This is a reproduction of a library book that was digitized

by Google as part of an ongoing effort to preserve the

information in books and make it universally accessible.

https://books.google.com
 NASA AMES RESEARCH CENTER

NASA SP-194
3 1769 00065 3819

to be removed

from Library

LIQUID

PROPELLANT

ROCKET
S

COMBUSTION
IC
UT

ADMI
NA

SP
AC
RO

AND
NIS

E
AE

TRAT

INSTABILITY
ION
L A
ON I
AT N

U.S.A

NASA IRDARY
EXTE
AMES RESEA
MOEFETY FIELD , CALIF
-

COPY
NO.

NATIONAL AERONAUTICS AND SPACE ADMINISTRATION

 SC
AMES RESEARCH CAT 3
MOF FET T FIE LD , CAL IF

JAN 17 1973
 NASA SP-194

LIQUID PROPELLANT ROCKET

COMBUSTION INSTABILITY

Editor
DAVID T. HARRJE
Princeton University

Associate Editor
FREDERICK H. REARDON
Sacramento State College

CS E
TI AC
U SP
ADMINI

NA ND
RO A
AE
STRATI
AL

ON
ION
NAT

Scientific and Technical Information Office 1972

U.S.A
NATIONAL AERONAUTICS AND SPACE ADMINISTRATION

Washington, D.C.
For sale by the Superintendent of Documents,
U.S. Government Printing Office, Washington, D.C. 20402
Price$5.50
Library of Congress Catalog Card Number 70–170324
Stock number 3300-0450
 PREFACE

This reference book originated in the concern of a number of engineers engaged

in the solution of problems of combustion instability for more effective communi-
cation between the various workers in this field . In December 1962 an ad hoc
working group was formed by the JANNAF Interagency Propulsion Committee
(then the Interagency Chemical Rocket Propulsion Group, ICRPG) to study the
extent of combustion instability problems in liquid propellant rocket engines and
to make recommendations as to their solution . This ad hoc group urged that a
more permanent working group be established to promote an integrated research
and technology plan , which could provide stability design criteria , and to promote
a better exchange of technical information among scientists and engineers inter-
ested in combustion instability in liquid propellant rockets. The ICRPG formed
a Working Group on Liquid Propellant Combustion Instability in January 1964.
Beginning that year, annual conferences have been held by the Working Group.
These conferences , the proceedings of which are published promptly in the form of
expanded abstracts (with illustrations) , have proven to be extremely effective in
enhancing the exchange of up-to-date information .
It was recognized from the beginning, however, that much of the theoretical
and experimental combustion instability information was scattered in numerous
progress reports and technical papers in various journals and conference proceed-
ings . In its first year, the Working Group recommended the preparation of a book
that would help to train new workers in the field , as well as providing a reference
for others. In 1964 a reference book committee was appointed by the Working
Group to outline the content . At the 1965 Working Group meeting, the committee
presented its recommended reference book outline, and means of implementing
the writing and publication of the book were discussed . Further deliberations by
the committee during 1966 resulted in the recommendation that a prime contract
be given to a general editor, someone well acquainted with the combustion in-
stability field , who could subcontract the variety of subject matter to a number of
specialized authors . This recommendation was adopted by the Working Group ;
work was initiated in 1967 under a NASA contract with Princeton University,
with Richard J. Priem, of the NASA-Lewis Research Center, as contract monitor .
Funds for the contract were provided by NASA, the Air Force , the Army, and
the Navy. The excellence of the work done by the Reference Book Committee is
evidenced by the fact that their suggested outline has been rather closely followed.
It is hoped that this reference book will prove to be useful to all workers in the
liquid propellant combustion instability field , whether they are engaged in re-
search, design, or development . The philosophy followed in compiling this book is
that the prime importance is to provide the main outline of the most significant
developments, both theoretical and experimental, with emphasis on fundamental
principles and relationships between alternative approaches. For detailed informa-
tion, the reader is supplied with an extensive list of references, which should help
guard against rapid obsolescence of the reference book, a danger faced by any
text in a fast-developing field .
1

iii
iv LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

There are four main parts to the book : ( 1) background information , including
an introduction to the phenomenon of combustion instability and a discussion of
pertinent aspects of the combustion and flow processes that take place in a liquid
propellant rocket engine (Chs . 1 to 3) , (2) analytical models of both low and high
frequency instability , with the theoretical basis of each model given first and the
use of the models in design and development following in a separate chapter (Chs.
4 to 6) , (3) a practical guide for designers , including aspects of excitation and
damping, with experiential information integrated as much as possible with the
results of theoretical studies (Chs. 7 and 8) , and (4) experimental aspects of the
study of combustion instability, that is, techniques used to identify and investigate
oscillatory processes in both research and developmental hardware , and methods
of rating the stability of a given engine (Chs . 9 and 10) .
The reference book is designed to allow the reader to quickly look up information
on combustion instability and related topics . The detailed index provided by the
authors and editors as well as the extensive table of contents should greatly aid
the reader in this respect. The General Nomenclature, supplemented by specialized
nomenclature when required , should provide the required information to interpret
the equations accurately. Each equation , figure , and table is uniquely numbered
by section to avoid confusion .
We are greatly indebted to the many authors and reviewers (whose names are
listed elsewhere in the book) for the generally high quality of their manuscripts
and their cooperativeness during the editorial process . Special thanks go to Robert
J. Hefner and L. Paul Combs , who took responsibility for compiling Chapters 9
and 10, respectively, and to Owen W. Dykema , who edited Section 7.4.

The Editors

David T. Harrje
Frederick H. Reardon
 CONTENTS

CHAPTER PAGE

PREFACE. iii

LIST OF EDITORS, AUTHORS, AND REVIEWERS. xvii

1 INTRODUCTION ... 1
1.1 LIQUID ROCKET ENGINE SYSTEMS . 1
1.1.1 Conventional Engines .. 1
1.1.1.1 Pressure-fed engines . 2
1.1.1.2 Pump-fed engines . 3
1.1.2 Advanced Engines .. 5
1.1.2.1 Aerospike engine .. 5
1.1.2.2 Staged combustion engine . 6
1.1.3 Performance Parameters ... 7
1.1.3.1 External performance parameters ... 8
1.1.3.2 Internal processes in rocket thrust chambers . 8
1.1.3.3 Real rocket performance calculations .. 14
1.2 COMBUSTION INSTABILITY ……. 14
1.2.1 Physical Manifestations .. 15
1.2.1.1 Damage.. 15
1.2.1.2 Effect on combustion efficiency . 16
1.2.2 Classification ... 16
1.2.2.1 Low frequency, chug.. 17
1.2.2.2 High frequency instability . 17
1.2.2.3 Intermediate frequency , buzz . 19
1.2.3 Initiation of Combustion Instability . 20

55888
1.2.3.1 Spontaneously initiated linear instability . 21

228
1.2.3.2 Induced or nonlinear combustion instability . 22
1.2.4 Dynamic Stability . ... . . 23
1.2.4.1 Dynamic versus statistical stability . 24
1.2.4.2 Dynamic stability in engine development programs . 26
1.2.4.3 Demonstrating dynamic stability in engine development programs . 27
1.3 HISTORICAL SURVEY .. 30
1.4 CURRENT STATUS ... 34

2 STEADY-STATE PROCESSES . 37
2.1 GENERAL DESCRIPTION OF COMBUSTION AND FLOW
PROCESSES . 37
2.1.1 Overall Description... 37
2.1.2 Conversion Time and Residence Time ... 38
2.1.3 Characteristic Length and Characteristic Velocity . 39
2.1.4 Gas-Phase Processes ... 39
2.1.5 Condensed-Phase and Gasification Processes . 40
2.1.6 Spray Combustion ....
2.1.7 Experimental Observations .. 42
2.1.8 Elaboration on Description of Spray-Combustion Models .. 42
2.2 INJECTION AND ATOMIZATION . 45
2.2.1 Manifold Flow .... 45
2.2.2 Jet Properties . . 46
2.2.3 Mechanisms of Atomization . 49
2.2.3.1 Liquid surface instability . 49
2.2.3.2 Liquid jet breakup, low velocity . 50
2.2.3.3 Liquid jet breakup, high velocity . 51
2.2.3.4 Summary of jet breakup results . 52

V
vi LIQUI PROP ROCK COMB INSTA
D ELLA ET USTI BILIT
NT ON Y
CHAPTER PAGE

2.2.3.5 Surface breakup .. 52

2.2.3.6 Liquid sheet breakup . 54
2.2.3.7 Secondary drop breakup . 55
2.2.4 Spray Description . ... . 55
2.3 SPATIAL DISTRIBUTION OF PROPELLANTS . 59
2.3.1 Mass Flux Distribution . 59
2.3.2 Mixture Ratio Distribution . 60
2.3.3 Mixing Processes... 63
2.3.3.1 Liquid phase mixing . 65
2.3.3.2 Droplet transport . 69
2.3.3.3 Vapor mixing . 70
2.3.4 Recirculation ... 72
2.4 LIQUID DROPLET VAPORIZATION AND COMBUSTION . 74
2.4.1 Droplet Heat-Up and Vaporization .. 77
2.4.2 Bipropellant Droplet Combustion .. 83
2.4.2.1 Envelope flame model for subcritical pressures-theory and experiment ... 84
2.4.2.2 Envelope flame model for supercritical pressures-theory and experiment . 91
2.4.3 Monopropellant Droplet Combustion ... 94
2.4.3.1 Monopropellant droplet decomposition in an atmosphere comprised
solely of inert gases or decomposition products-theory and experiment . 95
2.4.3.2 Monopropellant droplet fuel decomposition in an oxidizing atmosphere-
theory and experiment . 99
2.5 FLOW PROCESSES . 100
2.5.1 Core Flow .... 100
2.5.1.1 Effects of injector design . 100
2.5.1.2 Mechanical turbulence generation . 102
2.5.2 Boundary Flow ..... 102
2.5.2.1 Film or boundary coolant . 102
2.5.2.2 Wall effects .. 103
2.5.2.3 Off-design operation .. 103
2.5.3 Energy Release Distribution . 103
2.5.3.1 Element design ……….. 103
2.5.3.2 Combustion volume and length effects . 104

3 DYNAMICS OF COMBUSTION AND FLOW PROCESSES . 105

3.1 INTRODUCTION .... 105
3.2 FLOW IN PROPELLANT FEED SYSTEMS . 106
3.2.1 Feed System Acoustics .. 107
3.2.2 Component Dynamics .. 107
3.2.2.1 Lumped-parameter approach ... 107
3.2.2.2 Continuous-parameter approach . 110
3.2.2.3 Modal techniques.. 111
3.2.3 System Response... 112
3.2.4 Comparison of Analysis and Experiment . 114
3.3 INJECTION PROCESSES ... 115
3.3.1 Effects of Upstream Conditions . 117
3.3.1.1 Flow rate oscillations .. 117
3.3.1.2 Hydraulic flip .. 119
3.3.1.3 Injector vibration . 119
3.3.2 Effects of Small Perturbations in Chamber Conditions . 120
3.3.2.1 Atomization and jet breakup ..... 120
3.3.2.2 Mixing .. 120
3.3.3 Stream and Droplet Breakup by Shock Waves .. 126
3.4 COMBUSTION PROCESSES .. 128
3.4.1 Controlling Processes . . 128
3.4.2 Linear, Nonsteady Drop and Spray Burning. 129
3.4.2.1 The frequency spectrum.. 129
3.4.2.2 Response functions .. 130
 CONTENTS vii

CHAPTER PAGE

3.4.3 Nonlinear Drop and Spray Burning .. 136

3.4.3.1 Vaporization . . . .. 136
3.4.3.2 Effects of shock waves . 137
3.4.4 Special Effects ... 138
3.4.4.1 Monopropellant fuel . 138
3.4.4.2 Supercritical chamber pressure . 138
3.4.4.3 Liquid films on surfaces ... 138
3.5 WAVE PROPAGATION IN COMBUSTION CHAMBERS . 138
3.5.1 Linear Wave Motion . 139
3.5.1.1 General considerations .. 139
3.5.1.2 Application to specific combustor geometries . 141
3.5.2 Nonlinear Wave Motion . 147
3.5.2.1 Introduction .... 147
3.5.2.2 Longitudinal modes . 147
3.5.2.3 Transverse modes .. 150
3.5.2.4 Nonlinear transverse gas displacement . 151
3.5.3 Damping Effects .. 154
3.5.3.1 Liquid and solid particle drag . 154
3.5.3.2 Acoustic liners and nonrigid walls .. 155
3.5.3.3 Injector-face baffles .. 156
3.6 UNSTEADY FLOW IN EXHAUST NOZZLES . 159
3.6.1 Linear Nozzle Admittance Equation ... 160
3.6.2 Calculation of Admittance Coefficients . 162
3.6.3 Experimental Verification ... 165
3.6.4 Nonlinear Effects .. 165
3.6.5 Other Effects .. 166

4 ANALYTICAL MODELS OF HIGH FREQUENCY

COMBUSTION INSTABILITY. 167
4.1 INTRODUCTION ... 167
4.1.1 Scope of Current Analytical Models . 167
4.1.2 General Conservation Equations .. 168
4.2 THE SENSITIVE TIME LAG THEORY . 170
4.2.1 Basic Concepts . . . . 170
4.2.1.1 The sensitive time lag. 170
4.2.1.2 Theoretical approach ... 172
4.2.2 Linear Theory . ... 175
4.2.2.1 Governing equations . 176
4.2.2.2 Method of solution .. 176
4.2.2.3 Longitudinal mode solution . 178
4.2.2.4 Transverse mode solution.. 182
4.2.3 Nonlinear Theory . ... 187
4.2.3.1 Nonlinear combustion response . 190
4.2.3.2 Nonlinear wave motion ... 191
4.3 NUMERICAL INTEGRATION METHODS .. 194
4.3.1 Basic Concepts . . . .. 195
4.3.1.1 Approach and assumptions . 195
4.3.1.2 Governing equations .. 197
4.3.1.3 Burning rate models . 197
4.3.2 One-Dimensional Analysis .. 200
4.3.2.1 Simplification of equations . 200
4.3.2.2 Method of solution . 201
4.3.2.3 Typical results .. 202
4.3.3 Two-Dimensional Analysis .. 205
4.3.3.1 Circumferential surface (0 - x) model . 205
4.3.3.2 Transverse plane (r − 0) model ...... 207
4.3.3.3 Comparison with one-dimensional model .. 207
viii LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

CHAPTER PAGE

4.4 SIMILITUDES AND OTHER MODELS . 208

4.4.1 Response Factor Approach .. 208
4.4.1.1 Basic principles .. 208
4.4.1.2 Heidmann-Feiler analysis . 209
4.4.1.3 Dykema analysis . 214
4.4.1.4 Modal energy analysis .. 217
4.4.2 Similarity Approach .... 221
4.4.2.1 Similarity techniques .. 221
4.4.2.2 Analytical similitudes . 221
4.4.2.3 Empirical similitudes .. 222
4.5 COMPARISON OF ANALYTICAL MODELS . 226
4.5.1 Linear Models ..... 226
4.5.1.1 Characteristics of linear stability models .. 227
4.5.1.2 Comparison of linear calculations . . 228
4.5.2 Nonlinear Models ... 229
4.5.3 Concluding Remarks .. 231

5 ANALYTICAL MODELS OF LOW AND INTERMEDIATE

FREQUENCY INSTABILITY. 233
5.1 INTRODUCTION .. 233
5.2 CONSTANT COMBUSTION TIME LAG MODELS . 234
5.2.1 General Approach.... 234
5.2.2 Single Time Lag Model . 236
5.2.3 Double Time Lag Model . 236
5.3 VARIABLE COMBUSTION TIME LAG MODELS . 237
5.3.1 Low Frequency Instability . . .. 238
5.3.2 Intermediate Frequency Instability . 240
5.4 FEED SYSTEM RESPONSE . 241
5.4.1 Calculation of Injection Admittance . 242
5.4.1.1 Constant-area feed line .. 242
5.4.1.2 Line with area change .. 243
5.4.1.3 Orifice or valve .. 243
5.4.1.4 Pump ... 243
5.4.1.5 Manifold .. 243
5.4.2 Injection Admittance for Simple Feed Systems . 244
5.4.2.1 Pressurized-tank feed system.... 244
5.4.2.2 Constant-rate feed system . 246
5.4.2.3 Centrifugal-pump feed system.. 247
5.4.3 Stabilizing Effect of the Feed System . 249
5.5 ANALYTICAL METHODS OF SOLUTION . 249
5.5.1 Nyquist and Satche Methods .... 251
5.5.1.1 Example : single time lag model .. 253
5.5.1.2 Application to other engine systems .. 253
5.5.2 Stability Limit Approach .. 254
5.5.3 Other Analysis Methods . 257
5.6 ANALOG METHODS OF SOLUTION . 258
5.6.1 Mechanization of the Engine Model . 259
5.6.2 Representation of a Time Lag.. 260
5.6.3 Obtaining the Solution.. 260

6 USE OF ANALYTICAL MODELS IN DESIGN AND

DEVELOPMENT .... 265
6.1 INTRODUCTION ... 265
6.2 LOW AND INTERMEDIATE FREQUENCY MODELS . 265
6.2.1 Approach ... 266
6.2.2 Combustion Time Lag Modification . 267
6.2.3 Feed System Changes .. 270
 CONTENTS ix

CHAPTER PAGE

6.2.3.1 Injector impedance .. 270

6.2.3.2 Manifold capacitance . 271
6.2.3.3 Resonators .. 273
6.2.3.4 Feed line losses ... 273
6.2.4 Combustion Chamber Response.. 275
6.3 HIGH FREQUENCY : SENSITIVE TIME LAG MODEL . 277
6.3.1 General Approach... 277
6.3.2 Calculations Required . 278
6.3.3 Empirical Correlations of Combustion Response.. 280
6.3.4 Stability Prediction.... 284
6.4 HIGH FREQUENCY : NUMERICAL INTEGRATION METHODS .. 286
6.4.1 General Approach…….. 287
6.4.2 Calculations Required .. 288
6.4.2.1 Steady-state calculations . 288
6.4.2.2 Stability analysis .. 289
6.4.3 Correlation with Test Data.. 291
6.5 HIGH FREQUENCY : RESPONSE FACTOR ANALYSIS . 293
6.5.1 Heidmann-Feiler Approach .. 293
6.5.1.1 Process selection .... 293
6.5.1.2 Hydrogen-oxygen combustors . 294
6.5.2 Dykema Analysis . . . 297
6.5.3 Design Application of Modal Energy Analysis . 298
6.5.3.1 Design procedure . 298
6.5.3.2 Special considerations . 301
6.6 HIGH FREQUENCY : SIMILARITY RULES . 302
6.6.1 Stability Prediction Equations in Decision Making . 302
6.6.2 The Role of Pulsing in Stability Determination . 305
6.6.3 Stability and Efficiency . . 305
6.7 USE OF COMBINATIONS OF MODELS .. 306

7 DESIGN FACTORS AFFECTING EXCITATION... 309

7.1 INTRODUCTION ... 309
7.2 OVERALL COMBUSTION DESIGN CONDITIONS . 310
7.2.1 Chamber Pressure .... 310
7.2.1.1 General statement of the problem.. 310
7.2.1.2 Low frequency instability . 310
7.2.1.3 High frequency instability . 312
7.2.2 Contraction Ratio .. 314
7.2.3 Injection Density . 317
7.2.4 Axial Energy Release. 318
7.2.5 Transverse Energy Release Distribution . 322
7.2.6 Boundary Effects . 325
7.2.6.1 The boundary region ... 325
7.2.6.2 Boundary control factors . 329
7.3 EFFECT OF PROPELLANT COMBINATION ON STABILITY . 330
7.3.1 Cryogenic Propellants ... 331
7.3.1.1 Oxygen/hydrogen (LOX/LH₂) . 332
7.3.1.2 FLOX/light hydrocarbons .. 332
7.3.2 Storable Propellants . 333
7.3.2.1 Nitrogen tetroxide/50 % hydrazine-50% UDMH (N₂O./A-50) . 333
7.3.2.2 Other storable propellants .... 334
7.3.3 Cryogenic-Storable Combinations . 335
7.4 INJECTOR PATTERN .. 336
7.4.1 General Considerations .. 337
7.4.2 Unlike Impinging Jets . 339
7.4.3 Like-Impinging Jets .. 349
7.4.4 Coaxial Jets .. 357
X LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

CHAPTER PAGE
7.4.5 Other Injector Element Types .. 366
7.4.6 Summary of Conclusions .. 371
7.5 FEED SYSTEM COUPLING . 373
7.5.1 Injector Impedance.... 373
7.5.2 Coupled Resonances . 376
7.5.2.1 Lumped parameters . 376
7.5.2.2 Distributed parameters .. 377
7.5.3 Imposed Oscillations .. 378
7.5.3.1 Structural vibrations . 378
7.5.3.2 Pump blade wakes .. 378
7.6 POPPING AND SPIKING . 379

8 DESIGN FACTORS AFFECTING DAMPING. 385

8.1 INTRODUCTION ... 385
8.2 INJECTOR FACE BAFFLES . 386
8.2.1 Available Theory . 386
8.2.2 Blade Arrangement .. 389
8.2.2.1 Number of blades . 390
8.2.2.2 Symmetry . 394
8.2.3 Blade Design . 395
8.2.3.1 Baffle length . 395
8.2.3.2 Blade shape . 397
8.2.3.3 Blade cooling . 397
8.3 ACOUSTIC LINERS . 399
8.3.1 Liner Damping Theory . 400
8.3.1.1 Nonlinear analysis . 402
8.3.1.2 Flow effects ... 405
8.3.1.3 Parameters to evaluate liner designs .. 407
8.3.2 The Effects of the Environment on Resonator Behavior. 410
8.3.3 The Sizing of Resonators ..... 415
8.3.4 Number and Placement of Resonators . 417
8.3.5 Thermal Design Considerations .. 420
8.3.6 Further Acoustic Absorber Applications . 426
8.4 THRUST CHAMBER SHAPE .. 429
8.4.1 General Considerations . 429
8.4.2 Combustion Chamber. 430
8.4.3 Nozzle Convergent Section . 432
8.4.4 Injector Shape .... 434
8.4.5 Annular Combustion Chamber . 435
8.5 OTHER DAMPING EFFECTS . 438
8.5.1 Chamber Wall Materials . ... 438
8.5.2 Acoustic Damping by Condensed Phases . 441
8.5.2.1 Theory of particulate acoustic damping . 441
8.5.2.2 Role of particle size distribution. 445
8.5.2.3 Particulate damping in rocket stability analyses .. 445
8.5.2.4 Experimental studies of particulate damping in rocket motors . 446
8.5.3 Corner Effects ... 446
8.5.3.1 Stability considerations . 447
8.5.3.2 Radial winds ... 447

9 EXPERIMENTAL EVALUATION OF STABILITY BEHAVIOR.. 451

9.1 INTRODUCTION ... 451
9.2 SPECIALIZED RESEARCH COMBUSTORS . 451
9.2.1 Full-Scale Simulators . 452
9.2.2 Sub-Scale Simulators . 452
9.2.2.1 Pulse motor ... 453
9.2.2.2 Annular combustion chamber . 454
9.2.2.3 Wedge motor .. 454
9.2.2.4 Square motor . 454
 CONTENTS xi

CHAPTER PAGE

9.2.2.5 Two-dimensional motor ... 454

9.2.3 Basic Combustion Process Apparatus . 457
9.3 PRESSURE MEASUREMENTS . 459
9.3.1 Transducer Requirements ... 459
9.3.1.1 Hardware considerations . 459
9.3.1.2 Response characteristics . 460
9.3.1.3 Environmental exposures . 460
9.3.1.4 Range selection ... 460
9.3.2 Available Instruments .. 461
9.3.2.1 Transducers for combustion zone pressure measurements . 461
9.3.2.2 Pressure transducers for propellant systems .. 464
9.3.3 Location and Mounting .. 467
9.3.3.1 Transducer locations . 467
9.3.3.2 Mounting considerations .. 468
9.3.3.3 Special mounting techniques . 474
9.3.4 Signal Conditioning and Recording . 475
9.3.4.1 Signal conditioning .. 475
9.3.4.2 Recording of dynamic data . 477
9.3.5 Display Techniques . . . 478
9.3.5.1 Analog playback.. 478
9.3.5.2 Analog spectrum analysis . 480
9.3.5.3 Hybrid spectrum analysis . 481
9.3.5.4 Digital analysis ..... 483
9.4 OPTICAL MEASUREMENTS .. 484
9.4.1 Cinematography . . .. 485
9.4.1.1 Combustion zone photography 485
9.4.1.2 Exhaust plume photography . 486
9.4.1.3 Photographic techniques ... 487
9.4.2 Streak Photography .. 489
9.4.2.1 General principles . 489
9.4.2.2 Steady-state combustion distribution . 490
9.4.2.3 Application to combustion instability . 491
9.4.3 Electro-Optical Techniques .. 492
9.4.3.1 Fiber optics .. 493
9.4.3.2 AC radiometry . 495
9.4.3.3 Shock cone radiation .. 496
9.4.4 Shadow and Schlieren Methods . 497
9.4.4.1 Schlieren applications ... 497
9.4.4.2 Shadowgraph techniques .. 498
9.4.4.3 Silhouette photography ... 503
9.4.5 Holography of Liquid Rocket Engine Combustion . 503
9.4.5.1 Holography reviewed ... 504
9.4.5.2 Application of holography to rocket combustion .. 505
9.5 ACCELEROMETERS AND ACCELERATION DATA . 510
9.5.1 Selection of Appropriate Instrument . 510
9.5.2 Instrument Mounting .. 511
9.5.3 Acquisition of Data . 512
9.5.4 Typical Applications . 512
9.5.4.1 Detection and characterization of combustion instability .. 512
9.5.4.2 Determination of hardware resonance effects ..... 513
9.5.4.3 Estimation of hardware distortion during instability . 513
9.5.4.4 Detection of short-duration combustion perturbations . 514
9.6 OTHER MEASUREMENTS ... 514
9.6.1 Propellant Flow Rate ..... 515
9.6.2 Injection Characteristics . 516
9.6.3 Thrust and Other Performance Data .. 518
9.6.3.1 Measurable quantities related to performance and stability . 518
9.6.3.2 Effects of instability on measured quantities .. 519
xii LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

CHAPTER PAGE

9.6.3.3 Post shutdown observations ... 520

9.6.4 Sampling of Combustion Gases . 520
9.6.5 Thermal Measurements . 521
9.6.6 Acoustic Modeling . . . 523
9.6.6.1 Introduction and scope . 523
9.6.6.2 Acoustic damping coefficient . 523
9.6.6.3 Driving techniques ... 525
9.6.6.4 Applications . . .. 525
9.7 INTERPRETATION OF TEST DATA . 526
9.7.1 Resonant Combustion .. 526
9.7.1.1 Analog data reduction techniques .. 526
9.7.1.2 Resonant instability mode identification . 528
9.7.2 Low Frequency Instability .. 535
9.7.3 Combustion Disturbances . 536

10 STABILITY RATING... 539

10.1 INTRODUCTION .. 539
10.1.1 Purposes of Stability Rating Tests . 539
10.1.2 Experimental Approaches ... 540
10.1.2.1 Spontaneous instability methods . 540
10.1.2.2 Artificial initiation methods .. 541
10.2 EXPLOSIVE BOMBS . 542
10.2.1 Typical Designs .... 543
10.2.2 Effects of Design on Blast Amplitude . 546
10.2.2.1 Effect of explosive .. 546
10.2.2.2 Case design effects .. 547
10.2.2.3 Ambient medium effects .. 548
10.2.2.4 An empirical scaled blast correlation .. 548
10.2.2.5 Bomb case erosion and explosive heating . 550
10.2.3 Control of Shrapnel Damage. 552
10.2.4 Application to Combustors .. 555
10.2.4.1 Possibility of overbombing . 556
10.3 PULSE GUNS .. 557
10.3.1 Typical Designs . 557
10.3.1.1 Pulse guns with gun powder charges .. 558
10.3.1.2 Pulse guns with high explosive charges . 559
10.3.2 Effects of Design on Pulse Amplitude .. 559
10.3.2.1 Breech pressurization .... 559
10.3.2.2 Shock propagation in the gun barrel . 561
10.3.2.3 Shock expansion into the combustion chamber . 563
10.3.3 Effects of Combustor Operating Conditions ... 565
10.3.4 Application to Combustors .. 570
10.4 DIRECTED GAS FLOWS . 572
10.4.1 Typical Designs .... 572
10.4.2 Effects of Design Variables . 574
10.4.3 Applications . . .. 575
10.5 FEED SYSTEM PERTURBATION . 577
10.5.1 Siren (Continuous Oscillations) . 578
10.5.2 Pulser (Single Pulse Generator) . 580
10.6 OTHER RATING TECHNIQUES . 583
10.6.1 Liquid Hydrogen Temperature Ramping . 583
10.6.2 Variable Frequency Testing .. 586
10.6.3 Combustion Alterations .. 587
10.6.4 Pressure Level Changes . 588
10.7 COMPARISON OF RATING TECHNIQUES .. 588
10.7.1 Correlations Between Techniques ... 588
10.7.2 Limitations of Available Techniques . 589
10.7.2.1 Disturbance profile effects . . . 589
 CONTENTS xiii

CHAPTER PAGE

10.7.2.2 Access ports through chamber walls .. 589

10.7.2.3 Shrapnel damage . 591
10.7.2.4 Multiple pulses ... 591
10.7.2.5 Thermal initiation of detonation . 591
10.7.2.6 Acoustic interaction .... 591
10.7.2.7 External engine access . 591
10.7.2.8 Handling characteristics .. 592
10.7.3 Criteria for Selection of a Rating Technique .. 592
10.7.3.1 Program considerations ... 593
10.7.3.2 Engine design and operational considerations . 593

GENERAL NOMENCLATURE . 597

REFERENCES. 599

INDEX.. 629
 EDITORS , AUTHORS , AND REVIEWERS

Abbe, Charles J. Contributor, Sects. 6.4.3, 7.4.2,

Captain , USAF and 7.4.3
Formerly: Reviewer, Chapter 4
Senior Project Engineer
Air Force Rocket Propulsion Lab.

Agosta , Vito D. Author, Sects. 3.5.2.1 to 3.5.2.3

Professor of Aerospace Engineering
Aerospace Engineering and Applied Mechanics
Polytechnic Institute of Brooklyn

Bastress, E. Karl Author, Sects. 4.4.2 and 6.6

Director, Applied Sciences Group
.
Northern Research and Engineering Corp.

Bickford , Landis L. Author, Sects . 5.5.3, 6.2.3.4 , 7.5.2,

Supervisor, Control Dynamics and 7.5.3
Aerojet Liquid Rocket Company

Bloomer, Harry E. Author, Sect. 10.3

Aerospace Engineer
V/STOL and Noise Division
NASA-Lewis Research Center

Bracco , Frediano V. Author, Sect. 7.2.4

Member of the Research Staff Reviewer, Chapter 4
Dept. of Aerospace and Mechanical Sciences
Princeton University

Breen, Ben P. Contributor, Sect . 6.4.3

Vice President
KVB Engineering

Burge, Harland L. Contributor, Sect . 7.4.5

Manager, Applied Technology Dept.
TRW Systems

Campbell, David T. Contributor, Sects . 4.3.1 and 6.4.1

Manager, Propulsion Technology to 6.4.3
Advanced Programs Reviewer, Chapter 2
Rocketdyne

Campbell, John , Jr. Contributor, Sect . 7.4.4

Manager, Production Thrust Chamber Unit
Engineering Dept.
Rocketdyne

XV
xvi LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

Carpenter, Thomas W. Reviewer, Chapter 2

Assistant Professor
Aeronautical Dept.
California State Polytechnic College

Chew, Thomas J. C. Reviewer, Chapter 10

Senior Project Engineer
Air Force Rocket Propulsion Laboratory

Clayton, Richard M. Author, Sect. 7.2.6

Member of the Technical Staff Contributor, Sect . 7.6
Liquid Propulsion Section Reviewer, Chapter 10
Jet Propulsion Laboratory

Clinger, Eugene C. Contributor, Sect. 7.4.3

Engineering Manager
Liquid Rocket Division
Rocketdyne

Combs, L. Paul Editor, Chapter 10

Member of the Technical Staff Author, Sects. 10.1 to 10.4 and
Advanced Programs 10.7
Rocketdyne Reviewer, Chapter 9

Coultas, Thomas A. Author, Sects. 1.2.1 to 1.2.3 , 7.2.3,

Program Manager 8.3.6, 8.5.3, 9.2.2 , 9.4.1 , 9.4.2,
Combustion and Materials Technology and 9.4.3.2
Advanced Programs
Rocketdyne

Crocco, Luigi Author, Sects. 4.2 and 5.3.2

Robert H. Goddard Professor of Aerospace Propulsion
Aerospace and Mechanical Sciences Dept.
Princeton University

Culick, Fred E. C. Author, Sect. 4.5

Associate Professor of Engineering
Mechanical Engineering Dept.
California Institute of Technology

Dobbins, Richard A. Author, Sect. 8.5.2

Professor of Engineering
Division of Engineering
Brown University

Dykema , Owen W. Editor, Section 7.4

Member of the Technical Staff Author, Sects. 1.2.4, 4.4.1.3 , 6.5.2,
Applied Mechanics Division 7.4, 7.4.1 , and 7.4.6
The Aerospace Corporation Reviewer, Chapter 3

Erbs, Joseph E. Reviewer, Chapter 7

Project Engineer, Lance
Liquid Rocket Division
Rocketdyne
 EDITORS, AUTHORS , AND REVIEWERS xvii

Faeth, Gerard M. Reviewer, Chapter 2

Associate Professor
Mechanical Engineering Dept.
Pennsylvania State University

Fairchild, David A. Author, Sect. 10.5

Engineering Manager, Mechanical Design
Aerojet Liquid Rocket Co.

Feiler, Charles E. Author, Sects . 4.4.1.2 and 6.5.1

Head, Acoustics Section
V /STOL and Noise Division
NASA-Lewis Research Center

Fenwick, James R. Author, Sects . 3.2, 5.4.3, and 7.5.1

Member of the Technical Staff Contributor, Sect . 3.3.1
Liquid Rocket Division
Rocketdyne

Ford, Wayne M. Author, Sects . 9.4.1 and 9.6.4

Member of the Technical Staff
Advanced Programs
Rocketdyne

Garrison, Gary D. Author, Sects . 8.3.4 and 8.3.5

Assistant Project Engineer
Florida Research and Development Center
Pratt and Whitney Aircraft Co.

Goelz, Ralph R. Contributor, Sect . 8.5.1

Aerospace Engineer
Chemical Rocket Division
NASA-Lewis Research Center

Groeneweg, John F. Author, Sect . 2.2.4

Aerospace Engineer
V/STOL and Noise Division
NASA-Lewis Research Center

Hammer, Sandford S. Reviewer, Chapter 4

Associate Professor
Engineering Sciences Dept.
Hofstra University

Harris , George H. Author, Sects. 4.4.2 and 6.6

Operations Research Analyst
Management Sciences Division
Arthur D. Little , Inc.

Harrje, David T. Editor, Reference Book

Senior Research Engineer and Lecturer Editor, Chapters 1 , 2, 7 , and 8
Dept. of Aerospace and Mechanical Sciences Author, Sects. 1.3 , 1.4 , 3.3.1 , 3.3.2,
Princeton University 7.1 , 7.2.4, 8.5.1 , 10.3 , and 10.6.2
to 10.6.4
Contributor, Sects. 7.4.2 and 7.4.3
xviii LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

Hefner, Robert J. Editor, Chapter 9

Director of Engineering Consumer Products Group Author, Sects. 9.1 , 9.2.1 , 9.2.3 ,
Bell and Howell Company 9.3.3 , 9.3.5 , 9.4.1 , 9.6.1 , 9.6.2,
Formerly: 9.7.1 , 10.2 , 10.3 , and 10.7
Manager, Combustion Dynamics Dept. Contributor, Sect . 8.5.1
Liquid Rocket Operations Reviewer, Chapter 10
Aerojet-General Corp.

Heidmann, Marcus F. Author, Sects. 4.4.1.2, 6.5.1 , and

Research Scientist 10.4
Chemical Rocket Division Reviewer, Chapter 7
NASA-Lewis Research Center

Hewitt, Ross A. Reviewer, Chapter 3

Engineering Specialist
Combustion Dynamics Dept.
Aerojet Liquid Rocket Company

Howells , Edgar Author, Sect . 9.5

Senior Engineer
Research and Development
McGraw-Edison Power Systems Division
Formerly :
Member of the Combustion Devices Group
Bell Aerospace Company

Kesselring, Robert C. Author, Sects. 9.2.2, 9.4.2, 9.4.4.1 ,

Member of the Technical Staff and 9.4.4.3
Advanced Programs
Rocketdyne

Kosvic , Thomas C. Contributor, Sect . 4.3.2

Test Engineer
KVB Engineering

Lazar, Robert S. Reviewer, Chapter 1

Project Engineer
Naval Underwater Systems Center

Leeper, Charles K. Author, Sects . 4.4.1.4 and 6.5.3

Assistant General Manager
Research and Engineering
Aerojet Nuclear Company

Levine, Robert S. Reviewer, Chapter 1

Staff Scientist
Space Technology Division
NASA- Langley Research Center

Lewis, J. Dudley Author, Sects. 2.3.3 and 2.3.4

Superintendent , Liquid Engines Division
Rocket Propulsion Establishment
Great Britain
 EDITORS, AUTHORS , AND REVIEWERS xix

Lovingham , Joseph J. Contributor, Sect . 7.4.5

Director of Engineering
The McIntire Company
Formerly :
Chief, Systems Engineering
Reaction Motors Division
Thiokol Chemical Corporation

Lytle , Archie D. Author, Sect. 5.6

Principal Engineer
Bell Aerospace Company

Masters, Arthur I. Contributor, Sect. 7.4.4

Senior Assistant Project Engineer
Florida Research and Development Center
Pratt & Whitney Aircraft Company

Matthews, Birch J. Author, Sect. 9.4.5

Member of the Technical Staff
TRW Systems

McBride, James M. Author, Sects . 8.2.2 and 8.2.3

Supervisor Combustion Dynamics Contributor, Sect. 7.4.2
Aerojet Liquid Rocket Company

Miller, Irwin Author, Sects. 4.4.2 and 6.6

Senior Staff Manager
Management Sciences Division
A. D. Little Co.

Miller, Joseph Reviewer, Chapter 5

Manager, Propulsion Systems Engineering Dept.
TRW Systems

Mitchell, Charles E. Reviewer, Chapter 7

Assistant Professor
Dept. of Mechanical Engineering
Colorado State University

Monteil, Vernon H. Author, Sect. 1.2.4

Manager, Booster Propulsion
Applied Mechanics Division
The Aerospace Corporation

Morgan, C. Joe Reviewer, Chapter 8

Aerospace Engineer
Chemical Rocket Division
NASA-Lewis Research Center

Nestlerode , James A. Author, Sects . 3.2, 5.4.3 , and 7.5.1

Member of the Technical Staff Contributor, Sect. 3.3.1
Liquid Rocket Division Reviewer, Chapter 6
Rocketdyne
XX LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

Nicholls, J. A. Author, Sects. 3.3.3, 3.4.3.2, and

Professor 3.4.4.3
Dept. of Aerospace Engineering
The University of Michigan

Oberg, Carl L. Author, Sects . 8.3.6 and 8.5.3

Manager, Combustion
Advanced Programs
Rocketdyne

Osborn, John R. Author, Sect. 7.2.5

Professor of Mechanical Engineering Reviewer , Chapter 9
Purdue University

Phillips , Bert R. Author, Sects. 8.3.2 and 8.3.3

Aerospace Engineer
Chemical Rocket Division
NASA-Lewis Research Center

Powell, Walter B. Author, Sect. 1.1.3

Member of the Technical Staff
Liquid Propulsion Section
Jet Propulsion Laboratory

Priem, Richard J. Contract Monitor on Reference

Head, Rocket Combustion Section Book
Chemical Rocket Division Author, Sects. 4.3 , 6.4, and 6.7
NASA-Lewis Research Center

Proffitt, Robert L. Author, Sect. 9.4.3.2

Principal Scientist , Mechanics and Optics
Research Division
Rocketdyne

Ranger, Arthur A. Reviewer, Chapter 3

Assistant Professor
School of Aeronautics , Astronautics , and Engineering
Sciences
Purdue University

Reardon , Frederick H. Associate Editor, Reference Book

Associate Professor of Mechanical Engineering Editor, Chapters 3 , 4, 5, 6
Sacramento State College Author, Sects. 3.3.1 , 3.3.2, 3.4.4.1 ,
3.4.4.2 , 3.5.3.3 , 5.1 , 5.2 , 5.3.1 ,
5.4.1 , 5.4.2 , 5.5.1 , 5.5.2 , 6.1 , 6.3,
6.7, 7.3 , and 8.2.1

Rice , Edward J. Author, Sect. 2.2.3

Aerospace Engineer
V/STOL and Noise Division
NASA-Lewis Research Center
 EDITORS, AUTHORS , AND REVIEWERS xxi

Richmond, Robert J. Author, Sects. 1.1.1 and 1.1.2

Technical Assistant
Propulsion and Power Branch
Astronautics Laboratory
Marshall Space Flight Center

Rogero , Steve Author, Sect . 9.3.3

Senior Research Engineer
Instrumentation Section
Jet Propulsion Laboratory

Rosner, Daniel E. Author, Sect. 2.4

Associate Professor
Dept. of Engineering and Applied Science
Yale University and Consultant to Aerochem
Research Laboratories

Rupe, Jack H. Author, Sects . 2.2.1 , 2.3.1 , and

Research Group Supervisor 2.3.2
Liquid Propulsion Section
Jet Propulsion Laboratory

Sack, Larry E. Author, Sects . 3.2, 5.4.3, and 7.5.1

Member of the Technical Staff Contributor, Sect. 3.3.1
Liquid Rocket Division
Rocketdyne

Sanscrainte, Willard A. Reviewer, Chapter 9

Technical Director of Advanced Agena Rocket Engines
Bell Aerospace Company

Senneff, John M. Author, Sects . 8.3.6 and 10.2

Assistant Chief Engineer Contributor, Sects . 7.4.2 and 8.5.1
Combustion Devices Reviewer, Chapter 8
Bell Aerospace Company

Sirignano , William A. Author, Sects . 3.5.1 , 3.5.2.4,

Associate Professor 3.5.3.1 , 3.5.3.2, 3.6, 4.1 , 8.1 ,
Dept. of Aerospace and Mechanical Sciences and 8.4.1
Princeton University Reviewer, Chapter 8

Smith, Allan J. Jr. Author, Sects. 7.2.1 and 8.4.4

Research Engineer Contributor, Sects. 7.4.2 and 7.6
Georgia Institute of Technology Reviewer, Chapter 6
Formerly:
Design Engineer
Aerojet-General Corp.

