Numerical Modelling and Experimental Validation - Ducted Rocket Motor

Numerical Modeling of a Ducted Rocket Combustor
With Experimental Validation
Patrick W. Hewitt
Dissertation submitted to the faculty of the Virginia

Polytechnic Institute and State University in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
In
Aerospace Engineering
J. A. Schetz, Chair
W. F. O’Brien
B. Grossman
J. Marchman
J. Sparks
September 04, 2008

Blacksburg, Virginia
Keywords: Ducted Rocket, Ramjet, Combustion, Boron, Soot, Gas Generator

Numerical Modeling of a Ducted Rocket Combustor
With Experimental Validation
Patrick W. Hewitt
(ABSTRACT)
The present work was conducted with the intent of developing a high-fidelity
numerical model of a unique combustion flow problem combining multi-phase fuel
injection with substantial momentum and temperature into a highly complex turbulent
flow. This important problem is very different from typical and more widely known
liquid fuel combustion problems and is found in practice in pulverized coal combustors
and ducted rocket ramjets. As the ducted rocket engine cycle is only now finding
widespread use, it has received little research attention and was selected as a
representative problem for this research. Prior to this work, a method was lacking
domestically and internationally to effectively model the ducted rocket engine cycle with
confidence.
In the ducted rocket a solid fuel gas generator is used to deliver a fuel-rich multi-
phase mixture to the combustion chamber. When a valve is used to vary the fuel
generator pressure, and thereby the delivered fuel flowrate, the engine is known as a
Variable Flow Ducted Rocket (VFDR). The Aerojet MARC-R282 ramjet engine
represents the worlds first VFDR flown, and the first in operational use. Although
performance requirements were met, improvements are sought in the understanding of
the ramjet combustion process with a future aim of reducing the visible exhaust and
correcting uneven combustor heating patterns. For this reason the MARC-R282
combustor was selected as the baseline geometry for the present research, serving to
provide a documented baseline case for numerical modeling and also being a good
candidate to benefit from an improved understanding of the combustion process.
In order to proceed with the present research, experiments were first carried out to
characterize the gas generator particulate exhaust in terms of composition and particle
ii
size. Equilibrium thermochemistry was used to supplement these data to develop a gas
phase combustion model. The gas phase reactions and resulting particle definition were
modeling using the FLUENT Computational Fluid Dynamics (CFD) code for the
baseline GQM-163A Supersonic Sea Skimming Missile (SSST) operating conditions.
These results were compared to direct-connect ramjet ground tests in order to validate the
analysis tool.
Data were developed to understand the gas and solid phase fuel exhaust
characteristics at the propellant surface, exiting the gas generator injector, and following
secondary combustion with air. Particles were collected and analyzed from the fuel
generator exhaust. While exhibiting some variation with time in the firing, they were
roughly an average of 20 microns in diameter, in line with prior experience with
pulverized coal combustion experiments. A computational model was developed based
on coal combustion parameters using FLUENT. However, despite considerable effort,
the CFD analysis was not able to predict effective burning of the carbon particles to the
degree seen in testing. In addition, using equilibrium thermochemistry as a basis for
determining the carbon particle content in the fuel exhaust, the CFD analysis resulted in
trends in performance opposite to the test results. These facts led to a hypothesis that
there was actually a significant fraction of small particles or much less carbon produced
than equilibrium thermochemistry would predict. A parametric analysis was performed
replacing the 20 micron soot particles with fine fraction particles, representing a fraction
of the predicted equilibrium carbon soot being still in the gas phase as higher molecular
weight hydrocarbons, or in the form of sub-micron particles. When almost all particles
were replaced with fine fraction particles, the model was able to correctly predict
absolute values of combustion efficiency as well as trends for different injector
geometries. The presence of particles was apparent from the visible exhaust and
collection data, however they were found not to play a significant role in the combustion
process for this fuel and engine configuration.
The robustness of the computational model was also evaluated by examining the
effects of turbulence model, order of discretization, and grid size. Comparable trends and
results were seen for all cases examined.
iii
With the successful development of this modeling tool and an improved
understanding of the combustion process, future work is enabled to develop improved
combustor flow management and fuel injection schemes to improve existing designs and
develop new configurations. This research has served to advance the field of combustion
modeling by providing: 1) a solid ducted rocket combustion modeling tool considering
solid and gas phase combustion, 2) a correlation between primary combustion theory and
particulate exhaust sampling, 3) low length/diameter ratio ducted rocket combustor
modeling, and 4) combustor CFD coupled with solid particle tracking and combustion
models.
iv
Acknowledgements
As a lesson to those who believe that they have finally left the classroom behind, one
should avoid casual dining in Munich with the person of Dr. Joseph A. Schetz. With a
self-admitted cunning, I was lured into the current episode by this respected professor
after a 25 year hiatus. Having served as my mentor in round one, I was equally treated
for the second course with sound advice and encouragement, for which I will be eternally
grateful.
I also owe thanks to the Aerojet-General Corporation for allowing me to pursue this
goal while maintaining an active position with the company, although in a somewhat
diminished capacity. Specifically, Rick Yezzi and Adam Siebenhaar deserve credit for
enabling me to mix business and academic goals during this period, hopefully for the
good of all.
I would like to also thank my committee for helping with this endeavor by providing
guidance and useful inputs, and for tolerating the logistics associated with linking the
University and Industry for study and research.
Invaluable advice and support with FLUENT modeling was provided by Aerojet’s
Thong Nguyen and Ted Smith, who were responsible for implementing the requested
cases, and who responded faithfully to unreasonable schedule and technical demands.
And finally, none of this would have been possible without the understanding and
support from my wife Diana, who has been with me through all phases of my career, and
from Michelle, my daughter and fellow scholar.
v
Table of Contents
CHAPTER 1 - INTRODUCTION .............................................................................................................. 1
1.1 BASELINE DUCTED ROCKET COMBUSTOR FLOWFIELD ......................................................................... 1

