“dt
FOC fo, 10 = Vr (3-23)
This equation is a first-order linear ordinary differential equation that provides the
relationship between the inlet and outlet temperatures. It is important to note that in
this equation there is only one unknown, ‘i. The inlet emperature, 7,(f), is an input
variable and the one that forces the outlet temperature to change. In this example, we
‘want to study how T,() affects T(), so itis up to us to decide how this inlet temperature
will change. Thus it is not considered an unknown. In this chapter and the following
fone, input varibles are not considered unknowns because we have the fivedom to
change thom as we wish
‘To show that there is one equation with one unknown, we explicitly write
Ite ~ cso — voc, 22 (23
1 egy 1 unk (100)
Equation 3.2.3 is the mathematical model for this process. The solution of this dif
ferential equation yields the response of the outlet temperature as a function of time,
‘As just mentioned, the inlet temperature is the input variable, which is sometimes
refered to as the forcing funetion because it is the variable that forces the outlet tem-
perature to change. The outlet temperature is the output variable, which is sometimes
refered to as the responding variable because itis the variable that responds to changes
in the forcing funetion, or input variable,
AAs slated in the beginning of this example, we are interested in obtaining the transfer
function relating T() to 1:(2) To do so, we follow a series of steps that yield the desired
transfer Function; aller this example, we will formalize the procedure, We begin by
‘making variable change that simplifies development of the requied transfer fanetion,
‘Write a steady-state energy balance on the contents of the tank at the initial condi-
tions.
JoC,T,~ foC,T =0 oz
Subracing this equation fiom 1, 3-23 yields
P eve OL (025)
eG MIAO — Ts = sec ne oF
[Note that the derivative of the temperature is also equal to
aT = TT _ aM aT aT.
rn a dia 3-2 Thermal Process Example 85
‘which is the result of subtracting the righthand side of Bs, 3-23 and 3-24, This is
only a trick that proves helpful in the definition of deviation variables and the devel-
‘opment of transfer functions
[As presented in Chapter 2, we now define the following deviation variables;
1e=1) =7 (28)
r@- 1-7. (2.
where
T. T, = inital steady-state values of outlet and inlet temperatures,
respectively, °C
Tn, P(e) = deviation variables of outlet and inlet temperaturs, respectively, °C
Substituting Eqs. 3.2.6 and (3-2.7) into (3-25) yields
RoC 1300 ~ fo = voc, HO
ae 2a)
Bg, 3-28 fs the same as Eq, 3-2:3 except that iti writen in terms of deviation varibles,
‘The solution of this equation yields the deviation variable T)) versus time for a certain
input 1(0) IF the actual outlet temperature, 1{0), is desired, the steady-state value T
rust be added to Tf) in accordance with Fg. 326.
Deviation variables are used almost exclusively throughout contol theory, Thus the
caning and importance of deviation variables in the anaysis and design of process
control systems must be well understood. As explained in Chapter 2, their value indi-
cates the degree of deviation from some inital stady-stato value. In practice, this
steady-state value may be the desired value of the variable. Another advantage in the
use of these variables is that thir ial value, assuming we start from the inital steady
state, is zor0, which simplifies the solution of differential oquations such as Eq 3-28
by the Laplace transform.
Equation 3-28 can now be rearnged as fllows:
WoC, aCe ,
pea THO he
and we lt
Vol,
7-H B29
I, mest
+ me FLO=LO (32.9 86 Chapter 3 First-Order Dynamic Systems
‘The units of + are those of time, From Bg. 3-2.9 we see that for this example,
Lmilfkg im kg“C_
[m*/s][ke/m][0/ke°C)
Because Fg. (3-2.10) is a linear differential equation, the use of Laplace transform
yields
rsl'(s) — 1) + 1) = TS,
‘But the initial value of the temperature, T(O), is at F, so F(0) = 0. Performing some
simple algebraic manipulations gives
1
To) = ho) oan
1 tn
te"aa ba.)
Equation 3.2.12 is the desied transfer function, It is a firstonder transfer function
because itis developed ffom a firsorder differential equation. As we saw in Chapter
2, processes described by this type of transfer fimction are called first-order processes
or first-order Iugs. Equation 2-46 prosented the goneral form ofthis type of transfer
finetion. Inthe present example the tem, K, is unity
The tem transfer function atises from the fact that the solution of the equation
translates, or transfers, the input, [(f), to the output, P(). Transfer functions are further
discussed in Section 33.
‘As.a brio review of Chapter 2, let us assume that the inlet temperature, T,(?), to the
tank increases by ™ “C. That is, the inlet temperature experiences a step change of M
degrees in magnitude. Mathematically, this is written at
Te
cr, in terms of deviation variables,
where u(Q), as shown in Chapter 2, represents a step change of unit magnitude, 32 Thermal Process Example 87
Taking the Laplace transform, we obtain
M
Ty) ==
s
Substituting this expression for ['(s) into Eq, 32.11 rests in
1) *
sas +0
Using the method of parti fractions presented in Chapter 2 yields
re)
+
sat
Obtaining the values of A and B_ by partial fraction and inverting back to the time
domain, we get
r= Mae, 62.13)
TW Ti Md = 02.04)
The solutions of Eqs. 3.2.13 and 32.14 are shown graphically in Fig. 3-22. The
steepest slope of the response curve occurs at the beginning of the response; this is the
typical response of first-order systems to a step change in input. Note also that because
is unity, the total change inthe ourput is M-"C.
TM
rece | rw. 14,0
[aa Tae
Figure 322 Response of fintoier process to a step change in input variable 88 Chapter 3 FirstOnder Dynamic Systems
Section 2-4 presented the significance ofprocess time constant, 7. However, to review
syain, let = 7 in Fg, 32.13, which yields
1) = MCL =e = MC
1) = (632M
Thats, fora step change in input variable, the ime constant indicates the time it
takes the output variable to reach 63.2% oft total change; this is shown graphically
in Fig, 322, In five time constants, 57. the process reaches 99.7% of i toa change;
essentially the response is completed. Therefore, the time constant is related to the
speed of response ofthe process, "The slower a process responds to an input, the larger
the value of = The fister the process responds to an input, the smaller the value of
Tes important to realize that the time consent is composed ofthe diferent physical
propentes and operating parameters of the process, as shown by Eq. 329. That i, the
time constant depends on the volume of liquid in the tank (7), the heat capacities (Cy
and C,), and the process flow (f ), If any of these characteristics changes, the behavior
of the process also changes and this change is reflected in the speed of response of the
proces, or the time constant.