Sokolowski, Daniel E. Author, Sect. 10.6.1

Aerospace Engineer Reviewer, Chapter 5
Chemical Rocket Division
NASA- Lewis Research Center
xxii LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

Strahle, Warren C. Author, Sects . 3.1 , 3.4.1 , 3.4.2,

Associate Professor of Aerospace Engineering and 3.4.3.1
Georgia Institute of Technology Reviewer, Chapter 4
Formerly :
Member of the Professional Staff
Science and Technology Division
Institute for Defense Analyses

Szuch , John R. Author, Sects . 5.2, 5.5.1 , 5.5.2 ,

Project Engineer 5.6 , 6.2.1 , 6.2.2, 6.2.3.1 to
Advanced Systems Division 6.2.3.3, and 6.2.4
NASA-Lewis Research Center

Thibodaux, Joseph G. Jr. Reviewer, Chapter 1

Chief, Propulsion and Power Division
NASA-Manned Spacecraft Center

Tonon, Thomas S. Author, Sects . 8.3.1.2 and 8.3.1.3

NASA Fellow
Dept. of Aerospace and Mechanical Sciences
Princeton University

Valentine, Ralph S. Author, Sects . 2.5 and 9.6.3

Director, Engineering Research Dept.
Atlantic Research Corporation

Van Huff, Norman E. Author, Sect. 9.6.5

Manager, Design and Analysis Dept.
Engineering
Aerojet Liquid Rocket Company

Van Wyk, Rod Contributor, Sect . 4.3.2

Manager, System Analysis Section
Winchester Group Research
Olin Corporation

Varma, Ashok K. Author, Sect . 8.4.5

Guggenheim Fellow Reviewer, Chapter 6
Dept. of Aerospace and Mechanical Sciences
Princeton University

Vincent, Joseph Author, Sect. 10.5

Member of the Technical Staff
Liquid Rocket Division
Rocketdyne

Wanhainen, John P. Author, Sect. 8.4.2

Aerospace Engineer Contributor, Sect . 7.4.4
Chemical Rocket Division
NASA-Lewis Research Center

Waugh, Richard C. Reviewer, Chapter 5

Engineering Specialist
Combustion Dynamics Dept.
Aerojet Liquid Rocket Co.
 EDITORS, AUTHORS , AND REVIEWERS xxiii

Wesley, Robert D. Author, Sects. 9.3.1 to 9.3.5,

Section Supervisor , Experimental Instrumentation 9.4.3.1 , 9.4.3.3, and 9.6.1
Aerojet Nuclear Company

Wieber, Paul R. Author, Sect. 9.6.6

Aerospace Research Engineer
Physics and Chemistry Division.
NASA-Lewis Research Center

Williams, Forman A. Author, Sect . 2.1

Professor of Aerospace Engineering
Dept. of Aerospace and Mechanical Engineering Sciences
University of California, San Diego

Wood, Leroy M. Author, Sect. 9.4.4.2

Research Engineer
Propellants and Combustion Technology
Bell Aerospace Company

Wuerker, Ralph F. Author, Sect. 9.4.5

Member of the Technical Staff
TRW Systems

Zinn, Ben T. Author, Sects. 7.2.2, 8.3.1 , 8.3.1.1,

Professor of Aerospace Engineering and 8.4.3
Georgia Institute of Technology
CHAPTER 1

Introduction

The subject of this reference book is combus- interchangeably for the same application . Pres-
tion instability as associated with liquid propel- sure-fed designs are preferred for reaction control
lant rocket engines. Before the details of unstable and space propulsion systems where thrust and
combustion can be properly discussed, however, chamber pressure are low and where emphasis
the reader should have some knowledge of the is on simplicity for multiple restart capability.
liquid rocket engine systems involved, and the Pump-fed engines are used for high thrust , high
associated performance parameters, as well as chamber pressure applications typical of the
the combustion instability phenomenon itself. booster and upper stages of launch vehicles where
An understanding and knowledge of the past long durations are common . The propellant tanks,
accomplishments and the present status of the although large, can be kept light because their
problem should also prove helpful . It is the pur- strength must provide only for structural rigidity
pose of this introductory chapter to meet these and a pressure sufficient to provide the proper
needs. net positive suction head (NPSH) to the pump .
Subsequent chapters will treat much of this The more complex start transient can be tolerated
introductory material in far greater detail- because of a limited necessity for restarting the
exploring the mechanisms of instability, mathe- engine.
matical approaches for solutions of specific Although there are numerous possible pro-
problems, application of solutions to actual pellant combinations ,4483 three are basic to con-
engines , experimental measurements, and sta- ventional engines in current use. They are
bility rating techniques to name only a few of classified as earth storable, cryogenic and cryo-
the subject areas discussed at length. However, genic-storable. Earth storable implies that the
other portions of this chapter, such as the engine propellant is in the liquid state at standard
systems and performance parameters, will not conditions of temperature and pressure. Pro-
be treated again but will be only referenced . Even pellants which are gases at standard conditions,
with the material that is discussed later, the but through chilling have been liquified , are
emphasis given by each author to what he deems known as cryogenic propellants. Insulated con-
to be the most important aspects of the subject tainers are required for their storage and unless
most likely will receive a somewhat different cooling is provided , boiloff losses are inevitable .
emphasis in the chapters that follow. For this Cryogenic-storable combinations consist of one
reason, the reader with a particular need should cryogenic and one storable propellant.
evaluate the overall assessments discussed here The common storable combination consists
with the more specific information that follows of nitrogen tetroxide for the oxidizer and a
in the subsequent chapters . blend of 50 percent hydrazine and 50 percent
unsymmetrical dimethylhydrazine (UDMH) for
1.1 LIQUID ROCKET ENGINE SYSTEMS the fuel (the blended fuel is known as 50-50,
or Aerozine 50 and will be referred to as A-50) .
1.1.1 Conventional Engines *
Monomethylhydrazine (MMH) is sometimes
substituted for the A-50. These storable com-
Liquid rocket engine systems can be classified
into two broad categories, pressure-fed and
pump-fed . In general, these systems are not used * R. J. Richmond , Author.

1
2 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

binations are hypergolic (i.e. , combustion occurs Oxidizer manifold

from a spontaneous reaction as the propellants
come in contact) and thus find wide application
in space propulsion systems where restarts are
necessary. Apollo spacecraft use a number of Fuel manifold
rocket engines that employ these propellant
combinations in pressure-fed systems. Military Injector
pump-fed systems also use storables based on
"instant readiness " (no last minute tank filling Baffle
as would be required with cryogenics because of
boiloff losses) .
Currently, the most widely used cryogenic
combination consists of liquid oxygen for the
oxidizer and liquid hydrogen for the fuel . This
combination requires an ignition source such as
a spark arrangement. The performance of this
propellant combination is the highest of those Acoustic liner

under discussion . A typical engine using these Combustion

propellants is the pump-fed J-2 engine on the chamber
Nozzle
Apollo V system. Long term storage of liquid
hydrogen requires special insulation and careful
design because of the low temperatures involved Throat
(-423.3° F) .
The cryogenic-storable combination in wide FIGURE 1.1.1a.-Typical components of a thrust chamber
use is liquid oxygen and RP-1 (a fine cut of assembly.
kerosene) . This combination is used in a number
of booster engines-typical is the pump-fed F-1 propellants are reacted and which extends to the
engine of the Apollo V launch vehicle. The pro- inlet of the de Laval nozzle (combustion insta-
pellant combination provides good performance bility occurs within the combustion chamber) ,
combined with extremely low cost . Ignition is combustion instability damping devices such as
accomplished by igniter fluids such as triethyl- the baffle and acoustic liner, and the de Laval
aluminum (TEA or TEAL) , triethylborane nozzle which converts the thermal energy of the
(TEB), or mixtures (TEA/TEB) which are heated combustion products to kinetic energy
hypergolic with the oxygen. of the exhaust jet.
In future space engines another propellant
classification is becoming more prominent. Known 1.1.1.1 Pressure-fed engines.-The thrust levels
as space-storables, these propellants are cryo- of pressure-fed engines in current use range from
genics which under the space environment may about 5 to 22,000 pounds. Their propellant flow
be stored for sufficient periods to fill the role of systems are extremely simple. Basic components
current earth storables but with added per- consist of a propellant valve, injector, combustion
formance potential. Some typical examples are chamber, and nozzle. The propellant valve ad-
the light hydrocarbon fuels such as methane used mits the propellant to the combustion chamber
with fluorinated oxidizers (FLOX or OF2). through the injector which mixes and atomizes
Common to conventional engines in current use the propellant prior to reaction . The shape of the
is a thrust chamber assembly (TCA) as shown in combustion chamber is normally a right circular
Fig . 1.1.1a . The typical components are labeled cylinder to which is attached a converging-
and include the injector (which contains fuel diverging nozzle (de Laval type) . A typical
and oxidizer orifices to introduce the propellants schematic of a pressure-fed engine is shown in
into the chamber) , the manifolds for the pro- Fig. 1.1.1b . This represents the Apollo Service
pellants, the combustion chamber where the Propulsion System (SPS) which is used to propel
 INTRODUCTION $ 1.1 3

Fuel Oxidizer valves . Since the thrust level is directly propor-

tional to the ullage pressure for pressure-fed
systems, sophisticated pressure control is re-

Too
quired to achieve and maintain the desired thrust.
In contrast, the thrust level of pump-fed systems
Redundant Redundant is insensitive to normal tolerances in tank pres-
fuel ball oxidizer ball
valves valves sure . (See Ref. 593, pp. 173-263 for a detailed
discussion of pressurization systems .)
Since the propellant tanks must contain pres-
sure equal to the chamber pressure, the total
pressure losses in the flow system and the dy-
namic head associated with the propellant in-
jection velocity, they must be of rather heavy
construction compared to tanks for pump-fed
Ablative combustion systems. Therefore, pressure-fed systems are
chamber and nozzle
attractive for low chamber pressure applications .
Typical of such chamber pressures is the current
range from about 100 to 120 psia. With respect
to weight, as chamber pressure is increased be-
Fuel yond this range, pump-fed systems become more
Oxidizer attractive. In order to minimize tank weight,
great emphasis is placed on minimizing the
pressure drop across the feed line, propellant
valve, and injector . A further reduction in tank
pressure is realized by substituting ablative com-
bustion chambers and nozzles for regeneratively
cooled components , thereby avoiding the pres-
sure drop associated with the regenerative cooling
FIGURE 1.1.1b .-Service propulsion system propellant passages .
flow schematic.
Pressure-fed systems, as any other rocket
engine system, are susceptible to combustion
the command and lunar modules into lunar instability of the low, intermediate or high fre-
orbit and the command module back to earth. quency types . The various forms of instability
The feed system for supplying propellant to are discussed in Sect. 1.2. It is sufficient to re-
the engine consists of the propellant tanks, feed mark here that if pressure-fed systems seek to
lines, and prevalves in the propellant lines (the minimize tank weight via reductions of pressure
latter are usually located at the base of the tank drop , specifically, the injector pressure drop,
for positive containment of the propellant prior then low frequency instability becomes a par-
to engine start) . It is common for the inlet of the ticular problem. Here the coupling is between
main propellant valve to be considered the feed the feed system and the combustion process in
system-engine interface . The propellants are the engine. More will be said on this subject in
forced from the propellant tanks into the com- subsequent chapters.
bustion chamber by means of gas pressure in the
tank ullage space above the liquid propellant 1.1.1.2 Pump-fed engines.- Pump-fed engines
surface. There are several methods for providing are used for high thrust, high chamber pressure
this pressurizing gas ; however , the most common applications. Those in current use have thrust
is the stored gas technique using helium in high levels ranging from about 16,000 to 1,500,000
pressure vessels. Pressurizing gas is admitted to pounds at chamber pressures from about 300 to
the propellant tank ullage space through a line 1000 psia. The systems contain the same com-
containing a number of pressure regulators and ponents as the pressure-fed designs plus turbo-
4 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

pump and gas generator subsystems for pumping has no throat ; however, the turbine nozzles
the propellant into the combustion chamber. The through which the gas is injected into the turbine
typical pump-fed system operates on the high normally operate in a choked condition . There-
pressure ratio, low mass flow parallel* turbine fore, the gas generator, turbine manifold , and
cycle. The engine-feed system interface is at the nozzle assembly can be treated by the same
turbopump inlet flanges . analytical methods as the main combustion
In the feed system portion of the design , one chamber.
important consideration is to minimize the tank Ideally, the gas generator should supply a
weight. The tank must be strong enough to pro- completely reacted gas of uniform temperature
vide only structural rigidity and pressure suf- to the turbine . Combustion must be stable to
ficient to maintain the required NPSH ; the avoid vibration and pressure oscillations . Uni-
latter factor is important in pump design and form temperature is necessary to avoid local hot
considerable effort is devoted to minimizing spots which could overheat regions of the com-
NPSH during engine development. In some cases bustor body or the turbine manifold. A completely
effort has been so successful in this regard that reacted gas at the turbine manifold inlet is neces-
structural, rather than pressure considerations , sary for efficient turbine operation . If the gas is
govern the tank weight. not completely reacted at that point , the reaction
The turbopump subsystem (which is con- continues with rising temperatures as the gas
sidered part of the engine) consists of two pumps travels through the manifold . This is known as
(usually centrifugal) , a turbine, and in some cases gas generator "afterburning ." The gas that
a gearbox (now considered obsolete) mated travels the farthest before entering the nozzles
together to form a complete unit. The energy to is usually the hottest while that which enters the
drive the turbine is supplied by the gas generator nozzles almost immediately is the coolest . This
subsystem consisting of a propellant valve , in- results in the turbine operating at a lower average
jector and combustion chamber. Propellant is temperature than necessary. The injector has
tapped from the turbopump discharge lines, in- the largest effect on the uniformity and degree of
jected into the combustor through the gas gen- reaction attained, but the configuration of the
erator injector , chemically reacted and converted combustor body also is influential. The volume of
to gas. The gas is expanded through the turbine . the combustor body must be large enough to pro-
which drives the pumps . Since the operating vide adequate stay-time for the reaction to be
temperature limit of current turbine materials completed. In order to minimize the stay-time
is about 1500 ° F, the gas generator is operated and hence combustor size, mixing is enhanced
with excess fuel to avoid exceeding this limit . A mechanically by using L-shaped combustor bodies
schematic of the F-1 engine, a typical pump-fed along with choke rings and screens. Afterburning
system is shown in Fig. 1.1.1c. problems have been experienced primarily with
Since the gas generator is a combustion device , the LOX/RP-1 propellant combination. (See
similar in some respects to the main combustor Ref. 593 , pp . 273-304 for a more detailed dis-
and with many of the same inherent combustion cussion of pump-fed systems.)
problems, a brief discussion is warranted here. The main chamber and gas generator are sus-
The valve which controls the flow to the com- ceptible to the same types of combustion in-
bustor is normally mounted on the injector which stability mentioned in connection with the pres-
in turn is mounted on the combustor body. This sure-fed systems. Low frequency instability or
assembly is mounted directly on the turbine mani- chugging can occur when the combustion dy-
fold by means of a short duct. Since the gas namics are in resonance with the engine-feed
generator must provide gases at low temperature system dynamics. Normally the feed system up-
to be compatible with the turbine, cooling of the stream of the pumps is not considered in the
combustor body is not required . The gas generator analysis of chugging because of the isolation due
to the pumps. However, recent evidence has be-
* The word "parallel" refers to the fact that in this
cycle the turbine is in parallel with the main combustion come available which shows that this is not neces-
chamber. sarily the case. Consequently, pump-fed engines
 INTRODUCTION §1.1 5

Oxidizer
Fuel Fuel

Oxidizer turbopump Gas,generator

Main oxidizer
Fuel valve
Main oxidizer valve
turbopump

Turbine
J
Main fuel valve Main fuel
valve

Regeneratively cooled combustion

chamber and nozzle

Oxidizer
Fuel Gas cooled
skirt extention
Hot gas

FIGURE 1.1.1c.-F- 1 engine propellant schematic.

as well as pressure-fed engines must be considered turbine and the low pressure ratio, high mass flow,
in combination with their ultimate feed systems series* turbine cycles. There are several variations
before their low frequency stability can be as- of these two cycles.395 The conventional pump-
sessed . The combustion process could also couple fed system discussed previously is an example of
with the blade wake frequency of the turbopumps. a high pressure ratio, low mass flow, parallel
However, this rarely occurs because the blade turbine cycle. A variation of this is the so called
wake frequency usually is not matched to either tap-off cycle which taps gas from the main com-
the low or high frequency dynamics of the com- bustion chamber to drive the turbine, thus
bustion process. The blade wake frequency refers eliminating the gas generator subsystem .
to the pressure pulses present in the engine feed
lines due to the turbopump impellers and is de- 1.1.2.1 Aerospike engine. -One advanced en-
fined as the product of the pump speed and the gine concept under investigation is the toroidal
number of impeller blades. High frequency or aerodynamic spike or simply aerospike engine.³
acoustic instability can also occur. A schematic of a possible system designed to
operate at 1500 psia chamber pressure is shown
1.1.2 Advanced Engines in Fig. 1.1.2a . Gas tapped from the main com-
The preceding section discussed the two basic bustion chamber drives high pressure ratio, low
conventional engine systems. This section pre- mass flow, parallel turbines which puts this en-
sents two advanced systems being considered for gine in the same general cycle class as the con-
future applications. These are the aerospike and ventional engine. The aerospike differs from the
staged combustion systems. These systems, as conventional engine in the configuration of the
well as the conventional pump-fed systems, nozzle and combustion chamber. The combustion
operate on two basic power cycles which can be
categorized in terms of the turbine used . These * The series turbine cycle uses a turbine in series with
are the high pressure ratio, low mass flow, parallel the main combustion chamber.
6 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

Main oxidizer valve Liquid hydrogen inlet Hot gas control valve

Liquid LH2 Annular

Loxygen combustion
LO₂ inlet chamber

Hot gas Fuel

contol valve turbopump
Oxidizer
turbopump Regeneratively
cooled plug
4 Liquid oxygen
Liquid hydrogen
Turbine exhaust gas

FIGURE 1.1.2a.-Aerodynamic spike engine propellant flow schematic.

chamber is toroidal in shape, rather than cylin- ments probably will result in circumferential
drical, and instead of a conventional bell nozzle , partitioning of the combustion chamber. There-
the combustion chamber discharges the exhaust fore, each partitioned section might act as an
gases onto a regeneratively cooled, truncated plug individual combustion chamber whose dimensions
nozzle. The turbine drive gases, after expanding would not permit acoustic frequencies low enough
through the turbine, are discharged through the to couple with the feed system.
base of the plug to increase the base pressure and
nozzle performance. 1.1.2.2 Staged combustion engine. Another
The advantages of the aerospike engine are advanced engine concept being investigated for
that the nozzle provides continuous optimum future application is the staged-combustion
expansion to the local ambient pressure through- system68,69 employing a conventional bell nozzle
out its trajectory . The outer boundary of the with an extendible skirt. The skirt is extended at
exhaust jet is a freely expanding jet which upper altitude to improve the nozzle performance.
adjusts to the ambient pressure field . This re- This engine operates on the cycle which uses the
sults in a higher total integrated impulse over the low pressure ratio, high mass flow series turbine .
vehicle flight path . In addition , the combination
. This is sometimes called the staged- combustion
of the toroidal combustor and the aerodynamic topping cycle. A schematic of a proposed system
spike nozzle results in a compact short-length is shown in Fig. 1.1.2b . This system differs from
engine. Additional discussion of the aerospike the conventional system in both cycle and con-
engine is presented in Ref. 368 , pp . 10–15 . figuration. The engine consists of a primary com-
The aerospike engine is susceptible to the same bustion chamber, turbopump and turbine as-
types of combustion instability as the engines sembly, secondary combustion chamber and bell
previously discussed . In addition to the low and nozzle with skirt. The total fuel used by the en-
high frequency instability encountered in con- gine is pumped into the primary combustion
ventional systems, the aerospike engine by virtue chamber where it is combined with a small frac-
of the long path around its annular combustion tion of the oxidizer to produce a low temperature
chamber can encounter acoustic instability at (1500° F) fuel-rich gas. This gas is then expanded
frequencies low enough to couple with the feed through a low pressure ratio , high mass flow tur-
system. This is sometimes referred to as the "race- bine which drives the turbopumps. It leaves the
track" mode. It is similar to the coupling of turbine and enters the secondary combustor
acoustic modes with the feed system of large where it is combined with the remaining oxidizer
conventional engines. In practice this may not at the optimum mixture ratio prior to expansion
be a serious problem because structural require- through the bell nozzle.
 INTRODUCTION §1.1 7

Low- speed inducer

Oxidizer turbopump
Oxidizer
Fixed nozzle

Main chamber

Preburner

Fuel
Extendible skirt

Low- speed inducer Fuel turbopump

FIGURE 1.1.2b .-Staged combustion engine propellant flow schematic .

Staged combustion systems currently under ratio and expanded through the nozzle. The high
consideration operate at chamber pressures up chamber pressure permits increased specific im-
to 6000 psi for the primary combustor and at pulse because larger nozzle expansion ratios can
3000 psi and above for the secondary combustor. be used without corresponding increases in the
Due to these high pressures , regenerative cooling physical size of the nozzle exit . The extendible
of the secondary combustion chamber and throat skirt improves the performance further (at al-
requires special consideration . To accomplish this , titude) by providing additional expansion ratio.
combustion chambers must be constructed from Combustion instability can also occur in
materials with thermal conductivities much staged combustion systems. Both the primary
higher than those in current use. Transpiration
. and secondary combustors can encounter high
cooling, a technique of passing a controlled and low frequency combustion instability. In
amount of fuel through the chamber walls, is an addition, complex interactions can occur between
alternate cooling technique. It is not as desirable the primary and secondary combustors, the
as regenerative cooling because of the performance primary combustor and the feed system, and the
loss associated with poor combustion of the cool- secondary combustor and the oxidizer feed system.
ing fuel after it enters the combustion chamber.
Transpiration cooling, therefore, would be used 1.1.3 Performance Parameters *
only in the critical areas of the throat and com-
The following nomenclature pertains to Sect.
bustion chamber. The major portion of the nozzle 1.1.3:
downstream of the throat where the heat flux is
lower would be regeneratively cooled as is the cur- F Thrust
rent practice. ge Gravitational constant
The advantages of the staged combustion r Nozzle radius (perpendicular to axis)
system are that its cycle is more efficient than the α Streamline angle, measured from axis
high pressure ratio, low mass flow parallel turbine 8* Boundary-layer displacement thickness
cycle because all the propellant which drives the
turbines is ultimately burned at optimum mixture * W. B. Powell, Author .
8 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

VER Energy-release efficiency, in the characteristic velocity by simple me-

chanical means , and without any change in the
Isp with actual energy release
characteristic velocity or the vacuum specific
Isp with 100 % energy release
impulse.
0* Boundary-layer momentum thickness The external characteristics and performance
parameters of the rocket motor are illustrated
Subscripts :
and defined in Fig. 1.1.3a , along with an indica-
a Ambient tion of the accuracy which is acceptable in each.
e Nozzle exit Typical values of the vacuum specific impulse
Stagnation and the characteristic velocity can be seen in
vac Vacuum Table 1.1.3a.
1, 2, i Stream tube indexes
1.1.3.2 Internal processes in rocket thrust cham-
Superscript :
bers. If the performance of a real rocket motor
() Potential flow nozzle contour is to be analyzed and predicted, it is necessary at
the start to have a complete inventory of the
1.1.3.1 External performance parameters .-A processes and parameters which influence the
rocket motor is characterized to a great extent performance. Next it is necessary to have realistic
by the propellant mass flow which it consumes and workable analytical models for these proc-
and the thrust which it produces, and , to a lesser esses, and the requisite physical data as input .
extent, by the size, shape, and area ratio of the Finally, it is necessary to combine the treatment
exhaust nozzle, and by the pressure in the com- of the separate processes so as to consider all the
bustion chamber. processes and their interactions concurrently.
The fundamental rocket motor performance It is not sufficient to calculate the performance
parameter is the vacuum specific impulse (Ispvac) , or of an idealized reference model and then to sub-
pounds-force thrust, in a vacuum environment, tract supposedly linearly independent increments
per pound-mass per second of propellant flow . of performance corresponding to the separate
Vacuum specific impulse completely determines recognized sources of loss. It is not sufficient
vehicle performance in space flight applications . because there are important interactions between
Further, for a given propellant and nozzle shape ,
vacuum specific impulse is quite insensitive to
small changes in combustion chamber pressure,
the size of the rocket motor, the propellant mass
m
flow rate, or the thrust level.
A second rocket thrust chamber parameter,
the characteristic velocity (c*) , correlates propel-
lant mass flow. Experimental evaluation and Fpa
correlation of the characteristic velocity requires. Jhey
use of the stagnation pressure in the combustion
chamber; this is a quantity which cannot be Fvac = F + PA
measured directly , and which in many cases has
no single unique value . Thus the characteristic IsPvac= Fvac Measured data
(± 1/2 %)
velocity cannot be defined or measured as ac- m Fpa
ṁ
curately as can the vacuum specific impulse . How-
= PoA9
19c0,(±4%) Pa
ever, the characteristic velocity, even with an
A.
uncertainty of several percent , can be a useful
A+
parameter for sizing propulsion system com-
PPo Approx.
ponents . To achieve the required thrust the
propellant mass flow rate can typically be ad- FIGURE 1.1.3a .- External performance characteristics of
justed over a larger range than the uncertainty rocket motor (where the vacuum thrust is Fvac) .
 INTRODUCTION §1.1 9

TABLE 1.1.3a .-EXAMPLES OF PERFORMANCE CORRELATION AND PREDICTION

Low area ratio High area ratio

System é Ispvacy VER € Ispvac O/F c*,

(lbf-sec)/lbm (lbf- sec)/lbm ft/sec

Experimental Experimental Predicted

F2/H2 a1.9 343 1.00 a60 417 428.7 12.0 7850

P. = 50 psia
F = 1200 lbf
N2O4/A - 50 1.0 209.6 .98 $60 307 311.6 1.8 5550
P. = 100 psia
F = 1000 lbf
N2O4/A - 50 1.0 211.1 .99 b40 313.2 316.1 2.0 5310
P. = 105 psia
F = 8000 lbf
N2O4/A - 50 1.5 231.3 .97 b62.5 313.5 318.4 1.6 5490
P. = 97 psia
F = 20,000 lbf

a 15° Conical nozzle.

b Contoured nozzle (or bell nozzle) .

the several processes which occur in a real rocket A description of the listed liquid propellant
motor. rocket motor internal processes and a discussion
The internal processes which accompany the of their principal interactions follow:
production of thrust by a rocket motor are not as NONUNIFORM MIXTURE RATIO DIS-
evident as are the external characteristics, nor TRIBUTION : Propellant is injected into the
are they as yet completely understood . The motor at the upstream end of the combustion
.
principal internal processes in a conventional chamber. Often a low mixture ratio region is
rocket motor are shown on Fig . 1.1.3b and are
deliberately produced near the wall of the thrust
listed on Table 1.1.3b . Each of these processes
chamber in order to reduce the severity of the
has associated with it a loss in performance (com-
environment to which the thrust chamber is ex-
pared to the ideal one-dimensional isentropic
posed . Additionally, the injector propellant mani-
equilibrium performance) ; approximate percen-
tage magnitudes of the loss in specific impulse folding and the orifice pattern may be such that
there are variations in both local mixture ratio
corresponding to each of the processes are also
shown on Table 1.1.3b . While these identified and local density of mass injection across the
processes are of primary importance in the face of the injector, superimposed on any de-
analysis of steady-state performance of rocket liberately created distribution (see Sect. 2.3) .
motors, some of them are also known to be im- These propellant mass and mixture ratio dis-
portant in the analysis and prediction of com- tributions are best identified by non-reactive
bustion instability . fluid simulant flow tests, wherein the spray down-
10 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

Chamber wall

Equivalent inviscid flow boundary

Boundary layer TURA

AVEN
(friction and
heat transfer)

m2, (0/F)2 V₁, P, P

At

Arbritrary
exit surface
m₁, (0/F), Stream tubes

Droplet vaporization Supersonic flow region

combustion
(Energy release ) Transonic flow region

Additional energy release,

Injection , primary atomization , chemical kinetic rate limited
distribution of droplets equilibrium shift during
(stream tube formation) expansion through nozzle

Frac =" ↓ m₁Vicos αi + Sp₁ dA

FIGURE 1.1.3b .-Internal processes in the real rocket motor.

stream of the injector face is sampled . * Alter- of the stream tube rocket motor is the mass-
natively, a careful analysis based on the hydraulic averaged performance of each of its stream tubes,
flow characteristics of the injector propellant each presumed to expand through a nozzle having
passages and orifices can give a useful indication the shape and exit area ratio of the overall
of the delivered distributions. nozzle . The net effect of the variation in mixture
It has been found that regions of identifiable ratio from stream tube to stream tube is a de-
mass and mixture ratio which are larger than a crease in performance compared to the perfor-
typical molecular mixing distance (about inch) mance that would have been obtained at a uni-
tend to maintain their separate identity as they form average mixture ratio .
react and progress through the thrust chamber TWO-DIMENSIONAL NOZZLE FLOW :
and nozzle. The two-dimensional shape of the de Laval
Thus the rocket motor can be idealized as a nozzle affects the flow in two ways. Near the
group of separate, non-mixing , rocket motors throat, curving of the flow distorts the pressure
(stream tubes) operating in parallel, and con- distribution , leading to curved constant- pressure
strained to coexist within the overall chamber surfaces. This causes a decrease in the mass flow
and nozzle contour. through the nozzle, compared to one-dimensional
As a first order approximation, the performance sonic flow through the geometrical throat area ;
and the divergence of the exit flow results in a
* However, it is recognized that the chemical loss of axial momentum, and thus a loss in specific
reaction
can alter the non-reactive-fluid spray distributions. impulse.
 INTRODUCTION §1.1 11

TABLE 1.1.3b. -REAL ROCKET MOTOR PROCESSES stream of the throat. The supersonic flow for an
AND LOSSES
inviscid fluid can be developed by a method-of-
characteristics calculation, using as a starting
Process Typical loss point a transonic solution such as that mentioned
percent above. Such a calculation gives the pressure on
the surface of the nozzle downstream of the
Nonuniform mixture ratio distribution 0 to 5 starting line, and/or the pressure, density, ve-
(stream tubes) locity, and direction of flow through any chosen
Incomplete energy release 1 to 5 nozzle exit surface. The total thrust of the nozzle
Multi-phase flow (solid particles) Not treated here determined from either the starting line pressure
Two-dimensional flow (curvature and 0.1 to 3
and axial momentum flux plus the axial com-
divergence)
Finite reaction rates (kinetics) 0.1 to 10 ponent of the surface pressure forces in the super-
Boundary layer (friction and heat 0.5 to 5 sonic region of the nozzle, or from the exit sur-
transfer) face pressure and axial momentum flux, is the
same, and reflects the loss in axial momentum
due to divergence of the exit flow.
FINITE REACTION RATES : The gases
Various approaches to determining the effect in the combustion chamber of a rocket motor
of throat curvature on the pressure distribution are generally at a high enough temperature that
in the transonic flow field and on the mass flow some dissociation of molecular species has taken
through the nozzle are given in the literature. place. As the hot gases expand through the noz-
When two or more stream tubes coexist in a given zle, the pressure and temperature decrease. At
nozzle flow, additional constraints, beyond the the reduced pressure and temperature, the dis-
fundamental assumption that the static pressure sociated species tend to recombine , and to liberate
is everywhere continuous, are needed to define energy as they do so. However, these recombina-
the sonic surface and to determine the relative tion reactions are rate-limited , and are only
flow areas occupied by the stream tubes in the partly completed during transit through a typical
region of the nozzle throat. rocket nozzle . With a knowledge of the rate con-
Kliegel and Quan409 have presented an analysis stants for the particular reactions involved , and
of the flow of two concentric stream tubes within a specific nozzle size or time scale, the kinetic
a rocket nozzle. Propellant injection conditions effects can be incorporated in the nozzle flow and
were not specified . They concluded that the sonic performance calculation . The effect is always a
surfaces of the stream tubes must lie on a com- decrease in the otherwise attainable performance .
mon constant-pressure surface, since this condi- MULTI-PHASE FLOW: Some propellants
tion maximizes the total mass flow through the yield combustion products which contain solid
throat. particles in the combustion chamber, or which
Norton5196 studied the flow of multiple stream form condensed species during expansion through
tubes through a nozzle with the injection con- the nozzle. The magnitude of the effect of solid
ditions (mass , momentum, energy) specified for particle flow on the rocket motor performance
each stream tube, and found that the sonic sur- depends on the state and number of the particles
face was, in general, discontinuous. as well as on their drag and heat transfer coeffi-
In either case, if the properties of the gases in cients . The two-phase flow process effects on
two stream tubes differ, then the stagnation performance are complex, and the basic data.
pressures of the two stream tubes differ. This needed to analyze the process is difficult to ob-
causes the difficulty, which was mentioned earlier, tain . Some existing computer programs approxi-
in the definition and evaluation of the charac- mate the effect of two-phase flow in one-dimen-
teristic velocity. sional nozzles , but no two-dimensional treatment
The development of the supersonic flow field is available at the present time. For these reasons
and the divergence of the exit flow are determined no estimate of the magnitude of the two-phase
by the shape of the two-dimensional nozzle down- flow loss is given in Table 1.1.3b .
12 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

INCOMPLETE ENERGY RELEASE : formance is reduced as a result of the lower energy

Atomization , mixing , evaporation , and reaction available to the expansion process. The reduced
of the injected propellant do not occur instan- enthalpy also provides a different base condi-
taneously or completely in a rocket motor . Typi- tion from which the kinetic loss is reckoned . This
cally, most of the combustion is completed within empirical single-parameter model was selected
the combustion chamber, before the products of for interim use because it is simple and unam-
combustion enter the nozzle convergence section . biguous .
Some droplets may evaporate and some reaction While this simplified energy release model
may occur during transit through the nozzle. * cannot exactly duplicate the effect of the real
However, some propellant droplets may not distributed energy release process, it has proven,
evaporate within the confines of the thrust cham- empirically, to give useful overall performance
ber, and some evaporated molecules may not correlations and extrapolations when used with
find "partners" with which they can react. The several conventional propellant systems.
result in either case is an energy deficiency and a BOUNDARY LAYER FRICTION AND
small change in the composition and properties HEAT TRANSFER : The effect of surface.
of the combustion products, with a reduction in friction and of heat transfer from the combustion
the realized performance. gases to the wall is confined to a relatively thin
A distinction should be made between the layer of gases next to the wall. The core of the
stream tube process and the incomplete energy gas flow can be considered to be remote from
release process . A stream tube is a macroscopic viscous or heat transfer effects.
region of uniform mixture ratio, distinguishable The influence of the boundary layer on rocket
from other macroscopic regions having other motor performance is given by a model which
uniform mixture ratios. Energy release efficiency replaces the real flow with an equivalent com-
is a measure of the completeness of the reaction. pletely inviscid and isentropic potential flow. A
within a stream tube ; it is related to the degree displacement thickness, 8 *, is evaluated which
of mixing and vaporization of the drops pro- defines the size and shape of the potential flow
duced by the individual injector elements and to nozzle with respect to the real nozzle, so that the
the microscale processes by which molecules mix. mass flow through the potential flow nozzle is
and react. the same as that through the real nozzle. This is
The real distributed energy release process has shown in Fig. 1.1.3c . The excess momentum flux
not been completely described and modeled as through the equivalent potential flow nozzle
yet, although some analytical and experimental over the momentum flux of the real flow is given
investigations are under way and others have by a momentum thickness, 0, of the potential
been proposed . flow, such that an annular layer of the potential
In the absence of a capability of analytically flow of thickness @ has this amount of momentum .
describing the energy release process and pre- Then, as shown on Fig. 1.1.3c, the thrust of the
dicting its effect on performance, an empirical real motor is obtained by correcting the thrust
interim model and procedure23 has been found of the equivalent potential flow motor for the
useful. Incomplete energy release is modeled excess momentum flux and for the exit pressure
analytically by reducing the total energy of the acting on the displacement thickness. The pro-
system through a reduction in the stagnation . pellant mass flow as computed for the throat of
enthalpy. Prior to expansion , 100% of the pro- the equivalent potential flow motor is used with
pellant is presumed to be in thermodynamic the calculated real motor thrust to obtain the
equilibrium at the reduced enthalpy level. The specific impulse . The derivation of the expres-
chemical and fluid dynamic calculations are then sions for the displacement and momentum thick-
made in their usual manner . The calculated per- nesses from the basic heat transfer, friction , and
fluid flow relationships is given by Elliott, Bartz,
and Silver241 and by Alber6º¸
* A study by Mitchell491 has shown that combustion in
the supersonic portion of the nozzle flow, even at high A unique feature of the treatment by Elliott,
area ratio, can still contribute to performance. Bartz, and Silver is the simultaneous solution
 INTRODUCTION §1.1 13

Surface of real nozzle-

ré
Surface of potential flow nozzle
having equal mass flow
*
mreal =mpotential flow 8 Displacement thickness
nozzle 8 = Momentum excess thickness
21
DFB
L= |+
[+ 2 π r 8 e* p − 2 π r; Q₂ pé Vé cosx.