1.2 RESEARCH OBJECTIVES ........................................................................................................................ 7
1.3 LITERATURE REVIEW............................................................................................................................ 7
1.3.1 Ramjet Combustor Modeling ....................................................................................................... 8
1.3.2 Boron Particle Combustion ....................................................................................................... 10
1.3.3 Carbon Particle Combustion ..................................................................................................... 14
1.3.3.1 Diffusion Flame Soot Combustion...................................................................................................... 14
1.3.3.2 Carbon Particle Combustion ............................................................................................................... 14
1.3.3.3 Coal Combustion Research ................................................................................................................. 15
CHAPTER 2 – FUEL EXHAUST CHARACTERIZATION ................................................................. 19
2.1 THERMOCHEMISTRY ........................................................................................................................... 19

2.1.1 Gas generator Exhaust Composition ......................................................................................... 19
2.1.1.1 Equilibrium Calculations..................................................................................................................... 19
2.1.1.2 Non-Equilibrium Calculations ............................................................................................................ 20
2.1.2 Gas Generator Exhaust Temperature ........................................................................................ 27
2.1.3 Combustor Thermochemistry ..................................................................................................... 29
2.2 PARTICLE COLLECTION ...................................................................................................................... 31
2.2.1 Propellant Surface Particle Testing........................................................................................... 32
2.2.1.1 Window bomb testing ......................................................................................................................... 32
2.2.1.1 Particle Collection tests....................................................................................................................... 34
2.2.2 Gas Generator Exhaust Particle Testing ................................................................................... 36
2.2.3 Combustor Exit Exhaust Characterization ................................................................................ 41
CHAPTER 3 – RAMJET VERIFICATION TESTS............................................................................... 44
3.1 SSST PROGRAM DIRECT CONNECT TESTS.......................................................................................... 44