Up to now, the tank has been assumed to be well isultdthat is, there are nes
ligible heat losses to the surroundings. Consequently, there is not a heat loss term in
the enetgytalnce. Let us remove this assumption and develop the maematical mode
and the transfer functions that relate the outlet temperature, T((), to the inlet temperature,
T,(, and to the surrounding temperature, 7.(t).
[AS befor, using the same reference site for enthalpies and intemal eneeey, we stat
with an unseady-state eoergy balance
Focstie — 90 = soca = voc, 22
a
oC hMe, = VALE = 7.9] = FCT) = Yor, wo G25
beg. Fite 70)
wae
(i) = heat transfer rate to the suroundings, T/s
U = overall heat transfer coeicient, Jim?-K-S
A = heat transfer area,
T(t) = temperature of suroundings, *
‘The overall heat transfer coefficient, UJ, is a function of several things, one of them
being temperature, However, in this particular example, it is assumed to be constant
Because the mass of liquid in the tank and its density are also assumed to be constant,
the height of liquid is constant, and consequently the heat transfer area, 4, is also
‘constant.
an input variable 3-2. Thermal Process Example 89
[Fquation 3.2.15 provides the mathematical model of the process, To oblain the trans
fer functions, we start by introducing the deviation variables. This is done by fist
writing a steady-stale energy balance for this process at the initial conditions,
J0C,?, ~ UAT =T,] ~ foC,T = 0 6-246)
Subtracting Eq. 3.2.16 from Eq. 3.2.15 yields
foC AT Ae, ~ Ty) ~ VALT, - T) - (TA) T0.
aT) —F
)
at
= foc rte = 7] = ope, 247)
Please mote that the trick with the differential term (accumulation) has been done again.
Define a new deviation variable as
1g ~ rae - F. 248)
Substituting Eqs. 3-2.6, 3.2.7, and 3-2.18 into Eq, 32.17 yields
6249)
foC Ti) = UAIT(® ~ P40) = foC,T1e = Vo. a i
Equation 3-2.19 is the same as Eg, 3.2.15 excopt that itis written in terms of deviation
variables. This equation is also 2 frstorder linear ordinary differential equation. In tis
case there is sill one equation with one unknown, T(@, The now variable isthe su
rounding temperature T.(), which is another input. As this temperature changes, it
atfects the heat losses and consequently the process liguid temperature
Equation 3219 can be anianged as follows
Voc, ate,
Jol, + UA dt
UA
; fC, ,
1h ra NO pe amane
SoC, *UA*
go FT = KPO + KT) 229)
where
(3.22) 90 Chapter 3 FistOrder Dynamic Systems
+ dimensionless 223)
‘The right-hand side of Eq. 3220 shows the two input variables, Tf) and 1(0,
acting on the output variable, 1),
Taking the Laplace transform of Eq, 3-220 gives
rsh(s) = PO) + Ms = KTS) + KTS)
But the intial value of the temperature, T(O), is at T, so 1(0) ~ 0. Rearanging this
‘equation yields
ry) + —& (hs 6-228
a
wei
If the surrounding temperature remains constant, 7,0) = 7, then T,(Q) = 0, and the
transfer function relating the process temperature to the inet temperature is
6-225)
1 the inet liquid temperature remains constant, 7,() = T, then T,() = 0, and the tans
fer finction relating the process temperature to the sutounding temperature is
6-226)
If both the inlet liquid temperaure and the surrounding temperature change, then Ea.
3-224 provides the complete relationship
Equations 3-225 and 3.2.26 are the typical firstorder transfer functions presented
in Section 2-4, In this case, however, the steady-state gains (sometimes also called
process gains), K, and Ky, are not unity, as was the case in Tg, 3-2.12, To review
briefly the significance of the steady-state gains, let us assume that the inlet termperature
to the tank increases, in a step fashion, by Mf °C, that is,
Mu
Tey
‘The response of the temperature to this forcing function is given by
KM
POS Ger TeKiM|
ro eC
3-2 Thermal Process Example 91
fiom which
T= KM =e") (3-227)
T= T+ KMI=e) (3.2.28)
‘The output response is shown graphically in Fig, 3-23. The total amount of change
in T() is given by AyM, the gain times the change in input. Thus the guin tells us how
‘much the output changes per unit change in input, or how much the input affects the
‘output. ‘That is, the gain defines the sensitivity relating the output and input variables!
ean also be defined mathematically as follows:
SO _ A ouput variable (e228
AL ~ A input variable
“The gain is another parameter that describes the characteristics of the process. Con-
sequently, it depends on the physical properties and operating parameters ofthe proces,
as shown by Egs. 32.22 and 3.223. The gains in this process depend on the fow,
density, and heat capacity of the process guid J, p, and C,), on the overall heat wansfer
coefficient (U), and on the eat transfer area (4). If any of these changes, the behavior
of the process changes and is reflected in the gain
‘There are two gains in this example, The fist one, K, , relates the outlet temperature
to the inlet temperate. The other gain, K;, relates the outlet temperature 19 the sur
rounding temperature, The units of the gan tem mast be the units of the output variable
divided by the units of the input variable; this is obvious from Fa. 3-229.
[Note that the gain clearly indicates the proces characteristics. Inthe first part of this
example, the tank was assumed 10 be well insulate, and the gan, given by Eq. 3-2.12,
‘was uly. That i, in the steady sate, all the enengy entering with the infet steam exits
with the outlet stream, and the inlet and outlet temperatures are the same, This is not
re, %¢ ri,
Figure 323 Response of fistorder proves: 10 a sep change in inne varble 92 Chapter 3. First-Order Dynami
Systems
the case when the assumption of good insulation is removed and the tank is permitted
to transfer energy with the surroundings. Note that K, and Ky, given by Eqs. 3-2.22
and 3.2.23, are less than unity, which indicates that when the inlet temperature increases
by M degree, the outlet temperature does not increase by that much, That is, if the
energy in the inlet steam increases, then the energy in the ouet stream does not
increase as much because there is some energy transfer 10 the surroundings; this of
couse makes sense, Tt also makes sense that if VA fpC,,. then the inlet temperature
will have a greater effect on the ouflet temperature than will the surrounding temper-
ature, that is, Ky K,
Equation 3-2.24 shows that there is only one time constant in this process. That is,
the time it takes the outlet temperature to reach a certain percentage of its total change
duc to a change in inlet temperature is equal to the time it takes to reach the same
Percentage of the total change when the surrounding temperature changes,
It is always important, during the analysis of any process, to stop at some point to
check the development for possible errors, After the development of Eq. 32.20 is
tually a convenient point. A quick check can be made by examining the signs of the
equation to see whether they make sense in the real world, In Fq, 3-2.20, both gains
are positive, The equation indicates that if the inlet temperature increases, then the outlet
temperature also increases; which makes sense for this process. The equation also shows
that if the surounding temperature increases, then the outlet temperature increases. This
makes sense because when the surrounding temperature increases, the rate of heat losses
from the tank decreases, thereby inereasing the temperature of the contents of the tank.