- -
Freal =Fpotential flow [2 Tr₂O₂ cos de Pévér
pél
nozzle E

FIGURE 1.1.3c . - Boundary layer model and performance correction .

of the integral momentum and energy equations plete energy release and kinetics is a decrease in
for the thin boundary layer. This results in the the vacuum specific impulse. Further, due to the
losses due to heat transfer to the wall appearing thermodynamic nature of the nozzle expansion
indirectly, as part of the total boundary layer process, the percentage performance loss for a
loss, rather than appearing separately, as in given energy deficiency increases with the nozzle
earlier boundary layer treatments. Thus all area ratio, and the magnitude of this effect in-
interactions between shear drag and heat loss creases with increasing energy loss .
from the gases are taken care of within the boun- (2) The effect of two-dimensional flow on kinetic
dary layer model . Alber60 takes an overall look and other losses. Two-dimensional flow affects the
at this approach to the evaluation of the boundary magnitude and rate of change of the properties
layer losses and shows that the displacement of the expanding gas. Small nozzles and nozzles
thickness-momentum thickness approach is ex- having small radius of curvature throats and/or
actly equivalent to a correct accounting of the high expansion angles just downstream of the
axial component of the pressure forces acting throat may expand the combustion gases so
on the interior of the nozzle. rapidly that there is not enough time for rate-
INTERACTIONS BETWEEN PRO- limited equilibrium shifts to occur. Nozzle contour
CESSES : In principle, each process occurring curvature , located near the exit of a "bell " noz-
in the real thrust chamber interacts with each zle, can cause local changes in pressure, density,
other process. These interactions must be ac- velocity, and direction which enter into calcu-
counted for in a valid performance calculation lation of the boundary layer loss and the di-
procedure. Some of the more important of these vergence loss .
interactions are discussed here in general terms : (3) The effect of nonuniform mass and mixture
(1) The effect of incomplete energy release on ratio distribution on kinetic losses. Stream tubes
kinetic losses and on the nozzle expansion process. having different mixture ratios will produce com-
Incomplete evaporation , mixing, and reaction. bustion gases having different properties and
result in a loss of energy and a change in the temperatures, and thus the kinetic losses will be
composition of the combustion products, com- different in each stream tube.
pared to complete reaction . The temperature of (4) The effects of incomplete energy release,
the gases entering the expansion nozzle is reduced , kinetics, stream tube flow, and two-dimensional
and associated with this is a reduction in the flow on the boundary layer loss. The properties
fraction of dissociated species present. While used in calculating the boundary layer corrections
that part of the total loss directly ascribable to are those of the stream tube closest to the wall.
kinetics may decrease, the net effect of incom- The properties of the gases in this stream tube
 14 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

are determined by the local mixture ratio and C. Have conventional de Laval nozzles
by the energy release, kinetics, and two-dimen- d. Have conventional upstream-end propel-
sional interaction effects . lant injection techniques ; no mass addition
in the supersonic region
1.1.3.3 Real rocket performance calculations.- e. Are in steady-state operation
Calculation of the performance of a real rocket These restrictions correspond to the capabilities
motor must be based on an integrated physical of existing computer programs, and to processes
model that is a composite of models describing which are relatively well understood . Despite
each of the physical processes or effects occurring these seeming limitations, there is a wide and
in the real rocket motor . Separate inputs should important field of applicability for the developed
define each effect, and the calculated performance performance calculation capability ; i.e., most
should reflect the result of all interactions. rocket motors used for space propulsion appli-
If all of the real rocket motor processes are cations use propellants composed of the six listed
identified and appropriately modeled , and if the elements, produce more than 100 lbf thrust,
input information is available to accurately specify and have de Laval nozzles .
each process or loss , then the computed perfor- The rocket motor performance calculation
mance will be identical to the performance of capability developed by the Performance Stan-
the real rocket motor as measured on a test stand dardization Working Group is based on two com-
(presuming accuracy of the test data) . puter programs and on the use of the empirical
In practice , this degree of perfection has not model for the incomplete energy release process,
yet been achieved . Some processes have still not (discussed under Sect. 1.1.3.2) . The two computer
been adequately modeled , some needed physical programs were selected by the Working Group
data are not as yet known with sufficient ac- from among all those that were available to it
curacy, and present computational procedures in June 1967. These computer programs , selected
are limited . on the basis of technical validity, computation
Because of these present limitations, the capa- time, documentation , and other factors60, have
bility to predict the performance of the general since been modified and improved to meet the
case of a real rocket motor does not now exist . needs of a standardized performance calculation
However, a practical methodology for correlating and prediction procedure and to be compatible
and predicting the performance of a useful class with each other, and have been made available
of rocket motors has been developed by the as reference programs . 17,22,23,44,45 (These and other
Performance Standardization Working Group* related computer programs can be obtained
of the ICRPG23 , 555b. This class of rocket motors through CPIA (Chemical Propulsion Information
is limited to those that Agency) , The Johns Hopkins University, 8621
a. Use propellants whose combustion products Georgia Avenue, Silver Springs Maryland
are entirely gas phase, with no solid par- 20910. )
ticles . (Currently only the chemical ele- 1.2 COMBUSTION INSTABILITY *
ments C, H, O, Cl , F, and N can be han-
dled . ) Combustion instability problems have been
b. Are large enough so that the flow is not experienced during nearly every rocket engine
dominated by viscous effects (i.e. , above. development program . Since these problems
about 100 lbf thrust) severely impair the operation of the engine and
vehicle system, there is considerable incentive
* The Performance Standardization Working Group to seek an understanding of this undesirable
of the Interagency Chemical Rocket Propulsion Group phenomenon. Combustion instability results from
was organized in 1965 and has functioned through its coupling between the combustion and the fluid
three committees ; Overall Concepts, Theoretical Methods ,
dynamics of the system. Through this coupling,
and Experimental Measurements. The ICRPG Perform-
ance Evaluation Manual for Liquid Propellant Rocket oscillatory energy is supplied by the combustion
Engines23 is the product of joint efforts of the Overall
Concepts and the Theoretical Methods committees . * T. A. Coultas, Author, Sects . 1.2.1 to 1.2.3.
 INTRODUCTION §1.2 15

to sustain the oscillations. Only if the damping bration measurements often do not correlate
processes present in the system are sufficiently well with the corresponding chamber pressure
large to dissipate the oscillatory energy more measurements. Frequently, however, there will
rapidly than it is supplied, will the oscillations. be similarities in frequency as well as in severity
decay. Thus, combustion instability may be or amplitude (see Sect . 9.5) .
prevented by either increasing the damping or Temperature and heat transfer monitoring has
decreasing the coupling with the driving forces . also been successful in indicating the onset of
Several distinct types of instability have been combustion instability. Thermocouples buried in
observed and their physical manifestations have the chamber wall respond to the rapid increase
caused a variety of picturesque names to be in wall temperature . Faster response is afforded
generated for each of them. A common trait is by measuring the transducer coolant or the local
that all types of combustion instability are regenerative coolant temperature changes (see
characterized by chamber pressure oscillations , Sect. 9.6.5).
although the frequency and amplitude of these Combustion instability also causes oscillations
oscillations and their external manifestations of the axial position of the Mach diamonds in the
normally vary with the type of instability . exhaust plume and this can be detected by high
Oscillatory operation of a rocket engine is un- speed photography . The oscillations of the Mach
desirable for many reasons . One of the most im- diamonds will usually correspond in frequency
portant of these effects is the severe vibration . to the chamber pressure oscillations . Monitoring
Vibration levels in excess of 1000 g have been of the luminosity variations from the exhaust
experienced . Such vibration levels can impair the plume is another optical technique sometimes
operation of sensitive guidance components and used . These variations in luminosity may be
have severe effects upon payloads and even rela- very weak. For instance, it has been estimated
tively massive structural members . Another for one case that the relative amplitude of bright-
severe effect is the grossly increased heat transfer ness oscillations in the plume would be only 0.1
due to the oscillatory operation . This increase is percent of the relative oscillations in chamber
often sufficient to melt and destroy portions of pressure (optical measurements are discussed in
the rocket system. Other less drastic effects in- Sect . 9.4) . Flow rate variations and thrust varia-
clude decreased performance, uncontrolled im- tions have also been observed as an indication
pulse, variation in thrust vector, and the effects of combustion instability.
of oscillatory propellant flow rates.
1.2.1.1 Damage. In addition to the destruc-
1.2.1 Physical Manifestations
tive vibration, thrust magnitude and direction.
Combustion instability is manifested in many vibrations, and uncontrolled impulse caused by
ways. The most satisfactory method of detection an oscillating system, combustion instabilities.
and study of combustion instability is the mea- may result in extensive damage to the thrust
surement of chamber pressure (see Sect . 9.3) . chamber and injector itself. High frequency in-
Pressure measurements made in the propellant stabilities result in grossly increased convective
feed system show similar oscillations, and in some heat transfer coefficients in the chamber walls.
cases the amplitude measured here may be greater With the prevalent tangential mode instabilities
than that measured in the combustion chamber. this increase occurs at all axial positions in the
In the combustion chamber, frequencies from less chamber. Since the heat transfer rate is normally
than 100 to over 15,000 hertz have been measured highest near the nozzle throat, this is a very
at amplitudes of from 10 to 1000 percent of sensitive location. In one rocket engine develop-
steady-state chamber pressure. In addition to ment program it was found that combustion in-
pressure measurements, instabilities are often stability caused the nozzle to be neatly severed
shown by means of vibration measurements . The at the throat and dropped into the exhaust flame
very high vibration levels often measured have deflector. In another program, where considerable
given rise to the term "rough combustion ,' unreacted oxidizer was present near the injector ,
meaning in that case unstable combustion . Vi- together with the maximum tangential mode.
16 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

amplitude , increased heat transfer rates resulted Effort is also expended to assure that the
in chemically burning the thrust chamber at that propellants will be atomized into droplets suf-
location . This can become a chain reaction, burn- ficiently small that they will be completely
ing not only the chamber and the injector, but vaporized in the chamber prior to expansion
propellant lines and thrust stand structure as well . through the nozzle . Propellant vaporization is
Combustion instability is not always so dra- governed by heat and mass transfer between the
matically destructive. Lower frequency modes propellant droplets and the hot combustion gases.
of instability may do no damage at all . Even some Additional convective effects generated by the
high frequency instabilities are nondestructive acoustic or oscillatory pressure field will enhance
if the injector and thrust chamber are satis- the combustion efficiency by accelerating the
factorily cooled and sufficiently strong . In fact, at vaporization of the propellants. In spite of these
high frequencies, a quite sophisticated set of factors, which sometimes increase the combus-
instrumentation is often required to determine if tion efficiency, other effects of combustion in-
the combustion is unstable. At very high fre- stability may override them and decrease the
quencies, the amplitude of the oscillations may apparent or overall combustion efficiency.
be quite low and the damage incurred negligible The presence of low frequency or longitudinal
over the short periods of times (3 to 4 seconds) mode high frequency oscillations may result in
typically used to obtain performance data. a decided increase in the axial mixture ratio varia-
tions . This is particularly true with injectors
1.2.1.2 Effect on combustion efficiency. The having unequal injection pressure drops across
apparent performance of a particular rocket the two propellant systems. Here, a temporarily
system may either increase or decrease during a lowered chamber pressure will cause a much
combustion instability. At high frequencies, the larger quantity of one propellant to be injected
high transverse pressure and velocity gradients during the low pressure portion of the cycle than
enhance both factors which control the steady- of the other propellant. Thus, alternatively high
state rocket engine performance . These factors and low mixture ratio injection rates will result
are mixing (distribution) , and vaporization of the in wide variations in mixture ratio along the
propellants. In engines where either of these length of the chamber, resulting in poor per-
factors has not been very thoroughly optimized formance. Further, the grossly increased injection
for steady-state operation , it might be expected rates during the low pressure part of the cycle
that the performance would increase during an can result in a portion of both propellants being
instability. In contrast, for the case of an opti- exhausted from the chamber unburned . Thus ,
mally designed injector configuration displace- even though the mixing and vaporization may be
ment of the propellants may actually result in a very complete during one portion of the cycle ,
performance loss . Considerable design efforts (high chamber pressure and low injection rate)
are usually put forth to assure that the propellant the chamber is flooded during the low chamber
mixture ratio is made as uniform as possible at pressure portion of the cycle. Other less important
each point on the injector face. In spite of these losses occurring during combustion instability
efforts , however, combustion may occur locally include increased heat transfer and friction .
at mixture ratios considerably off design or As a rule of thumb, it may be stated that high
optimum value, particularly for propellants frequency instabilities tend to increase the com-
which vaporize at grossly different rates. It has bustion efficiency if the combustor is not initially
been found that if the propellants have not been a high performer, while low frequency oscillations
properly distributed within about an inch down- tend to decrease performance.
stream of the injector, any transverse mixture
1.2.2 Classification
ratio variations tend to persist, since gas phase
mixing is exceedingly slow. In the presence of a Several different classes of instability have been
strong acoustic field, however, the gas-phase identified and studied experimentally. Usually
mixing is considerably enhanced, minimizing the instability driving mechanisms differ among
this source of inefficiency. the classes such that different methods are re-
 INTRODUCTION $ 1.2 17

quired to control or eliminate the instability. time delay have met with mixed success . These
Historically, instabilities have been classified changes, even if successful in eliminating chug,
by their frequency range, but there is not a may decrease the system performance or bring
sharp dividing line between the so-called low, about a high frequency instability.
intermediate and high frequency classes. Classi- Other low frequency instabilities have been
fication of combustion instability, merely by its caused by coupling of the combustion process
frequency, has led to much confusion . It would with the injector structure. The injector may act
appear that a better method would relate the as a diaphragm and oscillate in an "oil can"
classes of instability to their effects, the most mode. This can cause nonuniform propellant
important coupling mechanisms, and to the injection and atomization which results in a low
devices used to eliminate them. frequency instability . Still other cases can allow
coupling between the combustion , or chamber
1.2.2.1 Low frequency, chug.- Of the various pressure, and the structural system. One instance.
types of combustion instability the low-frequency was found where pressure oscillations in the
type, or chug, also called putt-putt, groaning, propellant contained in the regenerative cooling
and motor-boating, is perhaps the easiest to jacket were being caused by chamber pressure
handle both from an analytical and experimental perturbations flexing the jacket wall structure
or developmental standpoint (see Chapters 5 with the resultant coupling causing a low fre-
and 6) . It is generally accepted that the fre- quency instability. Another rocket system in-
quency range which might be encountered in stability, of very low frequency (order of a few
the chugging mode is less than several hundred hertz) , is caused by propellant flow rate oscilla-
hertz . In this frequency range, the wavelength tions which result from pump amplification of
is usually much larger than characteristic di- the fluctuations of the pump inlet pressure (the
mensions of either the chamber or the feed sys- pump inlet pressure variations are due to the
tem . In some cases, however, there may be wave g-loading of the liquid column extending back
motion in the propellant feed lines. This insta- to the tank) . Although this "pogo " instability
bility usually begins with a low amplitude, si- is driven by thrust modulations that are trans-
nusoidal wave shape, growing in a linear fashion ferred to the structure, the combustion is per-
to higher amplitude . turbed so slowly as to remain essentially steady
Analytically, the chamber may be simulated and hence this is not generally considered a com-
by a lumped volume element, the combustion. bustion instability .
represented by a simple, constant time delay
and the propellant feed system resistance ne- 1.2.2.2 High frequency instability The most
glected, although feed system inertance and destructive type of instability is referred to as
capacitance may become important in the analy- high frequency instability, resonant combustion ,
sis. The combustion time delay is defined as the or acoustic instability . The latter is a generic term
time required for the liquid propellant to enter the derived from the observed correspondence in
chamber, travel at injection velocity to an im- frequency and phase between experimentally
pingement point, then be totally vaporized and observed chamber pressure oscillations and those
burned . Usually an empirical average can be calculated for the acoustic resonances of the
found for each propellant. A value which has chamber (see Chapters 4 and 6 ) . Both axial
often been used is simply the liquid flight time (longitudinal) and transverse (radial and tan-
from injector face to impingement point , usually gential) modes are included in this terminology .
for the least volatile of the propellants since this High frequency instability has also been called
constitutes a major portion of the total time lag. by such names as " screaming," "squealing,"
Methods of elimination of chug instabilities "organing," "screeching," and just plain "rough. "
include: increasing the pressure drop in the in- It is generally conceded that the effect of the
jector, increasing fluid inertance (i.e. , longer propellant feed system is usually unimportant
L/D in the injector or feed system) , decreasing in the study of high frequency instability . The
chamber volume, etc. Attempts to change the frequencies are often so high as to preclude coup-
18 LIQUID PROPELL
ANT ROCKET COMBUSTION INSTABILITY

ling with the relatively sluggish feed system. It than that required for sustaining the oscillations ;
should be noted, however, that in large combus- and (2) making changes in the dynamic energy
tion chambers the fundamental acoustic frequency losses or damping so that they exceed the energy
may be so low that the feed system can easily gains from the combustion response (see Chapters
couple. Combinations of resonant combustion 7 and 8) . Into the first category fall the very com-
and chug instability have also been observed . mon developmental attempts to achieve stability
In some cases the elimination of the chug by feed by varying the injector hole pattern , hole size ,
system changes has also resulted in elimination pressure drop, etc.
of the resonant combustion. In other cases , the Of the thousands of postulates or design criteria
opposite has been true. for achieving combustion stability, the following
An oscillatory source of energy is required for is an example of an injector design rule that has
sustaining an instability. For high frequency worked out quite well . In nearly every case, "the
instabilities, this energy must come from the stability of a rocket engine will be improved if
propellant combustion and is usually only weakly the two phases, i.e., liquid and combustion
dependent upon the feed system. Further, the gases , move at grossly different axial velocities ."
sustaining energy addition must be properly This rule of thumb indicates that if the most
time-phased with respect to the oscillating pres- volatile propellant is injected at higher velocity,
sure. In most high frequency instabilities the the engine will become more stable. Also , if the
coupling appears to be direct. Each wave af- less volatile propellant is injected at lower veloci-
fects the propellant combustion strongly enough ties , further increases in stability will be found.
so that sustaining energy is added directly to This "relative velocity " criterion is probably
that wave (i.e. , within a time no longer than responsible for the generally observed good sta-
the period) . Effects of secondary importance can bility characteristics of engines using gaseous
come from transient change in propellant in- hydrogen fuel . Like all generalizations in the
jection rates, in propellant impingement and study of combustion instability, there have been
atomization characteristics and from residual exceptions to the rule and certain limits must be
effects from one cycle influencing the amplifica- set . Often it is not practical for performance ,
tion of the next. In general, however, these simply compatibility, or other system reasons to increase
affect the equilibrium amplitude of the insta- the relative injection velocities (see Sect . 7.4) .
bility. The predominant effect of combustion chamber
SUSTAINING MECHANISMS : Sustaining baffles places them in this first category as well,
mechanisms which have been proposed for high because their stabilizing influence primarily re-
frequency include : loss of ignition , sensitive sults from simultaneously increased resonant
chemical preparation time, physical time delays, frequency (ie . , a shift to higher modes) and
detonation processes , pressure or temperature lowered acoustic displacement of the combustion
sensitive chemical kinetics, the "exploding" of gases within a baffle compartment. Some of their
droplets heated to above their critical temperature effectiveness may also stem from disruption of
and pressure and the shattering and mixing of wave propagation and droplet shattering. In
the streams , fans, or drops by the gas particle addition, there may be some effect of the baffles
motion. These are only a few of the more recurrent in energy dissipation due to vortex shedding, but
explanations which are advanced to explain the the extent of this contribution to effectiveness of
sustained combustion instabilities. Many of baffles is not currently known .
these will be discussed in detail in later chapters . Although it is known that engines may be
METHODS OF ELIMINATION : Two stabilized by the use of baffles on the injector
fundamental methods of eliminating high-fre- face, less well-defined criteria exist for defining
quency combustion chamber instability have how many baffles are required , i.e. , the necessary
been employed : ( 1 ) making changes in the pro- baffle spacing, or baffle length required to achieve
pellant spray combustion field or in the pressure
wave character* so that the combustion response *
Frequency or wave shape alterations may be accom-
to the wave motion releases less oscillatory energy plished through changes of the chamber geometry.
 INTRODUCTION $ 1.2 19

dynamic stability in the engine (see Sects . 3.5.3.3 or other types of acoustic absorbers (Sect. 8.2) .
and 8.2) . Although some empirical rules have Acoustic absorbers are also often considered to
been developed which seem applicable to some be a panacea. It is noted that if "enough" highly
propellant combinations over narrow ranges of absorbing resonators are inserted into a chamber,
injector variation , these rules are based only it will be stable. Similar to the case of baffles,
upon experience . When they fail in practice they questions remain as to the correct design criteria,
must be replaced by other rules which accom- and together with other operational factors it
modate the most recent failures of the old ones . is desirable to minimize the use of acoustic ab-
Generally, it has been thought that if a baffle is sorbers . It has also been found that the inclusion
made long enough to "shield" the region in the of particulate matter into the combustion gas acts
chamber wherein the major portion of the com- as an absorbing device by dissipation of oscillatory
bustion occurs, and if the baffle spacing is such energy through frictional processes associated
that the baffle cavity frequency is above about with particle drag. This damping method has
5000 Hz , the engine would be stable. Unfor- been widely acclaimed as the solution to high
tunately, there is much evidence which contra- frequency instability problems in solid propellant
dicts this rule . In one case it was found that motors (see Sect . 8.5.2) .
longer baffles made the engine even more un-
stable . In most cases , however, it has been found 1.2.2.3 Intermediate frequency, buzz . - Between
that if the baffles are made long enough, and if the two extreme types of combustion instability
they are spaced closely enough, the engine will is the intermediate frequency. It is unfortunate,
stabilize . but most of the combustion instabilities which
Baffles are not generally regarded as a panacea are not obviously either low or high-frequency
for promoting combustion stability. Even if are lumped into this intermediate category. This
sufficient and proper baffles did assure stability, propensity is so strong that often many chugging
it would still be desirable to minimize the length instabilities at higher than usual frequency are
and number of baffles used . The presence of referred to as buzz (see Chapters 5 and 6) .
baffles on the injector face represent a discon- The beginning of an intermediate frequency
tinuity in the most important combustion region . instability usual y shows a growing coherence of
It has been shown that baffles can have significant the combustion noise at a particular frequency
and deleterious effects on both combustion ef- with slowly increasing amplitude. There is usually
ficiency, and the effectiveness of nozzle thrust- wave motion in the propellant feed system. Al-
vector-control injection * . Furthermore, as the though there may be wave motion in the chamber,
baffle length is increased, the heat losses from the the phase and frequency does not usually corre-
combustion gases can become large enough to spond to an acoustic mode. If chug and buzz-
lower combustion efficiency, and the heat loads types are to be distinguished, it is by spatial
to the baffles may become prohibitive. Thus, it chamber pressure variations present during buzz
is desirable from the standpoint of cost , com- instability. The pressure wave shape is very nearly
plexity, performance, thermal compatibility and sinusoidal and one or both of the propellant feed
thrust vector control to minimize the number and systems may be highly coupled . Buzz-type in-
length of baffles used in a combustion chamber stabilities are not particularly damaging if they
to achieve dynamic combustion stability. remain at low amplitude, but may degrade per-
In the second category fall various types of formance, total impulse or thrust vector. In some
damping devices. Items of this nature have been cases the amplitude increases to the point of
noted to be effective when metal-walled com- triggering a high frequency mode.
bustion chambers were replaced with ablative In very large chambers there may be wave
chambers (Sect . 8.5.1 ) or lined with Helmholtz motion which approximates that of acoustic
instability. In one buzz instability the amplitude
* This effectiveness loss is because of mixture ratio and phase relationships in the combustion cham-
effects associated with the baffle which persists into the ber were the same as for a first tangential acoustic
nozzle. mode, but the frequency was 20 to 30 percent
20 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

lower than had previously been observed for this amplitude. The combustion chamber may be
mode. This particular instability was cured by either triggered unstable by some artificial dis-
the use of quarter-wave tubes placed on the turbance or the instability may develop spon-
propellant feed system. Some instances of longi- taneously .
tudinal high-frequency instability, although by The initiation of an acoustic instability is
definition an acoustic mode, probably should be frequently a nonlinear phenomenon ; that is,
called buzz . This is because this acoustic insta- there may be a threshold value of perturbation
bility has the linear buzz characteristics , is usually amplitude above which a sustained instability
feed system coupled, and may be eliminated by is caused, and below which the perturbation will
feed system rather than combustion changes. damp. Thus a single pressure disturbance can be
A type of instability which also falls into the amplified and result in sustained combustion in-
intermediate category is the so-called " entropy stability . A rocket engine's inherent stability
wave." Here axial mixture ratio gradients passing determines its ability to absorb large disturbances
the sonic plane in the nozzle may emit a pressure and yet return to its steady-state operation . This
wave which travels upstream toward the injector. was the impetus for combustion stability rating
The reflected wave influences the mixture ratio devices which provide artificial disturbances to a
and hence travels back downstream as an en- combustion chamber (see Chapter 10) .
tropy discontinuity at the gas velocity . The Prior to the introduction of stability rating.
overall effect is that of an intermediate frequency devices, a rocket engine's inherent stability was
oscillation . determined by reliance upon the occurrence of
Buzz is often encountered in development spontaneous stability. This required very many
programs on engines which are designed to tests and one system was deemed more stable
throttle over a wide thrust range . It is almost than another if, in many tests, the superior system
axiomatic that if the throttling range is to be exhibited fewer occurrences of combustion in-
very wide, buzz or chug will be encountered at stability. In other cases, the operational con-
some thrust condition . This is because the com- ditions were varied from the nominal. The system
bustion is given a continuously varying set of which remained stable through the widest ex-
conditions ; e.g., velocity of propellants , impinge- cursion of operating conditions such as mixture
ment time, atomization effectiveness, etc. It is ratio, chamber pressure, fuel temperature, etc. ,
nearly certain that at least one condition will be was alleged to be the more stable .
found which is favorable for coupling with wave The technique of relying upon spontaneously
motion in the propellant feed system. This is occurring combustion instability was desirable
particularly true if the engine is fairly large, al- in that the rating found was clearly associated
lowing many possible resonances in manifolds, with naturally occurring disturbances and the
domes, feed lines and other parts of the feed combustion was not changed because of the in-
system. sertion of foreign rating devices into the system .
This type of instability has been noted on However, using that approach , either a very large
many engines which were designed to be throttled . number of tests was required for rating, or the
Not only will the engine buzz at some operating rating obtained might not be typical of the oc-
condition , but often these oscillations will in- casional large trigger source (for instance, a hard
crease in amplitude sufficiently to initiate a high start) . Furthermore, there was no method for
frequency instability. determining the size of perturbation to which the
system would be stable. If the rating was ob-
1.2.3 Initiation of Combustion Instability
tained by changing the operating conditions, it
In addition to the characteristics of the fully- may have been obtained at conditions remote
developed instability discussed in the previous from actual operating conditions. At the off-
section, the manner in which the instability starts nominal conditions there might be completely
can also furnish an important clue to the diagnos- different driving and damping mechanisms present
tician. Knowledge of how the instability began than at the desired operating conditions. Because
may be as important as its mode, frequency and of these severe limitations, combustion stability
 2217
INTRODUCTION $ 1.2

rating devices have been developed to perturb or small perturbation analysis at the stability
stable, steady- state combustion upon demand, boundary.
and thus eliminate some of the disadvantages of Pressure perturbations on the order of a few
the naturally occurring instability rating ap- percent of the steady-state chamber pressure and
proach . exhibiting a sinusoidal shape are deemed linear.
Rating devices have helped to shorten the Transition to a distorted wave shape and ampli-
development cycle of rocket engines. Prior to tudes increased to above ten percent of chamber
extensive use of these devices it was possible to pressure imply a nonlinear regime has been
arrive at the flight test stage and encounter small reached . Should this latter wave shape be present
system changes which gave perturbations suf- from the onset of resonant combustion then a
ficiently different than those previously exper- nonlinear analysis is required.
ienced so as to cause a sustained instability. All classes of combustion instability are dis-
In addition to the methods of initiation , the tinguished from random turbulent fluctuations
beginning of the combustion instability is quite. or combustion noise by coherence of a particular
different for the two cases. The fully-developed frequency or set of frequencies and in most cases
instability may be remarkably similar even by a much greater amplitude . Turbulent or ran-
though this initiation stage is distinctly dif- dom chamber pressure oscillations in rocket
ferent . These characteristics are contrasted by engines may have amplitudes on the order of one.
a variety of name pairs ; such as, sinusoidal versus to three percent. It should be noted that these
steep-fronted waves, spontaneous versus trig- "random" oscillations, when spectrally analyzed ,
gered or pulsed, small amplitude versus finite show many preferred frequencies . These fre-
amplitude, and more generally, linear versus quencies, even in a stably operating system, can
nonlinear. This latter distinction is derived from usually be attributed to some system characteris-
the type of analysis which is believed applicable tics such as pump speed, propellant feed line
to the mathematical modeling of at least the length or combustion chamber dimensions .
onset of these instabilities. Because of this funda- The lower frequency instabilities such as chug
mental difference in the initiation characteristics , and buzz, and even most longitudinal mode high-
two rather distinct schools of thought have de- frequency instabilities, were historically found .
veloped (see Chapter 4) . to be linearly initiated . The high frequency non-
Linear instabilities are nearly always con- linear instabilities were not generally recognized .
sidered to be spontaneous, but devices have been This was probably due to the relatively crude
constructed which sometimes can enhance the pressure transducers which were available at
occurrence of linear instabilities . Nonlinear in- the time and the absence of rating devices . Fur-
stabilities, on the other hand, are always triggered ther, the early engines were usually characterized
by a finite disturbance. The disturbance may by low injection density and were prone to a
occur naturally, or may be artificially triggered linear, or slowly building instability . Exceptions
by a combustion stability rating device. to this may have been found on some start se-
quences which resulted in a finite perturbation
1.2.3.1 Spontaneously initiated linear insta- or "hard start . "
bility. Spontaneous instabilities require no initial Linear, low -frequency instabilities acting as
disturbance, but rather grow out of the noise triggers for high frequency instabilities are often
inherent in the combustion process. If an engine prevalent during engine start-up. As noted, an
is to experience a linear instability, it might be engine operating over a wide range of thrusts ,
expected to begin immediately upon reaching flowrates, and chamber pressures is much more
normal operating conditions, since no trigger is likely to experience an intermediate frequency
required. However, variations in test conditions instability. Thus, if an engine's start transient
as well as the closeness to a stability boundary is very gradual, i.e. , full thrust is achieved only
could delay that occurrence somewhat. It would after several seconds , some set of operating con-
appear that if an engine has experienced an in- ditions is likely to be prone to a linear buzz or
stability of this type, it should yield to a linear
. chug instability. This linear instability can grow
22 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

and trigger resonant combustion . To prevent affect the guidance sensor systems. Further, for
this occurrence upon engine start, the start se- small engines used for intermittent operation,
quence is typically shortened to a few milli- the resultant thrust or total impulse is severely
seconds (while still avoiding overshoot or hard altered from that of a smooth start. For larger
start) . In this manner, the linear growth of any engines , this ignition spike can result in a trigger
oscillations is curtailed since the " sensitive" for sustained resonant combustion which may
portion of the start is quickly passed through . cause subsequent hardware destruction .
NATURALLY OCCURRING TRIGGERS :
1.2.3.2 Induced or nonlinear combustion in- Hypergolic liquid bipropellant rocket engines
stability. It would appear that most of the com- may operate over a wide range of conditions ,
bustion instabilities which are encountered today ranging from a highly pulsed mode of operation
are initiated by finite disturbances. These finite to long duration firings. In either situation , the
disturbances may take the form of several types ignition and start transients must be compatible
of natural disturbances or artificial triggers which with the propulsion system. For example, high
are used to determine dynamic stability of an pressure spikes resulting from explosions of ac-
engine system. cumulated propellants may result in the complete
The naturally occurring disturbances have ac- failure of a radiation cooled chamber.
quired the descriptive names of " spikes" and Propellants are said to be "beneficially hyper-
"pops . " Although there is no universal agreement golic" if the chemical reaction initiated by con-
in the rocket industry, it is generally conceded tact of the two propellants is sufficiently energetic
that a "spike" is a significant chamber over- to establish steady-state combustion with a
pressure upon ignition of the engine. A "pop" smooth pressure buildup in a specified period of
is defined as a similar over-pressure, but occurring time. Because of low chemical reactivity, the
spontaneously during mainstage operation (i.e. , chemical reaction may build up slowly for a
operation at nominal chamber pressure). period of time allowing unreacted propellant
For many years the sources of pops and spikes accumulation to occur in the chamber. This period
were the subject of intensive research during of time may result in undesirable spiking when
rocket engine development programs. In more true ignition actually occurs because the propaga-
recent years, however, the source of the distur- tion flame front passes through premixed pro-
bance has become of less interest due to the pellants and builds into a detonation wave. Much
development of stability rating devices. Engine of the experience with spiking is clouded by in-
systems which have withstood a variety of ar- strumental problems. Very sophisticated and
tificial combustion perturbations and quickly high frequency response instrumentation is re-
returned to stable operating conditions have quired to make a careful study of spiking.431 A
been assumed to damp natural perturbations typical spike pressure amplitude may be ten
rapidly . This has been found to be the case in times steady-state chamber pressures, but with
many engine systems and natural perturbations. pulse duration of only a few microseconds . This
have thus been rendered innocuous, although spike may either decay in a single cycle or show
wherever possible the sources of such distur- linear damping, lasting through several cycles
bances are still removed. of an acoustic mode of the chamber. High fre-
Under conditions of high altitude operation, quency instability following a spike will usually,
hypergolic bipropellant combinations may char- but not always, damp very quickly as steady-
acteristically start with an extremely high state operating conditions are approached .
chamber pressure spike. This spike may be at- Pops are not usually severely damaging to
tributed to the explosion /deflagration of the combustion chambers, but like spikes they can
propellants collected in the thrust chamber and/ severely disrupt either programed impulse or
or accumulated on its walls during the period of delicate portions of missile systems. They are
ignition delay. These chamber pressure spikes also natural triggers for high-frequency instabili-
may be of sufficient magnitude to result in de- ties. Pops are most often noticed in medium to
structive thrust chamber failure, or adversely large-sized rocket engines and are not limited to
 INTRODUCTION $ 1.2 23

hypergolic propellants . Neither a comprehensive The pulse gun, as the name implies , is a device
knowledge of the source of pops nor information which resembles a gun . It consists of a breech into
on wave growth thresholds are known . Some pops which an explosive (or powder) is placed , usually
are thought to be caused by leaks in the injector. in a conventional cartridge case, a firing mecha-
It is possible that these leaks lead to liquid nism , a barrel, and usually a diaphragm to protect
propellant flowing in an uncontrolled manner the explosive from the direct rocket chamber en-
across the injector face, collecting in pockets , vironment. Like the gas pulse, the pulse gun
and (without the necessity of oxidizer being barrel is usually attached to the rocket chamber
present in the case of monopropellants such as such that the pulse is fired tangential to the
hydrazine) exploding . Likewise, initiating dis- chamber, but it too may be oriented in any
turbances could be caused by a collection of direction . This device fires upon command to the
storable propellants in cracks or crevices where firing mechanism which may be a mechanical
detonation of hydrazine (for example) is en- detonator to initiate the main charge. The ex-
hanced by partial confinement as well as catalytic plosive charge in a pulse gun may vary from 3
effects of and heat transfer from solid surfaces. or 4 grains of gun powder to over 100 grains of
Accumulation of hydrazine in pressure pickup high explosive. To some degree, the overpressure
ports has been detected by chemical analysis in the chamber caused by the pulse gun is related
after firing of N2O4/50 % hydrazine-50% UDMH to the charge size but the relationship with sus-
propellants. Apparently, the UDMH was selec- tained instability may not be so direct.
tively distilled away. Explosive decomposition The explosive bomb consists of three major
of film coolant, the possibility of a noise pulse of parts. There is the explosive itself, usually a high
unusual amplitude from the combustion zone, explosive such as RDX, some sort of a detonator
and disturbances in the feed system are all possible which may be either initiated thermally by the
sources of pops . Probably many disturbances rocket gases or may be exploded electrically from
occur that are too weak to cause pops . an external signal. Lastly, there is a case which
ARTIFICIAL TRIGGERS : To avoid the insulates the explosive charge and its detonator
many tests which may be required to determine from the environment of the rocket chamber and
stability by relying upon natural triggers, sta- furnishes some degree of containment for the ex-
bility rating devices have been developed . Al- plosive. Like the pulse gun , explosive charges
though many types have been employed , three from 2 to 100 grains are typically used. Unlike
techniques are most often used to perturb com- the pulse gun and gas pulse, the bomb is usually
bustion systems. These are the (inert) gas pulse, mounted inside the chamber and is not restricted
the pulse gun and the nondirectional bomb . The to a wall location . The perturbation from a bomb
first seems to act as a velocity perturbation to the is less directional than the pulse gun ; both can
combustion processes, while the pulse gun and give chamber pressure perturbations of 10 to 500
bomb give both a pressure and velocity perturba- percent.
tion . These devices give perturbations of a The presence of these devices may introduce
finite size (10 to 500 percent of chamber pressure) . spurious effects on combustion. It has been found
Thus the triggered instability is nonlinear and is that even though most of the bomb case is ex-
best described by the nonlinear theory. pelled after the bomb detonates, the bomb dis-
The gas pulse used on engines employing LOX turbs the combustion prior to detonation and
and RP- 1 propellants was found effective in the small case residual may have an effect. The
producing perturbations which triggered sustained pulse gun and gas pulse also introduce an ad-
instability. Recently, however, the gas pulse has ditional cavity into the chamber which may re-
been utilized on storable propellants such as sult in acoustic interaction .
N20, and 50 % hydrazine and 50 % UDMH with
practically imperceptible effect upon the com- 1.2.4 Dynamic Stability*
bustion . With the RP- 1 /LOX propellants the The ultimate objective of any rocket engine
gas pulse was successfully used to determine a
most sensitive combustion region . * V . H. Monteil and O. W. Dykema, Authors.
 24 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

development program is the successful flight . linear stability of the system. Control system
Since large costs, important program objectives theory217 defines a nonlinear system as stable if
and, in some cases men's lives are involved in the transients , resulting from a change of any
each flight, considerable efforts to guarantee the kind in the system operation, die out or if the
success of the flight are justified . In the area of amplitudes of subsequent sustained oscillations
combustion stability, this means the attainment are sufficiently small. This definition implies that
of high confidence, prior to conducting any the change in system operation can be of any kind.
flight, that the engine will not sustain damage and can be artificial or natural. The stability of
during the flight because of combustion insta- the nonlinear system is defined only by the
bility. Both theory and experiment indicate that system behavior subsequent to the transient . A
a condition referred to as dynamic stability will dynamically stable nonlinear system cannot
yield the necessary high degree of confidence sustain high amplitude , damaging oscillations
which will result in the engine performing stably. and will always return to acceptable operation
regardless of the transients which might occur.
1.2.4.1 Dynamic versus statistical stability.- The value of dynamic stability lies in the
In most cases, the coupled combustion gas dy- definition of the kind of steady-state operation
namic system of a liquid rocket engine appears that the system can sustain. In a nonlinear sys-
to be nonlinear and requires some kind of trigger, tem, large perturbations of steady-state operat-
natural or artificial, to initiate a high frequency, ing conditions may be required to gain confidence
high amplitude, damaging chamber pressure that the system cannot sustain damaging opera-
oscillation . As a result, a great deal of effort is tion . An unstable linear system, however, does
expended in attempting to determine what type not require a large system transient for oscilla-
and magnitude of triggers might occur in flight tion growth. If the steady-state operating con-
and in evaluating engine stability with artificial ditions become such that the system becomes
triggers of the assumed type. Many cases have linearly unstable, the instability will be immedi-
occurred, however, where no apparent trigger can ately apparent. Unfortunately, not all of the
be observed, and the oscillation appears simply operating conditions, which may control linear
to grow, out of combustion noise , to some limit stability, are known . Thus, a system naturally
cycle amplitude. It is apparent in these cases stable (no artificial perturbations) under many
that a trigger was not observable either because operating conditions in a number of tests may
the available instrumentation was not adequate still develop a destructive instability under ap-
to detect it or the system simply became linearly parently the identical operating conditions in-
unstable and no trigger was involved . The role curred in later tests.
of a trigger clearly is irrelevant in a linearly un- In summary, the concept of dynamic stability
stable system . involves the use of the system response to evaluate
The concept of dynamic stability on the other the stability of a system . To evaluate dynamic
hand, avoids any consideration of the possible stability the system is driven, by any means, into
types and magnitudes of naturally occurring high amplitude oscillations. If the oscillations
triggers. Dynamic stability is concerned only with subsequently decay rapidly to steady-state com-
the response of the system subsequent to transients bustion, then it is apparent that no high ampli-
in system operation and requires that the system tude, damaging, limit cycle oscillation (non-
return to normal operation after any and all tran- linear) exists (at least within the range of the
sients. The particular artificial transient which system disturbance) . A rocket engine which has
might be used to evaluate the system stability demonstrated this response over the range of
need not have a direct relation to naturally oc- operating conditions expected in flight should
curring disturbances . The significant factor in always remain stable in flight, even under rather
dynamic stability is the behavior of the system wide excursions and abnormal transients.
once the initial disturbance is removed . Although Another approach to gain some confidence in
dynamic stability is largely concerned with non- the stability of an engine is based on conducting
linear stability, confidence is also obtained in the a large number of stable tests and flights. This
 INTRODUCTION $ 1.2 25

confidence is called statistical stability. Statistical tensive component and system demonstration
stability indicates only that instabilities have program must be run or subsequent vehicles must
occurred rarely, or never, under the operating be flown with decreased confidence in the pre-
conditions tested . This does not rule out the pos-
dicted performance and reliability of the vehicle .
sibility that the system can sustain a nonlinear The loss of even one vehicle during a flight pro-
instability but may only indicate that the natural
gram because of unknown reliability of some com-
occurrence of a destabilizing trigger is rare. It
ponents or system is sufficiently expensive to
also does not indicate a broad range of stability
justify extensive efforts to assure the highest pos-
outside of the range of hardware and operating
conditions tested. sible confidence in the reliability of all components