3.1.1 Validation Ramjet Test Description ........................................................................................... 44
3.1.2 Test Data.................................................................................................................................... 48
3.2 DATA REDUCTION PROCEDURES ........................................................................................................ 53
3.2.1 Thermochemistry........................................................................................................................ 54
3.2.2 Air Mass Flow Measurement ..................................................................................................... 55
3.2.3 Flow Areas ................................................................................................................................. 55
3.2.4 Ramburner Nozzle Discharge Coefficient.................................................................................. 55
3.2.5 Pressure-Based Combustion Efficiency ..................................................................................... 56
vi
3.2.6 Thrust-Based Combustion Efficiency ......................................................................................... 56
3.3 BASELINE VERIFICATION TEST RESULTS ............................................................................................ 59
3.3.1 Equilibrium Combustion ............................................................................................................ 59
3.3.2 Test Data Used for CFD Analysis.............................................................................................. 65
CHAPTER 4 – COMPUTATIONAL FLUID DYNAMICS MODELING............................................ 68
4.1 GRID DEVELOPMENT .......................................................................................................................... 68

4.2 TURBULENCE MODELING ................................................................................................................... 71
4.3 GAS PHASE MODEL ............................................................................................................................ 73
4.4 PARTICLE COMBUSTION MODELING ................................................................................................... 81
4.5 MODEL PARAMETRICS ........................................................................................................................ 91
4.5.1 Turbulence Model ...................................................................................................................... 91
4.5.2 Higher Order Calculations ........................................................................................................ 92
4.5.3 Grid Dependency ....................................................................................................................... 93
CHAPTER 5 - CONCLUSIONS AND RECOMMENDATIONS.......................................................... 97
REFERENCES ......................................................................................................................................... 101
vii
List of Figures
FIGURE 1.1-1 SSST MISSILE CONFIGURATION............................................................................... 3
FIGURE 1.1-2 MARC-R282 RAMJET ENGINE .................................................................................... 4
FIGURE 1.1-3 MARC-R282 ENGINE HEATING .................................................................................. 5
FIGURE 1.1-4 SSST COMBUSTOR LAYOUT....................................................................................... 5
FIGURE 1.1-5 COMBUSTOR FLOW FEATURES................................................................................. 6
FIGURE 1.3.3-1 COAL PARTICLE REACTION RATE CURVE FIT .............................................. 18
FIGURE 2.1.1-1 POLYSTYRENE BEADS (136X)................................................................................. 21
FIGURE 2.1.1-2 FUEL PRODUCTS WITH NON-REACTING POLYSTYRENE ............................ 22
FIGURE 2.1.1-3 CHAMBER TEMPERATURE AND C* WITH NON-REACTING

POLYSTYRENE ........................................................................................................................................ 23
FIGURE 2.1.1-4 CHAMBER TEMPERATURE AND C* WITH NON-REACTING