Another check consists of examining the units of 7 and K. We know what each of them
should be, and the defining equations, Eqs. 3-221 through 3-223 in this example,
should confirm these expectations. This quick check builds our confidence and permits
us to proceed with the analysis with @ renewed hope of success.
Before finishing with this section, let us summarize the procedure we followed to
develop the transfer functions
1, Write the set of unsteady-state equations that describes the process, This is called
modeling,
Write the steady-state equations at the inital conditions,
Suibiract the two sets of equations, and define the deviation variables.
‘Obtain the Laplace transforms of the model in deviation variables
‘Obtain the transfer functions by solving the Laplace transform explicily for the
transformed output variable(s)
We {allowed these five steps in our thermal example. They constitute an organized
procedure that yields the transfer functions
DEAD TIME
Consider the process shown in Fig. 3-3.1. This is essentially the same process a8 the
cone shown in Fig. 3-21, In this ease, however, we are interested in knowing how T,()
responds to changes in inlet and surrounding’ temperate,
Let us make the following two assumptions about the ext pipe between the tank and
point 1. First, the pipe is well insulated. Second, the flow of liquid through the pipe is
ideal plug flow (highly turbulent) with no energy diffusion or dispersion so that dere
js essentially no backmixing of the liquid in the pipe. 3-3 Dead Time 93
|
Figure 331 Themal proces.
Under these assumptions, the response of T,() to the disturbances f(t) and 1(f) will
‘be the same as T(@) except that it will be delayed by some amount of time, That is, there
will be a finite amount of time between the intial response of T(t) and the change of
(0; this delay is shown graphically in Fig. 3-32. This finite amount of time has
developed because of the time it takes the liquid to move from the exit of the tank to
point 1 and is called a pure dead time, transportation delay, or time delay. It is repre-
sented by f and in this case can be easily estimated from
distance 24
Velocity ~ 771A,
JL
he sae 3a)
f
where
f = volumetric fow, mis
A, = cross-sectional area of pipe, m?
T= length of pipe, m
ooo
<1
hemp
Pe E fy +xM
L <<
mc | TOF nw
7
| Tie
Figure 5
Response of a thermal process ta step change in inlet temperature 94 Chapter 3 First-Order Dynamic Systems
Different physical variables tavel at different velocities:
Electric voltage and curent travel at the speed of light: 300,000 kmis, or
984,106 fuls
+ Liquid flow and pressure travel at the speed of sound in the fuid: 340 mvs, or
1100 fs
Temperature, composition, and other fluid properties travel at the velocity of the
fluid: typically, up to 5 mis (15 fUs) for liquids and 60 mvs (200 YS) for gases,
+ Solid properties travel at the velocity of the solid, such as coal in a conveyor, cake
im a filter bed, and paper in a paper machine
From this information, we can see that for the distances typical of indus process
contol systems, pure dead time is significant only for temperature, composition, and
Cher Mid and sod properties that are propagnted through space by the moving uid
or solid
Even when pure dead time (dead time due to transportation) is negligible relative to
the process ime constant, the response of many processes may appear 10 exhibit dead
time due to the combination of several firs-order processes in series, as we shall see
in Chapters 4 and 6, Tis pseudo-dead time cannot be easily evaluated ffom fandamentl
Principles and must be oblained empirically by approximation of the process response,
Methods to cany out such empirical evalation will be presented in Chapter 7.
Because dead time is an integral part of processes, it must be accounted for in the
transfer functions. Equation 2-18 indicates that the Laplace transform of a delayed
fincton is equal to the Laplace transform of the nondelayed function times the term
‘8: the term #78 js the Laplace tansform of dead time, Thus, ifthe transfer functions
relating T,(t) to T,(2) and T,(t) are required, using the assumptions stated at the beginning
of the section, the transfer functions given by Eqs. 3-2.25 and 3-2.26 are multiplied by
wor
=kee 32)
nel
and
Tis) | Kew :
TQ w+T 633)
‘At this point, it must be recognized that the dead time is another parameter that helps
define the characteristics of the process. Equation 3-3.1 shows that fe depends on some
physical properties and operating characteristics of the process, similar to K and 7,
If any condition of the process changes, then this change may be reflected in a
change in fy
‘Before concluding this section, we must stress that one of the worst things that can
happen to a feedback control loop is a significant amount of dead time in the loop. The
performance of Feedback contol loops is severely affected by dead time, as we will see
in Chapters §, 8, and 9. Thus processes and control systems should be designed to keep
the dead time’ to a minimum. Some steps we can take to minimize dead time include
putting the measurements as close to the equipment as possible, selecting rapidly re 3-4 Transfer Functions and Block Diagrams 95
sponding sensors and final control elements, and using electronic instead of pneumatic
instrumentation for processes with short time constanls
3.4 TRANSFER FUNCTIONS AND BLOCK DIAGRAMS
34,1 Transfer Functions
Chapter 2 presented the concept of transfer functions, This concept is so fundamental
to the study of process dynamics and automatic process contol that at this time we
briefly consider, once more, some of its important properties and characteristics
We have already defined a transfer fimetion as the ratio of the Laplace-ransformed
output variable to the Laplace-ransformed input variable. Transfer functions are usually
represented by
YG) _ Klaas + agus +. + mis + Dee
8 be
COX) Ba Da + bs D “
where
GG) = general representation of @ transfer function
Y(6) = Laplace transform of the output variable
X(9) = Laplace tansfoem of the input variable
K a's, b's = constants
fe = dead time
Equation 3-41 shows the most general and best way to write @ transfer function
When it is written in this way, K represents the gain of the system and has as units the
units of ¥() aver the units of X(Q, The other constants a’s and b's, have as units time)’,
where + is the power of the Laplace variable, s, multiplied by the particular constant;
this will render a dimensionless term inside the parentheses because the unit of 4 is
Thime, Notice that the coefficient of is 1
Note: In general, the unit of » isthe reciprocal of the unit of the independent variable
used in the definition of Laplace transform, Eq. 2- 1.1. In process dynamics and control,
the independent variable is time, so the unit of is Utime.