There are engines that are known to be non- and systems prior to demonstration testing and
linearly unstable, but have demonstrated a flight. The development of an engine to satisfy
reasonably good record of stable combustion in the dynamic stability criteria has been shown to
flight. Confidence that these systems will not en- be a means of assuring high confidence in the
counter a damaging instability depends entirely stable operation of the engine.
on the vast background of test data which indi- Probably the most vivid demonstration of the
cates that as long as the engine operates in a above discussion lies in the Atlas program . The
known region of test experience, there is a low Atlas booster injector was extensively tested and
probability that instability will result. Any small qualified at both the component and engine
change in hardware and/or nominal operating system levels prior to its first flight. The thrust
conditions can invalidate this vast background chamber assembly exhibited stable combustion
of supporting data and may result in instability and an incidence rate of instability of approxi-
and/or require the generation of an equally vast mately one percent . Despite this low probability
background of data with the new system to re- of unstable operation , two successive Atlas ve-
gain an equal confidence level. hicles developed spontaneous instabilities on the
There are many examples in the history of launch pad which resulted in a complete failure
liquid rocket engine development programs where of the missions . As a result, a time consuming and
a thrust chamber assembly which has exhibited expensive development effort had to be conducted
statistical stability for a large number of tests (in the middle of the flight program) to assure
suddenly begins to self-trigger high frequency stable operation. A 14-month special program
combustion instability in a high percentage of was necessary to develop a baffled injector, to
tests . The cause of the radical change in stability demonstrate dynamic stability, and to re-demon-
(if the cause can even be found) is often a small strate engine system operation , reliability and
change in hardware or operating conditions which, compatibility with the new injector. The Atlas
it was thought, would have little or no effect on booster with the dynamically stable, baffled in-
stability. When this change in stability occurs jector has never exhibited a combustion insta-
far along in a vehicle development program, when bility in hundreds of subsequent tests and
all hardware and operating conditions of the flights. Similarly, subsequent programs such as
vehicle/propulsion systems are "frozen ," the the F- 1 , and Titan III, Stages I and II, have
cost in terms of money and flight program delays developed dynamically stable injectors and have
required to correct this problem can be disas- encountered no incidences of combustion in-
trous. Not only must the stability be improved, stability in many hundreds of subsequent tests.
but the effects of the necessary changes (usually Furthermore, these engines have sustained mal-
to the injector) on the rest of the propulsion sys- functions such as baffle and chamber failures,
tem and vehicle must also be evaluated and foreign objects in the feed systems and extreme
changes made where necessary . Once the redesign excursions in operating conditions, yet these
is made, the data on performance, compatibility abnormal conditions have not resulted in com-
and reliability compiled from previous component, bustion instabilities . These same conditions
system and flight testing are no longer applicable might well have caused combustion instability
to the new system. Therefore either a second ex- in statistically stable engines.
26 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

1.2.4.2 Dynamic stability in engine develop- cept of dynamic stability faces the problem
ment programs.- Unfortunately, in the majority squarely and develops the desired stability early
of past rocket engine development programs, in a program. When the need for high confidence
combustion stability was defined as a requirement in the stable operation of an engine is recognized
late in the program, and in some instances, after in the preliminary planning stage, a decision can
flight testing had begun. In these cases the en- be made to satisfy simultaneously dynamic sta-
gine components were in an advanced state of bility, performance and compatibility from the
development, and the engine system had been very start of engine development . Theories and
extensively tested before the dynamic stability experimental evidence indicate that all of the
requirements were imposed. This had the effect variables of injector design, chamber configura-
of severely restricting design freedom in the tion, and operating conditions affect combustion
stability effort . Typical restrictions in past pro- stability, performance, heat transfer, and dura-
grams included requirements to maintain com- bility. If all these aspects of chamber design are
bustion chamber dimensions, propellant system considered simultaneously and early in the de-
hydraulic resistances, injector element type, velopment program , not only can a more optimum
chamber heat flux, and engine performance. Under solution for the total design be reached , but each
these restrictions a simple mechanical damping individual problem area can be solved more
device, the baffle evolved . Baffles could be quickly and surely.
mounted on existing injectors and thus solve the For example, the large thrust-per-element
stability problem with minimum effort and time. (LTE) injector design approach is often con-
These early injectors relied primarily on baffles sidered as a solution for combustion stability.
(generally consisting of an even number of Since an increase in thrust per element usually
blades extending radially from a central hub) to results in increased mean spray droplet size, one
stabilize the combustion . Later systems , with might expect a corresponding loss in performance.
greater freedom, employed baffles (generally an If a solution for an existing combustion insta-
odd number of blades) in conjunction with modi- bility problem is being sought far downstream
fications of the combustion process through the in a development program, little can be done to
use of larger injector orifices (see Sects. 3.5.3.3 avoid this performance loss and the loss either
and 8.2 for details) . must be accepted or the LTE solution for sta-
The prohibitive expense involved when major, bility cannot be used . If, however, a stable sys-
developed vehicle components must be changed tem is being sought at the outset of a develop-
to solve a stability problem clearly point up the ment program , then a small increase in the
need for early development of dynamic stability. combustion chamber stay time (chamber length
Early development of dynamic stability, demon- or contraction ratio) may restore the performance
strated by rigorous pulse testing, assures a broad with very little real penalty to the thrust chamber
framework of stability within which other aspects design.
of the vehicle may be developed with confidence.
Statistical stability cannot be developed early 1.2.4.3 Demonstrating dynamic stability in en-
since it requires a large number of tests on the gine development programs.- Perhaps one of the
final prototype hardware. It is exactly this as- most difficult problems encountered in early en-
pect of statistical stability which sets the stage gine stability programs was that of measuring
for costly time-consuming, difficult re-develop- the relative stability of any new injector design .
ment programs to obtain stability after all other It is obvious that an ideal evaluation of the
development is complete. All engine components stability of the system could be obtained if
must be completely developed before a meaningful every possible mode or mechanism of instability
statistical stability statement can be made. This could be artificially excited to produce clearly
approach often results in the acceptance of what- defined oscillations . The driving force could then
ever degree of stability which results at the end be removed to determine if the system would
of a development program . return to normal acceptable operation, or if it
In contrast with statistical stability, the con- would sustain oscillations of some objectionable
 INTRODUCTION $ 1.2 27

magnitude. Implementation of this concept , cited as overpressure criteria usually vary from
however, is difficult . At first the modes or me- 50 to 100 percent of chamber pressure . An over-
chanisms of instability were only vaguely the- pressure criterion usually does not yield consis-
orized and methods of introducing any sort of tent results because both the true magnitude as
planned disturbance in the system were non- well as the recorded indication of the overpressure
existent. However, it was obvious that a part of depend strongly on many unknown and un-
the nonlinear feedback system of a hot firing controlled variables. The problem of measuring
thrust chamber is the oscillation of the combustor shock-type pressure waves has been pursued for
gases in the various acoustic modes of the cham- many years. Instruments which are able to react
ber. Thus it appeared that the most convenient rapidly to a step change in pressure level are
method of evaluating the stability of an engine notoriously poor in measuring the absolute level
under development was to force the chamber of the step change itself. Careful calibrations of
pressure into oscillation , and to observe the sub- many commercial high frequency pressure trans-
sequent response . This method makes the im- ducers in a shock tube by investigators at JPL597
plicit assumption that an artificial disturbance. have shown overshoot from 56 to as high as 260
of chamber pressure is a direct means of triggering percent of the true step change in pressure. Varia-
a nonlinearly unstable system. In order to gain tions of this type are amplified by the sensitivity
the maximum confidence in the stability of an of the transducer to vibration , partially recessed
engine rated by this method, allowances must mounting, the type of transducer, the signal
be made for possible shortcomings of driving conditioning, and the recording system (see
just the chamber pressure. One approach would Sect . 9.3 ) . All of these problems make it extremely
be to produce very high amplitude chamber difficult to be sure that an adequate pulse was
pressure oscillations, over a broad frequency supplied and, therefore, whether dynamic sta-
spectrum. A few attempts to drive the system bility was adequately demonstrated .
with a sinusoidal oscillation proved that the In the transmission of the pulse from the charge
energy requirements for such a system were far to the measuring instrument, distance is an im-
beyond that attainable with any existing or portant factor. Normally in a gaseous medium , for
contemplated driving devices. The most direct a given charge size , the local pressure level will
alternative method of performing this operation be proportional to the inverse of the distance.
subjects the chamber to a very short duration from the charge for the three-dimensional case.
pulse of sufficient amplitude to excite any or all In a chamber filled with droplets , which cause
of the acoustic modes. This pulse could be pro- "diffusion" of a pressure wave, the pressure
vided only with some sort of explosive device. decays much more rapidly. On the other hand,
Two types of explosive pulsing techniques it has been observed that the chamber reaction
have been used for stability rating of rocket en- zone may respond violently even to small pulses,
gines. These are the pulse gun and the non-direc- producing overpressures which are many times
tional bomb . Both were described in Sect. 1.2.3 greater than that produced by the same explosive
in connection with nonlinear combustion in- charge in a pressurized , passive chamber. This
stability. A more detailed discussion of explosive combustion zone response to the pulse is known
pulsing techniques can be found in Chapter 10 . to be affected by the magnitude of the pressure
The majority of engine development programs pulse and the local combustion conditions. Thus
concerned with dynamic stability have used the actual overpressure that a transducer will
bombs to generate the explosive pulses and thus record varies widely depending on the location
further discussion in this section will be limited of the instrument relative to the charge, and
to bombs. on the response of the intervening combustion
.
For demonstration of dynamic stability in an zone.
engine development program, it is necessary to Because of the difficulty of reproducibly gen-
define the pulse in some terms which will yield erating and measuring overpressure levels, many
consistent results . Pulse overpressure levels and engine development programs have used an ex-
bomb size have been used as criteria. The levels plosive pulse charge size criterion to demonstrate
28 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

dynamic stability. One obvious upper limit to As a result of the above discussion , a reasonable
the charge size on a given engine is excessive test program can be established to adequately
structural damage . For solid wall thrust chambers demonstrate dynamic stability and establish
of the Atlas and Titan types a 200 grain charge confidence that an engine will not self-ignite
appears to be near the structural limit . After instability in flight . A typical dynamic stability
repeated pulse testing with 200 grain charges criterion intended for this purpose is outlined
some minor bolt stretching and baffle cracking below. The general approach followed in this
have occurred . criterion is to strive to develop dynamic stability
Another limit on charge size is the effect of early in a program, with a few candidate in-
the pulse on propellant flow rates into the cham- jectors . These tests are designed to screen out
ber. An excessively large pulse may change the unstable configurations and to establish the dy-
flow rates sufficiently that the new operating namically stable candidate. Use of bombs as
conditions are quite different from those which large as are feasible, and the location near the
one wishes to test for stability. During the short injector face and the wall is the approach often
period before the desired flows are re-established employed for this initial screening . The possibili-
the amplitude of the pulse induced chamber ties of flow stoppage and/or a more sensitive
pressure oscillations may be greatly reduced . charge size or location are evaluated with the
The degree to which flows are changed by the selected candidate . Development of performance
pulse, and the time before nominal flow are re- and compatibility may proceed using the most
established are functions of the particular cham- stable candidate injector. The stability of the
ber and feed system under test and few generali- final prototype configuration is then evaluated
zations can be made with respect to a limit on over the limits of the anticipated flight operating
bomb size. There have been cases where a large conditions. A final test of stability in the engine,
charge will not induce instability while a smaller with small bombs to limit hardware damage,
one will induce it (see Sect . 10.2.4.1 ) . This has evaluates possible flight- configured engine sys-
been attributed to the effect of the large pulse tem effects of stability. Criteria very similar to
on the propellant flow rates. A simple test series , this have been followed successfully for Stages
with a range of charge sizes, can be run to es- I and II on Titan III and Agena. Much of this
tablish that there is no charge size, small or large, type of stability testing has also been accom-
which can initiate instability in the engine . plished on the Atlas, Thor, H- 1 and F- 1 engines.
The location of these charges in the chamber The following is an example of dynamic stability
should be such that the charge is most likely to criteria for a large engine :
initiate oscillations in the preferred and/or
damaging modes of the chamber. Theory and I. General Criterion
experiment indicate that the transverse modes An injector shall be considered dy-
are both the most probable and damaging. The namically stable if the amplitudes of
best charge location to initiate oscillations in all driven oscillations resulting from all of
transverse modes simultaneously is near the the required explosive pulse tests at-
injector and near the chamber wall . Here again tenuate to 5 percent of mean chamber
an anomaly apparently exists , as in the case of pressure within 40 milliseconds.
the effect of charge size on propellant flows. A Explosive
II.
few isolated cases have been reported where an
The explosive shall be composition C-4
.
engine can be triggered by a bomb located near
or a commercially available RDX equiva-
the throat, but not by the same bomb, under the
lent.
same conditions, from any other location . Again,
the bomb location in the corner between the in- III. Bomb Locations
jector and the wall should still be considered the Bombs shall be located at each of three
the prime location to demonstrate dynamic sta- positions as specified below:
bility and a few tests should be made to eliminate A. Radially between the wall and the
the possibility of any location anomaly. mid-radius of the chamber and axially
 INTRODUCTION $ 1.2 29

not further from the injector than one- ratio in steady-state operation . The
quarter of the distance from the injector maximum and minimum test conditions
to the throat. are defined as the estimated extremes of
B. Radially between the wall and the operation in flight.
mid-radius of the chamber and axially V. Hardware
in the convergent nozzle section. Pulse tests shall be conducted in solid
C. Radially on the centerline of the wall (or water-cooled) combustion cham-
chamber and axially not further from the bers, except for engine pulse tests which
injector than one quarter of the distance shall be conducted on flight- configured
from the injector to the throat. hardware.

IV. Test Conditions VI. Required Explosive Pulse Tests

Pulsing shall be accomplished at 5 con- The required explosive pulse tests are
ditions of chamber pressure and mixture listed in the following table :

Test conditions
Bomb size,
Test grains Location Hardware
Chamber Mixture
pressure ratio

Initial screening of candidate injectors (each injector type)

200 A Nom. Nom . Solid wall

12 200 A Min. Max. Solid wall

Candidate injector evaluation (at least one injector)

12316

25 A Nom. Nom. Solid wall

50 A Nom . Nom. Solid wall
100 A Nom. Nom . Solid wall
4 200 Α Nom. Nom . Solid wall
5 200 B Nom . Nom. Solid wall
200 C Nom . Nom. Solid wall

Prototype verification (each of two injectors)

12345

200 Α Nom. Nom. Solid wall

200 A Max. Max. Solid wall
200 A Min. Min. Solid wall
200 A Max. Min. Solid wall
200 A Min. Max. Solid wall

Engine verification (each of two engines)

1 50 A Nom. Nom. Flight-configured

30 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

1.3 HISTORICAL SURVEY * to include instability observation via circum-

ferential as well as axial slits. Levine and Lawhead
In the history of liquid propellant rocket de-
and their associates at Rocketdyne developed a
velopment the point where combustion instability
two-dimensional motor in which the entire cross
became important was where these phenomena
section (in the form of a 2- D strip) of a high-
were first recognized as being responsible for
thrust chamber could be observed and where
rocket failures. Combustion-oriented oscillations
attention could be concentrated on the important
in pressure could easily go undetected using the
combustion phenomena immediately downstream
steady-state instrumentation that was standard
of the injector. † Still later, Rossman et al.606 at
in the 1950 era and before. When combustion
Bell Aerosystems used the window principle for
instability amplitudes were sufficiently high to
studies of the droplets that were generated in the
generate audible signals, the chance for detection
actual combustion environment . One of the
of the instability was enhanced .
early studies that utilized such observation
The recognition of different classes of oscilla-
techniques, together with pressure instrumen-
tion , with different responsible mechanisms, was
tation, was conducted by Tischler et al.462 at
also an important milestone in the 1950 era. Two
NACA Lewis Research Center. That study
theoretical papers were a direct result of interest
pointed out the complex situation of combined
generated by a Bureau of Aeronautics symposium
modes since two types of wave motion were
on the subject of combustion instability in liquid
clearly shown in combination.
propellant rocket motors (Naval Research Lab-
Other investigators were also involved in the
oratory December 7-8 , 1950) . One analysis by
Summerfield673 considered the effect of both in- improvement of the instability recognition and
observation techniques. The combined efforts
ertia in liquid propellant feed lines and com-
of researchers at Princeton , MIT, and NACA
bustion chamber capacitance with a constant
Lewis resulted in the Li-Liu differential pressure
combustion time lag, and was applied to low
transducer.449 This strain gauge-type pickup with
frequency instability. Crocco173 advanced the
a water-cooled catenary diaphragm was the
concept of the pressure dependence of the time
forerunner of later miniaturized versions using
lag , which was applied to low frequency phenom-
the strain gauge concept . Industrial development
ena in monopropellant and bipropellant motors,
of the capacitance-type pressure transducers ,
and high frequency instability with combustion
which have proved even more successful over
concentrated near the injector face. This is not
the years in combustion research, were also in an
to say these Princeton professors were the first
embryo stage during the early 1950's with higher
authors dealing with the subject , since Gunder
frequency, more rugged, smaller versions to
and Friant300 had provided early insight to the
come. The piezoelectric-type transducer, which
problem and the time lag theory apparently had
has gained wide usage in the past few years, did
been suggested by the Von Karman group at
not contribute to these early studies.
JPL in the late 1940's .
Another experimental research that was being
Experimental investigations of combustion
carried out during this early era involved basic
instability were in the early stages at a number
measurements of the fundamental times as-
of industry, government, and university labora-
sociated with the combustion process to test the
tories in the early 1950's . For example, the use
sensitive time lag theory. Information on chamber
of observation windows as a means to study
behavior was being sought by varying fuel and
combustion phenomena were used by Altseimer65
oxidizer flow rates sinusoidally, and then mea-
at Aerojet for steady-state observations, and by
suring the resultant fluctuations in chamber
Berman et al.95,96 at General Electric to observe
pressure. This technique did provide basic in-
unsteady phenomena. Later these techniques
formation on the total time lag469 (see Sect.
were extended by Ellis and Pickford242 at Aerojet
6.2.2) which had significance for low-frequency
instability. However, in order to check the high
* D. T. Harrje, Author, with contributions from the
ICRPG Working Group on Liquid Rocket Combustion
Instability . See Sect. 9.2.2 for details .
 INTRODUCTION §1.3 31

frequency mode theory, frequency limitations ties to the chamber length required for efficient
of the apparatus necessitated replacement by a steady-state combustion. A similar model was
stability-limits-testing technique that was used later developed by Lambiris, Combs and Levine421
by Grey and Harrje¹80 in 1958 at Princeton to at Rocketdyne to better describe combustion
confirm the sensitive time lag theory as applied phenomena as encountered in large engines. The
to the longitudinal mode. This was achieved by use of streak photography, variable-length test-
holding the combustion process constant and ing , static pressure profiles and shock tracers
varying the chamber length (hence the frequency) . have all added to our knowledge of the axial
Similar techniques were applied by Harrje and steady-state combustion profile . Taking these
Reardon187 via a sector motor to confirm the rocket combustion studies up to the present,
transverse mode behavior two years later. it now appears possible to check out the validity
In late 1958 a widely publicized disagreement of various combustion mechanisms by a " direct"
occurred between Zucrow-Osborn, 774 and 114
method.11
Crocco.174 The point in question involved the Before leaving the subject of injection studies,
results of the tests just mentioned , which in- which were often directly related to the combus-
dicated a definite time lag behavior, and those tion processes, some specific topics should be
results from gas rocket studies at Purdue which cited . Included in the group were studies of spray
argued against the necessity of a time lag . Sev- fluctuation , impingement angle, liquid phase
eral years passed before the theory of Sirignano646 mixing, dynamic characteristics, drop-size dis-
and the experiments of Glassman et al.113 involving tribution , spatial characteristics, droplet break-
the wave shape and the importance of the up , etc.
Arrhenius rate function in gas-rocket-type com- The steady-state model developed at Lewis
bustors helped to clarify the differences between Research Center was followed in 1962 by a non-
the two combustion devices . linear theory using a one-dimensional model to
During the 1950's and early 1960's the AFOSR predict combustion instability limits. Priem and
Contractors Meetings provided the forum for Guentert563 were the originators of this theoretical
most liquid rocket instability discussions just as approach which is described in Chapter 4. The
today the ICRPG serves this need . One constant model uses various mechanisms from the injec-
point of discussion was whether nonlinear or tion-combustion studies just discussed . A dis-
linear theoretical analysis should be used for turbance is inserted into each model and the
the high-frequency instability problem. Torda,697 computer is used to evaluate whether stability or
then at Brooklyn Polytechnic Institute, was one instability results.
early advocate of the nonlinear approach. Priem Considering the actual rocket hardware, al-
and others later supplied the concepts applied though a relatively large thrust United States
to rocket engines. The subject of nonlinear theory liquid rocket engine was available in the early
will be discussed more fully in a moment. 1950's (i.e. , the Redstone) , the era of the large
While arguments as to the best way to proceed engine was really closer to 1960. A number of
theoretically on the instability problem were in engines in the 100 to 200 thousand -pound thrust
progress, the question remaining in the back- class had been developed by that time and the
ground was how much was really known about greater than million-pound thrust designs were
steady-state combustion. Along these lines in- being studied (F-1 and M- 1 ) . Large engine feed
formation on the factors influencing mass and system associated instability had already been
mixture ratio, injector element design, and other experienced and solved for the case of the Red-
propellant flow phenomena were being studied stone engine. The approach followed for high
by Rupe615 and others at JPL . Related studies frequency instability was that incidences could
were being conducted by Heidmann , 336 Ingebo , 375 be eliminated with controlled start sequencing
Morrell, 499 Priem559 and others at Lewis Research (to avoid triggering such modes) , and by careful
Center. These studies formed the background selection of injection patterns and types , e.g. ,
for the Priem-Heidmann propellant vaporiza- LOX/hydrocarbon injectors avoided unlike im-
tion model564 which relates the propellant proper- pingement injector designs which were more
32 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

susceptible to spontaneous resonant combustion . much attention in recent years because of the
Life with this approach was hazardous as was ability to damp transverse modes. However, the
shown by two Atlas mishaps described in Sect. design of the subsonic portion of the nozzle is also
1.2.4.1 . Thus in 1961 the Atlas booster493 became important for stability. The majority of nozzle
the first production engine to adopt a different studies have been theoretical , with Crocco ,
stability approach- use of a baffle so as to avoid Culick, Reardon , Sirignano, and Zinn the prin-
both spontaneous and triggered forms of resonant cipal contributors. The nozzle may serve as the
combustion. main damping source for longitudinal mode or
Techniques to control combustion instability may provide slight amplifications in certain
using baffles were first used in 1954 by Male and transverse mode situations . Only in the longitu-
Kerslake461 at NACA Lewis . Numerous studies dinal case have experiments been made.
at Aerojet, Rocketdyne, and Princeton also The merits of being able to determine the
helped to point out the benefits and limitations. relative stability of candidate injector designs
of the baffle approach. As explained in Sect . 8.2, by rating devices were discussed at length in
the mechanisms by which stability is accom- Sect. 1.2.4.1 and the details of the techniques
plished using the baffle are still not completely can be found in Chapter 10. These techniques
clear- indeed more than one mechanism is proba- were under development in the late 1950's pri-
bly responsible. The criticality of injection con- marily at Aerojet and Rocketdyne. Stemming
ditions near the baffle was found to be especially from such studies a 220-grain bomb was devel-
important in establishing the desired damping. oped for the Atlas MA-5 dynamic stability
A few years after the addition of the baffle to verification in 1961.493 Studies of stability rating
the Atlas engine there was strong sentiment that techniques using bombs , pulse guns and gas in-
further instability research was unnecessary jection have been conducted more recently
since the high-frequency problem had been (1965-1968) by Combs et al. at Rocketdyne
solved . This same argument was used again a as well as at other organizations . Evaluation of
few years later for the acoustic liner approach to the techniques was the prime purpose of the
instability suppression . Unfortunately, it has studies as discussed in Chapter 10. The rating
not turned out to be that simple for either device. techniques are the standard means of evaluating
to achieve universal stability and both devices. dynamic stability, for example, the Titan IIIM
have preferred applications. Stage I baffled injector was designed for dy-
*
Liner design has had the advantage of a namic stability and was proved successful from
reasonable theoretical base, particularly with the first test (1964) . Two programs that have
recent additions that treat nonlinear effects . yielded a great deal of information on the de-
However it has been found that typically the velopment problems in achieving dynamic sta-
design problem is made difficult by the actual bility were the F-1 program20 conducted at
chamber environments that are present . Experi- Rocketdyne and the GEMSIP program " at
mental studies in liner design for rocket motors Aerojet. An ad hoc committee formed for the
were initiated by Blackman and Lewis in the F- 1 problem played an important role of trans-
early 1960's at Pratt & Whitney. That work has mitting information to and from both of those
been extended by Garrison . Other experimental programs.
work has been principally conducted by Phillips To achieve dynamic stability more than damp-
at NASA Lewis, Oberg at Rocketdyne, and ing devices and rating techniques were required ;
Harrje at Princeton . The theory of resonators predictions of the stability limits were necessary.
as related to combustion applications was clari- Reardon's empirical correlations650 based upon
fied by the work of Ingard . Nonlinear theoretical the Crocco sensitive time lag theory were applied
extensions using the jet-flow model were done to the design of various engines. Modifications
by Sirignano . of the Priem theory by Dynamic Science and
Both baffles and liners have been receiving Rocketdyne proved helpful in other design ap-

* See Sect . 8.3 for individual references. † See Sect. 3.6 for individual references.
 333
INTRODUCTION $ 1.3

plications. 91,136 Another correlation approach con- criticality of the impingement processes has been
ceived in the early 60's by Dykema238 also al- studied by a new approach at JPL618 and offers
lowed stability predictions to be made on new an explanation for the popping phenomena
hardware designs . This later approach is based (Sect. 7.6) .
on droplet parameters controlling the response, The advent of the laser and the ability to
a subject which was further studied by Strahle665 overpower combustion light has encouraged re-
at Princeton . In another analysis with the em- newed study of the difficult problem of droplet
phasis on understanding the conditions conducive measurements both cold and hot . Weurker et
to instability in LOX/LH2 engines, Feiler and al. * at TRW have been the leaders in this area.
Heidmann260 formulated a response model which Use of improved paraffin droplet techniques has
has proven quite successful. also shown considerable promise in the study of
Chapters 4 and 6 supply the background and droplet distributions (Dickerson et al.224 at
application data on these approaches. In addi- Rocketdyne) .
tion to NASA's interest, AFRPL has been deeply Another area of recent interest has involved
involved in this important aspect of designing further development of damping devices . Rocket-
for stable operation through in-house testing dyne has concentrated on the use of slots in the
efforts, extension of computer programs for the injector face, often in conjunction with baffles ,
Priem approach136 and encouragement in the a design which was used for the LM ascent en-
publication of a manual on how to apply the gine.522 A similar approach involved the use, at
sensitive time lag theory.650 Bell Aerosystems, of a toroidal cavity resonator
Damping devices have not been necessary in with apertures extending into the chamber at the
all rocket motors to achieve the desired stability injector-chamber interface. This design proved
for manned flight . One example is the Agena completely successful on the PBPS axial engine.
engine with a successful history of 250 launches , Similar designs had also shown considerable
over 350 space firings and nearly 3000 ground promise in the Bell LM ascent engine develop-
tests which have demonstrated that a reliable ment . In an even more recent program, Aerojet
system can be maintained without producing and Aerospace have been performing tests on a
disturbances frequently assumed to a general relatively small number of damping devices
condition of rocket engine combustion . The vari- placed across an injector face. As few as three
able-thrust TRW LM ascent engine has met all Helmholtz resonators324 placed in the walls of
the dynamic stability criteria without requiring research chambers at Princeton had indicated a
damping devices . The merits of the later injector marked stability improvement-preferred con-
design approach together with a competitive ditions were those in which a short baffle limited
Rocketdyne module design (see Sect . 7.4.5 ) are the frequency spectrum with damping accom-
currently being tested on larger diameter hard- plished by the resonators.
ware where the stability demands are more dif- The nature of the tangential mode is still not
ficult to meet. completely clear. Tests in recent years by
Designs of recent years have extended engine Clayton15 at JPL have revealed shock-type waves
operation to higher pressures, above the critical in 11 -inch diameter hardware . Other observa-
conditions of the propellants . Experimental pro- tions in similar diameter hardware have revealed
grams have been evaluating these effects on the peaked Maslin and Moore wave predictions.
stability. Such studies have been active at Aero- Phase differences have also been shown where
jet, Rocketdyne, Penn State, Illinois Institute of the wave leads near the nozzle entrance in the
Technology, and elsewhere. JPL study whereas it leads near the injector
It has not been only the newer ranges of pres- based on the measurements of others . Explana-
sure operation that have required additional tions have been offered which involve detona-
study. The effect of reactive jets of the hyper- tion-like processes,48 however, the frequency has
golic propellants which cause the jets to deflect been shown to be limited in all cases to that
rather than mix has also received recent atten-
tion at JPL364 and Dynamic Science.777 The * See Sect. 9.4.5 for references.
34 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

predicted from acoustic theory. These wave data. ment has led a somewhat clouded history. Unlike
are necessary if the proper theoretical analysis performance, which has always received its proper
is to be applied . Wave growth studies have been share of attention , instability has often in the
made by Agosta in the past and remain a sub- past been something that was only spoken of in
ject of active interest. guarded terms certainly it was not to be ad-
In this review of the history of combustion vertised. In recent years this state of affairs has
instability the emphasis , following a brief dis- vastly improved and frank and open discussions
cussion of the earlier works, was to concentrate of current instability problems have taken place
on the high frequency modes or resonant com- and even are part of the annual ICRPG Com-
bustion. While resonant combustion is perhaps bustion Conferences .
the most dramatic way to destroy an engine A factor in determining the heartbeat of cur-
(burnout in a matter of milliseconds) problems rent research into combustion instability phenom-
in both low and intermediate frequency insta- ena is the status of engine development programs.
bility have plagued many engine development Currently liquid rocket development programs
programs. The early monograph by Crocco and are at a very low ebb ; hence the concern for the
Cheng179 served a need in clarifying the basic health of the associated instability research . Will
mechanisms associated with low frequency in- the answers be ready when the next generation
stability. An active program of research into low
. of engines are developed? This cyclic tendency
frequency phenomena has been maintained for is not unique to the rocket field nor has it failed
many years at NASA Lewis with Wenzel, Szuch, to be felt before in instability research. Just prior
Dorsch, Priem* and others participating . Studies to the advent of Sputnik (1957 era) marked one
in this area are still in progress though they are previous low point. Another hesitation occurred .
few in number. with the development and use of the baffle on
Intermediate frequency instability has had the large engines, in the early 1960's. However, re-
most uneven history . Often the receptacle for those occurring engine problems on various thrust units
unstable phenomena which could not be readily subsequently spurred even more work in these
explained by the high and low frequency theory, areas of research and development .
much uncertainty continues to exist as to the To summarize the status of the instability
importance of this type oscillation. Scala in 1956 problem as of today it is necessary to subdivide
analyzed entropy wave-type instability but, as the problem into several principal categories :
explained by Crocco in Sect . 5.3.2 , this insta- (1) the understanding of the processes funda-
bility is relatively rare in a pure form. In the past mental to liquid rockets such as atomization ,
few years, Fenwick et al. at Rocketdyne, McCor- mixing, vaporization , chemical kinetics, etc. ,
mack at Dartmouth and investigators at Prince- (2) the application of instability theory to hard-
ton have looked at the bunching effects and the ware design primarily as a guide in predicting
basic jet frequencies as responsible mechanisms stability trends, (3) the application of damping
for intermediate frequency phenomena .† Many devices such as baffles, liners, slots, etc. , which
of these jet characteristics had been observed in do not attempt to alter the combustion but
earlier investigations. The Rocketdyne analysis rather seek to absorb energy or restrict wave
was used with success in solutions of some H-1 motion, (4) the ability to detect the phenomena
development problems of buzz in 1966 . of combustion instability, and (5) the develop-
ment of techniques to rate liquid rocket engines
1.4 CURRENT STATUS for stability.
Progress in category one, the fundamentals of
Combustion instability research and develop-
liquid rocket combustion processes, has been
understandably slow. This is because of the ex-
* See Chapter 5 for individual references.
treme difficulty in attempting to observe and
† See Sect. 3.3 for individual references.
measure these physical and chemical processes
D. T. Harrje, Author, with contributions from the
ICRPG Working Group on Liquid Rocket Combustion which often take place simultaneously at high
Instability. speed and in environments that rule out conven-
 INTRODUCTION $ 1.4 35

tional instrumentation . An example of such diffi- tools to the theoretician in his battle with the
culties are the studies of unsteady droplet burning instability problem . Unfortunately, because of
rates (as discussed in Sect . 3.4.2) where direct ob- such factors, a crash program to devise the ulti-
servation is often impossible, especially if one mate theory has never been feasible-constant
wishes to closely simulate the actual environments reevaluation of the current tools available and
encountered in high injection density, high cham- how they relate current knowledge of the actual
ber pressure rocket combustors . Hence simulation rocket combustion instability requires both time
must be attempted by other means such as and patience .
scaling, substitution of propellants (for such im- Before leaving the subject of theory status,
provements as optical clarity) , laboratory evalua- mention should be made of low and intermediate
tion of only a portion of the processes involved , frequency theory as well as that pertaining to
etc. resonant combustion . Although often regarded
Perhaps it will never be possible to solve the as a problem substantially less difficult, instability
combustion instability problem by the direct in the low and intermediate ranges continues to
approach of attempting to understand in suf- appear in a surprising number of engine develop-
ficient detail the steps through which an element ment programs. Often this is due not to the lack
of propellant passes from its point of injection of theory or the inability to analyze the problem
into the combustor until it is converted into com- but rather because of the numerous factors that
bustion products. However, this basic effort can influence the occurrence of these instabilities
needs to be continued because knowledge of how and new parameters that enter the picture be-
the physical and chemical processes are influenced cause of different operating regimes . A recent
by injector element design and operating condi- example is the problem associated with pres-
tions, and how they are altered by factors such surization gases that dissolve in the propellants
as pressure and velocity perturbations is essen- causing enhancement of lower frequency oscilla-
tial to the successful application of instability tions . Further study is currently being carried
theory. Should the key be found in these funda- out to provide the proper model. Another exam-
mental studies the dividends would be tremen- ple is the spray bunching phenomenon associated
dous think what it would mean to be able to with intermediate frequency oscillations. Studies
directly predict stability from parameters such in that case have yielded important insight so
as droplet size (primarily a function of the injec- that problems of that type could be overcome .
tion orifice diameters and differential pressure) In category three, immediate solutions to the
and the unsteady burning rate factors (from a instability problem are being sought using damp-
knowledge of propellant choice, injection density, ing devices. Success has been achieved in almost
axial and transverse combustion distribution , every case . Progress to date has also been good
etc.). in understanding how these devices function,
In category two, the theories as currently used particularly in the case of liners where theoretical
act more to indicate trends and thus point the models have kept pace with the applications.
way to logically move in achieving stability. Cor- Based on usage, certainly baffle proponents need
relations using the theories as outlined in Chap- to make few apologies . Even without a good
ters 4, 5 and 6 can all point to considerable suc- theoretical base, experimentation with the con-
cesses in this regard . Through refinements in our trolling design parameters has achieved a high
knowledge via the more fundamental studies , degree of success. Were it not for the heat transfer
future theoretical models hopefully will improve penalties and the design complications, the quest
these stability correlations . We have moved a for alternate solutions would have gained little
long way from the dark days of the early fifties . impetus. Recently more effort has been directed
Incorporation of nonlinear effects, with the re- toward better explaining the fluid mechanical
sultant clarification of phenomena that pre- role played by baffles so that less empiricism will
viously were poorly understood , is one milestone be required in future applications.
along the road . Computer developments and With regard to liners and the closely related
mathematical advances have provided powerful acoustic slots, application has been rather limited
36 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

although the tests that were performed have been In those applications where chamber pressure
generally quite encouraging . As mentioned in measurements still are not feasible (e.g., certain
the previous section, limited liner concepts and flight-weight hardware) other primarily optical
slots have been successfully applied to developed techniques, often using recent advances in tech-
hardware. In most applications it is required that nology, have been used with considerable success .
the gas properties (speed of sound based on These approaches are described in Sect . 9.4.
temperature, molecular weight, and ratio of Last of the categories is the status of rating
specific heats) possess a degree of uniformity so methods to ascertain the tendency toward com-
that the theory may be successfully applied . The bustion instability of an engine. Stability rating
trend in optimization is to choose the best lo- techniques have improved over the years to a
cation for these damping devices so that designs point that rating of combustors over a wide thrust
can be minimized in size, complexity and cost. range can be achieved with a high degree of con-
Acoustic liners for ablative chambers is one ap- fidence. Problems may still exist in the very low
proach being investigated . The theory of acoustic thrust range where the rating technique may
slots, nonlinear regimes of operation , and opti- alter the normal combustion environment or pro-
mum configuration based upon fundamental flow vide a source for damping after the disturbance
considerations are typical of other areas of study. has been generated (the physical size of the bomb
The last two status categories are concerned can be the problem in the first instance, the port
with detection and ratingcombustion insta- of the pulse gun the cause in the second) . How-
bility incidences. Detection has vastly improved ever, these are the exceptions rather than the
over the years. Where there were at one time rule and the ability to rate engines for stability
problems in evaluating instability occurrences has become quite refined as discussed in connec-
at relatively low frequency, today measurements tion with Dynamic Stability (Sect. 1.2.4) and
at tens of thousands of cycles per second have Stability Rating (Chapter 10) .
been achieved, thus allowing not only better high Thus it might be said at this time that a
frequency measurements, but accurate recording good working balance has been achieved between
of shock-type wave forms as well. One of the the theoretician and experimentalist, the de-
principal factors in this success is the wide use velopment engineer and the researcher in the
of relatively small piezoelectric transducers with field of combustion instability. Each group has
ablative protection, helium or water cooling . This made a contribution toward the goal of providing
stable combustors and in the understanding of
together with tape recording advances (flat re-
the controlling factors. Not all the problems have
sponse to 80 kHz or better is now possible) has
been solved but steady progress has been made
allowed these and other miniaturized probes to
on all fronts. Future needs will require new in-
make unsteady pressure measurements that were sight ; however, the immediate problem of engine
previously impossible. Although further im- stabilization has been achieved in almost every
provements in pressure measurements will un- case. Alternate approaches have evolved so that
doubtedly take place in the future, the current important factors such as performance, chamber
status is extremely good. compatibility, and stability can exist in harmony.
CHAPTER 2

Steady-State Processes

2.1 GENERAL DESCRIPTION OF COMBUSTION that the propellants are usually injected from a
AND FLOW PROCESSES * distributing manifold ( Sect . 2.2.1 ) through orifices
into the combustion chamber in the form of liquid
In a very general sense, the character of the
jets (Sect. 2.2.2 ) . In some manner, the jets must
time-dependent perturbations that can exist in a
atomize, i.e. , break up into small droplets , and
system depends on the steady-state properties of
the droplets must vaporize. Atomization is often
the unperturbed system. Therefore, every aspect
achieved by causing two (doublet ) or three (trip-
of the steady-state combustion and flow processes
let ) streams of like ( or sometimes unlike ) liquids
in liquid propellant rocket motors is potentially
to impinge;* impinging jets produce thin liquid
relevant to combustion instability. For this reason ,
sheets or fans which disintegrate rapidly, first
it is important to have as thorough a knowledge
into ligaments and then into droplets (Sect . 2.2.3 ) .
as possible of steady-state motor processes before
Except in monopropellant systems, the reactive
undertaking a study of combustion instability.
fuel and oxidizer vapors must intermix ( Sect.
Five textbooks can be recommended as sources
2.3.3.3 ) . The mixed vapors react, and finally the
for background material : Ref. 63, Ref. 78 (es- hot product gases flow out of the combustion
pecially Ch . 7 ) , Ref. 536 (especially Ch . 26 ) ,
chamber through a choked nozzle .
Ref. 535 (especially Ch . 3 ) , and Ref. 740 ( es-
Numerous variations of this sequence of events.
pecially Ch . 11 ) .
can occur. For example, with some injector designs
A complete knowledge of the sequence of events
the liquid propellants may be partially or totally
that occurs in a steadily operating liquid propel-
mixed, atomized or vaporized before they enter
lant rocket motor has never been attained . Liquid
the combustion chamber, by contouring the in-
propellant combustion processes are quite compli-
ternal injector geometry suitably, by injecting
cated and defy precise analytical description . A
gases into injector passages in a controlled manner,
host of comprehensible subprocesses are of direct
or by other methods ( Sect . 2.3.3.1 ) . Some pro-
relevance to the overall combustion sequence as
pellants (monopropellants and bipropellant com-
shown in Fig . 2.3.3a . It is details of these sub-
ponents with monopropellant characteristics ) may
processes to which subsequent sections of this
experience appreciable condensed-phase com-
chapter are devoted . In the present section , we
bustion reactions before mixing or vaporizing ;
shall attempt to discuss how these subprocesses
in other systems liquid-phase mixing may occur
might be integrated into a description of the over-
and give rise to condensed-phase or surface re-
all process. The integration is necessarily impre-
actions. Heterogeneous reactions may also be of
cise ; except for the simplest models, it consists
importance in the absence of liquid-phase mixing,
merely of patchwork.
particularly for hypergolic propellant combi-
nations which ignite spontaneously upon contact,
2.1.1 Overall Description

As a brief introduction to the kinds of com- * Other techniques that have been used for liquid /liquid
bustion and flow processes that occur in liquid systems include showerhead, splash-plate and swirl
atomizers. Gas/liquid injectors often employ coaxial
propellant rocket motors, it can be stated first
elements where liquid in the central jet is sheared by a
surrounding annular gas. Impinging gas and liquid jets
* F. A. Williams, Author. are also used .