POLYSTYRENE AND DIFFERENT ASSUMPTIONS FOR POLYSTYRENE HEATING ............. 23
FIGURE 2.1.1-5 POLYSTYRENE DECOMPOSITION PROCESS .................................................... 24
FIGURE 2.1.1-6 THERMO-GRAVIMETRIC ANALYSIS RESULTS FOR POLYSTYRENE........ 26
FIGURE 2.1.2-1 CALCULATED CHARACTERISTIC VELOCITY, TEST DC-1............................ 29
FIGURE 2.1.3-1 TYPICAL ENGINE OPERATING TEMPERATURE.............................................. 31
FIGURE 2.2.1-1 WINDOW BOMB APPARATUS................................................................................. 33
FIGURE 2.2.1-2 WINDOW BOMB IMAGE OF FLAME FRONT ...................................................... 33
FIGURE 2.2.1-3 PARTICLE COLLECTION BOMB............................................................................ 34
FIGURE 2.2.1-4 PARTICLE COLLECTION BOMB RESULTS......................................................... 35
FIGURE 2.2.2-1 GAS GENERATOR EXHAUST SAMPLING TEST SETUP ................................... 36
FIGURE 2.2.2-2 GAS GENERATOR EXHAUST PARTICLE PHOTOMICROGRAPHS............... 37
FIGURE 2.2.2-3 GAS GENERATOR PARTICLE SIZES, 0-1 SECOND SAMPLE .......................... 39
FIGURE 2.2.2-4 GAS GENERATOR PARTICLE SIZES, 2-3 SECOND SAMPLE .......................... 40
FIGURE 2.2.2-5 GAS GENERATOR PARTICLE SIZES, 10-11 SECOND SAMPLE ...................... 40
FIGURE 2.2.3-1 RAMJET EXHAUST SAMPLING TEST SCHEMATIC ......................................... 41
viii
FIGURE 2.2.3-2 RAMJET EXHAUST SAMPLING TEST SETUP..................................................... 42
FIGURE 2.2.3-3 RAMJET EXHAUST SAMPLES ................................................................................ 42
FIGURE 3.1.1-1 AEROJET RAMJET TEST FACILITY ..................................................................... 45
FIGURE 3.1.1-2 SSST HEAVYWALL DIRECT-CONNECT INSTALLATION ............................... 45
FIGURE 3.1.1-3 SSST HEAVYWALL TEST HARDWARE ................................................................ 46
FIGURE 3.1.1-4 SSST HEAVYWALL FUEL INJECTORS ................................................................ 47
FIGURE 3.1.2-1 DIRECT-CONNECT VITIATED AIR MASS FLOWRATES ................................ 48
FIGURE 3.1.2-2 DIRECT-CONNECT AIR TEMPERATURE ........................................................... 49
FIGURE 3.1.2-3 DIRECT-CONNECT RAMBURNER STATIC PRESSURES................................. 50
FIGURE 3.1.2-4 DIRECT-CONNECT GAS GENERATOR CHAMBER PRESSURES .................. 51
FIGURE 3.1.2-5 DIRECT-CONNECT FUEL FLOWRATES.............................................................. 52
FIGURE 3.1.2-6 DIRECT-CONNECT GROSS THRUST.................................................................... 52
FIGURE 3.2-1 RAMJET ENGINE STATION NOMENCLATURE .................................................... 53
FIGURE 3.3.1-1 DC-1 TEMPERATURE RISE COMBUSTION EFFICIENCY................................ 60
FIGURE 3.3.1-4 PRESSURE-BASED TEMPERATURE RISE COMBUSTION EFFICIENCY..... 62
FIGURE 3.3.1-5 FUEL/AIR COMBUSTION EFFICIENCY DEFINITION...................................... 63
FIGURE 3.3.1-6 FUEL/AIR COMBUSTION EFFICIENCY – BASED ON PRESSURE ................. 64
FIGURE 3.3.1-7 RAMBURNER PRESSURE LOSSES ........................................................................ 65
FIGURE 3.3.2-1 EFFECT OF BORON COMBUSTION ON RAMJET FLAME TEMPERATURE 66
FIGURE 3.3.2-2 TEMPERATURE DECREASE BY NEGLECTING BORON.................................. 67
FIGURE 4.1-1 BASELINE GRID AND EDGE BOUNDARY CONDITION DESIGNATIONS ....... 69
FIGURE 4.1-2 BASELINE .05 IN. NOMINAL GRID – DUMP REGION.......................................... 70
FIGURE 4.1-3 BASELINE .05 IN. NOMINAL GRID – DOME REGION .......................................... 70
FIGURE 4.1-4- BASELINE .05 IN. NOMINAL GRID – NOZZLE REGION..................................... 71
FIGURE 4.3-1 COMBUSTION TOTAL TEMPERATURE FOR SIMPLIFIED GASEOUS FUEL 76
FIGURE 4.3-2 COMBUSTION EFFICIENCY FOR SIMPLIFIED GASEOUS FUEL ..................... 77
ix
FIGURE 4.3-3 PREDICTED CONTOURS OF STATIC TEMPERATURES FOR CASE DC-3 –
GAS PHASE ONLY ................................................................................................................................... 79
FIGURE 4.3-4 PREDICTED CONTOURS OF FUEL MIXTURE FRACTION FOR CASE DC-3 –
GAS PHASE ONLY ................................................................................................................................... 80
FIGURE 4.4-1 THERMOCHEMISTRY FOR GAS PHASE AND CARBON PARTICLES ............. 83
FIGURE 4.4-2 PREDICTED MIXTURE FRACTION FOR CARBON PARTICLE FUEL FOR
CASE DC-3 ................................................................................................................................................. 84
FIGURE 4.4-3 PREDICTED CASE DC-1 PARTICLE TRACES......................................................... 85
FIGURE 4.4-4 PREDICTED TEMPERATURE PROFILES FOR TESTS DC-1 AND DC-3............ 85
FIGURE 4.4-5 PREDICTED GAS PHASE COMBUSTION PROFILES FOR TESTS DC-1 AND
DC-3............................................................................................................................................................. 88
FIGURE 4.4-6 COMBUSTION EFFICIENCY AS A FUNCTION OF FINE CARBON.................... 90
FIGURE 4.5.1-1 MODEL RESULTS FOR REALIZABLE K-Ε MODEL........................................... 92
FIGURE 4.5.2-1 SECOND ORDER MODEL RESULTS ...................................................................... 93
FIGURE 4.5.3-1 FINE GRID CONVERGENCE HISTORY – BASELINE UNDER-RELAXATION