The transferfunction completely defines the steady-state and dynamic characteristics,
or the total response, of a system described by a linear differential equation. Itis
characteristic of the system, and its terms determine whether the system is stable or
‘unstable and whether its response to a non-oscillatry input is oscillatory. The system,
for process is said to be stable when its output remains bound (finite) for all times for
1 bound input. Chapter 2 presented some discussions on stability and how it is related
fo terms in the transfer function. Chapters 6, 8, and 9 treat in more detail the subject
of stability of process systems.
The following are some important properties of tansfer functions
1. In the transfer fimetions of real physical systems, the highest power of y in the
numerator is never higher than that in the denominator. In other words, 1 > m.
2. The transfer function relates the transforms ofthe deviation of the input and output
variables from some initial steady state, Otherwise, the nonzero initial conditions
‘would contribute additional terms to the transform of the output variable 96 Chapter 3 First-Order Dynamic Systems
3. For stable systems, the steady-state relationship between the change in output
variable and the change in input variable can be obtained by
lim Gs)
This stems fiom the final value theorem, presented in Chapter 2
lim Y(t) = lim s¥(s)
mone
fim YO = lim G(X,
fim Ye) = tim GE)Iim 3X(9)]
lim Y@
Jim G¢s)}llim Xo)
‘This means thatthe change in the output variable after a very long time, if bound,
can be oblained by multiplying the transfer function evaluated at s— 0 times the
final value of the change in input.
3-4.2 Block Diagrams
A very useful tool in process contol is the graphical representation of transfer Functions
by means of block disyrams. This section offers an introduction to block diagrams and
block diagram algebra,
All block diagrams are formed by a combination of four basic clements: amrows,
summing points, branch points, and blocks; Fig. 34.1. shows these elements. The
arrows in general indicate flow of information; they represent process variables or
control signals, Each arrowhead indicates the direction of the flow of information, The
summing points represent the algebraic summation of the input amows, E(s)
R(s) = ©(3). A branch point is the position on an arow at which the information
RS Goal Me
aca #0] Bianen
wow Veta, a
Figure 34.1 Elements of a block dia-
gan, 34 Transfer Functions and Block Diagrams 97
Tee, Fe!
Figure 3-42 Block diggram of Bg, 3.2.12,
branches out and goes concurently to other summing points or blocks. The blocks
represent the mathematical operation, in transfer fanetion form such as G(s), which
is performed on the input to produce the output. The amows and block shown in
Fig, 3-4. represent the mathematical expression
M(s) ~ GASES) ~ GGIR(S - CO)
Any block diagram can be handled, or manipulated, algebraically. Table 3.4.1 shows
some rules of block diagram algebra. These rules are important any time a compli=
cated block diagram is simplified. Let us look at some examples of block diagram
algebra.
Bees
Draw the block diagram depicting Eqs, $2.12 and 3-224,
Equation 32,12 is shown in Fig. 34.2. Equation 3-2.24 may be drawn in at Teast
to different ways, as shown in Fig, 3-43, Often the diagram with fewer blocks is
preferred because itis. simpler.
‘The block diagrams of Eq. 3-224 show graphically that the total response of the
system is oblained by algebraically adding the response due to a change in inlet tem-
perature to the response due to a change in sunounding temperature, This algebraic
addition of responses due to several inpuls to obtain the final response is a property of
linear systems and is called the principle of superposition. This principle also serves as
the basis for defining linear systems. That is, we say that a system is linear if it obeys
the principle of superposition,
10,78 nO.
Sore, $ rw
0 He
Figure 3-43 Block diagram of Fg. +224, 98 Chapter 3 First-Order Dynamic Systems
Table 34.1. Rules for
Block Diagram Algebra
1 YS)= XS) Xs) = X63),
2 Asecite a Geamive pes
Ys) = Gals} Gls) X(9) = 7
2 Diu Pope
YS) = Gls) [XA(s} — Xols), = Gls) Xs) — Gls) Xafs}
14} foo)
4, Blocks in Parallel:
Y65} = Gils) + GyLS)] XG}
GAs; X6) + Gx) XG}
xis Yoo Xe) Yo)
MSG foioeeun
Eel
5, Positive Feedback Loop:
Ys} = Gls XS) + Gx
Ga
Ye Xt)
™O few
Gis —__ x05
Gis oe”
aa _| Ye!
exiidga|
ae
6, Negative Feedback Loop:
Y(S) = Gls) XU) + G65) M9) = 34 Trae Pts an ck Digan 99
Determine the transfer functions relating Y(s) to. Xy(s} and X,(s)ffoen the block diagram
shown in Fig, 44.4. That is, obtain
Yo gy UD.
x) Xs
Using rule 4 of Table 3~4.1, the block diagram shown in Fig, 3-44a can be reduced
fo that of Fig. 3445 (please note that reduction is used in this context to mean
x0)
Ku
G,-6;
Xa.
Slee,
Ye=(G__UXsls
XQ)
BS aaa
@
Figure 3-44 Block diagram for Example 3-42. 100 Chapter 3 FitstOnder Dynamic Systems
Simplification and that it consists of reducing the number of blocks). Using rule 2,
Fig. 3-44h can be futher reduced to Fig. 3-4.4¢ Then
Y¥(s) = Gs(G, — G)X, (5) + (Gy — DXAL5)
from which the two desied transfer fnetions can be determined. They are
wish
Kn AG - G)
Example 3-42 has shown a procedure to reduce a block diagram to a tansfer func-
tion, This reduction of block diagrams is necessary in the study of process contol, as
will be clear in later chapters. In these chapters, numerous examples of block diagrams
of feedback, cascade, feedforward, and multivariable control systems are developed
Let us look at che reduction to transfer functions of some of these block diagrams.
EXAMPLE 343
Figure 3-45 shows the block diagram of a typical feedback contol system, From this
diagram, determine
ce) Os)
Lo os,
Figures 3-456 through d show the different reduction steps using the rules in Table
3-4.1, Finally, from Fig. 3.4.54, we obtain the transfer functions as
Cs, [Link],
Cs] 1 + CCGG. 42)
fan} GG.
oe) GG, _ 3)
Ls) 1+ [Link] “oe 3-4 Transfer Functions and Block Diagrams 101
oe ee
feb Hore
;
‘
Byles,
ee “( +04 7 Ss
os
J oy
His)
| Traceone|
ce) aaa | 5 cy
So aaa
1
Figure 3-45 Block diagram o
feedback control system.
Example 3-43 shows how to reduce a simple feedback contol system block diagram
to transfer functions, These types of block diagrams and transfer fictions will become
usefl in Chapters 6, 7, 8, and 9, when feedback contol is diseussed.