37
38 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

and notably at interfaces between impinging fuel H; Total (thermal + chemical) enthalpy
and oxidizer jets or fans. Thorough gas-phase added to gas from drop of kind j
mixing of fuel and oxidizer may not be achieved, per unit mass of drop vaporized
either by accident or by design (Sect . 2.5 ) , e.g. , k Specific reaction rate constant
poor mixing occurs near the chamber walls when mi Time rate of increase in mass of drop
fuel-rich streams are intentionally sprayed there of kind j
for purposes of cooling. Condensed phases may be Qi Number of drops of kind j produced
present in the equilibrium mixture of combustion per unit volume, per unit mass
products when utilization is made of propellants range, per unit time
containing either metals (which produce refrac- q Heat flux vector
tory oxides ) or substantial amounts of carbon R Gas constant per unit mass
(which itself condenses ) . There are regions in N° Universal gas constant
many motors, especially between spray fans, where Ui Diffusion velocity of species i
reaction products recirculate back to the vicinity
. VLi Velocity of drop of kind j
of the injector instead of proceeding directly to the 10i Mass rate of production of species i
nozzle, with the result that a fraction of the gases by gas-phase reactions
remains in the combustion chamber for a long time X Position of drop in chamber
(Sect. 2.3.4 ) . A number of examples may be cited nc Combustion efficiency
of chambers which operate at pressure levels above Tc Overall conversion time
the critical point of one or more of the liquid Tr Residence time
propellants ,601 under these conditions the sharp Ω;; Mass of species i added to gas by a
distinction between liquid and gas disappears and vaporizing drop of kind j per unit.
the usual descriptions of atomization and vapori- mass of drop vaporized
zation processes require modification (Sect .
Subscripts :
2.4.2.2 ) .
This list of complications , illustrating the non- i Index denoting chemical species :
systematic ways in which departures from the i = 1 , ... , N
normal combustion sequence often occur , could j Index denoting droplet kind ; j = 1 ,
be extended farther. However, it is less instructive M ( e.g. , fuel, oxidizer)
to pursue a discussion of complications than to
elaborate further on simpler models for the com- 2.1.2 Conversion Time and Residence Time
bustion sequences. The present paragraph serves
as a warning not to apply existing models uni- A helpful crutch in considering the complex
versally. sequence of combustion processes is to ascribe to
The following nomenclature pertains to Sect . the propellant combination an overall conversion
2.1 : time Te, which is the time required , from the
instant of injection, for a representative element
A, n Constants in vaporization rate ex- of propellant to produce equilibrium combustion
pression, Eq. 2.1.5-1 products. Sophisticated variants of this conversion
a, b, a, ß Constants in size distribution func- time appear in the "time-lag" descriptions of com-
tion, Eq . 2.1.6-1 bustion instability179 ( Sect . 4.2) . In discussions of
B Pre-exponential (frequency) factor steady-state motor operation , it is relevant to
in reaction rate compare 7. with the residence time 7, of a repre-
DI First Damköhler number sentative element of propellant in the chamber.
Eact Activation energy Obviously the optimum motor design is achieved
Fi Drag force per unit mass on drop of when the first Damköhler number213 DI = Tr/Te is
kind j unity, since for Dr < 1 the residence time is too
G (TL) Drop size distribution function short for the conversion processes to be completed
gi Number of drops of kind j per unit in the chamber, whereas for Di >1 the excessive
volume chamber volume may cause flow friction and
 STEADY-STATE PROCESSES $2.1 39

weight penalties. The description of the steady- injector design , etc.; they typically lie in the range
state combustion sequence, which arises from the 10 to 70 inches.
Tc
Te and T. concepts, is ill-defined and physically un-
satisfying because nothing is said about the history 2.1.4 Gas-Phase Processes
of events that occur during the interval re, thereby
making it impossible, in general, to compute Additional hypotheses concerning the nature of
either T. or 7, in a simple way. One must introduce the gas-phase conversion processes are needed for
simplifying hypotheses concerning processes that obtaining a correspondingly simple expression for
Te. If it is assumed (in addition to instantaneous
occur during 7. for these concepts to become use-
vaporization) that gas-phase mixing processes are
ful . Some simplified models lend themselves natur-
ally to description in terms of Te, while others do also rapid, so that the gases in the chamber are
not. essentially premixed combustibles , then 7. is de-
termined by the chemical kinetics (both mecha-
2.1.3 Characteristic Length and Characteristic nisms and rates) of the gas-phase reactions . For
Velocity example, approximating the gas-phase reactions
as a one-step, first-order process with specific.
One useful simplifying assumption is that in- reaction rate constant
jected propellants vaporize very rapidly, since
then both Tc
7 and 7, can be investigated by con- k= exp ( - Eact/R°Te) (2.1.4-1 )
sidering only gas-phase processes . Under this con-
(where B is the frequency factor, Eact is the acti-
dition , if the material in the combustion chamber
vation energy and ° is the universal gas con-
is approximated as an ideal gas, with gas constant stant) , one finds by solving the elementary first-
R and specific heat ratio y, at stagnation con- order chemical kinetic equation that the ratio of
ditions, then 7, is simply related to the chamber
the mass of reactants present at time t to the
temperature Te and to the characteristic length L*
mass of reactants present at the time of injection
of the motor (the ratio of the chamber volume to
(t = 0 ) , a ratio which clearly equals the com-
the throat area) . This correspondence can be seen
. bustion efficiency ne of a chamber with residence
by first noting that 7, is the ratio of the mass time t, is given by
contained in the chamber to the mass per second
flowing out the nozzle, then using the well-known nc = 1 - e-kt (2.1.4-2 )
isentropic formula relating the mass flow through
Strictly speaking, this result yields an infinite
a choked nozzle to stagnation conditions ; the chemical conversion time 7. ( a property shared by
result is78
many widely differing models, which makes it
Y+ 1Y+1/2(y-1) more appropriate to discuss the results of such
Tr = L * (YNT.) - 1/2
2 models in terms of 7. rather than L* ) , but com-
bustion will be reasonably complete (e.g. , nc =
L* (~ + 1 \ r+1/7−1 0.99) in a time that is expressible in terms of the
-1/2 (2+1) (2.1.3-1 ) parameters B,་ Eact and Te ; e.g. ,

Te≈4.61B1 exp (Eact/ R°Te) ( 2.1.4-3 )

where the characteristic velocity c * (the product
of chamber pressure and throat area divided by since In (1-0.99 ) = - 4.61 . This result provides
the mass flow rate ) has been introduced in the last an illustration of how, with a sufficient number
expression . This result for 7, enables one to in- of assumptions, the conversion time can be related
terpret the Damköhler criterion for optimum . to parameters of a more fundamental nature .
motor design in terms of an optimum character- Much more can be said about gas-phase chemi-
istic length L*. Irrespective of the validity of the cal kinetics, about gas-phase mixing and about
assumption of short condensed -phase residence models that disregard the presence of condensed
times, it may be stated that experimentally, opti- phases in the chamber. The best description of
mum values for L* are observed to exist and to premixed gas-phase kinetics depends intimately on
vary appreciably with propellant combination , the propellant combination ; for some propellants
40 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

(e.g. , hydrogen-oxygen ) the kinetics are under- the ratio of the condensed-phase to gas-phase 7.
stood well, 122 while for others practically no infor- contributions to increase as the chamber pressure
mation exists . It generally is found that the gas- is increased, and it also provides a basis for intro-
phase reaction time decreases when either the ducing a "pressure-sensitive time lag" into de-
chamber temperature or the chamber pressure is scriptions of combustion instability179 (Sect . 4.2 ) .
increased. Since this division of the conversion time is rather
Descriptions of gas-phase mixing processes imprecise, it is instructive to consider condensed-
(Sect. 2.3.3.3 ) depend strongly on chamber shape phase models which are more explicit, less phe-
and on injection -pattern geometry ; relevant mo- nomenological and more mechanistic .
lecular transport properties are typically known Aside from descriptions dealing with homo-
more accurately than chemical kinetic parameters, geneous condensed-phase chemical kinetics, the
but turbulent mixing processes, which generally simplest model that focuses attention on the con-
occur in motors, are difficult to describe prop- densed phase was stated first by Probert.568 It is
erly 536,740 Mixing-process models for rocket com- assumed that the liquid propellant jets break up
bustion have not received much attention and instantaneously into droplets and that gases
deserve to be studied further. evolved from droplets by vaporization , mix and
One-dimensional models for premixed gas flow react instantaneously to produce equilibrium com-
in constant-area motors, accounting for nonzero bustion products. The droplets are assumed to
Mach numbers in the chamber through relevant move at a velocity VL in the direction of the
mass, momentum and energy conservation equa- chamber axis, and the rate of vaporization of
tions, can be formulated in terms of the algebraic these droplets is taken to control the conversion
equations of "diabatic" flow (constant-area flow time. Thus, the model is essentially one of spray
with heat addition ) . Such models can be extended combustion or spray evaporation .
to variable-area chambers and in particular to A number of different choices are possible for
throatless chambers which produce supersonic flow describing the spray vaporization processes . For
through thermal choking (e.g. , purely diverging monopropellant systems it may be assumed that
reactors ) ; the description then involves obtaining the chemical heat release occurs in the spherically
solutions to ordinary differential equations.289,748 symmetrical gas flow region near each droplet.
Instead of dwelling further on these and other Then monopropellant droplet combustion theories
elaborations of descriptions of gas-phase processes, (Sect . 2.4.3 ) should be used for describing the
we shall proceed to discuss models that consider vaporization rates . If it is assumed that the
condensed phases . One reason for doing so is that chemical heat release occurs in gas regions far
for most rocket motors the conversion time Tc is removed from each droplet, in that case simple
appreciably longer than would be expected on the vaporization theories (Sect . 2.4.1 ) should be used
basis of gas-phase chemical kinetics alone ; thus , for describing the vaporization rates .
if gas-phase processes are dominant , mixing must In bipropellant systems ( Sect . 2.4.2 ) there is a
be of importance . great variety of limiting cases. If one of the
propellants is much more volatile than the other,
2.1.5 Condensed - Phase and Gasification
then it is reasonable to assume that the volatile
Processes
propellant vaporizes instantaneously after injec-
A straightforward extension of the preceding tion and that the spray model refers to vapori-
ideas is to assume that 7. is the sum of two terms , zation of the less volatile constituent. In bipro-
a time lag associated with condensed-phase proc- pellants utilizing liquid oxygen and hydrocarbon
esses and a time lag associated with gas-phase fuels , it can usually be assumed that the oxidizer
processes.78 It can be argued approximately and vaporizes instantaneously, while for hydrogen-
qualitatively that the rates of the condensed-phase oxygen systems the hydrogen can often be as-
processes are likely to be relatively unaffected by sumed to be gaseous at the injector exit.440 When
the chamber pressure, whereas the rates of the one of the propellants vaporizes quickly, the
gas-phase processes are pressure dependent, gener- vaporization rate of the less volatile constituent
ally increasing as pressure increases. This causes can be described on the basis of the theory of
 STEADY-STATE PROCESSES §2.1 41

burning of a fuel droplet in an oxidizing atmos- G (r ) bri exp ( -ar ") (2.1.6-1 )
phere (or of an oxidizing droplet in a fuel atmos-
where a, b, a and B are constants. Other functional
phere, as the case may be) ( Sect . 2.4.2 ) . Alterna-
forms have also been used (Sect. 2.2.4) .
tively, if chemical reaction times are long com-
A partial differential equation with independent
pared with the time for an element of vapor to
variables r and t (or ri and the axial coordinate
diffuse away from the vicinity of a droplet (but
x ) can be written for the time ( or space ) evo-
still short compared with the time for the droplet
lution of the distribution function G. This equa-
to vaporize) , the droplet vaporization rates can
tion has been referred to as the spray equation .
be described on the basis of the theory of non-
If the values of the constants a, b, a and ẞ are
reactive droplet vaporization (Sect . 2.4.1 ) . If both
.
assumed to be known from measured atomizer
propellants are of comparable volatility, then
characteristics, and if the quantities VL, A, and n
spray descriptions can be developed for which
are treated as known constants, then the spray
condensed phases of two different chemical types
equation can be solved for the spatial development
are taken into account.740 Droplet vaporization
of G in the combustion chamber . From this so-
during the heat-up period can also be analyzed.
lution , one can calculate ne for a chamber of a
Whatever description is adopted for the vapori-
specified length, thereby obtaining a rather de-
zation process , analysis of this process yields an
tailed description of the combustion processes
expression for the time rate of change of droplet
implied by the spray combustion model. It is
radius r , which can usually be approximated
found that, for a given chamber length, large
reasonably well* by an equation of the form740
vaporization rates ( large A) , small droplets ( small
drL гL) , small injection velocities ( small VL ) and spray
= _ - ArL-n (2.1.5-1 )
dt distributions that are as nearly monodisperse as
possible, all favor high combustion efficiency.740
where the value of A which is independent of гL, This relatively simple type of spray combustion
depends on local gas-phase and droplet properties model is amenable to improvement by taking into
and where the constant n generally lies in the account the effect of the spray on the gas phase .
range 0 <n < 1. It is not proper to specify VL and A in advance ;
instead, these quantities will vary with axial dis-
2.1.6 Spray Combustion tance in a manner dictated by mass , momentum
and energy conservation for the two-phase system .
In order to complete the description of a spray in one-dimensional flow. We shall not delve now
combustion model , one might assume for sim- into the details or results of the improved analyses,
plicity that at any given axial position in the except to say that it becomes desirable to account
chamber all droplets are of the same size (i.e., for differences between droplet and gas velocities ,
that the spray is monodisperse) . It is obviously so that droplet drag coefficients become relevant
more realistic to account for differences in the
parameters . † Studies of the coupling between the
diameters of the various droplets, but the system gas phase and the spray have led to a considerable
is so complex that this can be done only statisti- amount of information on propellant vaporization
cally. Thus, it is convenient in such models to
as a design criterion for rocket-engine combustion
introduce a droplet size distribution function. chambers. The primary fallacy in these descrip-
G (r ) , such that G (r ) dr, is the probable number tions probably is the assumption of one dimen-
of droplets with radius between гL and r + dri . sionality; if this assumption is deleted , then it
A four-parameter functional form for G (r ) , ca- becomes exceedingly difficult to develop a compre-
pable of correlating experimentally measured size hensible, detailed description of spray combustion .
distributions, is743 A few calculations have been made for sophisti-

* This relationship, as usually derived , does not consider † See Refs. 53, 535, 656 , 660 and 745 and see Sects.
forced convection or droplet stripping. However, these 2.3.3.2 and 2.4.1 for drag coefficients.
processes can be correlated approximately by the same See Refs. 117, 132, 343 , 369, 373 , 420, 557, 558, 559,
equation with a suitable relative velocity dependence in A. 564, 565, 651 , 738 and Sect . 7.2.4.
42 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

cated spray combustion models in which properties largest droplets to those expected for the gas ."420
are not invariant in planes normal to the motor The motion pictures show axial striations of
axis. * The choice of the transverse variations is luminosity extending downstream from the like-
based on the droplet distributions expected to be on-like doublet fans, sometimes beyond the throat
produced by atomizers that have particular rela- of the nozzle.440 They indicate that intense heat
tive geometrical positions, consistent with some release occurs where the fuel and oxidizer fans
simple injector designs . Cold- flow measurements first meet and that the more volatile propellant
of spray characteristics can help in providing up- often vaporizes quickly, producing a downstream
stream boundary conditions and in extending the region in which droplets of the less volatile pro-
calculations to include a wider class of injectors . 356 pellant (e.g. , hydrocarbon fuel in hydrocarbon-
Electronic digital computers are used in the calcu- liquid oxygen systems ) burn in the gaseous atmos-
lations, and it is difficult to draw generally valid phere provided by the volatile constituent. Down-
conclusions from the results . Faced with such stream intensity peaks coincide with the axes of
complexities in attempts to develop improved the fans of the less volatile propellant, thereby
theoretical descriptions of the complete com- providing support for the use of droplet com-
bustion sequence, one is motivated to seek guid- bustion models in downstream regions . It was
ance from laboratory experiments on rocket inferred qualitatively from these studies that the
motors. upstream jet-impingement and fan-intersection
regions are of critical importance to phenomena
2.1.7 Experimental Observations † of combustion instability.440 Processes occurring
in these regions are certainly the most complex
The most revealing laboratory experiments that and the least understood of the processes occurring
have been reported consist of observations made.
anywhere in the chamber. Thus, it is likely to be
by Levine and coworkers on two-dimensional
quite difficult to develop adequate analytical
motors with transparent walls . 420, 421 , 424, 440 Both
descriptions of the steady-state combustion and
streak photography and high-speed cinema- flow processes that are most relevant to com-
tography have been employed . The natural lumi- bustion instability.
nosity of the combustion process has been ob-
served, and strong backlighting has been used in
2.1.8 Elaboration on Description of
an effort to make condensed phases visible. Many
Spray-Combustion Models
measurements were made with liquid oxygen as
the oxidizer and alcohol as the fuel, since this In spite of this unfavorable outlook, analytical
combination facilitates observation in some re- work is continuing on the development of steady-
spects. However, a number of observations were state spray-combustion models, with the specific
reported on liquid oxygen and hydrocarbon fuels objective of obtaining models that will be useful
such as kerosene , on liquid oxygen and gaseous or in calculations of combustion instability (Sects .
liquid hydrogen, and on storable hypergolic combi- 4.3 and 7.2.4) . It is therefore of interest here to
nations such as nitrogen tetroxide and unsym- look more closely at the bases of these models and
metrical dimethyl hydrazine. at how subsequent material in Chapter 2 bears on
These experiments show highly heterogeneous their development .
conditions in the combustion chamber and do not Models for steady-state combustion of a spray
appear to conform to the assumption of one- generally begin with the assumption that the
dimensional flow. The streak photographs are spray is a dilute collection of spherical droplets .
interpretable in terms of the presence of droplets The term "dilute" means that the volume occu-
of differing sizes moving with differing velocities ; pied by condensed material can be neglected in
slopes of streak traces can be correlated with comparison with the volume occupied by gas- an
velocities ranging from those expected for the excellent approximation except possibly in the
immediate vicinity of the injector . In hypothe-
* See Refs. 165 , 166 , 356 , 421 and Sects. 2.3 and 2.5. sizing that droplets are spherical, consideration of
† See also Sects . 2.3.3.2, 2.3.3.3 and 2.3.4 . the injection process and of most atomization
 STEADY-STATE PROCESSES $2.1 43

processes is ruled out. Thus the information in Here ρ is the gas density, V is the gas velocity,
Sects. 2.2.1 , 2.2.2 and 2.2.3 affects spray com- Y is the mass fraction of chemical species i in
bustion theories only through upstream boundary the gas, U is the diffusion velocity for species i ,
conditions.* An exception to this statement is w₁ is the mass rate of production of species i in
that atomization through aerodynamic shattering the gas phase by homogeneous chemical reactions,
and atomization or coalescence through inter- and there are N different gaseous species in the
droplet collisions, can sometimes be included as system. The quantity ;; is defined as the mass of
break-up or coalescence criteria, by assuming that chemical species i added to the gas by a vaporizing
the interaction time is short enough for the conse- droplet of kind j per unit mass of droplet vapor-
quent droplet deformations to be negligible ; thus, ized . The introduction of ;; affords the option
material in Sect . 2.2.3 sometimes appears directly of either considering simple vaporization (in which
in spray combustion models ( Sect . 4.3.1.3 ) . case, for the example of a fuel droplet, ;; would
If VL; denotes the velocity of a j droplet of be unity when i denotes fuel and zero otherwise )
mass m, then counting droplets of kind j leads to followed by homogeneous combustion , which is
the steady-state spray equation accounted for through w , or hypothesizing the
a existence of a diffusion flame in the thin boundary
· layer surrounding the droplet (in which case, for
( m;g; ) +▼ • (VL¡g ; ) = Qj
am
the example of a fuel droplet, i ; would be zero
j = 1 , ... , M (2.1.8-1 ) when i denotes fuel, negative when i denotes
oxidizer, and positive when i denotes reaction
Here the size distribution function g; (m , x) , de-
products, with the values of ;; for oxidizer and
fined as the number of droplets of kind j per unit
products determined by the stoichiometry of the
volume at position x per unit range of mass about
diffusion flame ) . The material in Sect. 2.4 is rele-
m, possesses an m dependence that can be de-
vant to the determination of Nij.
scribed by formulas inferred from either Eq.
Overall mass conservation for the gas can be
(2.1.6-1 ) or equations appearing in Sect . 2.2.4
obtained by summing Eq. (2.1.8-2 ) over all
(which contains additional information on distri-
species N. Since conservation of mass in the
bution functions ) . The quantity m; (m , x ) is the
gasification process implies
time rate of increase in mass of a droplet of kind j
N
and mass m at position x ; its approximate func-
ΣΩ;; = 1 j = 1 , ...
. ) M
tional dependence on m can be inferred easily i= 1
from Eq. (2.1.5-1 ) . A more elaborate form indi-
cating the dependence on droplet and gas prop- since conservation of mass in homogeneous chemi-
erties is given in full in Sect . 2.4 . The source term cal reactions implies
Q; represents the number of droplets of kind j per
unit volume per unit range of m produced per Σω ; = 0
i= 1
unit time by shattering, droplet collisions, etc .;
the information in Sect . 2.2.3 must be used in and since the definition of diffusion velocities
obtaining Q. There are M different kinds of implies
droplets (fuel, oxidizer ) in the system . N
Conservation of chemical species i in the gas Y₁U₁ = 0
can be expressed, for steady flow of a dilute i=1
spray, as N
M 00 we find (using Y = 1 ) that
i=1
V• [p (V + U₁) Y; ] = w ; − Σ [ ® m‚ª‚g ;dm
j=1 0
M
i = 1 , ... , N (2.1.8-2 ) ▼ • ( pV ) ==
= Σmg,dm ( 2.1.8-3 )
j1
* Aid in establishing upstream boundary conditions
can be gleaned from Sects. 2.3.1 , 2.3.2 , 2.3.3.1 and 2.3.4 The derivation demonstrates that Eq. ( 2.1.8-3 )
as well as 2.2.1 , 2.2.2 and 2.2.3. is contained in the set ( 2.1.8-2 ) .
44 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

To derive an equation for conservation of mo- quantity H; is defined as total (thermal plus
mentum of the gas, it is reasonable for dilute sprays chemical ) enthalpy added to the gas from a droplet
in rocket motors to assume that forces experienced of kind j per unit mass of droplet vaporized . Its
by droplets consist solely of aerodynamic forces value depends on processes occurring in the bound-
(skin-friction and separation drag ) exerted by the ary layer adjacent to a droplet and can be inferred
gas. If F; ( m , x ) denotes the drag force per unit from discussion given in Sect. 2.4. Applying energy
mass (i.e., acceleration ) exerted on a droplet of conservation across the boundary layer, one finds
kind j by the gas , then the steady-state mo- that H; is the total enthalpy leaving the surface
mentum conservation equation for the gas can be of a droplet of kind j per unit mass vaporized and
reasoned to be therefore energy conservation for the droplet can
M be expressed as
-- - mFig¡dm
pV.VV = −Vp — V • ≈− Σ [
j=1 °
0 a
m; (mh; ) + VL ; • V (mh ; ) = ṁ¡H¡
am
M ∞
-Σm; (VL - V) g , dm ( 2.1.8-4 ) j = 1 , ... , M (2.1.8-7 )
j- 10
where p is the hydrostatic pressure, T is the where h; (m , x) is the total (thermal plus chemi-
cal ) enthalpy per unit mass for a droplet of kind j .
viscous stress tensor of the gas , and the last term
Equations ( 2.1.8-1 ) through (2.1.8-7) can be
accounts for the momentum carried to the gas by
the material that vaporizes from the droplets . viewed as 5M +N +4 equations in the 5M +N + 4
unknowns g , (pYi) , V, VL , h and hj. When
Information appearing in Sects. 2.3.3.2 and 2.4.1
expressions for gas-phase transport fluxes*
is useful for obtaining F; in terms of droplet and
gas properties . The steady-state motion of the (Ui , T, q) , homogeneous reaction rates† (w ) ,
droplets is described by the equations thermodynamic properties ( p , ... ) and droplet
behavior ( m;, Qi , Nij , Fj , H;j as obtained from
a
m; Sects. 2.2.3 , 2.3.3.2 and 2.4 ) are appended to the
· VLj + VLj • VVL¡ = Fj
am set, and when upstream spray distribution func-
tions and a sufficient number of other upstream
j = 1 , ... , M ( 2.1.8-5 )
boundary conditions and boundary conditions at
in which the first term accounts for the possibility the chamber side-walls are known , then in princi-
that the droplet velocity is size dependent . ple a complete description of the spray combustion
In terms of the total (thermal plus chemical ) process can be obtained from these equations . To
enthalpy per unit mass for the gas h and the develop such solutions is the fundamental ob-
gas-phase heat flux vector q, which includes jective of steady-state theories for spray com-
energy transfer by heat conduction , diffusion and
. bustion .
radiation , the steady-state equation for conser- Since it is clear that the system of equations is
vation of energy for the gas can be written as complicated, solutions have been obtained only
by introducing further simplifications. Except for
▼ • [µV (h + V²/2) ]
the studies mentioned at the end of Sect. 2.1.6 ,
= V· q - V • ( T.V) the first simplifying assumption always has been
M 00 one-dimensional flow. This reduces the number of
- Σ m (F¡ • VL¡ ) g¡dm independent variables from four (m and x ) to
two (m and x) . Next , unless one introduces highly
M 00 simplifying assumptions concerning the fluid me-
- Σ [* m; ( H; + V² ;/2 ) g;dm (2.1.8–6 ) chanics ,740 the usual practice has been to eliminate
j=1 0
m as a variable by considering only monodisperse
where V2 V.V and VL2 = VL; VLj. The last two
terms account for the work done on the gas by
* These quantities generally have been neglected en-
the droplets and the energy added to the gas by tirely since typically they are small in the core flow.
the material vaporizing from the droplets . The † Only occasionally have these been included.
 STEADY- STATE PROCESSES $2.2 45

sprays,,656,660,738 by introducing a small number ( 2 The applicability of the results to real rocket
to 5 ) of discrete droplet sizes ,558, 559, 564, 565 or by motors should always be judged through com-
accounting for a size distribution in a somewhat parison with quantitative and qualitative obser-
ad hoc manner.745 The result is a set of ordinary vations such as those reported in Sects. 2.3 and
differential equations in x which at least can be 2.5.
programmed for computer solution.
After reducing the system to ordinary differ- 2.2 INJECTION AND ATOMIZATION
ential equations, investigators have adopted two 2.2.1 Manifold Flow *
paths to obtain solutions. One group has simplified
further until it becomes possible to obtain ana- In all but the very simplest of injection schemes,
lytical solutions.53,656,660, 740,745 Another group has such as a single element , it is necessary (or con-
programmed the system for solution by digital venient ) to distribute the propellant to the in-
computer. * The first approach leads rapidly to jection orifices by means of a manifold. In most
some general conclusions such as those cited in current schemes this manifold consists of a number
Sect . 2.1.6 . The second approach affords the possi- of connected passages fed by the single outlet of
bility of simultaneously including a greater num- a shutoff valve, and terminating in the orifice
ber of phenomena. The second approach has pro- array.
duced a number of rather general conclusions for Consider first the case where the injector con-
specific propellant combinations. These include421 sists of a large number of relatively small orifices.
a. A uniformly-distributed, one-dimensional The mass and mixture ratio distributions (Sects.
treatment of the injection region is un- 2.3.1 and 2.3.2) may then be characterized on a
realistic . scale defined by the flow from the individual
b. Calculation of droplet acceleration in a orifices, assuming the interaction of jets and sprays
high-velocity, combustion gas stream must is unimportant for this situation . Here the mani-
include the influence of droplet deformation fold design , based on distribution capability , plays
on the aerodynamic drag coefficient . a dominant role in determining the essential fea-
C. The effects of forced convection on heat tures of the early reaction region in the chamber.
and mass transfer processes must not be If the manifolding results in a poor distribution of
neglected . propellants ; performance, chamber capability and
d. Droplet breakup processes must be con- even stability can be adversely affected . For such
sidered, particularly if combustion chamber situations it is essential that the overall discharge
length for attaining high efficiency is to be coefficient from valve exit to orifice exit be
computed. matched (or controlled in a specified way if the
e. The combustion rates of monodisperse discharge from individual orifices is not intended
sprays compare well with the combustion to be uniform) so as to provide the desired local
flow.
rates of corresponding polydisperse sprays.
If the largest droplets' velocities are of As the scale of the element increases, the mass
interest, however, the monodisperse spray and mixture ratio distributions within individual
elements are detectable in the overall combustion
model does not supply that information
and cannot be used. process. Under these circumstances, the inter-
f. Both fuel and oxidizer sprays should be action between manifold flows and jet properties
considered, particularly if details in the must be taken into account . The turbulence within

injection region are to be computed . the manifold and, in particular, the local velocities
g. The changes of chamber pressure with com- in the vicinity of the orifice entrance (e.g., the
bustion progress and gas acceleration are so-called cross-velocity ) , can directly influence the
easily incorporated in a model for machine stability (both spatial and temporal ) , contiguous-
solution, and should not be neglected .
* J. H. Rupe, Author.
* Refs. include 132, 420, 557, 558, 559, 564, 565 and Note that changes in combustion length can mask
738. such effects.
46 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

ness, and symmetry of the effluent jets.705 As these elements . However, such analyses tend to de-
jet properties are degraded by adverse conditions generate rapidly through a series of compromises
within the manifold , the mass and mixture ratio that are required to facilitate manifold fabrication .
distributions associated with each element become Thus, the manifold design is dominated by em-
ill-defined, unsteady, and nonreproducible. This piricism with consideration given to minimizing
results in the initial conditions for the so -called pressure drop, cross-velocities, and dead spots .
steady-state combustion process also becoming The problem has received only superficial docu-
indeterminate. Thus, it is seen that in addition mentation in the open literature (for example ,
to the process of propellant distribution , the mani- Ref. 705).
fold also has a role in determining the flow prop-
2.2.2 Jet Properties *
erties within the orifice itself and at the orifice
exit. Although the orifice configuration613,616 (Sect . Insofar as liquid propellant rocket injectors are
2.2.2) can be utilized in certain instances to relax concerned, the term "jet " refers to the effluent
such manifold requirements as low cross-velocity flow from an injector orifice . In most cases of
at the orifice inlet, the available compromises practical interest the jet is in the liquid phase†
(e.g., an orifice more than 10 diameters long ) are and is conceived as a steady, continuous stream
often unacceptable since they are believed by some that is aligned along a particular reference line
designers to introduce insurmountable difficulties relative to the exit orifice geometry, and exhibits
in fabrication . a marked density discontinuity with its surround-
In addition to the requirements that the mani- ings . Although many geometrical variations are
fold provide the essential control of mass distri- possible, the one receiving wide acceptance is the
bution -either uniquely when jet properties are "cylindrical" jet . This jet is characterized by an
inconsequential or in conjunction with the orifice. axis of symmetry , a free boundary, and a velocity
geometry when jet properties are a factor- it is profile within the flow. The extent to which these
also necessary to consider the absolute magnitude conceptual qualities are realized is to a large
of the pressure loss sustained in the manifold . measure dependent upon the extent to which the
itself and the absolute volumes contained within flow within the cylindrical portion of the orifice
the flow passages . These are not free variables can eliminate the deleterious effects introduced
since they are, in general, elements of a flow by the hydrodynamics of the manifold flow and
system and must be considered from that view- the transition acceleration from the manifold to
point . Unfortunately, specific guidelines for mani- flow within the orifice boundaries .
fold design have not as yet been elucidated ; how- The orifice configurations that are most common
ever, it seems safe to assume that the objectives in rocket injectors are quite similar to those that
of small volumes and low pressure losses are both have been utilized for many years in control
desirable but often incompatible. Under steady- and /or measurement devices.464 This family of
state conditions, manifold volume has little sig- orifice configurations is illustrated in Fig. 2.2.2a
nificance and hence can be increased to produce. and the essential features of each are as follows :
relatively quiescent conditions at the orifice entry (a) represents the conventional sharp edged orifice,
and thus maximize jet stability. Conversely, the (b) is a rounded or contour approach orifice ,
problems associated with flow transients at engine (c ) is a higher LD, square-edged orifice, which
startup and shutdown , and propellant holdup in produces a cylindrical jet as in (a ) and ( b ) ,
the manifolds are aggravated as the volumes while (d ) is that same orifice where the jet has
increase.
The manifold problem can be treated to a first * J. H. Rupe , Author.
order analytic approximation by calculation of † A notable exception is hydrogen, which in most cases
the equivalent hydraulic resistances of each pas- is injected into an environment above its critical pressure
sage. This basic concept is illustrated in most so that the only possible distinction between liquid or
gas is based on whether or not the temperature is less
texts on hydraulics (see Refs . 234, 621 , 709 ) .
than or greater than the critical temperature . Other
This approach is clearly an iterative procedure exceptions are found in the high chamber pressure designs
that must account for the interaction of all flow receiving attention currently.
 STEADY -STATE PROCESSES $2.2 47

ature, cross currents and resultant contraction the

discharge coefficient may be as low as 0.90.464
In the case of the venturi orifice ( e ) both
divergence angle and length affect the ability of
— (a )
the orifice to flow full.464 Chamber conditions are
also very important in determining flow in the
divergent section ; especially when this orifice is
operated as a cavitating venturi181 to limit com-
munication from chamber to feed system. The
- ( b)
discharge coefficient is therefore also easily
altered . 464
It is interesting to note that modern rocket
injector designs are still dominated by the con-
—-— ( c ) figuration exemplified by (c ) and (d ) and hence
are characterized by the flow properties just de-
scribed . The transition between flow condition
(c ) and (d ) is known as " hydraulic flip" and is
sensitive to injection pressure, orifice length/di-
ameter ratio , orifice entrance configuration , and
-(d )
the fluid properties .
In an effort to delineate bounds on operating
conditions for which hydraulic flip would occur ,
Northrup studied the phenomenon in 1951.519 This
work was extended by Wright761 who attempted to
- (e )
determine the effect of cross -velocities on the
separation conditions. In some very recent
studies383 these experiments were verified and an
analytic model was developed as a tool to aid in
FIGURE 2.2.2a.-Jets emanating from several typical predicting the conditions for which separation
orifice configurations.
might occur. The interrelated effects of orifice
L/D ratio , manifold cross -velocities and pressure
drop on the discharge coefficient of typical orifices.
reattached to the orifice wall, and finally (e)
(and of the hysteresis effect exhibited ) are illus-
represents a venturi orifice .
trated in Figure 2.2.2b from the work of
For the sharp-edged orifice flow conditions of
Northrup.3,519 , 318
(a ) and (c ) the discharge coefficient Ca lies be-
Clearly this concern with dual flow configur-
tween 0.6 and 0.7 . * In ( d ) , the flow reattaches to
ations is related to the generally recognized need
the wall allowing the orifice to flow "full." The to control discharge coefficients and maintain the
result is agitated and divergent flow at the exit.
stability and direction of the jet configuration as
(bushy or broomy flow) , and the discharge coeffi- well. This is critical in those instances where the
cient rises to between 0.8 and 0.85. A still higher
secondary flows become important, as exemplified
Ca is achieved with (b ) when the contour is not
in the sprays of impinging jets .
too abrupt and L/D is small ( ≈0.25 to 0.5 ) , i.e. ,
Correlations relating spray properties to the
for a good, smooth, rounded orifice Ca≈0.97 . For
degradation of jet properties have not yet been
the best designs Ca = 0.99, while with poor curv-
devised . However some insight into the magnitude
of these effects is given in Ref. 617 where it is
* A procedure for determining the discharge coefficients shown that mass distributions in sprays of "well-
for "non-cavitating" flows through short, sharp-edged
controlled" impinging doublets exhibit gross vari-
orifices has been developed by Hall, 304 who showed reason-
able agreement with experimental measurements for that ations when the velocity profiles are not identical.
particular flow regime . Similar illustrative experiments are described in
 48 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

x- Decreasing flow o-Increasing flow

.9
0.052 inch dia . hole 0.040inch dia . hole
L/D ratio = 3 L/D ratio =1
.8

Cross-velocity Cross-velocity
.7 O ft/sec Oft/sec

.6
6.
ficient

.8

5 5
,Cd w

.7
coef
Flo

9
6
∞

10 10
.7

.6
.9

.8

20 20
.7

.6
0 20 40 60 80 100 0 20 40 60 80 100

Pressure drop , psi

FIGURE 2.2.2b . -Typical variation of flow coefficient vs pressure drop at different cross-velocities for orifices with and
without flow discontinuities.