The tcansfer fanetions given by Eqs. 3-42 and 3-43 are referred to as closed-loop
transfer functions. The reason for this tem will become evident in Chapter 6, Looking
at Fa, 34.2, note that the numerator is the product of all of the transfer functions in
the forward path between the two variables related by the transfer function, ((s) and
C(s), The denominator of this equation is 1 plus the product of all the transfer fane-
tions in the control loop shown in Fig, 3.4.54 Inspection of Eq. 3-43 shows that the 102 Chapter 3 First-Order Dynamic Systems
‘numerator is again the product of the transfer functions in the forward path between
L(3} and C(s). The denominator is the same as that of Eq, 3-42. If there had been more
than one forward path between input and output, the development would have shown
the numerator to be the algebraic summation of the product of the transfer functions in
each forward path
EXAMPLE 3-44
Consider another typical block diagram as shown in Fig, 3-46 Chapter 10 shows that
this block diagram depicts a cascade control system. For now, simply determine the
following transfer fimetions
CO) yy
Ris) Ls)
‘The block diagram of Fig, 3-4.6« can be thought of as being composed of two closes
Toop systems, one inside the other (in practice, this is exactly what it is), Figure 3-4 6h
and c show the steps to reduce the block diagram of Fig, 3-4,6a: rule 6 is applied twice
From Fig 3-4,6¢, the following transfer functions are obtained:
Gas)
and
cs
Us) 1 6,GG,+ [Link],
G45)
We have leamed how to develop several tansfer functions (Eqs. 342, 3.43, 3.44,
and 3-455) fiom block diggrams. We have not intended, however, to give their signif.
jcance; this will be done in the chapters where control systems are presented
‘A useful recommendation is to write, next to each amow, the units of the process
Variable or control signal that the axow represenis. This makes it fairly simple to
recognize the units of the gain of a block, which are the units of the output arrow over
the unis of the input erow. This procedure also helps avoid the algebraic summation
of arrows with different units. It is extensively illustrated in Chapuers 6, 8, 9, 10, 11,
12, and 13.
‘As mentioned at the beginning of this section, block diagrams are a very helpful tool
in process control. They show the ow of information in a graphical way, identify the
input and output signals (or variables) in a system, and show the occurrence of loops
and parallel paths, We will lean more about the logic of block diagrams and get more
practice drawing them as we continue our study of process dynamics and contol, Later
chapters make use of block diagrams to help analyze and design control systems, 344 Transfer Functions and Block Diagrams 103
LG,
aly Laci ~\_ | A
ese eee ee
a ou)
BEA fo. Toes
¥
Le,
Re
how
Figure 3-46 Block diagram of a cascade control system Lod Chapter 3 First-Order Dynamic $
mo. % mst %
40, etn
£40. stig
pin
Figure 381 Gas ves
3:5 GAS PROCESS EXAMPLE
‘Consider the gas vessel shown in Fig. 3.1. A fan blows air into a tank, and from the
tank the air flows out through a valve. For purpases of this example, let us suppose that
the air flow delivered by the fan is given by
£4) = D16m¢0)
where
Fil) = gas flow in soffmin, where sef is cubie feet at standard conditions
‘of 60°F and 1 atm
m.() = signal to fan, Ye
‘The flow through the valve is expressed by
A.) = 0.00506m, (1) PUTA = PM
where
£48, = gs flow, sein
‘mu(d) = signal to valve, %
p(t, = pressure in tank, psia
1\(t) = downstream pressure fiom valve, psia
“The volume of the tank is 20 AY, and it can be assumed hat the process ecurs iso
thermally at 60°F. The inital steady-state conditions are
FH R<8sctm p= 40 pia py= 1am ii 50%
We want to develop the mathematical model, transfer functions, and block diagram
that relate the pressure inthe tank to changes in the signal to the fan, m(); in the signal
to the valve, m, (2); and in the downstream pressure, (0)
‘We must first develop the mathematical model for this process, An unsteady-state
‘mole balance around the control volume, defined as the fin, tank, and outlet valve,
provides the starting relation, That is
Rate of moles into _ Rate of moles out _ ale of accumulation
contol volume of control volume ~ of moles in control volume 3-5 Gas Provess Example 105
5, jn equation form,
~ ~, dr
Bio ~ bran = 22 Gs)
Tea. 3 unk. [£00 4(0, 000]
where
B = molal density of gas at standard conditions, 0.00263. Ibmoles/scf
at) = moles of gas in tank, lomoes
‘The fan provides another equation
FAG = 0.16m,¢) G52)
2 eg, 3 unk.
[Note that because mf) is an input variable, itis up to us to decide how it will change.
‘Thus it is not considered an unknown,
‘The valve provides still another equation:
LAD) = V.00S06M,CE, PLOT
653)
3 eq, 4 unk [p(0)
«The signal.) and downstream pressure p(t) are other input variables and thus are
not considered unknowns.
Because the pressure in the tank is low, the ideal gas equation of state can be used
to relate the moles in the tank to the pressure.
DAV = n(ORT bs)
4 oq, 4 unk
‘The set of Eqs. 35.1 through 3-5.4 constitutes the mathematical model for this pro-
cess. The solution of this set of equations describes, considering the assumptions taken,
hhow the pressure in the tank (the output) responds to changes in m.(), m,(), and Bilt)
(the inputs),
So far we have completed the first step of the procedure, outlined atthe endl of Section
3.2, Before proceeding to the second step, we must realize that the expression for (0,
Bq. 3.53, is a nonlinear equation. The Laplace transformation can be applied only 10
linear equations, Thus, before continuing to the second step, we must linearize all the
nonlinear terms. This linearization is done using Taylor Series expansion as presented
in Chapter 2, 106 Chapter 3 First Onder Dynamic Systems
Because (0) = flm(0), pd, pd), its linearization is done with respect to m,(),
p(t), and p(t about their steady-state values m,,.p, and P,
a7 + LO. _ AO, _ 0) ie
LAO) + am (m,(t) — m,] + ap) He p+ ap) Late, Dy
or
1 fae C60 = + CLO F.+ Cile- Fr OSS)
were
At, « —
am 0.00506 Jap — 7.) (35.6)
= 0.00506m (12) = pop = pi) (35.7)
0.005067 (12) —[Link]""-7) 05.
and
5.)
Now there is a set of linear equations (Eqs. 3.5.1, 35.2, 3-5.4, and 3-5.5) that describes
the process around the linearization values of ti, Dy a0d
To. simplify this set somewhat, solve for n(®) in Eq. 354 and substitute it in
Bq, 351
ae
Bio - A100 = EO
+5.19
RT dt uy
With this simple substitution, the st of equation is reduced to tree equation, Eqs
35.10, 552, nd 3-55, with three unknowns, f(0), fi and pC.
We can now proceed with the next two stops ofthe procedure, which cll for writing
the steady-state equations, subtracting. them ffom thir respective counterparts, and
defining the required deviaton variables
First we write a steady-stte mole balance around the tank. 3-5 Gas Process Example 107
Subtracting this equation fiom Eq. 35.10 gives
VT
AU ~A ~ also -71 = pO os.)
Defining the following deviation variables
F=f -f FDL -f, PO. we BP
and substituting these variables into Eg. 3-5.11 yield
Vv apo,
ia = 05.12
pF It, ~ PELE (5.12)
‘Writing the steady-state equation for the fan and subtracting it from Eq. 3-5.2 give
Fd = 0.1604(0) 65.5)
where (0 = mit) —
From Eq, 355, afler subtraction off, fiom both sides of the equation,
FA) = CMA) + CPO) + CPO 65.14)
where
Mat =m) i,
Put = pil =
Recapping what has been done, there are now three equations, Eqs, 3-5.12 through
3.5.14, and three unknowns, F,(), Fd, and P(Q. All of these equations and variables
are in deviation form,
We now proceed with the last two steps of the procedure. Substituting Eqs. 3-5.13,
and 3-5.14 info Eg. 3-5.12, taking the Laplace transform and rearranging, yield
K, Ks
Pos) A i) — SE Msn AS Piss 058
whee
0.16 psi fei Vn ato
rors Ops Rise. ™ O19
‘The desired transfer fimetions can now be obtained,
Ps) -_K 5.07)
Ms 1 108 Chapter 3 First-Order Dynamic Systems
Pi)
Mis eT bss)
fa 35.19)
oT
(3)
Because the steady-state values and other process information are known, all gains and
the time constant can be evaluated as
K, = 0615 psite —-K, = 0619 psite —K, = -O611
242 min
All of the transfer functions are of fist onder. Fig. 3-52 shows the block diagram for
this process
‘Afler considering the presentation in Chapter 2 about transfer finetions and their
response (0 inputs, and after what has been presented in this chapter, we should have
good fecting for the complete response of any first-order system. We know by ana-
Iyzing Eq. 3.5.17 that ifthe signal to the fan increases by 10%, then the pressure in the
tank wil wlimately change by + (10\(K1! psi. We also know that 63.2% of the chang,
‘or 0.63 10)(K,). will occur in one time constant. Ths response is shown graphically in
Fig 3-53, Remember that X, isthe gan that Mr, has on P(t) and that gives how
fis P(t) responds to 2 change in Mi)
[Equation 3-518 indicates that if the signal to the valve inercases by 5%, then the
pressure in the tank will decrcase by (5)K, psi, The negative sign in font of the gain
indicates this type of response. Certainly it makes sense that if the signal to the valve
increases, opening the valve and thus extracting more gas from the tank, then the
pressure in the tank should fll
‘Equation 3-5.19 indicates that if the downstream pressure from the valve increases
by 3 ps, the the pressure in the tank will decease by (39K psi. That is, if P, (t) changes
by *3 psi, P(t) will change by (3)K,, From a physical point of view, however, this
does not make any sense. If the downstteam pressure increases, then the flow through
the valve decreases, increasing the pressure in the tank. Where is the discrepancy?
Reviewing the definition of Ky, we see that it depends on C, , and on the basis of Eq
3.57, itis obvious that C, is negative. Thus K- is negative (K, =~ 061 1), and con
sequently, the pressure in the tank actually increases
igure 35.2 Block disgram for gas
proves, 3-6 Chemical Reactors 109
+10 rob
mid % Mio. %
ca °
Time
Ps i0k, ok;
632k,
ith, Pit os
e o
Figure 35.3 Response of pressure to signal to fan
At this point, we should reformulate the procedure for oblaining the transfer fune-
tions. This is necessary because we now realize tha linearization of nonlinear terms is
fan important step in the procedure.
1. Write the set of unsteady-state equations that describes the process. This is called
modeling
2. Linearize the model if necessary.
3, Write the steady-state equations at the initial conditions
Subtract the two sets of equations, and define the deviation variables.
‘Obsain the Laplace transform of the linear model in deviation variables.
CObiain the transfer functions by solving the Laplace transform explicitly for the
‘transformed output varable(s)
3-6 CHEMICAL REACTORS
36.1 Introductory Remarks
“The example presented in this section involves @ chemical reaction, Because the soi
hiometies of the reactions are given in moles, the balances done in chemical reactors
are usually mole balances, either on a specific component i or on tolal moles. The
problem however, i that mole balances cannot be writen using the equations presented
in the introduction to this chapter. That is, taking the reactor as contol volume,
Rate of Rate of Rate of accumulation
component i= component i of component i
finto reactor out of reactor in reactor
Moles are not necessarily conserved in chemical reactions. Consider, for example, the
action 2A +B 3S + P, Under steady-state operation, the moles of reactant A
exiting the reactor are not the same as those entering (reactant A is consumed!). Sim. 110. Chapter 3 First-Order Dynamic Systems
‘larly, 3 moles of reactants are used, whereas 4 moles of products are formed, so the
{otal moles are not conserved, either. Remember, however, that the total mass is always
conserved.
‘Therefor
the mole balance equations must account for the production or depletion
of moles due to reaction, The unsteady-state component mole balance that accounts for
this production or depletion is written as
Rate of Rate of Rate of Rate of accumulation
component i - component i + production of = of component i
ino reactor out of reaclor component jn reactor
‘The rate of production of component i in the reactor is usually given by
Rate of production of component #= yy,V, moles of component itime
where
the stoichiometric coefficient of component i in the reaction
volume of reacting mixture
rate of reaction of the key component in the resetion, This rate (always
moles of key _component_formedieacted
Positive) is usually given im Fe of reacting mixture)
‘An important term in these definitions is Key component. The key component may
bbe any component-reacant or productin the reaction. The stoichiometic coefti-
cient, v, of the chosen component is made equal to 1. A postive y, indicates production
of component i; a negative ¥, indicates depletion of component i, Thus the rate of change
of any component i is expressed as a multiple of the rate of reaction of the Key com
ponent, the volume of the reacting mixture, and the number of moles of component i
changing per mole of Key component reacting.