Ref. 288 where truly identical jets whose center- The detailed properties of a family of jet flows
lines were misaligned in a predetermined fashion have been presented in some detail in Ref. 616 .
were utilized to form a series of sprays and the However, these studies failed to generalize on the
resulting mass and mixture ratio distributions relationship between orifice configuration and
were determined. When taken together the results manifold disturbances to the resulting jet prop-
show that relatively small changes in direction erties. Thus, although definitive information on
and/or momentum (or its distribution within the steady jets formed by particular orifice geometries
jet ) will produce marked changes in the properties is available, the coupling between jet properties
of the sprays that are formed . Thus, it must be (as determined by injector design ) and the conse-
concluded that control of mass and mixture ratio quent steady-state combustion environment has
distributions on a scale that is determined by the yet to be elucidated .
discrete properties associated with a given element Although this chapter is primarily concerned
must stem from adequate control of jet properties. with the steady-state properties of the injec-
 STEADY-STATE PROCESSES $2.2 49

tion/combustion process, it should be clear from For proper combustion it is necessary to atomize
the above that the attainment of truly steady jets the propellants and thereby provide a great in-
is dependent upon conformance to at least one of crease in liquid surface area to accelerate burning.
two essential criteria ; i.e., "quiescent manifolds" Hopefully, in this process the appropriate distri-
or "near fully-developed-flows at orifice exits." bution of the propellants will be maintained so as
Since common practice does not satisfy either of to achieve the desired mixing. Atomization in-
these criteria, then it must also be concluded that volves the breakup of liquid jets and/or sheets .
in most instances the early combustion region is In this section the processes involved in liquid
only quasi-steady and it is this quasi-steady en- stream and drop breakup are discussed with a
vironment that is perturbed by and ultimately brief review presented of the analytical treatments
coupled to the combustion chamber dynamics. available to describe the simpler breakup mecha-
The fact that variations in combustion chamber . nisms .
conditions, ( e.g. , pressure , vibration , etc. ) can Several breakup mechanisms may be important
perturb these quasi-steady injection properties in in the atomization process with the mechanism
a significant manner, not necessarily related to dependent upon the type of injector . A highly
combustion or combustion instability, can intro- idealized version of the atomization process may
duce considerable difficulty in correlating unsteady be described as follows : the liquid is ejected from
"combustion" observations . an orifice as a jet, as a sheet , or as two or more
jets which may impinge to form a sheet . Dis-

2.2.3 Mechanisms of Atomization * turbances are present in the liquid which deform
the surface. Typical disturbances include jet tur-
bulence , gas bubble formation , injector orifice
The following nomenclature pertains to Sect.
imperfections, aerodynamic effects from ambient
2.2.3 :
gas, or injector vibration . Forces due to surface
Ak Coefficients in disturbance func- tension, or more importantly to aerodynamic
tion series pressure caused by a relative velocity difference.
Cd Discharge coefficient between the fluid and the ambient gas, act upon
f (x , r, 0, t) Disturbance function . the surface deformations causing them to grow.

g Spatial distribution of disturbance If growth is sufficient, the jet or sheet may be

function severed or at least an element of fluid may be
k Transverse wave number ripped from the liquid surface. The severed fluid
Lb Breakup length may form a drop or may form an unstable liga-
lw Wavelength ment which will later break up into drops and
n Longitudinal wave number, 2π/ lw finally, if the aerodynamic force on the drops is
S Complex disturbance growth rate , sufficiently large, the drops can be shattered into
λ+iw smaller drops .
tb Breakup time For some simple breakup mechanisms theo-
tsep Time for separation of fluid from retical treatment has been successful. In the fol-
jet, fastest-growing disturbance lowing sections the general analytical approach
y Thickness of annulus, coaxial in- and its application to some simple breakup models
jector element are discussed .
λ Growth rate of disturbance
Kinematic viscosity 2.2.3.1 Liquid surface instability . -A stable,
Disturbance frequency steady flow system is assumed to exist as a liquid
<
3

jet or sheet. Superimposed upon this stable system

Subscript : is an assumed disturbance of the form
max Conditions associated with fastest-
growing disturbance. f(x, r, 0, t) = estg (x , r, 0) ( 2.2.3-1 )

where x, r , are the spatial coordinates, t is time,

* E. J. Rice, Author. and s may be a complex number ( s = λ + iw ) .
50 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

The nonlinear equations of mass, momentum , low velocity jets where capillary forces are pre-
or energy conservation are linearized by assuming dominant. At high jet velocities jet turbulence
the disturbance is small compared with the stable and aerodynamic forces are controlling. In the
system dimensions (jet or sheet ) . These pertur- transition region theoretical breakup analysis is
bation-type equations are then substituted into of little use since both mechanisms are of com-
the appropriate conservation equations . This re- parable importance .
sults in differential equations in the spatial dis- The failure to predict drop size distributions is
turbance function g (x, r, 0 ) . Solutions for the a further limitation to the theoretical approach.
spatial disturbance function are then sought which In some limiting cases average drop sizes are pre-
will satisfy liquid surface boundary conditions . dicted but not distributions. Drop size distri-
The boundary condition may be expressed as a butions must be determined experimentally and
surface pressure which can include surface tension , the dependence upon system properties must be
shear stress, and aerodynamic pressure. The empirically correlated . This will be dealt with in
spatial disturbance function g may include the Sect. 2.2.4.
velocity potential function , and/or the stream In spite of all its limitations , theoretical analysis
function, or just the disturbance amplitude func- of liquid jet and sheet breakup provides much
tion , depending upon the approach used . useful information in several limiting cases . The
If the solution yields a positive real part of the following sections will discuss the results of some
complex growth rate (s in Eq . ( 2.2.3-1 ) ) the of these analyses .
disturbance will grow and thus be unstable. The
disturbance wavelength for maximum instability 2.2.3.2 Liquid jet breakup, low velocity. -The
is then found . This may be determined from breakup of a liquid jet at low velocity cannot be
maximum disturbance growth rate or else from truly called atomization since an increase of sur-
maximum disturbance growth, if the rate is a face area does not result . However , it is an inter-
function of time. The predominant disturbance esting limiting case and has been used to produce
wavelength then allows inferences about drop size, sprays of uniform drop size.206
breakup length, and breakup time. Rayleigh578 first derived the equation for the
Several limitations to the theoretical analysis stability of a non-viscous jet subjected only to
surface tension forces. Weber723 extended the anal-
should be realized . First, the disturbance ampli-
tude is assumed small to linearize the differential ysis by including the jet viscosity . Both analyses
equations of motion. At breakup, the disturbance. were for the case of symmetrical disturbances on
may approach or exceed the fluid system dimen- the jet surface ( see Fig . 2.2.3a) .
sions (e.g. , jet diameter) and thus the linearized Rayleigh's578 analysis yielded the following re-
equations may not be valid. Second , to obtain lationship between the wavelength of the dis-
closed form or explicit solutions an assumption of turbance which will exhibit the maximum growth
either very long or short wavelength disturbances rate (real part of s in Eq. (2.2.3-1 ) ) and the jet
must be made which means the solution is not radius :

necessarily valid except over a limited wavelength lwmax = 9.02rj (2.2.3-2 )

range. Third, in determining the predominant. For comparison , Weber's723 analysis, which in-
disturbance wavelength it is usually assumed that cluded the liquid viscosity, yielded
disturbances of all wavelengths initially have the
1/2
same amplitude. This may be violated due to 3μι
lwmax = 8.89r; ( 1+ ripLS (2.2.3-3 )
such phenomena as a nonuniform turbulence spec- V2
trum, cyclic flow oscillation or separation at a
with the maximum growth rate given by
boundary, or perhaps a mechanical oscillation of
the orifices .
Another limitation to theoretical breakup analy- Amax = [√Spur +GALT,] (2.2.3-4 )
S T
sis involves the overlapping of competing breakup
mechanisms . For instance, theoretical breakup The slight difference between Eqs . ( 2.2.3-2 ) and
analysis is successful in predicting drop size for (-3) with μL = 0 is due to the approximation of
 STEADY-STATE PROCESSES $ 2.2 51

He considers the solution for an arbitrary dis-

turbance (f) of the form

f= Akeinx+st cos (k☺) (2.2.3-8)

k

where k = 0 , 1 , 2, .... The equation for the growth

rate is

FIGURE 2.2.3a.-Symmetric disturbance . Sn2

s = λ + iw = − vn² + √√ ( vn²) ² + ( 1 - k² - n²r;2)
2pLri
Bessel functions in the derivation leading to Eq.
(2.2.3-3 ) . The following inferences as to breakup (2.2.3-9)
time, breakup length, and drop size are given for Examination of Eq . ( 2.2.3-9 ) suggests that
the results of Weber. The parallel set due to Eq.
(2.2.3-2) can be obtained with only small differ- k²< 1- (nr ; )2 (2.2.3-10)
ences by setting viscosity equal to zero . The is a necessary condition for A to be positive .
breakup time for the jet can be estimated by Now nr; = 2πr; /lw is a small number because of

SPLY , 6μLF the long wavelength consideration . Thus , insta-

1
tb≈ = + (2.2.3-5 ) bility results only with k = 0. This is the axisym-
Amax S S
metric case as seen in Eq . ( 2.2.3-8 ) and shown
and the breakup length is then : in Fig. 2.2.3a .
For k = 1 (stable ) the disturbance is such that
Lb≈Vjtb (2.2.3-6)
the jet cross-section remains circular but is dis-
With the assumption that the jet breaks at one placed from the undisturbed jet axis . This type
wavelength interval and this forms one drop, a of disturbance is shown in Fig. 2.2.3b .
mass balance yields the following for the drop size : A low velocity jet thus breaks up only from
axisymmetric disturbances whose wavelength is
71/6
3μι greater than the jet circumference .
dL = 1.88d1; [ 1 + 3x (2.2.3-7)
VPLSdj .
2.2.3.3 Liquid jet breakup, high velocity. - The
From Eqs. ( 2.2.3-5 ) and ( -7) it is seen that more interesting (and complicated ) case of
high surface tension ( S) causes more rapid jet
breakup with high jet velocity can result in true
breakup, while for low viscosity fluids the drop
atomization with the production of very small
size is independent of S and depends only upon
drops and a large increase in liquid surface area.
jet diameter, dj. For viscous fluids the breakup
High jet velocity in this case really means a high
time and the drop size are seen to increase.
relative velocity between the jet and its surrounding
Equations (2.2.3-2 ) through (-7 ) are valid
atmosphere. This can be accomplished by injecting
only for disturbances whose wavelengths are
a high velocity jet into a low velocity atmosphere
greater than the jet circumference ( lw > rd ; ) . For
or by surrounding the jet with a high velocity gas.
this case the surface tension is the driving force
The main distinction between this case and the
causing the instability, and the viscous terms
previously discussed low velocity jet is that aero-
represent damping forces. The jet tends toward a
dynamic forces must now be considered acting on
configuration with less surface area and thus less
the jet surface.
surface energy. Short wavelength disturbances
(lw < d; ) represent stable configurations since the
surface area is increased . The surface tension is
thus a restoring force tending to maintain the
jet's original shape.
Levich439 provides an excellent discussion of the
stability of a low velocity liquid jet subjected to
arbitrary deformations with large wavelengths . FIGURE 2.2.3b. -Asymmetric sinuous disturbance.
52 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

Weber723 considered the case of jet breakup liquid surface increase, and energy dissipation in
with atmospheric effects . The results of his analy- the jet. From Eqs . ( 2.2.3-13 ) and (-14 ) it is seen
sis show the disturbance maximum growth rate that , at high relative velocities, the time for
(Amax) to increase and to shift to shorter wave- separation of individual drops is much less than
length with increasing relative velocity. This the time for complete breakup of the jet. The
would indicate a faster disintegration of the jet atomization of the jet thus commences almost
and reduced drop size . For more detail on immediately upon injection into the atmosphere
Weber's723 results see Ref. 24 , p . 1-40 . and continues throughout the jet length.
Due to the complicated nature of the equations For high viscosity fluids Levich439 shows that
when aerodynamic effects are considered , Levich439 p(AV) 2- Sn
separated the analysis into four limiting cases. 12 (2.2.3-15 )
PLV
He considered long and short wavelength dis-
turbances for both high and low viscosity fluids. Thus higher growth rate disturbances are obtained
The following is a summary of that analysis . for long wavelengths (smaller n ) . He also shows
For short wavelength disturbances and low that the time for jet breakup to increases without
liquid viscosity, the following expression was ob- bound with increasing viscosity, and thus con-
tained for the wave growth rate : cludes that small drops (from small lw ) cannot
2 be produced by viscous jets .
In²
S= [p ( AV) ² — Sn ] (2.2.3-11 ) Considering the case of long wavelength dis-
PL turbances on the surface of a high velocity jet ,
with an arbitrary disturbance as in Eq. (2.2.3-8 )
Thus, s can be real for
and for small fluid viscosity, the disturbance
2TS growth rate was shown439 to be
(2.2.3-12 )
P(AV)2

For very high relative velocities extremely short - pn¹r;2 (AV)2 In

Sn2
wavelength disturbances can be unstable. It should S= + ( 1— k² - n²r ; ² )
be recalled that for low velocities only disturbances 2PL ) 2pLrj
(
with wavelengths larger than the jet circumference (2.2.3-16 )
were unstable. If the drop size produced is of the where growth rate is real for
order of magnitude of the disturbance wavelength ,
high velocity jets can produce very small drops in W
comparison to low velocity jets. pn²r;³( AV) 2 In
πι
In contrast to low velocity jets the separation (2.2.3-17 )
k²≤1— (nr ; ) ² +
of fluid from the jet surface, due to short wave- S
length disturbances, does not sever the jet . Final Thus, for large relative velocity, values of k
breakup of the jet is produced by the cumulative greater than zero can produce instability. It should
effect of separation of small quantities of fluid.
be recalled that for low jet velocity instability
The time for separation of the fastest growing
resulted only with a symmetrical disturbance
wavelength is
(k = 0 ) . An inspection of Eq . ( 2.2.3-16 ) reveals
S PL that even though a sinuous disturbance (k = 1 ,
tsep≈ (2.2.3-13 )
(AV) 3 3
p³ see Fig. 2.2.3b) may grow, the symmetrical dis-
turbance (k = 0, see Fig. 2.2.3a ) will grow faster
while the breakup time for the jet is and should thus dominate the jet breakup .
At sufficiently high velocities the second term
rj PL
tь≈ (2.2.3-14) in Eq. (2.2.3-16 ) can be omitted and thus
AV
Ρ
Equation ( 2.2.3-14 ) was derived using an energy s = n²r;AV In (2.2.3-18 )
2PL (1 )
balance involving the energy transferred to the
jet by the surrounding gas, energy consumed in which shows the disturbance will grow for long
 STEADY-STATE PROCESSES §2.2 53

wavelength ( w > πr ; ) . In Eq . ( 2.2.3-18 ) the type fluids . Fluid viscosity increases the drop size and
of disturbance* (k = 0, 1 , 2 , ...) no longer ap- jet breakup length (see Eqs. ( 2.2.3–5 ) , ( −6 ) and
pears. All types thus grow equally fast . (-7) ) .
Levich shows the time for jet breakup of low At intermediate jet-gas relative velocities , where
viscosity, high velocity jets to be aerodynamic forces must be considered, short
wavelength disturbances are unstable for low
rj
tb≈ (2.2.3-19) viscosity fluids . This results in formation of small
AV ρ
drops which are shed from the jet surface. Long
A comparison of Equations (2.2.3-19) and (-14 ) wavelength disturbances are also unstable with
shows that the breakup times are of the same the symmetrical case having the fastest growth
order of magnitude. This implies that , for a high rate. For sufficiently high relative velocities the
velocity jet of low viscosity, small drop size sinuous and higher order transverse† disturbances
atomization and large drop size fragmentation can become unstable. These would result in jet
have probabilities equal in order of magnitude. fragmentation forming large drops (see Eqs .
For long wavelength disturbances on a high (2.2.3-12 ) and (-16 ) ) .

velocity, high viscosity jet , the growth rate was For very high jet velocities small droplet shed-
shown439 to be ding from the jet surface continues but the char-
acter of long wavelength disturbances changes.
p( AV) ² ( nr ; ) 2 All transverse and symmetrical disturbances grow
In (2.2.3-20 )
4μι ( ) equally fast . Sinuous or higher order transverse
disturbances may cause jet fragmentation with
the jet breakup time to be
the formation of large drops (see Eq. (2.2.3-18 ) ) .
5μι Increasing fluid viscosity reduces the fluid shed-
tb = (2.2.3-21 )
P(AV)2 ding from the jet surface and increases the drop
size for that fluid which is severed . High viscosity
and the wavelength for maximum growth rate fluid jets break up due to long wavelength frag-
to be mentation with resultant large drop sizes (see
lwmax = 6rj (2.2.3-22) Eq . (2.2.3-15 ) and (-20) ) .
Some support for the theoretical results is ob-
High viscosity thus reduces the disturbance.
growth rate and increases the time for jet breakup tained from the experimental observations on jet
breakup given by Grant and Middleman.296 At
(and thus the breakup length) . Large drops due
low jet velocities they observed symmetrical wave
to jet fragmentation are produced with very
breakup. Higher velocities caused the appearance
viscous jets. A number of experimental studies
of transverse waves which damped and sym-
have provided further information on high relative
metrical waves continued to break the jet . A
velocity breakup . These include publications by
Ingebo, 379 Clark 146 and Morrell , 500, 501 further increase in jet velocity produced jet
breakup by transverse waves . For even higher
2.2.3.4 Summary of jet breakup results. -This velocities , and a turbulent jet , surface atomization
prevailed.
section provides a brief review of the more im-
portant conclusions obtained from the theoretical
2.2.3.5 Surface breakup- The theoretical jet
jet breakup discussion in Sects. 2.2.3.2 and 3.
breakup models reviewed in Sections 2.2.3.2 and
Equation numbers are given for reference back to
these previous sections . 3 considered the deviation, due to a disturbance of
Low velocity jets break up by fragmentation a liquid jet from its original cylindrical geometry.

due to the growth of symmetrical waves on the jet

surface. The diameter of the resulting drops is † This should not be confused with transverse acoustic
about twice the jet diameter for low viscosity modes of instability in a combustion chamber. The defini-
tion implied here derives from Eq . (2.2.3-8 ) and defines
the jet cross-section shape under the influence of the
* Also see Eq. (2.2.3-8) . disturbance.
54 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

A low viscosity jet , under the influence of a high the length of the high gas velocity region . A higher
relative velocity gas environment, formsshort liquid velocity results in more rapid escape from
wavelength surface disturbances which are inde- the high gas velocity region. A larger liquid jet
pendent of jet diameter (see Eq . 2.2.3-12 ) . In diameter reduces the fraction of liquid atomized
the following analysis a simplified approach ig- by high gas velocity surface stripping. Mayer's
noring the actual jet geometry is thus used. analysis assumes a steady velocity difference be-
Mayer470 considered the case of wind-induced tween the gas and the liquid, thereby eliminating
capillary waves on a flat, deep liquid surface . He the separate effects of liquid and gas velocities
assumed that when a wave grows to an amplitude and the finite quantity of high velocity gas avail-
comparable to its wavelength, the wave crest able for surface atomization . The importance of
severs and forms droplets of size proportional to finite liquid jet diameter is discussed, but it does
the wavelength. A functional form for the drop . not appear in Mayer's analysis .
size distribution generated by the wave breakup The qualitative discussion of the concentric
was obtained which enabled the derivation of an tube injector was presented here since, with proper
average drop size given by simplifying assumptions, the system may be ame-
2/3 nable to analytical treatment .
μLVS/PL
dL = 9 ( 16 ) 1 /3B (2.2.3-23 )
P(AV)2 2.2.3.6 Liquid sheet breakup. -Dombrowski and
Johns230 considered the stability of asymmetric
where B≈0.3 . Excellent agreement was shown
waves (see Fig. 2.2.3b) on a viscous liquid sheet
between this theoretical drop size¹70 and empirical
under the influence of aerodynamic forces caused
correlations of jet breakup in high speed gas
by a steady cocurrent gas flow. They obtained
streams obtained by Weiss and Worsham.724
wave number and growth rate expressions which
Adelberg49, 50, 51 has extended Mayer's analysis
are viscous effect multipliers of the inviscid so-
to include the effect of fluid acceleration on the
lutions of Squire.664 Also they allowed the sheet
surface disturbance wave speed . He shows that
thickness to vary with time (following the fluid
these acceleration waves (as distinguished from
particles) .
capillary waves of Mayer's analysis) are con-
The drop size produced by the breakup of a
trolling if the jet is experiencing a large acceler-
radially flowing liquid sheet (sheet thickness
ation .
varies inversely with distance) was obtained in
An empirical correlation on the mass median
the following manner. When the disturbance with
drop size produced by a concentric tube injector
the wavelength for greatest total growth (not
was presented in Ref. 352. The correlation , con-
necessarily greatest growth rate) reaches a critical
sidering only those parameters actually varied,
amplitude , the wave breaks at the crests and
can be reduced to
troughs. Wave fragments of one-half wavelength
Vi dj³ are formed which contract into ligaments. The
di~ (2.2.3-24)
AV y ligaments are assumed to break into drops ac-
cording to Weber's criterion (Eq . 2.2.3-7) .
where y is the gas annulus thickness . The exponent The resultant drop size expression of Dom-
on AV is seen to agree well with that of Eq. browski and Johns230 is quite lengthy and the
(2.2.3-23 ) (i.e. , 1 versus - ) . reader is referred to the source for the details .
The conditions in a concentric tube injector However, the following should be noted here. In
differ from those assumed by Mayer in the fol- Ref. 230 the theoretical drop size ( after adjust-
lowing ways. A thin annulus of gas of high velocity ment by a suitable constant ) was compared to that
is injected concentrically around a liquid jet of obtained experimentally from fan sprays and the
low velocity. The high velocity gas stream mixes agreement was favorable. The effect of the spray
with the surrounding gas environment . When this variables upon drop size depends upon which of
mixing region has penetrated to the liquid jet , four domains is dominant. These four domains
the liquid jet thereafter experiences a decelerating can be defined by whether or not the viscosity
gas environment . A thinner gas annulus decreases is effective in the sheet breaking into ligaments
 STEADY-STATE PROCESSES $2.2 55

and in the ligaments breaking into drops . The that a drop will shatter if the surface deformation
.
limits on the exponents for the two extremes of at the stagnation point is roughly equal to the
7
low and high viscosity are : viscosity 0 to 201 drop radius . This occurs if the Weber number
surface tension + to , relative velocity -3 exceeds six for a low viscosity drop and ten for a
to , liquid density - to ≈ , and gas high viscosity drop in the case of a suddenly
-
density to ≈12. applied steady gas velocity.
It may be possible to utilize the analysis of If the result is rounded off and the surface
Dombrowski and Johns230 to obtain theoretical deformation is equal to the drop radius , the drop
drop size expressions for other atomization devices. breakup time can be estimated by
besides fan sprays. However, the sheet thickness
dL
as a function of time must be known . Perhaps tb≈ (2.2.3-26)
the excellent analysis of Hasson and Peck326 can 2 (AV) Ρ
provide the starting point for impinging jet in- for a low viscosity drop with a suddenly applied
jectors and the water bell analysis of Taylor683 steady gas velocity.
for swirl atomizers .
It has been pointed out357 that in the actual
Spray formation from impinging liquid jets re- atomization process, a newly formed drop is not
sults from breakup of the resulting liquid sheets. subjected to a steady relative gas velocity. The
The empirical dropsize correlation by Ingebo375 is
drag force between the gas and the liquid will
one which has been used often for the common
reduce the relative velocity. If the drop breakup
case of impinging liquid jets. More recent dropsize
time is sufficiently large, the Weber number may
correlations have been obtained by Dickerson , 224 be reduced below critical before the drop shatters.
using a molten wax technique. In the latter work
Photographs of drop breakup with a suddenly
spray size distributions for various types of im-
imposed gas velocity were presented by Rabin ,
pinging jet elements were empirically correlated Schallenmuller, and Lawhead.569 Two types of
in terms of orifice sizes and injection velocities.
drop breakup were shown, shear- and bag-type
Work with liquid sheet breakup in a unique in- breakup. With both types the drop is first flattened .
jector design is being conducted by Riebling 592 If bag-type breakup occurs the center of the disk
blows out into a bag which breaks into very small
2.2.3.7 Secondary drop breakup. - A liquid drop
droplets while the ring breaks into much larger
removed from a sheet or jet may be exposed to
drops . With shear-type breakup the edges of the
the aerodynamic pressure effect of a high relative
flattened drop are sheared into sheets and liga-
gas velocity . If this pressure is sufficiently large
ments which then break into drops. In Refs. 210
to overcome the restoring force of the drop surface
and 515 the drop breakup time was experimentally
tension, the drop will disintegrate into smaller
shown to be of the same form as (but ten times
droplets. The Weber number357 defined as
the value of) Eq . ( 2.2.3-26) . Additional references
p(AV) 2dL on secondary droplet breakup include Refs. 222,
We = (2.2.3-25)
28 223 and 755 .

is considered as an effective indicator of the neces- 2.2.4 Spray Description *

sary gas dynamic environment for secondary
breakup . The final form of the liquid phase in the chain
of events leading to combustion is often a col-
Hinze357 has analyzed the forced deformation
lection of roughly spherical droplets called a
of a liquid sphere under the influence of the pres-
spray. † A common objective of experimental drop-
sure distribution caused by turbulent gas flow
let measurements and atomization analyses is a
around the sphere. He considered the two cases
of a suddenly applied steady velocity and a
gradually increasing velocity as may occur with * J. F. Groeneweg, Author.
For some cryogenic propellants, such as liquid oxygen ,
falling raindrops. His results are presented as the
this concept of a spray may be inadequate since sub-
deviation of the liquid surface from the original stantial phase change may occur while the liquid is in
sphere at the stagnation point. Hinze concludes ligament or jet form.
 56 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

description of the spray which allows the pre- estimates of f must be used . Statistical uncertainty
diction of droplet propagation within the com- is reduced as sampling effort increases. While the
bustion chamber and, ultimately, the energy re- time variable is not a consideration in this section
lease profile (see Sects . 2.4 and 7.2.4 ) . The purpose because only steady-state conditions are con-
of this section is to outline the type of information sidered, unsteady behavior is of interest for insta-
necessary for spray description and to briefly bility analysis (Sect . 3.3 ) . Conservation equations
review the existing level of knowledge of steady- may be written for f,743 and are the basis of pre-
state spray properties. dicting downstream behavior based on the specifi-
The basic factor influencing all approaches to cation of an initial density function at formation .
spray description is summed up by the statement The function f could be considered to depend
that spray formation is a random process . Fixing on other variables than those listed above , such
all injection parameters which can be controlled as temperature, however, existing knowledge of
such as flow rates, fluid properties and injection even velocity and position dependence is scarce.
geometry does not produce drops of a single size Drop sizes have received most of the emphasis in
traveling at the same velocity at a particular both experiment and analysis . However, detailed
position. The fact that droplet births are dis- measurements299 have shown that, even immedi-
tributed throughout a region of space and are the ately after formation at a particular position ,
result of fluctuating gas-liquid interactions means drops of a given size move with different velocities
that a statistical treatment is required . indicating that velocity should be treated on an
The following nomenclature pertains to Sect . equal statistical basis with drop size . Spray prop-
2.2.4 : erties are also strong functions of position 277, 589
due to droplet-gas transport processes . From an
DL Drop diameter
Djk analytical viewpoint the key positions are those
Generalized mean drop diam-
where spray formation has just been completed
eter
since they locate the initial conditions required
Dm Mass median drop diameter for analysis .
D30 Volume mean drop diameter
The bulk of existing spray data is composed of
Dmax Maximum drop diameter
various types of drop size distributions and mean
D32 Volume-surface (Sauter) mean.
drop sizes. Analytical representations of these
drop diameter
quantities are obtained by various integrations
f (DL, VL, x, t) Spray density distribution
of f over the droplet variables. Integration over a
f. Spatial drop size distribution
particular droplet variable reduces the amount of
fF Flux drop size distribution
information available and ignores the distributed
bgo General drop number distri-
nature of the variable in favor of the simplification
bution
realized by dealing with a mean value. The re-
X Drop position
mainder of this section ignores time dependence
α , B, d, n, K, 4 Parameters in empirical distri-
bution functions since only steady-state conditions are considered ,
and spatial dependence is implicit .
A basis for spray description is provided by the Two general types of drop size distribution
concept of a spray density function , f(DL, VL, x, t ) , data are available corresponding to two kinds of
which specifies the number of drops in the ranges sampling methods used . They are the spatial
of dDL, dVL, and dx, about the size , DL, velocity distribution f , and the flux (temporal) distri-
VL, and position x at a time t . This function bution fr given by
theoretically represents the results of a spray
sampling operation in which sample size ap- dV (2.2.4-1 )
f =ft dv₂
proaches infinity while increments in droplet prop-
erties, ADL, AVL and Ax (the category sizes con-
sidered) approach zero . A theoretical prediction r = √ √ ₁f dVL (2.2.4-2)
of f from a knowledge of injection parameters
does not now exist (Sect . 2.2.3) so experimental Spatial distributions representing the number of
 STEADY-STATE PROCESSES $2.2 57
57

drops per unit volume per unit size are estimated peak (unimodal) , but some experimenters have
from instantaneous samples of the droplet popu- found definite bimodal characteristics . 101 , 218, 299, 339
lation in a given spray volume ; e.g. , high-speed Partial characterization of distributions is fur-
photography. Flux distributions which specify the nished by various mean sizes. A general set of
number of drops per unit size crossing a unit area means is defined by507
per unit time are estimated by collecting a sample
of all drops passing a plane in the spray during
a known sampling time ; e.g. , frozen wax224,724 or S ( DL ) 'g dDL
immiscible fluid collection cell methods . 465 The (Djk) j-k = (2.2.4-3 )
distinction between f, and fr is usually ignored in
the literature, but differences between the two √ ( D1 ) kg dDL
can be large depending on the distribution of drop
velocities.299 In the remainder of this section the where g is a number distribution and the inte-
symbol g will be used to refer to size distributions gration is over the complete range of sizes. If
in general without regard to type . Size distribu- k = 0 the denominator simply assures normali-
tions are usually normalized to make the integral zation . Two commonly used means are the volume
over all sizes equal to one. In that case, any mean, D30, and the volume-surface (Sauter ) mean ,
physical magnitudes of liquid concentrations or D32 . Two additional parameters used to character-
fluxes are lost unless they are purposely specified ize distributions are the mass median , Dm, and a
as the normalization factor. maximum drop size, Dmax .
Data which result from individual measure- Since a theoretical prediction of the form of g is
ments of diameter produce a number distribution lacking, a variety of empirical forms have been
of the type shown schematically in Fig. 2.2.4. used to fit size distribution data . The equations
Uncertainties are greatest at very small sizes are basically of two types :
where measurement resolution is limiting and at
g ( DL ) DL exp ( -BDL ) (2.2.4-4)
very large sizes where measured frequencies are
particularly sensitive to sample size . Mass distri- g (DL) exp ( y²) (2.2.4-5)
butions obtained by weighting the ordinates of with
the number distribution by DL are a useful form
DL
which often allows reasonable extrapolations to y = 8 ln (2.2.4-5a)
D,
be made at the size extremes. The particular
or
examples shown in Fig. 2.2.4 each have a single
7DL
y= ln (2.2.4-5b)
Dmax -DL.

1.6 ×10-2 where a, B, 8, n, K and are parameters . The first ,

Fraction
number
micron

Eq. ( 2.2.4-4) , may be called a generalized gamma

mass
per
of
or

function, 450 and specialized conditions on the pa-

1.2 Number rameters lead to forms having names such as
Nukiyama-Tanasawa and Rosin-Rammler. The
second group consists of logarithmic transfor-
Mass
mations of the normal distribution ,390 Eq.
(2.2.4-5 ) . Equation (2.2.4-5a) gives the log-
normal distribution whose range extends to infi-
nite sizes while the upper limit form, Eq.
(2.2.4-5b) , provides for a maximum size. For any
particular case of Eqs . (2.2.4-4 ) to (-5 ) , the
means given by Eq. (2.2.4-3 ) may be calculated
50 100 150 200 250
Drop diameter, microns as functions of the distribution parameters. 97, 507
Mass medians are usually determined from a
FIGURE 2.2.4 .- Typical drop size distributions. cumulative plot and maximum size from either
58 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

the largest observed or a drop stability criterion mum drop sizes. In principle, an infinite number
(Sect . 2.2.3 ) . of means given by Eq. ( 2.2.4-3 ) are equivalent
A comprehensive empirical approach would be to the specification of the distribution . In practice ,
to choose a distribution equation capable of fitting. investigators have usually only measured and
the data and to correlate its parameters with in- correlated one or two size indicators for a given
jection conditions . The labor of such an approach set of conditions . A sampling of the literature
has limited the number of such attempts and their (see Table 2.2.4 ) confirms the intuitive trends of
scope . For example, the parameters in Nukijama- drop sizes which vary directly with orifice size ,
Tanasawa forms of Eq . ( 2.2.4-4 ) have been corre- liquid viscosity , and surface tension ; and inversely
lated for impinging jets224,375 and cross-current with relative velocity between the gas and liquid .
injection379 in air streams ; and the parameters in Additional information for impinging jet arrange-
the upper-limit form of Eq. ( 2.2.4-5 ) have been ments may be found in Refs. 224 and 375. Maxi-
correlated for a swirl atomizer589 and a collection mum drop diameters measured for injection into
of size data from several injector types.506 In the a cocurrent gas flow have been found to depend
absence of parameter correlations, distribution. on the acceleration or deceleration of the gas
equations are fitted to specific data sets or the stream in addition to the liquid and gas ve-
distribution data are simply plotted and tabu- locities.376 The dependence of size on gas density
lated . 101 , 277, 339 Direct measurements of spray char- cannot be approximated by a simple exponent
acteristics in a rocket combustion environ- since a switch in dependence from inverse to direct
ment377, 378,373 are scarce because of the associated is observed as gas density increases. 218, 229 Specific
experimental difficulties. Indirectly, distribution injector types and their relation to droplet sizes are
characteristics based on the log-normal form have discussed in Sect . 7.4.
been inferred from an analysis of rocket perform- In summary, spray description requires a sta-
ance based on a vaporization model.564 tistical treatment, and the concept of the spray
Other than size distributions, most of the exist- density function furnishes a basis for analysis.
ing data is in the form of mean, median, or maxi- Spray data available as input to this model are

TABLE 2.2.4- VALUES OF EXPONENTS SPECIFYING THE DEPENDENCE OF DROP SIZE ON INJECTION VARIABLES

Conditions Drop size Orifice Surface Liquid Liquid Velocityb Ref.

correlated diameter tension viscosity density

Single jet crosscurrent D30 12 4 1 -1 -3(g) 379

injection into heated air Dmax 0.42 0.29 0.29 -0.29 -0.87(g)

Single jet, costream and

contrastream injection into Dm €512 } - % (r)
-4 724
heated air

Correlation for pressure, D32 0.65 0.20 0.35 -0.35 -0.55(r) 506
swirl and impinging jets Dmax .52 .18 .48 -.48 —.66 (r)

Swirl nozzle injection into D32 0.53 0.21 0 -0.48 -0.50(L) 219
stagnant gas

For example, D30 = di1/2S1/4uL / PL /4-3/4.

b g, gas ; L, liquid ; r, relative.
• Not varied in experiment.
 STEADY-STATE PROCESSES $2.3 59

mainly limited to specific size distributions or ing (primarily diffusion in a smoothly operating
correlations of mean sizes with injection con- combustion chamber) if the desired throat con-
ditions . dition is to be met . The problem in general is to
achieve (other factors being equal) a uniform
2.3 SPATIAL DISTRIBUTION OF PROPELLANTS mass flux throughout the throat cross section with
a minimum chamber length.
The following nomenclature pertains to Sect .
The initial propellant distributions produced by
2.3 :
real injectors are basically three-dimensional ar-
C Concentration of gaseous propellant rays of droplets formed from ligaments emanating
Em Mixing efficiency factor, defined in Eq. from many sources. The complexity of such sys-
(2.3.3-1 ) tems imposes the necessity of reducing these more-
F (C) Concentration distribution function or-less known properties of a single spray to the
G Local flow rate per unit area desired mass flux distribution thru some arbitrary
g Spreading factor or intuitive artifice.
ΚΑ Sampling area correction factor In certain analyses544,421, 64 the discrete prop-
Mj Momentum ratio function , defined in erties of the individual elements are reduced to
Eq . ( 2.3.3-2 ) an assumed uniform ( or other prescribed ) distri-
m Local total mass flow rate, mi + m2 bution of particles of known size near the injector
n Number of samples with < T face so that the droplet-gas interaction process
n' Number of samples with > T can be described . Although this process results in
P, q, A Constants in F (C ) expression , Eq. a tractable combustion model, the contribution of
(2.3.3-8) the injector (to the extent that it does not satisfy
St Lagrangian scale of turbulence the assumption ) with regard to mass flux distri-
Vrms Root-mean-square lateral velocity bution is lost .
W Point-source strength An alternate approach613 emphasizes the contri-
α Impingement angle bution of individual elements as determined from
B (p, q) Beta function measurements on non-reactive sprays . Such meas-
Φ Mass flux fraction , mi/ (m₁ + m2) urements serve to define, strictly from geometrical
¥ Ripple factor, Cmin/Cmax considerations , the mass flux distribution at the
axial station where the boundary enclosing some
Subscripts :
T 95% of the total mass flux for the element is a
Overall or total spray conditions
proportionate share of the chamber cross section .
1, 2 Indices specifying propellant compo-
At that station the radial component of velocity
nents (e.g. , oxidizer and fuel)
is presumed to go to zero- hence a mass flux for
2.3.1 Mass Flux Distribution * the bounded surface can be specified .
In the case of elements composed of identical
The mass flux distribution in the transverse like-on-like impinging jets, these distributions can
plane is of considerable interest in liquid rocket be characterized to a reasonable approximation
engine design . If a uniform mass flux distribution by a two-dimensional Gaussian distribution within
is to be achieved by the time the nozzle throat is essentially elliptical boundaries . This is the general
reached, i.e. , all propellants to be burned prior to description that is utilized by Reardon et al.583 in
that location (thus providing uniform product portraying the element orientations for like-
temperature and density) , then considerable at-
tention must be given to the initial mass flux † The discussion here will concentrate on designs which
distribution achieved at (or near) the injector. seek to achieve uniform mass flux across the injector face.
Although the initial mass flux distribution can be The use ofnonuniform mass flux distributions for purposes
somewhat less than uniform , such nonuniformities such as : improved stability, modified heat transfer,
specialized research modeling, etc. , encourages other
cannot exceed the capabilities of secondary mix-
mixing phenomena which include recirculation and radial
winds. These are discussed in Sects. 2.3.3, 2.3.4, 7.2.5 ,
* J. H. Rupe, Author of Sects. 2.3.1 and 2.3.2. 8.5.3 and 9.2.2.
60 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

doublet sprays from which interactions and/or The performance level that is achieved,703, 145, 271
coupling effects with combustion chamber dy- the compatibility of the gas side boundary
namics are inferred . This representation is spe- flows609,168 with the wall, as well as the stability of
cifically appropriate for the distributions produced the combustion process583 are intimately related to
by doublets comprised of identical jets. this parameter. Although it is undoubtedly true
On the other hand, for non-identical jets (in- that there is no one particular mixture ratio distri-
cluding like-jets of differing momenta ) the mass bution that is best for all possible requirements ,
distributions resulting from impingement are sym- the intuitive arguments for achieving a specified
metrical only about the plane including the jet distribution in a controlled and reproducible man-
centerlines. The phenomenon is illustrated in Fig. ner are rather overwhelming. Also as is the case
2.3.1a where the normalized mass flux for two with mass flux distribution ( Sect . 2.3.1 ) , it is
different element configurations are portrayed.613 logical from the standpoint of performance to
It is noted that only in one of these cases are the strive for a uniform distribution at the nozzle
jets identical. As the jet momenta diverge from throat, and to accomplish this objective with the
a ratio of unity, the resultant becomes increasingly minimum possible chamber length . Since the
non-planar and hence can make a unique contri- modification of an initial mixture ratio distribution
bution to the overall mass flux distribution . by secondary mixing processes (Sect . 2.3.3 ) is a
This concept of an element mass distribution relatively ineffectual process , 168, 318 it is apparent
being utilized to form a composite for a complete that the injection scheme must play the most
injector is illustrated in Fig . 2.3.1b613 which por- important role in the determination of this cham-
trays the mass distribution for an injector con- ber property .
sisting of 6 elements . It should be noted that in Attempts to measure mixture ratio distributions
the actual "model" (of which this is a repro- in actual rocket engines by direct sampling tech-
duction ) the absolute magnitude of mass flux is niques date back at least twenty years. 73, 125, 448
proportional to the density of the photographic However, until the advent of the high speed mass
negative, which can be used to provide a quanti- spectrometer, these studies were necessarily
tative description of the mass flux distribution at limited due to the extreme severity of the environ-
the model plane. A computer model has recently ment being sampled and the dependence of a
been developed356 to describe the three-dimen- composition determination upon an inferential
sional spray mass flux patterns formed by arrays relationship with, for example, temperature.73 The
of injector elements as in a complete injector. additional complications arising from slow sam-
The model uses empirical cold flow data613,356 to pling rates and the necessity for direct analysis
characterize the contributions from individual of the sampled gases precluded the wide accept-
elements . Collisions of spray droplets from ad- ance of such techniques. However, several new
jacent elements are considered on a probability efforts628,758 that utilize high speed mass spec-
basis . Development injectors, as well as research trometers for real time analysis of local com-
injectors, are now characterized by similar meth- bustion products show promise of yielding the
ods . 224, 258 Normally full injectors are tested rather much needed information . No data have been
than individual elements in order to characterize , published to date but the apparatus and tech-
non-reactively, the element-element interactions . niques are in the advanced stage of development .
In lieu of a direct measurement of mixture ratio
2.3.2 Mixture Ratio Distribution distribution and/or a correlation with injector

Of the several parameters that serve to char- geometries, there are several alternate procedures
acterize the combustion process in a liquid pro- that have been devised to describe this parameter.
One method uses the flow rate from individual
pellant rocket engine, the mixture ratio distri-
bution is certainly one of the most important.
an equivalent representation would be to specify the
* It should of course be recognized that the description individual mass flux distributions rather than their quo-
of mixture ratio serves only to separate the mass flux tient so that the discussions relating to scale of the element
distributions of the individual propellants. Obviously and secondary mixing are equally applicable here.
 61
STEADY-STATE PROCESSES $2.3

30°

60
45° 60° 75° D-ARM 15° 45°
0 G00810
G - 00055 G - 00980
OXIDIZER G 0.1195
G 0.0085
G 00115- G 0.1440
30° G 00155 G 0.1885
G 0.0210 G0.2180
G -0.0270 -6.0.2345
G - 00360 G - 0.2600
G = 0.0450 G - 0.2745
G 0.0570 75°
15°

C-ARM O 0° A-ARM

75° 15°

FUEL
TEST CONDITIONS
SIMULATED CORPORAL PROPELLANTS a 44 deg
(CCI4 + KEROSENE AND ZnCl2) L/d 100
60° ORIFICE DIAMETER € 85.01 30°
FUEL 00986 in. 2.80
OXIDIZER 0.173 in. m, 0.329 lb/sec
0.921 lb/sec
SPRAY TIME 085 sec ■ 1.250 lb/sec

45° 15° 0 75°

30° B-ARM
(a)

30° 45°
19

45 60° 75° D-ARM 15

G - 00037 G - 0.0945
G - 00094 G 01134
30° G - 0.0189 G 0.1513 60°
G 0.0283 G-0.1891
G 0.0378 FUEL -G - 0.2269
G 00472 -6-02648
G - 0.0567 -G - 0.3026
G - 0.0756 -G - 0.3394
75°
15

C-ARM 0 O A-ARM

75° TEST CONDITIONS

SIMULATED N204 -N2H4 PROPELLANTS a- 45 deg
(CCI4 + KEROSENE AND H2O) L/d 100
ORIFICE DIAMETER E - 73.64
G /Ao FUEL 0.173 in. 1.23
OXIDIZER OXIDIZER 0.173 in. 0.493 lb/sec
-0.609 lb/sec
60° SPRAY TIME 0.5 sec .1.102 lb/sec
30°
45 30° 15 B-ARM
(b)

(a) On a spherical surface for nonsimilar jets.