‘To demonstrate father the application of this component mole balance, conser the
reaction previously given. Assume that for this reaction, the rate is experimentally
moles of B
determined w be rq = kes(Deals ees amy WHEE Bib the key component,
Therefore, Y= — 2, v= = 1 m= Sand y,
‘An unsteady-stato mole balance on component A is writen as
Rate of moles of _ Rate of moles of , dnt,
A into reactor” A out of reactor dt
where n,(0) is the moles of A accumulated in the reactor
‘An unsteady-state mole balance on component S is written as
Rate of moles of Rate of moles of Ange,
into reacior Sout of reactor * 3707 = gy
where ng(t) i8 the moles of $ accumulated in the reactor, 3-6 Chemical Reactors 111
‘The unsteady-state total mole balance is writen as
Rate of total Rate of total Rate of Rate_ of accumulation
moles into - moles out + production = of total moles
reactor of reactor of total moles in reactor
and
Rate of production of total moles = y,r,V, molestime
where v, = >! vj For the partielar reaction at hand, vy = Ue+ Up + Uy Uy 7
B41-2-1=1
Simi, the energy balance must also account for the energy given off or taken in
by the reaction, The energy balance equation is usually written as
fa of coy _ Ras of coy _ RH SF SMY Rae of acoulon
ino odor ~ ot of ecer MW = Sy i ear
‘A usual reference state for the enthalpies and intemal energy is the pure components
in the phase (liquid, gas, or solid) in which the reaction takes place, a temperature of
25°C, and the pressure of the system, Using this reference state, we can write
Rate of energy associated with reacion = VrAH,, energyime
where AH, is the enthalpy of reaction evaluated at 25°C in eneryyimole of key eom-
ponent,
36.2 Chemical Reactor Example
Consider the chemical reactor system shown in Fig, 3-6.1. The reactor is a vessel where
the “wellknown” reaction A — B occu, Let us assume that the reaction occurs at
constant volume and temperature. In addition, let's assume constant physical properties
and that the reactor is well mixed. The rate of reaction is given by the expression
rat) = ke5()
ce alt, MORE ogi,
fo.
Figure 3-6.1 Isothermal well-mixed chemical reactor,
© 112 Chapter 3 FirstOrder Dynamic Systems
where
te of reaction of component A, kmoles of AJ».
constant of reaction, m?/kmoles-£
concentration of component A in reactor, kmoles of A/m
nA),
k
eld)
The objective is to develop the mathematical model, find the wansfer functions, and
aw the block diagram, relating c4(t) and c44(t) to the inputs f(t and cu.
Our proveue calls for fist developing the mathematical model, Remember, in our
way of doing things, those input variables, f(t) and c(t) in this ease, are not considered
unknowns. The contol volume includes the valve and reaetor. For this process, an
unsteady-state mole balance on component A, of the type presented at the beginning
of this seetion, provides the first equation
Rate of moles of Rate of moles of Rate of change of Rate of accumulation of
component A into — component A out + component A in = moles of component A
control volume of control volume control. volume in control volume
of, in equation form,
Aieult — Lead + (avr) = vy LA ben)
Veg, 2 unk [ex(0) (0)
The rate-fseaction expression provides another equation:
rab = kei (3-62)
2 eq, 2 unk,
Baquations 3-6.1 and 3-6.2 constitute the mathematical model for this process. Writing
this model is the frst step in our procedure, The second step calls for linearizing the
nonlinear terms in the model
Linearizing the fist two terms of Fg, 346.1 and Eq, 3-62 around the initial steady-
state values off, Ex, and E, yields
Hiodealt) = fEq + Eulf ~F) + FleaAt) ~ Ea (3-63)
Hocalt) ~ Fe. + EA HO ~F) + Felt) — (3-64)
TAQ SF, + UE Mel) = Ee) (3-6.5)
Substituting Eqs. 3-63, 3-64, and 3-65 into Eq, 346.1 yields
fal * &
HO ~F) + Heal
= flea ~
eA
) Vy = WEE Meat, ~ Ee
aes)
(5-66) 3-6 Chemical Reactors 113
Equation 3-6,6 is the equation that describes the process around the linearization
values, We can now proceed to obiain the transfer functions, Writing a mole balance
at the initial steady state and subtracting it from Eq. 3-666 yield
Fc. - #evety = 0
CaF Ut + FCA - EAPO) 7 (66.7)
where FU) = OF Call= cults By, and CAD = c4lth~ 4
From Eq, 3-67,
nt |
CAs) ae Fs. 36.8)
whore
6-69)
and
Cys) _
Cy) eT Bot)
To obtain the relationships for cf). assuming ideal plug fw and no reaction oc-
curing in the outlet pipe, we can state
Caall) = ext = fa)
of, in terms of deviation variables,
Cudt = CAE) B60)
66.10) 114° Chapter 3. First-Order Dynamic Systems
Fo nwt
‘ wo)
PA cx cul
Paes
Figure 3-6.2 Block diagram for welled isother
‘mal chemical reactor
where
fc ~ dead time between the reactor outlet and point 1, seconds
1 distance between the reactor outlet and point 1, m
A, ~ cross-sectional area of pipe, me
The Laplace transform of Eg. 36.11 gives
Cus) = EAS, 66.13)
‘Thus, from Eqs. 3-69, 346.10, and 3-6.13, the final desired transfer fnetions are
Cuss) Ker
we ES (0.14)
and
CdS) = Kero
Cue eT b-b5)
Figure 3-62 shows two different ways to draw the block diagram for this reactor
3:7 EFFECTS OF PROCESS NONLINEARITIES
‘A most important characteristic of processes is their linear or nonlinear behavior. To
understand what these terms mean and appreciate their significance, consider the ther: 3-7 Effects of Process Nonlinearities 115
‘mal process presented in Section 3-2. In this particular process, because the flow! is
considered constant, the gains, K, and Kz, are constants over the complete operating
range, That is, their numerical values (given by Eqs. 3-222 and 3.223) do not ever
change, no matter what the process operating coniition, The value of the time constant
1. Fg, 3221, is also constant for this system, The fact thatthe parameters that describe
the characteristics of this process are constants means that the behavior of the process
is also constant. That is, the process will behave in the same manner, sensitivity and
speed of response, at any operating condition. Processes that exhibit this characteristic
ate called linear processes.
In Section 3-1, we noted thatthe controller mast be tuned, or adapted, to the process
to obtain adequate control performance, Because the behavior of a linear process is the
same over the complete operating range, if the contoller i optimally tuned at one
‘operating condition, it s also optimum at any other operating condition. This is certainly
an ideal operation and the one we could hope for
However, consider now the gas process presented in Seotion 3-5. In this process the
sais K,, K,, and K, 1 given by Bg, 35.16 depend on C, ,C, and Cand the rumerical
values ofthese terms depend on the values of p , p-., and Mm, around which the linea
zation of the nonlinear function f(t) was done, Therefor, the numerical values of K, ,
K,, and K, also depend on where the linearization was performed. The numerical value
of the time constant, as also given in Eq, 35.16, also depends on Cs. This means that
the values of the terms that describe the process charateristis, and thus the process
‘behavior islf, depend on the operating condition. The process behavior changes as
the operating conditions change! Processes that exhibit these characterstis are called
nonlinear processes. Nonlinearity is « characteristic of most chemical processes
To demonstrate graphically the effect of the process nonlinearities, two different
cases are shown. Inthe first cas, the pressure inthe tank was allowed to vary between
25 psia and 70 psia, while Keeping the process flow constant and maintaining other
process conditions at their steady state. The signal to the outlet valve was allowed to
‘vary to keep the process flow constant. This is the case when it is desired to run the
process at different pressures even though the process flow has not changed, Figure
3-711 shows how Ki, Ko, Kx, and 7 vary as the pressure in the tank varies, Figure
3-7. La shows that K, varies by a factor of 4, and Fig. -7.1b shows that K; varies by a
factor of 10, Similarly, Fig. 3-7.1d shows that r varies by 2 factor of 5. Kis not affected
so much, as showin in Fig. 3-7.1e
Another interesting case occurs when the signal to the fan varies, thus varying the
process flow through the tank, while the pressure is kept constant. This could happen
‘when a pressure control system is having to react to upsets, the upset being the signal
to the fan in this case. The signal to the outlet valve was allowed to change to match
‘the outlet gas flow to the inlet flow provided by the fan and to Keep the pressure in the
tank constant, Figure 3-7.2 shows how Ky, Ko, Ky and 7 vary as the signal to the fan
varies. All of these figures show the nonlinear characteristics of this simple process
The nonlinear behavior of processes is very detrimental to their contol. As the pro-
cess behavior changes with operating conditions, the controller should be re-tuned, or
re-adapted, to maintain optimum contol performance. Often, the best we can do is tune
the controller so that its performance is best atthe design operating point and acceptable
over the expected range of operating conditions; tuning methods are preseated in Chap-
ter 6, Techniques have been developed to permit the contoller to re-tune itself, auto 116 Chapter 3 FirseOrder Dynamic Systems
rar
Presuy ian,ala Presa tan, ai
ta ®
0.0
al 9
os| a
,,f2 06 rose 7
* cee 6
69 5
' ~. ab
os Dl 3
ox C1 eae
‘ @
1 Gains and time constant as a function of pressure in tank.
™ 40 60 ao
Sia to fan,
@
20
200
5 a5
fe 0
eae °,
oa a (rr
« @,
Figure 3-7.2 Gains and time constant as function of signal to fn, 38 Additional Comments 117
‘matically, as the process characteristics change. These techniques are refered to as sel
tuning, or adaptive tuning, and are presented in Chapter 15. Computer control systems
provide the necessary computing power for the realistic application of the technique.
Although it is not presented in Sections 3-5 and 3-6, the dead time also depends on
the operating conditions. Equation 3-3.1 shows that if the process flow varies, then the
dead time will also vary. Thus all the terms that describe the process behavior are
functions of the operating conditions,
Process nonlinearities are certainly not a desirable characteristic, but they are unfor-
tunately a realsic and very common one.
3-8 ADDITIONAL COMMENTS
11s now important to analyze what we have done ffom a more “general” point of view.
Tf we Took atthe form of the transfer functions that have been developed in the diffrent
examples (Fqs. 42.12, 32.25, 32.26, 45.17, 3518, 3519, 369, and 36,10), we
see that they ate all ofthe frm
Yo) _K
X@) 741 bau)
Laplace transform of the output variable
Laplace transform of the input variable
Im Section 2-4, this Eq. 3-8.1 was defined as the standard form of the transfer funtion
for a first-order system. The distinguishing characteristic of this form is that the second
term of the denominator is unity. This is the form of all first-order systems regardless
of whether they are thermal, uid, reacting, mechanical, or electrical systems. ‘This is
important because it says thatthe behavior of any system, no matter what type, described
by Eq, 3-811 is the same; they all respond the same way to forcing finetions. The
‘meaning of gain, K, and time constant, 1, is the same for all of them.
Sometimes dead time is present, and in this case the transfer function becomes
Xs) Ke
10. 8.2
Xo) “7 (62)
Equation 3-82 is more general than Eq. 3-8.
‘One of the most important terms in the study of aulomatic control isthe time constant,
-r We have developed several expressions for r Eqs. 32.9, 3-221, and 3.5.16, These
equations are all analogous, that is, they are all of the form
capacitance
‘conductance G63)
The capacitance is a measure of the abiliy of the proces to accumulate the quantity
conserved (mass of energy), The conducance is a measure ofthe abiliy of the process
to regulate itself 118 Chapter 3 First-Order Dynamic Systems
Process
Thermal
Ges
Reacting
For example, for the thermal system of Section 3.2, the time constant, Eq. 3.2.9, is
1
VoC. _ capacitance °C
$0C, ~ eondvconce' Te
°c
‘We can also write this expression for 7, assuming that C,
Cp, Which is a good as-
sumption for liquids, as
Tt
‘This expression clearly shows the accumulation, V, and flow, f, terms. Table 3-811
presents the analogous expressions for the processes shown in this chapter. A. process
that has not been presented here, but is given as an exercise in Problem 3-1, is that of
‘mixing, or blending, This process is similar to a reacting process in which no reaction
‘occurs. Its expressed by assigning the reaction constant, f, a value of zero in the time
‘constant in Section 3-6, In this case,
v vom
FF UCV ~ F ms
‘Another comment we wish to make concems the method used to obtain the desired
‘tansfer fimetions and block diagrams. As you have undoubtedly noticed, the procedure
first of all requires a good knowledge of process engineering. ‘The steps followed to
‘obtain the transfer fictions were outliaed in Section 3-5,
‘You must also have noted that most of the time, the developed equations that describe
the process are nonlinear. We have linearized them to be able to obtain the desired
transfer fimctions. These transfer functions deseribe the process in a region close to the
Table 3-8.1 Time Constant Analogy for Different Processes
Time
Vacble Constant Capacitance Conductance
VpC, Tis
Temperature 3.29)
r foc, 29) $6.36
Voc. Us
Temperature HP ig, 3.221) +0
iperatun ioc, + UA Eq, ) f0C, + UA.
v
También podría gustarte