(b) On a spherical surface for similar jets .
FIGURE 2.3.1a.-Mass flux distributions for typical doublet elements .
 62 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

VIEW LOOKING UPSTREAM

INJECTOR REFERENCE
FUEL: CORPORAL MARK
OXIDIZER: SFNA
MIXTURE RATIO = 2.80

ER
DIZ
OXI

) P
Y
(T
EL
FU (TYP)

CHAMBER
BOUNDARY

IMPINGEMENT-POINT PLANE AT STATION 2.767 in. INCHES

MODEL PLANE AT STATION 8.597
Gmax =0.738 lb/(sec in. 2) AT STATION 16.45 in.

FIGURE 2.3.1b . - Mass flux distribution for a six-element array of unlike doublets .

orifices (based in general on an extrapolation from been successful in the measurement of mixture
non-reactive calibration fluids to propellants ) , ratio gradients.102
while a second utilizes experimental measurements The flow rate method is the most common
of mixture ratio distributions within sprays formed means to estimate local mixture ratio but neces-
by elements or entire injectors but with non- sarily results in an arbitrary definition of the
reactive fluid . In at least one case705 the mixture region of the chamber that is characterized by
ratio has been inferred from the erosion pattern the stated value. Conn et al.168 imply a circular
on an ablative wall. Another technique using section while Harrje583 prefers an ellipse. Each of
thermocouple rake measurements of the sprays these is undoubtedly a valid representation when
formed by propellants (and pseudopropellants ) the scale of the element is small. On the other
injected at different temperature levels has also hand, when the distribution within the element
 STEADY-STATE PROCESSES $ 2.3 63

becomes significant, then a procedure that defines correlation with element configuration that in
local properties becomes more meaningful. The turn provided a so-called " uniformity criterion . "'612
results obtained with one such procedure613 are This criterion has been used to improve rocket
illustrated in Fig . 2.3.2 where the mixture ratio combustor performance . 145,610
distribution for two typical doublet elements are These sampling techniques have also been useful
superimposed upon their mass flux distributions in characterizing mixture ratio distributions of
(the same mass flux distributions shown in Fig . other element types244,591 as well as complete in-
2.3.1a) . It is particularly interesting to note that jectors, and to show that uniformity of mixture
regions of relatively nonuniform mixture ratios. ratio is indeed a prerequisite to high perform-
(associated with significant mass fluxes ) dominate ance . 389, 609
these distributions even though the mixing effi- It should be noted that the properties deduced
ciencies characterizing the entire sprays were from nonreactive sprays of unlike doublets have
relatively high . restricted applicability for highly reactive pro-
For cases of no element-element interaction a pellants135b, 246, 389, 415,777 due to a phenomenon re-
composite array of such elements would yield a ferred to as reactive stream separation . For certain
mixture ratio distribution analogous to the one propellants and for specific element types ( e.g.
shown in Fig. 2.3.1b and these distributions would unlike doublets ) preliminary bounds for this oc-
then provide the appropriate initial conditions currence have been established in terms of oper-
required for the analysis of steady-state com- ating pressure, jet velocities and diameters , pro-
bustion.554,762 The mathematical model356 previ- pellant temperatures, and propellant reac-
ously mentioned in connection with spray mass tivity.415,430 However, it seems clear that these
flux likewise describes mixture ratio distributions gross effects should be taken into account when
for liquid-liquid propellant systems. In effect, this prescribing the size of the element for which the
model represents a numerical means of describing initial mass and mixture distributions of unlike
information such as shown in Fig. 2.3.1b , with impinging hypergolic elements are presumed
additional (analytical) predictions as to how this known . 389, 619
mixture ratio pattern would vary with axial lo- Although the analogous problem of combustion .
cation . Allowance is taken for inter-element drop- effects on the mixture ratio and mass distributions
let collisions, but not for gross distortion of spray of elements composed of like-on-like sprays have
patterns by adjacent elements . not been studied extensively and therefore have
Extensive current application of the technique not as yet been quantitatively described , it would
originally advocated by Rupe613 to describe mix- seem reasonable to assume that similar effects
ture ratio distribution of full ( liquid-liquid) in- may appear if fuel and oxidizer from closely
jectors has provided empirical verification of the spaced adjacent elements come into intimate con-
general validity of this method in application to tact prior to full atomization into spray droplets.
performance analysis. 224, 258 In some cases, however, Thus it is seen that the mixture ratio distributions
element-to-element interaction substantially dis- associated with highly reactive (hypergolic ) sys-
torts the distribution that would be anticipated tems require further elucidation.
by simple superposition of individual element
patterns. 2.3.3 Mixing Processest
In at least one instance617 detailed measurements
Before detailed consideration is given to the
on nonreactive sprays have been used to determine mechanism and characteristics of the various
a single quantity, i.e. , a mixing efficiency, to
mixing processes which occur and play an im-
represent the degree of uniformity of mixture
portant role in rocket engine combustion , it is
ratio and to utilize this parameter* to define a
pertinent to review briefly their overall significance
and relative importance.159, 355, 269
* The parametric relationship for maximized mixing
uniformity as applied to unlike doublets is discussed in
Sect . 2.3.3.1 , see Eq. 2.3.3-2. The modified versions of † J. D. Lewis, Author. British Crown copyright. Re-
that relationship for other unlike element types are dis- produced by permission of the Controller of Her Britannic
cussed in Refs . 244 and 591. Majesty's Stationery Office.
64 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

60° 75 D-ARM 15° 30

0
G 00058-
G00309-
30° G 00881
G 0.1468-
G02056- OXIDIZER

-r/r+1) 0.58940, 1.43

15 -0.66310, 1.97 75
0.73680,2.80
-0.76310, 322
-0.78940, -3.75
+1)-076310, 3.22 -0.81570, 442
0.78940,375-
0.81570, 4.42
C-ARM O -0° A-ARM

FUEL
75°
TEST CONDITIONS
SIMULATED CORPORAL PROPELLANTS
(CCI4 + KEROSENE AND ZnCl2)
OORIFICE DIAMETER
FUEL 0.0986 in.
OXIDIZER 0.173 in.
60° a . 44 deg 30°
L/d 100
E- 85.01
r/r+ 1 ) Q73680, r2.80 (NOMINAL)
VALUES OF G NORMALIZED TO
1.0 lb/sec
30 15° B-ARM 60°
(a )

60° 75° D-ARM 15 30°

8
G000612
G 0.04571 45°
45 G 0.07145-
G010005-
G0.12862 FUEL VALUES OF G NORMALIZED TO
AG Q15720- m1.0 lb/sec
G 0.18517 60°
G 0.21437 +1) 0.59730,1.48
G 0.24293- 0.54545,1.20
G 027151- 0.49730, 0.99
G0.30010
15 75°

C-ARM A-ARM

75 . 15°
TEST CONDITIONS - / +10.44210, = 0.790
SIMULATED N204 -N2H4 PROPELLANTS 0.38680, 0.630
(CCI + KEROSENE AND H₂O) -0.33160, r = 0.495
ORIFICE DIAMETER -0.27630, 0.380
FUEL 0.173 in.
OXIDIZER 0.173 in. 30°
a 45 deg OXIDIZER
L/d 100
Ar+1) 054545, 1.20 (NOMINAL:)
E -73.64
45- 30° 15° B-ARM 75° 60° 45
(b)

(a) On a plane surface for nonsimilar jets.

(b) On a plane surface for similar jets .
FIGURE 2.3.2 . - Mixture ratio distributions for typical unlike double elements .
 STEADY-STATE PROCESSES $2.3 65

It has long been recognized from practical de- processes under rocket combustion conditions in
sign requirements that chemical kinetic processes , the three categories enumerated above, it is im-
based on pre-mixed vapor-phase reactants , are portant to recognize the complexity involved and
neither rate controlling nor do they exert a domi- that any theoretical or experimental appraisal of
nant influence upon combustion characteristics in one process in isolation is virtually impossible to
most liquid bipropellant rocket engines during achieve in practice .
steady-state operation . It may be noted that
monopropellant combustion is far more susceptible 2.3.3.1 Liquid phase mixing .- There are obvi-
to control by kinetic effects . In common with ous attractions in attempting to achieve a high
other high intensity systems, bipropellant rocket proportion of oxidant/fuel mixing in the liquid
combustion performance is largely determined by phase, since this technique offers the greatest
a complex network of interdependent physical prospect of exercising design control where it is
processes as are indicated schematically in Fig . likely to be most effective, at the source of the
2.3.3a . This diagram illustrates the general case ensuing chain of combustion processes. Although
and it will be appreciated that under any given several attempts to obtain liquid-liquid mixing
set of operating conditions, propellant combi- before injection have been reported , these have
nation, or injector design configuration some of usually been unsuccessful owing to the onset of
the reaction paths may assume greater or lesser uncontrolled reactions during transient off-design
importance and may even be absent altogether. operation caused by the high sensitivity of the
Nevertheless, it can be seen that for all bipro- premixed propellants. Consequently, attention has
pellant combinations and at every selected oper- been directed mainly towards methods of liquid
ating condition , adequate mixing of fuel with phase mixing within the combustion chamber
oxidant is an essential criterion for the attainment itself. Exceptions to this generalization are pro-
of efficient combustion , i.e. , maximum heat release vided by design solutions of the "recessed-cup"
in minimum volume. and similar types where mixing and often partial-
As indicated in Fig. 2.3.3a, it is possible for reaction take place in small pre-combustion cham-
propellant mixing in the rocket engine to occur bers on the injector face. In practice , liquid phase
under one , or more, of the following three con- mixing has been found to enhance the rates of
ditions : heat release and combustion efficiencies with some
1. in the liquid phase hypergolic propellant combinations* ; whereas,
2. by droplet transport in a heterogeneous en- with non-hypergolic propellants there has often
vironment been a tendency to provoke unstable com-
3. in the vapor-phase bustion 539, 318, 446
By comparison with combustion systems for So far, no authoritative theoretical model has
air-breathing engines , the rocket has the ad- been derived to describe liquid phase mixing
vantage of being able to utilize liquid - phase mixing processes, especially those in the presence of highly
under prescribed circumstances, whereas the exothermic chemical reactions, produced by typi-
higher gas temperatures of the rocket usually cal injection systems. Instead, there have been a
prevent all but rudimentary attempts at enhance- considerable number of experimental investiga-
ment and control by aerodynamic methods . Other tions , among which those at the Jet Propulsion
practical considerations apart from combustion Laboratory (e.g. , Refs . 617 , 612 , 615 and 244 )
efficiency may also impose design constraints es- are of particular significance. These studies have
pecially affecting the injector configuration . For provided quantitative data and empirical corre-
example, although rapid mixing in the liquid and lations under nonreactive or essentially isothermal
vapor-phases is a prerequisite of good perform- conditions and attempts have been made to relate
ance, at the same time there is often a requirement these to measurements of the overall combustion
to inhibit mixing processes in the boundary layer
in order to reduce heat transfer rates to acceptable *
Reacting stream separation246,389,415 may preclude
values by film cooling or other means. liquid phase mixing for common hypergolic propellant
Therefore, in the following discussion of mixing combinations under certain operating conditions.
66 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

PROPELLENT
INJECTION

LIQUID STREAM ATOMIZATION VAPOUR STREAM

IMPINGEMENT FORMATION

DROPLET DROP-VAPOUR VAPORIZATION DROPLET

COLLISION MIXING HEATING

RECIRCULATION
TRANSFER
LIQUID PHASE LIQUID - VAPOUR VAPOUR PHASE SUPER CRITICAL
MIXTURE MIXTURE MIXTURE MIXTURE

ATOMIZATION DIFFUSION TURBULENT

MIXING

HEAT
VAPORIZATION

LIQUID PHASE GAS PHASE

REACTION REACTIONS

TURBULENT.
MIXING

INTERMEDIATE FINAL
REACTION PRODUCTS REACTION PRODUCTS

TURBULENT
MIXING

EXPANSION IN
EXHAUST NOZZLE

DELAYED RECOMBINATION
REACTIONS PROCESSES

EXHAUST JET

FIGURE 2.3.3a. - Combustion and allied processes in liquid propellant rocket engines.

performance and to qualitative visual observations (b) impinging sheet ; these are similar to im-
of the flame shape or intensity. pinging jets except that propellants are
Propellant mixing can be produced by a wide formed into thin sheets by fan sprays or
variety of injector designs, the most common of other methods before converging upon
which include
each other
(a) impinging jet ; whereby single or multiple
plain cylindrical streams of each propel- (c) splash plate ; propellants are injected
lant converge at a common point in free through plain jets to impinge on suitably
space near the face of the injector positioned plate. Some liquid phase mixing
 STEADY-STATE PROCESSES $2.3 67

occurs on the surface during film for- to present in any general manner. However,
mation limited comparisons have been possible by the
(d) vortex ; plain jets of one propellant im- adoption of a mixing efficiency factor, Em,617 de-
pinge on a cylindrical surface film of the fined as
second, thus providing a low resultant
axial momentum Em= 100
0 { 1- [ ← KAM ( T - 6)
MTOT

Typical mixing and resultant flow patterns pro-

Β΄ ΚΑm (φτ- φ)
duced by representative impinging-jet , splash +Σ (2.3.3-1 )
ήτ ( φτ - 1 )
plate and vortex injectors are illustrated in Fig.
2.3.3b derived from Refs. 246 and 111. Liquid where
phase mixing can be seen to be confined to a very n number of samples with < & T
small volume and its effectiveness is known to n' number of samples with > T
depend on the destruction of the oncoming fluid KA factor for sampling area correction defined
momentum in a regular and controlled manner ; as the ratio of the area of equivalent
hence the advantages of surface impingement spherical surface represented by the sam-
techniques which do not depend on close design ple to the cross sectional area of the
tolerances and are least prone to flow disturbances. sample tube
The wide variety and range of complexity of m local mass flow rate of the spray
injector designs has made the analysis and in- mT overall mass flow rate of total spray or
terpretation of experimental results very difficult component stream

Oxidant
Fuel
Combustion gases

IMPINGING JET

ANG..

SPLASH PLATE "VORTEX" TYPE

FIGURE 2.3.3b.-Relative locations of fuel, oxidizer and combustion gases associated with splash plate and vortex-
type injectors.
68 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

& local mass flux fraction = m₁/ ( ṁ1 + m2) practical combustion conditions, provided mixing
фт overall mass flux fraction rates are large compared with chemical reaction
= ṁT₁ ( ṁT₂ + ṁT₂) rates246 ; for example, when using stabilized fuming
nitric acid and various alcohols or kerosene/amine
and subscripts 1 and 2 are used to identify the
mixtures. Conversely, with propellant combina-
two components (e.g. oxidant and fuel) .
tions which give rise to rapid liquid phase reactions
This mixing factor , based on the variation of
(e.g. , oxidizers such as chlorine trifluoride or
local mixture ratio and weighted in accordance
nitrogen tetroxide with mixed hydrazine deriva-
with the proportion of the spray represented by
tive fuels) the impingement mechanism can be
each sample, has been used by various workers to
strongly influenced . Photographic studies246 have
compare sprays produced by differing injector
shown violent evolution of gases at the impinge-
configurations and operating conditions, and is
ment interface sufficient to blow apart , or sepa-
generally preferred to other expressions relating rate, the streams of fuel and oxidizer and thus
the distribution of samples about a mean value.
inhibit further mixing and atomization . With such
For example, in the simplest case of the impinge-
systems it is, therefore, probable that the uni-
ment of two unlike jets in free space it has been
formity of mixture distribution is not only de-
possible to show612,615 that for a range of geo-
pendent on the ratio of the momenta in the two
metrical and operating conditions the mixture
streams but also on the relative momentum be-
distributions are dependent on stream momenta.
tween them. Highly reactive propellant combi-
At any given condition the most uniform mixture
nations, however, can be made to give good com-
distribution ( i.e. Em was a maximum) was ob-
bustion efficiencies by design features which nullify
tained with equal momenta in the two jets, i.e. ,
the disrupting influence of gas evolution i.e. , the
p1u2d1 = p2u22d2 use of multiple interlacing injection elements or
or such devices as the recessed-cup injector.
-1 Injector designs like the splash plate and vortex,
p1u12d ,
M; == 1+ = 0.5 (2.3.3-2) which lead to liquid phase mixing on a solid
p2u22d2 ]
surface, have the advantage of avoiding the
in which p is the fluid density , u the axial injection necessity for accurate alignment inherent in im-
velocity, and d the jet diameter ; while M; is a pinging systems and to some extent mitigate
convenient expression for representing large vari- against the effects of spontaneous reaction . How-
ations in momentum ratio. A typical correlation ,
100
from Ref. 615 , obtained from experimental studies
of a simple unlike doublet system with an im-
80

pingement angle of 60°, is shown in Fig. 2.3.3c. 80

Similar trends of variation in mixing efficiency
with liquid stream momenta have also been ob-
60

60
served for coplanar triplet injectors and concave Em
surface deflectors .592, 252 These and other similar
investigations have also shown the dependence of 40 A₁/A₂
- 1.0
mass flux distribution (spray patternation ) and -0.670
in some cases of spray quality (drop-size and -0.655 and 0.656
-0.592
20

drop-size distribution ) , on injector geometry and 20

- 0.439
jet momenta. In addition , there is evidence237 -0.391
under nonreacting conditions, that the miscibility -0.261
O
0.2 0.3 0.4 0.5 0.6 0.7 0.8
of the two fluids has little or no effect upon the 2
1+ pi ui ² di
mixing efficiency and distribution of fluid com- M; = 2
P2 uz² dz
ponents.
Cold flow tests with effectively inert fluids ,
FIGURE 2.3.3c . -A correlation of the dynamic character-
similar to those described above, are capable of istics of free liquid jets and Em for an impingement
simulating the liquid- phase mixing behavior at angle of 60°.
 STEADY-STATE PROCESSES $ 2.3 69

ever, these designs lend themselves to an infinite methods of propellant injection which provide
variety of geometrical configurations, scale effect intentional liquid stream impingement as a pri-
and operating criteria with the result that no mary mixing device may still rely on spray mixing
effective characterization of their design features as a secondary process, see Fig . 2.3.3b. In addition ,
has been possible up to the present time . Never- for many injectors (e.g. , like-impinging) , each
theless, a number of qualitative design rules of propellant is atomized independently through a
an ad hoc nature have been accumulated for par- multiplicity of separate injector elements so there
ticular systems as a result of intensive develop- is no direct liquid mixing. Primary mixing of
ment experience. A typical procedure is the fol- propellants then takes place through intermingling
lowing : during the development of an injection between unlike sprays as a result of ( 1 ) natural
system, several design configurations are sub- expansion in the spatial distributions , ( 2 ) droplet
jected to cold flow mixture and mass distribution dispersion by turbulent eddy diffusion in the sur-
measurements over a range of operating con- rounding gas and ( 3 ) forced mixing by deliberate
ditions. Selected configurations are then subjected spray impingement.
to open burner tests for photographic recordings As in the case of vaporization rate predictions,
to be made of the flow patterns and flame shape , an intimate knowledge of the initial spray char-
then finally to detailed determination of their acteristics is an essential prerequisite of any at-
combustion performance and heat transfer prop- tempt to predict and understand the subsequent
erties under normal rocket engine conditions.¹¹¹ droplet mixing processes. Numerous experimental
By this means it has been possible to formulate determinations of the spray properties under cold-
empirical design rules on the effects produced by flow conditions have been reported in the liter-
changes in stream momenta, impingement po- ature (see Sect . 2.2.3 and 2.2.4 ) for atomizers of
sition, injector geometry, etc. , for the particular the types used as injector elements in rocket.
systems studied . Some of the rules may be of engine applications. These studies have included
more general application . measurements of the drop -size distributions , mass-
Unfortunately, this comparatively tedious and flux distributions (patternation ) and velocity dis-
costly approach* is likely to remain for some time tributions near the point of spray formation for a
the best available method for the acquisition of variety of injector element designs and scales with
detailed knowledge on any given type of injector, a range of fluid properties, ambient densities and
where complex liquid and gas phase mixing proc- operating conditions. In many instances it has
esses are involved . The desirable alternative , a been possible to correlate results satisfactorily by
comprehensive theoretical model of the various empirical , or semiempirical, relationships for a
chemical and fluid dynamic processes, is unlikely particular range of circumstances and these re-
to be realized until a better understanding of the lationships are of use in theoretical models of spray
fundamental mechanisms has been achieved. combustion processes.
Both vaporization and mixing processes in
2.3.3.2 Droplet transport. - The initial com- sprays are then dependent on the ballistics and
bustion characteristics of most liquid-fueled high- trajectories of individual drops, or groups of
intensity systems depend largely on the related droplets, which may be determined from em-
processes of atomization and spray mixing origi- pirically derived drag coefficients, CD, using an
nating at the injector. This is especially true of expression of the form740
the liquid bipropellant rocket engine where the dVL CDPAV AV
injection process is employed both as a means of = 0.75 (2.3.3-3)
dt PLdL
imparting a large surface area to the liquid volume
for vaporization purposes and for encouraging in which AV is the relative velocity between
spray dispersion as an aid to the mixing of fuel droplets of diameter dɩ and liquid density PL, and
with oxidant, see Fig. 2.3.3a . Previously discussed surrounding gas of density p . In practice it has
been found that the drag coefficients of liquid
* Even this approach is much less costly than hot droplets , especially under evaporating conditions
firings. and high Reynolds number, depart markedly from
70 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

those derived for solid spheres , i.e. Stokes Law in In addition to the time-averaged spatial distri-
which bution and propellant mixing processes attributed
CD = 24Re-¹ (2.3.3-4) to the liquid injection mechanism ; it may also be
necessary to consider temporal variations in mass-
where Re is the Reynolds number based on droplet flux , and hence in mixing, caused by periodic
diameter, relative droplet to gas velocity and the
disruption of the liquid streams during the atomi-
appropriate gas properties. Correlations which
zation process .446, 341 Small- scale combustion studies
allow both for liquid evaporation374 and for droplet have confirmed the presence of these periodic
distortion at high relative velocities569 with burning variations in terms of light emission and temper-
and non-burning fuel droplets are
ature at a point, and good correlation has been
CD = 27Re-0.84 0 <Re <80 obtained with measured frequencies. It may be
= 0.271 Re0.217 80 <Re <10 (2.3.3-5) noted that these periodic fluctuations in mixture
=2 Re >104) ratio also give rise to apparent anomalies in the
composition of gases sampled from the combustion
The importance of recognizing the effects of zone which suggest unrealistic departures from
evaporation and drop distortion , implicit in these chemical equilibrium among the sampled con-
correlations of CD with Re , has been demon- stituents.200
strated419 in a comparison of experimentally de- Finally, a simplified analytical treatment of
termined burning rates with theoretical predic- forced mixing between unlike sprays has been
tions for droplet evaporation under typical rocket considered in the formulation of a semi-empirical
combustion conditions . combustion model421 based on the use of adjacent.
It is now theoretically possible by statistical sprays of fuel and oxidizer produced by impinging
methods to provide a comprehensive model for jet "like-on -like" injection elements. Mixing was
the subsequent droplet transportation mecha- assumed to take place instantaneously at the
nism743 and thus derive a complete distribution of boundaries of "wedge" shaped regions containing
mass and composition within the combustion fuel or oxidant sprays whose configurations were
chamber, including the probability of droplet col- determined from cold flow experiments. A com-
lisions producing either agglomeration or liquid parison with experiments in a two-dimensional
phase reactions . However, the complexity of the combustion chamber showed that the theoretical
problem has, so far, defied rigorous analysis and model, which included spray mixing as a rate
in practice numerical solutions have only been controlling process near the injector face, was
obtained after drastic , or perhaps oversimplifi- better able to describe the measured velocity
cation of the mathematical expressions and the distributions than one where vaporization alone
physical model. Instead, the majority of work on was considered.
spray mixing under rocket combustion conditions None of the studies of spray mixing and droplet
has been by inferential observations based on the transport processes, whether theoretical or experi-
flame patterns produced in semi-transparent mental in nature, have yet considered adequately
chambers (e.g. , Refs . 246 , 111 , 419 and 421 ) and the effects of operation at high pressures when the
on direct measurements of composition or tem- critical temperature of either, or both, droplet
perature profiles.4445 systems could be exceeded.
Small-scale studies of single and multiple fuel
sprays under simulated rocket combustion con- 2.3.3.3 Vapor mixing.-It has long been recog-
ditions445 have shown that nonuniform mixture nized from high-speed motion and streak pho-
distributions in the combustion products could tography that some nonuniformity in gas compo-
be related to cold-flow measurements of mass-flux sition and temperature distributions persist.
distribution . The technique has also been used to throughout the rocket combustion chamber and
demonstrate that the additive effects of adjacent into the expansion nozzle despite efforts to pro-
sprays of the same component, measured under mote fuel and oxidant mixing in the liquid phase
cold-flow conditions, are reproduced in the com- by direct impingement or spray interpenetration .
bustion product distributions. Moreover, these striations in mixture strength
 STEADY- STATE PROCESSES $ 2.3 71

have been demonstrated to continue far beyond. solutions obtained at various locations down-
the point at which vaporization processes were stream from the point sources by assuming values
complete, 445 so that final mixing of the propellants for the scale and intensity of main stream turbu-
was taking place through turbulent diffusion lence . Results of this analysis were presented in
processes in the vapor phase. In practical systems terms of a "ripple factor," y, defined as the ratio.
where it becomes possible to achieve good atomi- of the minimum to the maximum concentrations
zation , for example by the use of blast atomizers * across any given plane normal to the direction of
with oxygen /hydrogen or H.T.P. †/hydrocarbon flow. This ripple factor was found to depend on
propellant combinations, it is probable that vapor the ratio of downstream length to the separation
phase turbulent mixing becomes the dominant distance between sources and on the intensity,
rate-controlling mechanism in the steady-state but not the scale , of turbulence . For the selected
combustion process. Although a high degree of values of separation distance and turbulence in-
turbulent mixing is clearly desirable if high per- tensity it was concluded that turbulent mixing
formance is to be achieved ; there is also evidence rates were large (at least in the principal com-
that excessive mixing is unnecessary and can , bustion zone) compared with those of vapori-
under certain circumstances, degrade the com- zation and, therefore, that the latter process was
bustion efficiency.214 dominantly rate-controlling.
Turbulent mixing between gas streams has re- Experimental studies of vapor phase mixing
ceived considerable attention in connection with under rocket combustion conditions have used
air-breathing combustion systems of all types and direct gas sampling methods,445, 447 the addition of
an extensive coverage of the subject is available a tracer,350 or spectroscopic measurements129 to
in the published literature. Less attention has evaluate changes of composition profiles along the
been paid to this topic under rocket combustion length of the chamber. In the particular case of
conditions ; but a theoretical analysis becomes oxygen/hydrogen combustion with concentric jet.
feasible if the intensity and scale of the turbulence injection systems , radiation measurements129 con-
are known, together with the configuration and firmed that turbulent mixing controlled the extent
aerodynamic properties of the gaseous sources . A of reaction if liquid oxygen was injected through
further assumption has to be made, or empirical a central jet surrounded by an annular stream of
data have to be used, on the influence that com- gaseous hydrogen . However, if the propellant
bustion can have on physical properties and on streams were interchanged, a condition which
the turbulent intensity. Although it has been might be expected to enhance mixing near the
postulated that discrepancies between theoretical injector by expansion of the gaseous core, it was
and measured flame propagation rates were due found that propellant vaporization appeared to
to "flame generated turbulence''397 and this con- become rate controlling. A similar result was ob-
cept has received wide acceptance, serious doubts tained by assessing the variation in combustion
about its validity have recently been expressed.437 efficiency with chamber length349 and this trend
.
A simple analytical treatment100 of turbulent. has also been observed by other workers at
mixing under rocket combustion conditions was differing engine scales.
based on equations developed for the injection of The influence of turbulent mixing on combustion
gas from a point source into a uniformly turbulent performance has been analyzed 351 using as a basis
gas stream in the absence of chemical reactions . the earlier analytical treatment of the decay of
Molecular diffusion processes were ignored and concentration profiles.100 Numerical solutions for
eight propellant combinations and a range of
* Blast atomizers may be defined as those in which the chamber geometries were expressed in terms of a
primary source of energy for liquid breakup is derived mixing parameter, TL/S , the product of turbu-
from a suitably oriented jet of gas, usually at high-velocity. lence intensity and the ratio of chamber length to
H.T.P. (High Test Peroxide) . Concentrated solutions
injector element spacing. Reasonable correlation
of hydrogen peroxide in water, (usually 85 to 100 weight
was established with experimental measurements
percent hydrogen peroxide) which are used either as a
monopropellant or as an oxidizer in a bipropellant system and this "mixing model" enabled conclusions to
for rocket propulsion applications. be reached on the effects of maldistribution on
72 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

combustion performance and optimum perform- vortex generation and nonuniform recirculation .
ance mixture ratio. However, an apparently anom- Under rocket combustion conditions the "pockets"
alous result was the decrease of turbulence in- of gas flowing past a point can vary in composition
tensity along the chamber axis, from values as between the extremes of pure fuel and pure oxidant.
high as 10 percent near the injector to about and it has been found200 that these variations about
2 percent 8 inches downstream. the mean value were best described by the beta-
More recently, as part of an analysis of con- distribution function , F ( C ) , given by
vective heat transfer within the combustion cham-
F (C) ACP-1 ( 1 - C ) -1 (2.3.3-8 )
ber, a better description of experimental results
on turbulent mixing processes has been achieved.531 for 0 < C < 1
In this theoretical model it is assumed that the
where p and q are constants limited to positive
root mean square lateral velocity, and not the
values and A is a normalization constant defined
turbulence intensity, remains constant throughout
as
the chamber length . Thus, as combustion takes
place and the mean axial velocity increases , the
A-1 = CP-1 ( 1 — C ) -¹dC ( 2.3.3-9 )
turbulent intensity will fall away in a manner fo
closely similar to measured observations . The
same basic mixing theory as that developed for = ẞ ( 1 , p , q)
hydrocarbon/air combustion637 was used to predict in which the concentration , C , is expressed as the
the concentration , C, derived from a point source ratio of oxidant-equivalent moles of fuel to the
of strength, W, at a radius, r, from the longi- total oxidant equivalent moles of both fuel and
tudinal axis of the source, given by oxidant , i.e., С = (2nc + nн) /( 2пc + nн +no) .
W This analysis was successful in explaining ap-
C == • exp ( −r²/4g ) (2.3.3-6) parent departures from chemical equilibrium in
4ng 447
samples collected over finite time intervals.4
Moreover, as a corollary, it was concluded that
in which g is a spreading factor defined by
conventional gas sampling methods and a com-
g = StVrmst + S [ exp ( − Vrmst /St) -1 ] ( 2.3.3-7 ) plete chemical analysis could be used to assess
the magnitude of mixture ratio fluctuations at a
where S is the Lagrangian scale of turbulence, point.
Vrms the root mean square lateral velocity , and t 2.3.4 Recirculation
the time after injection . The adoption of a con-
stant value of r.m.s. lateral velocity, Vrms, has The significance of recirculation as a flame-
been found by computation to lead to a final holding, or piloting, mechanism in air -breathing
turbulence intensity of the order of 1.5 percent combustion systems of all types is well known
with complete combustion and thus is in good and has been the subject of a number of investi-
agreement with experimental observations . gations reported in the published literature . These
Most experimental and theoretical studies of have ranged from aerodynamic studies of the
vapor phase mixing have been concerned with conditions under which recirculating flows and
changes in composition , which may be described vortices can become established in confined ducts ,
in terms of mixture strength profiles, over sections usually of very simple geometrical shape, to as-
of the combustion chamber normal to the direction sessments of the effects of recirculation and en-
of gas flow. However, nonuniformities may also trainment on the mixing between coaxial streams .
be present in an axial direction which give rise Experiments have been carried out both with
to random, or periodic, fluctuations in composition reacting and non -reacting constituents and, al-
about a mean value at any given point. It has though much of the work is not strictly applicable
already been mentioned that these fluctuations to rocket conditions, many useful qualitative ob-
can originate from the spray breakup mechanism . servations and generalizations emerge.205, 77, 84
(see Sect . 2.3.3.2 ) , but they can also be caused by In the liquid propellant rocket engine, gaseous
gas dynamic processes near the injector face, e.g. , recirculation phenomena are normally confined to
 STEADY-STATE PROCESSES $2.3 73

a narrow region of the combustion chamber near can be clearly identified with local flow patterns
the injector where they can be induced by en- and velocities .
trainment and momentum exchange between the Most experimental investigations of recircu-
injected propellants and the surrounding gas. lation under rocket combustion conditions have
These entrainment effects give rise to high local been by qualitative observations, using high-speed
turbulence" and a reverse flow of gas which, photographic techniques, as part of more general
depending on its source, consists either of cool studies of the flame patterns in transparent com-
unburned propellant vapor or of hot combustion bustion chambers . Moreover, few attempts have
products. Moreover, the recirculation flow pat- been made to confine attention to the recirculation
terns may be localized around each individual zone itself in order to relate spray geometry or
injector element, or else may be of larger scale propellant momenta to the resulting flow re-
giving rise to "cross winds" over appreciable areas gimes . 201,275 One detailed study of recirculation as
of the surface of the injector. Recirculation can, a flame piloting mechanism689 did , however , in-
therefore, be either beneficial or detrimental to vestigate in a systematic manner the significance
the stable combustion processes and reliable oper- of selected design and operational variables on the
ation of the engine according to the scale and overall performance of the system. In this es-
intensity of the flow, also depending on the compo- sentially experimental program, the effects of basic
sition and temperature of the gases taking part. chamber geometry and propellant properties, in-
When injector designs and the rates of chemical cluding mixture ratio, were assessed in terms of
reaction are such that only cold unreacted pro- mixing profiles derived from gas sampling and
pellant vapors form the recirculating flow, it has analysis, both for reacting and non -reactive con-
been stated159 that the resultant eddy can have ditions. Nevertheless , further work of this nature
but little effect on the combustion process. While is needed before reliable generalizations can be
this may be substantially true for highly localized made and the importance of recirculation in the
effects in which the injected stream only entrains more general context of vapor-phase mixing can
its own vapor ; other evidence is available to indi- be established .
cate that if unlike vapors become entrained then In addition, limited studies of the flow regime
mixing is improved and there is an increase in the near the injector face have been made using flow
initial rate of heat release. visualization techniques similar to those developed
In the case of recirculation of combustion for application to air-breathing combustion sys-
products, there are claimed to be three inde- tems . Of these, possibly the most interesting is
pendent ways in which they can assist com- that in which the phase change produced by flash
bustion.159 First , they are able to provide a con- vaporization is used to simulate the density
tinuous heat source for ignition purposes and, if changes under combustion conditions,360 although
not balanced in composition , are capable of further it is evident that considerable development of this
reaction with the propellant vapors . Secondly , technique is necessary before it can be regarded
they enable combustion processes to be brought as being of general usefulness.
into the "core" of each spray system. Finally, the The importance of the momenta of the injected
resultant evolution of gaseous combustion prod- propellants in determining the nature of the re-
ucts within the sprays may accelerate the spatial circulation has been demonstrated experi-
expansion process, thereby leading to more rapid mentally -526 by subjecting a typical impinging jet
spray mixing. Nevertheless , despite this awareness injector to a wide range of differential pressure
of its beneficial effects as a flame piloting mecha- (i.e. , mixture ratio ) conditions at a constant
nism, there has apparently only been one instance, chamber pressure. Temperature measurements at
in a thermal ignition H.T.P./kerosene rocket en- selected points on the surface of the injector
gine, where deliberate design provision was made showed the anticipated approximate relationship
to make use of this phenomenon . On the other with the gas temperature and mixture ratio as
hand, an excessive reverse flow of hot combustion long as the total momentum of the oxidant flow
products can have detrimental effects in causing exceeded that of the fuel. A complete reversal of
overheating of the face of the injector, 116 which this trend was observed at very fuel-rich mixture
74 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

conditions as fuel momentum became dominant , and Agoston,752 F. A. Williams740 and A. Williams.741
Fig. 2.3.4 , when surface temperatures rose rapidly The following nomenclature pertains to Sect.
despite the continuing fall in the mean com- 2.4 :
bustion temperature . Further evidence from these a Stoichiometric coefficient
investigations showed that when oxidant mo- B Dimensionless interfacial mass transfer
mentum dominated the recirculation process the rate (transfer number)
reverse flow was fuel-rich in composition , but the B Pre-exponential factor
converse was true when the fuel sprays provided b Stoichiometric coefficient
the main entrainment mechanism . C Constant in Eq. ( 2.4.1-8 )
Therefore, empirical work has shown that re- CD Drag coefficient
circulation can provide an important contribution DI First Damköhler number
to the stability and intensity of the combustion D Fick binary diffusion coefficient
process in the injection region. Although no clear dL Droplet diameter
proof exists, it is also logical to assume that dis- Eact Overall activation energy of homogene-
turbances in the recirculating flow regime can ous reaction
produce related variations in the heat release rate he Heat of vapor phase combustion per
in the primary mixing region and thus establish unit mass of oxidizer (Sect . 2.4.2) or
a possible driving mechanism for combustion of fuel (Sect. 2.4.3)
instability. k Reaction rate constant
Le Lewis number, Pr/Sc
m" Mass flux per unit area, m/πгL²
2.4 LIQUID DROPLET VAPORIZATION
n Overall reaction order, a +B
AND COMBUSTION *
Pe Peclet number, Re Pr, based on di
Far from literally atomizing (or "moleculariz- req
Equivalent initial release radius at pre-
ing") liquid propellants, rocket motor injectors vailing pressure (supercritical com-
produce nonuniform sprays of droplets with the bustion)
median size typically containing some 1015 mole- Ist Stoichiometric oxidizer/fuel mass ratio
cules/droplet . Owing to low liquid phase diffusion Su Laminar flame speed with respect to
rates and temperatures this initial state of mo- unburned gas
lecular aggregation is ordinarily incompatible with √∞
Relative velocity (absolute value) be-
high volumetric chemical reaction rates. Hence tween droplet and gas , | V - Va |
rapid droplet vaporization is essential to compact α Reaction order with respect to oxidizer
and efficient combustion . The combustion be- vapor
havior of such sprays is indeed a statistical conse- B Reaction order with respect to fuel
quence of the vaporization /combustion behavior vapor
of its interacting individual droplets, each of which K Apparent evaporation coefficient ,
finds itself in a rather extreme and time-varying -d ( d₁² ) /dt
local environment (even during stable combus- V Kinematic viscosity, μ/p
tion ) . In this section, we briefly discuss the results. T Dimensionless time, Dt/reff2
of fundamental research on individual droplet Χ Stoichiometry parameter,
vaporization , with emphasis on the significant (Yox, /YF, L) / Ist
effects of localized gas phase chemical reaction in
Abbreviations :
augmenting the vaporization rates of monopro-
pellant and bipropellant droplets (fuel in oxidizer bp Boiling point
vapor or vice versa ) . While some overlap is in- B-S Burke-Schumann
evitable, attention will be directed at important DS Distributed-source
experimental and theoretical developments since fct() Function of ( )
the comprehensive reviews of Spalding,661 Wise LOX Liquid oxygen [O2 ( L) ]
PS Point-source
* D. E. Rosner, Author. QS Quasi-steady
 STEADY-STATE PROCESSES $2.4 75

760 200

740-
Povo2Ao
Liquid oxygen Povo²

720- 150
Temperature

X
PfVq2Ap
Kerosene

/ec
lb
700-

s
X
C
°

Apv2
Momentum
6
680- 100

•
Maximum
temperature

660

640- 50

620

ON

600 10
1,0 1.5 2.0 2.2
Mixture ratio by weight [ LOX /kerosene ]

FIGURE 2.3.4 . - The influence of liquid propellant momentum on injector face temperature.

Gas cr At critical point

33

(g)
(L) Liquid comb With combustion
μ Micron (10-6 m) d Droplet
(-) Average value of ( ) D Pertaining to diffusion
Concentration eff Effective
[ ]
Subscripts : eq Local equilibrium
ad Adiabatic ext At extinction
AE Auto-extinction H Pertaining to heat conduction
ΑΙ max Maximum (B -S limit)
PU

Auto-ignition
bp Boiling point P Product
BO Blow-off u Unburned mixture
8ུྪབ་
LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

སྒྲ
*
ཙ
ཚ Vapor While such laboratory experiments can provide
vap Vaporization valuable insight into the possible life-history of a
At wall (vaporizing surface) droplet in a rocket motor combustion chamber,
Wet-bulb direct use of these data is precluded by the vast
At infinity (far from droplet ) gap in environmental conditions (pressure level ,
droplet size, relative flow velocities, turbulence
To fix ideas it is useful to examine the experi- level ) ordinarily separating well-defined labora-
mentally observed behavior of an isolated fuel tory experiments from rocket chamber conditions .
droplet when suddenly confronted with a stag- For this reason considerable attention has properly
nant, hot oxidizer ( containing) gas (see Figs . been focused on acquiring an understanding of the
2.4a and 2.4b ) . Such experiments518 typically physical and chemical phenomena governing each
involve photographically following the rate of stage in the droplet lifetime . In the absence of
diameter decrease of a fuel droplet suspended on such understanding it would clearly be impossible
a thermocouple junction. One notes the existence. to ( 1 ) generalize (correlate ) the results of avail-
of the following distinct regimes : ( 1 ) transient able experiments, ( 2 ) reliably extrapolate the
heat-up interval during which the droplet surface results to environmental conditions anticipated
temperature rises to the (nonreactive) "wet bulb" in rocket combustion chambers, and (3) correct
temperature with some change in droplet diameter , for artifacts of the experimental configuration .
dL (2 ) a vaporization period during which dr2 Even in the presence of such understanding,
decreases linearly with time with the droplet however, the application of such fundamental
temperature remaining nearly constant, and (3 ) data on droplet behavior to rocket motors con-
an "augmented" vaporization period (d12 again tinues to tax the ingenuity of designers.
linear in time ) associated with the presence of a While radically new experimental techniques
diffusion flame surrounding the droplet . Heat have apparently not been introduced , methods
generated by the localized gas phase reaction of used successfully in the past have been extended
fuel vapor and ambient oxidizer drives the surface to provide interesting new data on the drag
temperature up to near the fuel boiling point , and coefficients of vaporizing and burning bipropellant
sustains the enhanced vaporization rate until droplets, vaporization and burning times for
droplet consumption or extinction.

2 QS - Comb
dLo

Kvap

2
d

Kcomb
Pre - Heat QS - Evap

tign tlife
t

FIGURE 2.4a . -Square of droplet diameter versus time ; fuel droplet placed in hot stagnant oxidizer-containing gas.
 STEADY-STATE PROCESSES $2.4 77

T ( p)
'bp
Twb( comb)

-Twb (vap)

Tw

Pre Heat QS - Evap QS - Comb

O
tign tlife

FIGURE 2.4b .- Droplet surface temperature versus time ; fuel droplet placed in hot stagnant oxidizer-containing gas.

bipropellants at extremely high (supercritical ) long enough to ensure that unconsumed droplets
pressures, and monopropellant burning rates and do not emerge. Accordingly, several authors have
flame structure (in ambient environments which considered the design consequences of deliberately
react with the primary decomposition products of simplified one-dimensional steady flow models in
the droplet ) . Previous theoretical models of the which propellant vaporization is the rate-con-
combustion of bipropellant and monopropellant trolling process . In the most complete numerical
droplets have been considerably extended , in study of this type,565 it is assumed that ( 1 ) all
part with the help of new mathematical techniques droplets in the initial spray have the same initial
and/or approximations. These extensions are of velocity and never shatter or coalesce, and ( 2 )
three types : i.e. , those which ( 1 ) increase the properties of the bulk gas flow (resulting from
accuracy of absolute predictions (burning rate, propellant vaporization ) correspond to instan-
flame shape, etc. ) , (2) extend the domain of taneous stoichiometric combustion . The results of
applicability of previous results ( e.g. , to higher single drop studies play an important role in such
pressure levels , incipient forced convection ) , and models since the initial spray distribution is
(3 ) enable a qualitative understanding of the represented by several different groups of drop
effects of physical and chemical kinetic parameters sizes (each containing a fixed percent of the initial
(e.g. , on flame zone structure, burning rate, mass of the spray) whose histories are then
environmental conditions enabling droplet ignition
. computed from the laws governing droplet heat
or leading to extinction , etc. ) . As a byproduct , transfer, mass transfer and drag.
this work has delineated the limitations of pre- Most experimental droplet evaporation studies
vious theoretical models of droplet burning, and have been carried out using the above-mentioned
highlighted aspects of droplet burning having "captive droplet " technique (e.g. , ≥500μ droplet
direct relevance to rocket motor combustion suspended on one (or more ) thermocouple
instability. In what follows we outline these junction (s ) or a silica filament ) , in which forced
advances against the voluminous background of convection or natural convection flows of hot
previously available droplet results . gases augment the droplet vaporization rate.
Smaller droplets can be accommodated using
2.4.1 Droplet Heat-Up and Vaporization
"free-flight" techniques in which droplets are
Combustion chamber "stay-times" must be projected,363 or fall freely through hot gases
78 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

produced by a furnace or flame.240,757 In the latter nearly linearly with instantaneous droplet
case drag measurements can also be obtained for diameter.
vaporizing droplets by simultaneously recording Several alternative, but essentially equivalent ,
droplet position-time data.240 As noted below, this quantitative formulations of the above-mentioned
information is required to predict droplet trajec- phenomena have been reported . Our choice here
tories, instantaneous velocity lags (between local is based on conciseness and ease of generalizing
gas and droplet ) and, hence, instantaneous the results to cases in which combustion occurs .
vaporization rates in combustors . In each case one considers a single droplet , and
Experimental studies on single-component expresses the facts that : ( 1 ) the time rate of
droplets have led to the following important change of its energy content is equal to the
observations and conclusions, 112, 248, 249, 302, 562 which instantaneous difference between the convective
form the basis of the quantitative model outlined heat input and the heat required for surface
below: ( 1 ) in a quiescent environment the time to vaporization , ( 2 ) the time rate of change of
achieve the droplet wet bulb temperature is droplet mass is governed by the rate at which its
proportional to dro² and can be an appreciable vapors can diffuse away through the background
portion of the total vaporization time, ( 2 ) owing gas, and (3 ) the product of instantaneous droplet
either to small droplet size, or internal circulation mass and acceleration is equal to the net aero-
within large droplets , actual drop temperature dynamic (drag) force acting on the droplet.
histories are well represented by neglecting Explicitly, the resulting coupled equations
internal temperature gradients * ( compared to governing heat transfer, mass transfer and drag
external gradients ; i.e. , the "infinite liquid thermal for a quasi-spherical droplet of instantaneous
conductivity approximation ") , ( 3 ) for droplets of diameter d₁ ( = 2гL ) are, respectively,
high volatility the unidirectional mass transport
d NUH
process causes the heat transfer, mass transfer , [( ri) PLCLTL ] = ( 4πrL²) R (T - TL)
dt dL
and drag coefficients to fall below those appro-
priate to solid (nonvaporizing ) spheres at the - (4гL2 ) m'hy (2.4.1-1 )
same Reynolds number, ( 4 ) despite the transient
1 d
nature of the droplet heat/mass transfer problem, ( TYL³µL) = m "
step-by-step application of relations strictly (4πг12) dt
derived for steady state conditions usually leads
to accurate predictions†, (5 ) after the droplet Nup (Yv.w- Yv.00
= Dp ( 2.4.1-2 )
temperature reaches the "wet bulb" temperature di 1 - Yv ,w
(i.e., when heat transferred from the hot gas
dVa
environment is used almost exclusively to vaporize = d ( πrL² ) CD
PL ( TTL³) = ½µ ( V — Va) | V - Va│
liquid) the droplet surface area decreases nearly dt
linearly with time (see Fig. 2.4a ) ‡ corresponding (2.4.1-3 )
to a mass rate of volatilization , m , decreasing
Here the station subscripts L, w, refer respec-
tively to the liquid, the gas/vapor mixture at
* Internal absorption of IR radiation can produce the liquid/gas interface (Tw = TL ) , and the
unusual temperature profiles within droplets (see, e.g. , ambient stream ("far" from the droplet) ; the
H. C. Hottel et al.363) , however, calculations indicate that remaining symbols are defined in the Nomen-
this effect is probably negligible in most rocket motors and
clature. The subscript d (Eq . (2.4.1-3 ) ) refers to
industrial furnaces . It is also clear that the present discus-
the droplet, unsubscripted variables pertain to the
sion does not apply to droplets capable of undergoing
exothermic liquid phase reaction. gas, V - Va = V. being the instantaneous
† Limitations of this "quasi-steady" (QS) approxima- absolute value of the relative velocity between
tion will be discussed later on. droplet and gas. The dimensionless heat transfer ,
Systematic departures from the "d2-law" occur in the
mass transfer, and drag coefficients Nuн, Nud ,
presence of non-negligible convective contributions (free
or forced), since convection causes diameter-dependent and CD appearing (and essentially defined ) in
heat transfer augmentation. Eqs . (2.4.1-1 , -2, and -3 ) depend not only on the
 STEADY-STATE PROCESSES $2.4 79

local Reynolds number* Re VdL/ , but also Yv.w = Yv.eq (TL) . Once solved for TL.wb, the value
on the dimensionless mass transfer rate per unit of Yv.eq (TL.wb) can be inserted in Eq . ( 2.4.1-2 )
areat: to provide the quasi-steady vaporization rate
m"dL Yv.w-Yv.co during the period TL ≈ TL.wu (see Fig. 2.4b ) .
B= = (2.4.1-4 ) In the extreme case in which T. greatly exceeds
DoNud 1 - Yv.w
the boiling point of the liquid, * examination of
Physically, this is largely due to distortion of the Eq . (2.4.1-5 ) reveals TL,wb-Tbp, hence
radial profiles of vapor mass fraction Yv, tempera-
Cp (To - Tbp)
ture, T, and tangential velocity due to convective Bvap≈ (2.4.1-6)
hy
transport associated with the radial mass transfer
from the droplet. Thus, the simplest possible This limiting result , when combined with the
forced convection correlations which include these appropriate Nu ( Re , Pr, B ) relation , enables m"
essential phenomena would be of the form‡ to be calculated without the need to explicitly
Nuн ( Re, Pr, B) , NuD (Re, Sc , B ) , and CD ( Re, B ) , consider the vapor diffusion equation . For this
where property values are introduced as some reason, this extreme is (somewhat misleadingly)
suitable "mean" condition representative of the called "heat transfer controlled."
local environment within the boundary layer Early work on droplet vaporization in forced
(see below) . convection environments was based on "syn-
Before presenting recent experimental and thesized" correlations of the separable form : 657, 661
theoretical results for the dimensionless transfer
coefficients, some comment is in order on the In (1 + B )
Nu ( Re, Pr , B ) =
: NUB-O (Re, Pr )
(frequently invoked ) simplifications arising when B
( 1 ) the convective heating of the droplet and
(2.4.1-7)
evaporative cooling are nearly in balance ( i.e. ,
when TL TL , wb ) , and ( 2 ) Le≈1 . Then Eqs. where, NuB-0 (Re, Pr) constitutes the corre-
(2.4.1-1 ) and (2.4.1-2 ) can be combined to sponding solid sphere Nusselt number, commonly
provide a simple implicit equation for TL,wb,661 taken ast
viz :
NUB-O (Re , Pr ) = 2 ( 1 + CRe¹/2Pr1/3) ( 2.4.1-8 )
Yv.eq ( TL, wb) -Yv.∞
Bvap , wb = However, the logarithmic "driving force"
1 - Yv.eq (TL, wb)
In ( 1 + B ) is known to be exact ‡ only in the
Cp (To - TL.wb) stagnant limit (Re-0, NUB-0-2 ) . Moreover ,
= (2.4.1-5 ) available constant property solutions to the
hv (TL, wb)
conservation equations indicate that ( 1 ) the
where we have introduced the assumption of local In ( 1 + B ) driving force overestimates laminar
equilibrium at the gas/liquid interface, viz. flow mass transfer rates at high Reynolds number
and large B, ( 2 ) except as an empirical approxi-
* The Grashof number must be considered in the mation , it is unlikely that the transfer coefficients
presence of appreciable bouyancy. can be written in the separable form fct ( B ) •
† Sometimes referred to as the "transfer number. " 661
fet ( Re, Pr) at all Reynolds numbers, and (3 )
The second part of Eq . 2.4.1-4 follows from a mass balance
at the liquid/vapor interface by invoking the conditions
of ( 1 ) no vapor phase chemical reaction , (2 ) Fick's law * This is defined by pv. eq (Tbp) = p , where p is the
of diffusion, and (3 ) ambient ( "foreign" ) gas impenetra- prevailing total pressure .
bility into the droplet. This semi-theoretical form has been suggested by
In the absence of gas phase chemical reaction , the 280 Ranz and Marshall,577 and, most recently,
Frössling,28
differential equations governing heat and mass transport Lee and Ryley.435 Recommended values for the constant C
are similar , leading to the expectation that Nuн (Re, Pr, are 0.276, 0.30 and 0.37 , respectively.
B) and NuD (Re, Sc , B) will have similar functional forms. The (implicit ) law Nu/Nuв_o = [ln ( 1 + B ) /B] is
For gas mixtures relevant to combustion the diffusivity equivalent to the (explicit ) law99 (utilized by Priem , 562
ratio D/ R /(p cp) ] (or Lewis number, Le = Pr/Sc) is fre- El Wakil et al.248,249 ) Nu/NuB_0 = B ' / [exp (B ' ) -1 ] , where
quently near unity, in which case Nuн≈NUD. B' = m'dL/( Dp NUB-0)
 80 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

even in the absence of mass transfer Eq. ( 2.4.1-8 ) free-fall. These authors state that the use of
overestimates the incipient effects of small [In ( 1 + B) ]/B in the ordinate was somewhat less
forced convection when (Re < 1 ) . * successful than (1 +B) -1, again suggesting that
Despite these "theoretical" objections, experi- the driving force In ( 1 + B ) somewhat under-
mental correlations of the above-mentioned estimates the "blowing effect" (thereby over-
separable form have proven reasonably successful estimating mass transfer rates) when laminar
and, although there is a need for simple design forced convection is significant.‡
formulae, few practical alternatives have been Corresponding drag coefficient data, correlated
suggested . The meaning of "reasonably successful" in the same way, are shown in Fig. 2.4.1b . Here
can be judged from the recent experimental work the " standard sphere drag" curve is CD (Re ) for
of Eisenklam et al . ,240 who photographed 25-2000μ solid inert spheres (B = 0 ) . With regard to the
droplets of six distinct liquids falling freely drag on bluff objects it should be commented
through unheated air. Figure 2.4.1a shows heat that the reduction factor [ln ( 1 + B ) ] /B has no
transfer coefficients† inferred from observed theoretical basis, even in the Stokes flow limit :
evaporation coefficients Kvap [where Kvap= Re<< 1 . A self- consistent formalism for calculating
-d (d₁² ) /dt ] during the TL≈TL.wb portion of the the correction to the Stokes drag law for a
vaporizing spherical droplet with no internal
* circulation has recently been outlined by Fendell
Improved semi-empirical forms of Eq. 2.4.1-8 can be
developed for small Peclet numbers (Pe = Re⚫ Pr) and B = et al.266 who have calculated the reduction in
O based on the fact that the proper small Pe behavior has CD • Re for several illustrative cases. Unfortu-
been shown to be Nu = 2 [ 1 + (1/4) (Pe ) + ··· ] , with
nately, however, simple closed-form corrections do
the next term being of order Pe² In Pe.
In the presence of appreciable property variations
(due to any cause) a " correlation" is of no use if an author However, a driving force like B/( 1 + B) would appear
has not carefully stated the conditions under which to be unreasonable on theoretical grounds when B→ ∞
property values are to be (or have been) inserted . In Fig. since there is no evidence that transfer rates would become
2.4.1a, based on the work of Eisenklam et al.240 all property independent of B at sufficiently large B. The maximum
values are those of the surrounding medium (air) evaluated value of Bvap in the experiments of Eisenklam et al.240 was
at the temperature (TL ,wb + T∞) . 3.

20

Evaporating drops
Nu (1 + B ) = 2 + 1.6 Re "/2
10 • Heptane
Pentane
O Methyl alcohol
5

x Ethyl alcohol
Nu
B
(1+
)

Nu ( 1 + B )= 2 + 1.6 Re 1/2

ΤΟ Evaporating drops
+ Water (temp . 200-1000 °C )
5 ▲ Benzene

2
10-2 10-1 10° 10 ' 102
Reynolds number , Re

FIGURE 2.4.1a .-Correlation of heat transfer data for a vaporizing droplet (after Eisenklam et al.240) .
 STEADY-STATE PROCESSES $2.4 81

1000

Standard Cp curve

O Methyl alcohol
× Ethyl alcohol

X
Δ Benzene
100

O
• Heptane
Pentane
(8 +1)

+ Water

10

X
1.0 ‫لس‬ ‫لسي‬
0.1 1.0 10.0 40.0
Re

FIGURE 2.4.1b .- Correlation of drag data for a vaporizing droplet (after Eisenklam et al.240) .

not emerge, hence additional computations will completely vaporize a droplet in a particular high
be required to establish trends . temperature environment : (1 ) increases as the
Before considering extreme conditions which tax square of the initial droplet diameter, ( 2 ) di-
the validity of Eqs . ( 2.4.1-1 , 2 , 3 ) , it is instructive minishes for fuels of high volatility (low Tbp,
to briefly examine the simplest possible case, low latent heat ) , and ( 3 ) depends only weakly on
viz , the evaporation constant for a small droplet the prevailing pressure level. *
preheated to near the wet bulb temperature, It is now appropriate to consider several circum-
suddenly placed in a high temperature stagnant stances in which the previous equations fail to
gas. Then, for Le≈ 1 , the previous "quasi-steady" apply circumstances which, it turns out, are
equations yield661 often intimately connected with unstable rocket
motor combustion . Thus, essential modifications
dL.02
Kvap = are to be expected when either ( 1 ) parameters or
tvap
boundary conditions (p , Yv ... , T. , V . , ...)
8 Cp ---
= In (2.4.1-9) * In Eq . (2.4.1-9) only Tьp and hy noticeably depend on
) in [ 1+
2 (Cp
- PL - Tip) ]
(T₂ hy pressure level. However, these effects are slight (at pres-
sures well below the critical pressure of the fuel) , act in
This equation quantitatively expresses the often opposing directions, and appear within the logarithmic
verified behavior that the time, tvap, required to term .
82 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

undergo appreciable percentage changes in times. Since some droplets can apparently reach their
not large compared with r12/D, ( 2 ) the droplet critical points prior to significant fractional
temperature is driven up to near the thermo- vaporization loss or velocity equilibration , it is
dynamic critical temperature of the liquid , or likely that droplet distortion , shattering and,
(3) exothermic chemical reaction ( either vapor ultimately, simple vapor/gas diffusion will then
decomposition, or fuel vapor/oxidizer vapor govern the rate of mixing . This class of "near-
combustion) occurs sufficiently close to the critical" phenomena may bear directly upon high
droplet surface to augment the heat transfer rate pressure rocket motor combustion instability, in
and, hence, vaporization rate (see Fig. 2.4.1a ) . view of the greatly increased pressure dependence
Items ( 1 ) and ( 2 ) both violate the fundamental of the "droplet" vaporization rates in this region .
quasi-steady assumption underlying Eqs. Analogous remarks apply to droplet combustion
2.4.1-1 , -2, and -3 according to which a truly at supercritical pressures (see Sect . 2.4.2 ) .
transient, moving boundary problem ( e.g., a The problem of enhanced droplet vaporization
shrinking droplet ) has been approximated by an rates in the presence of localized exothermic
"equivalent" steady problem with radial mass chemical reaction (within the droplet boundary
transport.740 Item ( 2 ) is intimately related to this layer ; see Fig . 2.4a ) is taken up in Sect . 2.4.2 ,
approximation since the radial interface velocity however it is appropriate to raise here two related
cannot be neglected ( compared to the radial gas questions : ( a ) Under what set of environmental
velocity at r = rL) when the liquid density is not conditions is ignition ultimately possible (for a
much greater than the prevailing gas mixture. particular droplet ) ? (b ) How long will it take
density , 742,434 On the face of it this conclusion before ignition occurs? The latter question is
would appear to cast doubt on calculations of the closely related to the transient heat-up/vaporiza-
type reported by Wieber,738 who used equations tion calculations already outlined. Indeed
essentially equivalent to ( 2.4.1-1 , 2 , 3) to show Williams742 has shown that the so-called droplet
that at supercritical pressures* heptane and "ignition -delay" is usually of the same order of
oxygen droplets can attain their critical tempera- magnitude as the time to attain the wet-bulb
tures following injection into 2780° K gas . How- temperature. However, tign for a particular fuel
ever, it is interesting to note that during transient may display ( 1 ) a dependence on oxidizer mass
heat-up an evaporating droplet can actually fraction Yox. in the surroundings , (2) a sensitivity
expand due to decreasing liquid density (see Eq. to ignition-promoting additives and ( 3) can be
2.4.1-2) as , in fact, Wieber's calculations reveal. appreciably shorter than the time to achieve
In such cases the absolute interface velocity need TL TL.wb. Accordingly, one alternative approach
not be large as PL pw, hence the accuracy of the for estimating a lower limit to tign is to estimate
QS approximation may not degenerate badly.§ the time required to first form a locally combustible
mixture (T, YF, Yox combination ) * within the
droplet boundary layer. Priem et al. ,562 and El
Wakil and Abdou,247 who have made such cal-
At the critical temperature of a substance, parent
"liquid" and the "vapor" with which it is in equilibrium culations, term this the physical ignition delay.
are equally dense -hence the two phases become in- To obtain the total ignition delay one must
distinguishable. As a consequence, both the latent heat of generally add on a chemical ignition delay, i.e. ,
vaporization and surface tension vanish as TL- Ter. The
the time required from the beginning of perceptible
equilibrium "vapor " pressure at Ter is called the critical
pressure, per (see Sect. 2.4.2 ) . homogeneous reaction to the establishment of the
* The critical pressure itself can be exceeded without envelope diffusion flame. Unfortunately, reliable
significant departures from the QS/finite latent heat estimates of the chemical ignition delay cannot be
droplet model. What is relevant is the supercritical pressure
level at which no subcritical wet bulb temperature (satis- * For this purpose it is often helpful to note that for
fying Eq. 2.4.1-5 ) exists.601 Le≈1 the following simple interrelation between Yox,
Only when PL = const . does Eq . (2.4.1-2) specialize to YF, T exists within the boundary layer :
the frequently used form -p (dr₁/dt) = m ” . T -T Yox, -Yox YF - YF..
§ A more quantitative statement must await additional
Tw - To Yox, -Yox,w YF , w - YF ..
calculations and/or measurements.
 8883
STEADY-STATE PROCESSES $ 2.4

made based on presently available data and Controlled droplet combustion experimentation
theory. has led to the establishment of the following
important facts and patterns. † ( 1 ) Combustion in
2.4.2 Bipropellant Droplet Combustion the immediate vicinity of a droplet can be of two
general types, depending primarily on the relative
An important consequence of spray size dis-
tributions and unequal droplet evaporation rates velocity V. between gas and droplet . At low V ..
for combustion chambers is that a fuel droplet can one observes distorted "envelope" flames char-
find itself in an oxidizer-rich vapor (or vice versa ) . acterized by a contiguous vapor phase diffusion
This introduces the possibility of localized flame zone wrapped around (but not contacting)
the droplet . In this envelope flame, vapors
exothermic vapor phase chemical reaction , with
attendant increases in the heat feedback and emanating from the fuel droplet react with oxygen
diffusing inward from the environment providing a
hence vaporization rates of individual droplets
localized source of heat and reaction products.
(see Fig. 2.4a-b ) . To answer the question of
whether such processes play an important role in At larger V. the quasi-steady state configuration
becomes that of a "flame-holding" droplet
determining bipropellant spray combustion effi-
ciency, space requirements, and stability, it is supplying fuel vapor into its wake, where com-

necessary to first consider the laws governing bustion then occurs due to mixing with ambient
gas. Ultimately, at sufficiently large V , "blow
such "heterogeneous combustion ," the possible
off" (complete extinction ) occurs . ( 2 ) In the
vaporization rate augmentations due to gas
envelope flame regime, for nearly stagnant
phase combustion in the droplet boundary layer or
conditions , d² again decreases linearly with
wake, and the dependence of these vaporization
rate augmentation phenomena on liquid/vapor time, * however, the apparent evaporation coeffi-
properties and environmental conditions . cient ( ie., the combustion constant comb)
In addition to obvious modifications of the exceeds Kvap (for the same environmental condi-

suspended drop and falling drop techniques tions ) by a factor depending primarily on the
heats of combustion and vaporization, ambient
already mentioned , a fruitful experimental method
temperature level and ambient oxidizer mass
for elucidating steady "burning" rates of in-
dividual droplets (with and without convective fraction. (3 ) Droplet burning rates are remarkably
insensitive to the fuel chemistry (structure ) and
flow) and reaction zone structure has been the
pressure level , suggesting rate control due to
"simulated-captive droplet" technique, in which a
wetted porous sphere plays the role of the droplet * physical processes ( diffusion/heat transport)
(see Fig. 2.4.2a ) 57,753 Such experiments are rather than chemical kinetic factors. ( 4 ) The drag

carried out by feeding liquid reactant into the on a burning droplet (with either envelope or
sphere at the rate determined (upon ignition ) wake flame) is significantly less than that on a
by the physical and chemical parameters of a solid sphere of the same size in the same environ-
system under study-the major assumption ment, and (5 ) quasi-steady state droplet com-
being that simulation of internal droplet phe- bustion is not possible for all combinations of
ambient oxidizer mass fraction, temperature ,
nomena (circulation heat conduction/radiative
heat absorption /emission ) is not essential to pressure level and droplet size ( e.g. , in a gas
mixture of particular composition and tempera-
model steady state droplet combustion . While
practical considerations have limited this tech- ture, combustion cannot be achieved if the pres-
nique to sphere diameters above about 2000μ, sure levels and/or droplet diameters are too
much of the experimental data on isolated small ) .
"droplets" (discussed below) has been obtained
with its aid . † References include 57 , 58 , 240, 290, 292 , 293 , 303, 363,
518, 658, 680, 753 and 756. For a recent review and bib-
liography see Ref. 741 .
Some of the earliest heterogeneous combustion data Again this implies that the instantaneous mass vapori-
were obtained using a simple variant of the simulated zation rate, m , varies nearly linearly with droplet diameter.
drop technique, viz . fuel-wetted, fabric- covered solid Thisis for p<<per (see Ref. 124) ; phenomena peculiar
(nonporous) spheres . 659 to p >per will be dealt with later on in the present section.
B
84 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

-Wetted porous sphere

(Alundum )

Flame
zone
Fuel

Product Product
vapor
vapor

Oxidizer Oxidizer
vapor vapor

Oxidizer
Oxidizer

Fuel

FIGURE 2.4.2a . -Fuel-fed porous sphere technique for simulating droplet combustion (after Wise, Lorell, and Wood75 ) .

2.4.2.1 Envelope flame model for subcritical compatible with the requirement that the con-
pressures -theory and experiment. -As a starting ductive heat feedback from the vapor phase
point in the iterative process of building a theory diffusion flame be adequate to supply the latent
compatible with the above-mentioned facts, con- heat of vaporization at the droplet surface .
sideration was focused on the simplest case of a Without making any restrictive assumptions con-
spherically symmetric system-i.e. , a fuel droplet cerning the vapor phase flame zone structure,
suspended in an unbounded oxidizer containing when Le = 1 and constant mean thermodynamic
gas mixture with no free or forced convective transport properties are assumed, it can be shown
distortion of the composition and temperature from the governing conservation equations that a
profiles in the vicinity of the droplet. Upon linear combination of temperature and oxidizer
ignition, the droplet supplies fuel vapor which mass fraction satisfies a boundary value problem
meets and reacts with oxidizer vapors in a narrow identical to that governing the corresponding
gas phase diffusion flame zone concentric with
290
(enveloping ) the droplet surface (Godsave, "
* In this figure , and in the following discussion, the
Spalding658 et al . ) . In this quasi-steady model droplet is considered to be fuel, and the surroundings
(see Fig. 2.4.2b) * flame zone radial position and contain oxidizer. However, the treatment clearly applies
the fuel vaporization rate adjust themselves to be equally well to the converse case .
 STEADY-STATE PROCESSES $2.4 85

Tf

T (r)

T∞
Yp (r)
Yox,
TL
rf Yox (r)

YF(r)

Fuel
droplet Flame zone

Fuel
Product
vapor
vapor

Oxidizer
Product vapor
vapor

FIGURE 2.4.2b. Diffusion flame model of subcritical pressure bipropellant droplet combustion in the absence of convective
distortion.

pure droplet vaporization problem. In this way, ·

vided one replaces Bvap =cp (T - Tw) /h, with†
one can immediately write the quasi-steady mass
Cp ( T - Tw) Yox, .he
burning rate in the general form.740 Bcomb = + (2.4.2-2)
hy hy
m = 4πгL Fortunately , when combustion occurs the assump-
(
Cp
tion TwTbp (p ) can be justified even in unheated
Cp (T - Tw) + (Yox, -Yox, w) he surroundings ,742 hence all quantities in Eqs.
In 1+ Bet Yox , the Yox. )h.] (2.4.2-1 , -2 ) are estimable. As an important
immediate consequence, Eqs . ( 2.4.1-9) and
(2.4.2-1 ) (2.4.2-2) show that in a stagnant environment
the enhancement in apparent vaporization con-
Here he is the heat released (per unit mass of
stant [-d (d2) /dt ] due to combustion (see Fig.
oxidizer consumed ) in the vapor phase reaction 2.4a) is approximately:
between fuel and oxidizer. While (as indicated
above) Eq. (2.4.2-1 ) applies regardless of the Kcomb In (1 + Bcomb)
distribution of chemical heat release within the (2.4.2-3)
Kvap In (1 + Bvap)
droplet boundary layer, in the absence of addi-
which indeed exhibits the experimentally observed
tional information its quantitative use is limited
to the case when the homogeneous chemical
reaction rates are adequate to prevent oxidizer † An interesting (but rarely stated) interpretation of
from "fighting its way" to the surface, i.e. , when this Le = 1 result is that the evaporation rate of an ignited
droplet in oxidizer-containing gas at temperature T. is
Yox, w =0. Then Eq. ( 2.4.2-1 ) becomes identical
the same as the evaporation rate of a noncombusting
to the expression for steady-state droplet vaporiza- droplet (of the same size) in combustion products of final
tion (in the absence of chemical reaction ) pro- temperature T + (Yox ho/Cp) .
86 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

trends with T. , Yox,, he and h,. This simple contours, Fig . 2.4.2b) . As will be noted below,
treatment also rationalizes the observed in- this thin-flame limit is not merely an irrational
sensitivity of burning rates to chemical factors computational approximation —it represents a self-
and pressure level, since In ( 1 + Bcomb ) does not consistent asymptotic limit * [for the cases of
show order-of- magnitude changes for a wide class very large (a ) bimolecular forward rate constant ) ,
of organic fuels in oxygen- containing gases, (b) pressure level and/or, ( c ) droplet size ] from
moreover In ( 1 + Bcomb ) exhibits only a weak which the maximum possible (diffusion limited )
dependence on pressure level (for reasons identical. droplet burning rate, mmax , can be predicted .
to those already discussed for Bvap) *. To embrace the fact that droplet combustion
Before discussing chemical kinetic phenomena , simply does not occur for all combinations of
ignition and extinction , it is appropriate to stress droplet size andand environmental conditions
here that even in the absence of effects peculiar to (p, Yox. , T. ) it is necessary to explicitly
particular experimental conditions (e.g. , natural introduce chemical kinetic limitations into the
convection around large droplets at high pres- envelope flame model. Despite the resulting
sures) the previous expressions often provide only complexity, only by these means does it appear
a guide to orders-of-magnitude and trends . This to be possible to decide whether (a ) the lower
is due principally to the extreme property varia- limit map (no chemical reaction ) or the upper
tions,399 and departures from unit Lewis number limit max ( flame sheet envelope) is more appro-
which can occur in real casest (e.g. , oxygen priate in a particular set of circumstances, or (b )
droplets burning in a hydrogen environment ) .399, 361 neither limit yields a sufficiently accurate pre-
For this reason several methods of accounting diction . Even if a detailed knowledge of the
for property variations have been suggested ; (multi-step, branching chain) chemical kinetics of
however, useful results usually require simplifying fuel vapor/oxidizer vapor systems of practical
the kinetic model to the extent that all chemical interest were available, specific numerical pre-
reaction is considered confined to a flame sheet of dictions would be costly and tedious, hence a
negligible radial thickness . In this extreme, great deal of semi-quantitative and qualitative
sometimes called the Burke-Schumann128 limit , ‡ information has recently been gleaned from
oxidizer and fuel vapor do not co -exist (intermix) — simplified but general theoretical treatments263, 398,
instead, both concentrations vanish at the flame 452,682 of evaporation as enhanced by single step
sheet , with their gradients standing in the irreversible vapor phase reactions of the form :
stoichiometric ratio to one another (see dashed aOX + bFproducts . In these treatments the
* Early
measurements on large captive droplets292,303 reaction rate (per unit volume ) is assumed to be of
and porous spheres57 in nearly quiescent oxidizing environ- the mass-action form : rate a k (T) • [ OX ] [F ],
ments revealed more sensitive pressure dependencies than which defines the forward rate constant k (T )
expected from hv (p ) and Tьp (p) . However, these depar- (usually taken to have an Arrhenius temperature
tures from the simple theory were shown to be compatible
dependence, i.e. , B exp ( -Eact/ T ) ) and the
with induced (natural convection) flows which distort
individual reaction orders a, B. If Tf.ad, is the
the quasi-steady flame shape and augment heat /mass
transfer rates. adiabatic flame temperature for the ambient
† However, Eq. (2.4.2-1 ) , together with a reasonable gas/fuel combination, and n is the overall order
choice of mean properties, has provided remarkably (n = a +B) then inspection of the governing con-
successful absolute predictions of Kcomb for hydrocarbon/air
servation equations indicates that the normalized
systems ; see, e.g. , Refs. 293 , 292. Recent computer solu-
tions of the transient variable property burning equations magnitude of the combustion-enhanced vaporiza-
(in the thin-flame limit ) suggest that this agreement only tion rate should be of the general form :
occurs after an initial transient which is not short com-
pared to the total droplet lifetime [see Kotaki , S. and * See Refs . 261 , 263, 279, 398, 452 and 682 .
Okazaki , T. , Int . J. Heat Mass Transfer, Vol. 12, pp . 395– Full advantage is usually taken of the Le = 1 (or
609 , 1969, and Nuruzzaman, A. S. M. , and Beer, J. M. , "Schvab-Zeldovich" ) approximation which enables first
Comb. Sci . Tech. , pp . 17-24 , 1971 ] . integrals (algebraic interrelations between YF(r), Yox (r) ,
After the investigators who first applied "flame-sheet" Yp (r) , T (r) ) to be used to reduce the overall coupled
concepts (in 1928) to treat steady-state diffusion flame problem to the integration of one highly nonlinear ordinary
shapes in ducts. differential equation.
 STEADY-STATE PROCESSES $2.4 87

exists at each value of pn/2dL. However, in the

m- mvap
= fct (Dr) (2.4.2-4 ) more frequently encountered case (large activa-
mmax -mvap
tion energy) m /mvap VS. pn/2dL reveals a domain
where Di represents the appropriate* Damköhler of multiple-valued (non-unique or non-monotonic)
number (ratio of characteristic diffusion time to behavior,263 within which (in particular) both
chemical reaction time) , given by high ṁ (near equilibrium ) and low m (near
inert ) solutions* exist for a particular pn/2dL.
k(Tf.ad) (aMox + bMF This situation is quite reminiscent of earlier work
D₁= Pond² (2.4.2–5)
4D∞∞ Mox"MF on the combustion of carbon (a true heterogeneous
reaction ) wherein it was shown that "ignition "
and fet (D1) has the limits fct ( 0 ) = 0 , fct ( ∞ ) = 1 .
and "extinction" correspond to the system
The behavior of m for large but finite Dr has been
suddenly jumping from the lower branch to the
investigated by the method of matched asymptotic
upper branch and vice versa. To fix ideas, con-
expansions , which (in the present application )
sider the near equilibrium combustion of a large
offers a systematic technique for obtaining
(captive) droplet at high pressure . As the pressure
corrections to max to account for incipient flame
level is continually reduced only slight departurest
zone broadening. * Coupled with a regular param-
from max will be observed . However, these
eter perturbation for DI« << 1 and numerical
departures will increase until point AE is attained
methods for intermediate D1, it is possible to
( Fig. 2.4.2d ) . At only slightly lower pressure
obtain the behavior of fct (D1 ) in Eq. ( 2.4.2–4 ) .
levels "low pressure extinction " (auto-extinction)
To bring out the implications of these theoretical
will occur, i.e. , the only "available" quasi-steady
studies for practical systems the resulting behavior
state will be that corresponding to only slight
is shown on log scale in Figs . 2.4.2c, d in the form
(usually negligible) combustion enhancement. ‡
of the enhancement ratio m/mvap vs‡ pn/2dL,
Similar comments apply for extinction due to the
implying that such parameters as ambient oxidizer
droplet size becoming insufficient (at constant
concentration , temperature level, fuel identity,
pressure level ) . Analogy with previous ignition
etc. , are held constant.
studies would also suggest that droplets originally
Two qualitatively different types of behavior
vaporizing in accord with the "near-inert " branch
are obtained,263 depending primarily on the ratio.
will "spontaneously ignite" (auto-ignite ) when
Eact/RTf.act. When this ratio is sufficiently small
the pressure level brings pn/2dL above (pn/2dL) ai ·
(as for vapor phase reactions of very low activa-
In this way, the domains of possible droplet
tion energy) then m/mvap VS. pn/2dL is single-
combustion receive their explanation in terms of
valued (monotonic) , i.e., one and only one quasi-
the relevant gas phase kinetics, within the confines
steady burning rate between mvap and max
of a quasi-steady spherically symmetric theory.
Even for this simplified kinetic scheme, however,
* The appropriate Damköhler numbers for the near
extensive computations and parametric studies
frozen (virtually no reaction ) expansion is similar to
Eq . 2.4.2-4 but based on k evaluated at TL , wb or T∞ , would be required to map out similar curves for
whichever is higher . For droplets in a convective environ- various ambient temperatures and oxidizer levels .
ment the characteristic diffusion time will be shortened,
hence, if all other things were equal, D, would be reduced
appreciably due to relative motion between droplet and * In all cases of this type examined theoretically to
gas. In practice convection produces flame distortion and date, the intermediate (3rd eligible steady state) is stat-
ultimately, flame "blowoff" (see Fig. 2.4.2e and accom- ically unstable and hence should be physically unrealizable .
panying discussion) . † The effects of incipient flame zone broadening on the
Slight reversibility (large but finite equilibrium con- burning rate have been shown398 to be intimately connected
stant) , also produces flame zone broadening, and the with the reaction order, a, and are extremely small when
effects can be treated by similar mathematical tech- α
a≤1 .
niques. 144,263 Tarifa680 has presented data on the dependence of
This abscissa is proportional to D1/2 for a particular Kcomb on pd for gasoline, and p°.55d for kerosene . Also
system (in view of the pressure dependence of the diffusion included are data on the pressure dependence of the
coefficient) . When the overall order is 2 , the abscissa minimum droplet diameter allowing combustion (for
becomes simply the pressure-diameter product. kerosene, gasoline, n-heptane and ethyl alcohol) .
88 LIQUID PROPELLANT ROCKET COMBUSTION INSTABILITY

10

B - S Limit

m m max
map mvap

2
pn/dĹ

FIGURE 2.4.2c . -Homogeneous chemical kinetic effects on quasi-steady droplet combustion rates ; single-valued case (low
activation energy) .

10

B - S Limit

AE
m mmax
To
p m
Unstable nv ap
vap
branch

AI

pn
/ 2
dĹ

FIGURE 2.4.2d . - Homogeneous chemical kinetic effects on quasi-steady droplet combustion rates ; multiple-valued case263
exhibiting auto-ignition and auto -extinction (high activation energy) .

This, together with the realization that only negligible radial thickness (a "modified flame.
trends are usually sought , has motivated the surface" ) . The quasi-steady, constant property,
development and exploitation of less rigorous , coupled linear* differential equations governing
but more computationally convenient approximate flame location, temperature and composition
methods. In one of these, due to Peskin and
Wise,546 the nonlinearity of the Arrhenius tem-
* In this model the nonlinearity is artificially removed
perature dependence is retained but the chemical
from the differential equations and placed in the boundary
reactions are artificially confined to a zone of conditions on each zone.
 STEADY- STATE PROCESSES $2.4 89

profiles, and burning rate , m , are then solved, Eq. 2.4.2-8 ) † provided the extinction criterion
leading to implicit but closed-form transcendental (see below) is not contravened . This appears to
relations between m, D, and the parameters of the be rigorously true at very low values of Re - Pr,
problem. Peskin et al.545 have used these results to (at which the semi-empirical Frössling form of
illustrate the existence of auto-ignition and NuB-0 (Re) breaks down ) since Fendell et al.267
auto -exti