...................................................................................................................................................................... 94
FIGURE 4.5.3-2 FINE GRID CONVERGENCE HISTORY – AGGRESSIVE UNDER-

RELAXATION........................................................................................................................................... 94
FIGURE 4.5.3-3 FINE GRID CONVERGENCE HISTORY – GAS PHASE COMBUSTION ONLY
...................................................................................................................................................................... 95
FIGURE 4.5.3-4 FINE GRID CONVERGENCE HISTORY – GAS PHASE COMBUSTION ONLY
...................................................................................................................................................................... 96
UNLESS OTHERWISE NOTED, ALL IMAGES ARE PROPERTY OF THE AUTHOR.
x
List of Tables
TABLE 1.3.2-1 METALLIC DUCTED ROCKET FUEL ADDITIVES .............................................. 10
TABLE 1.3.3-1 COAL PARTICLE REACTION RATES (FIELD DATA).......................................... 17
TABLE 2.1.1-1 ARCADENE 428J GAS GENERATOR EXHAUST PROPERTIES ......................... 20
TABLE 2.1.3-1 EQUILIBRIUM THERMOCHEMISTRY INPUT PARAMETERS ......................... 30
TABLE 2.2.2-1 SSST PROPELLANT EQUILIBRIUM EXHAUST SOLIDS ..................................... 38
TABLE 2.2.2-2 BORON ANALYSIS RESULTS .................................................................................... 38
TABLE 3.1.1-1 HEAVYWALL TEST HARDWARE FLOW AREAS ................................................. 47
TABLE 3.3.1-1 RAMJET TEST DATA COEFFICIENTS .................................................................... 60
TABLE 4.3-1 GAS GENERATOR THROAT PRODUCTS FOR NON-REACTING BORON ......... 74
TABLE 4.3-2 ADJUSTED THROAT MASS FRACTIONS................................................................... 74
TABLE 4.3-3 GASEOUS FUEL ONLY MASS FRACTIONS............................................................... 75
TABLE 4.4-2 FUEL SOLID PHASE SUBSTITUTION ......................................................................... 89
xi
Nomenclature
Hv = Volumetric heating value

Hg = Gravimetric heating value
A/F = Air to fuel mass ratio
K = Reaction rate
q = Rate of removal of carbon per unit external surface
pg = Partial pressure of gas
x = Particle diameter
c* = Characteristic exhaust velocity
$ = Solid phase compound
* = Liquid phase compound
γ = Ratio of specific heats
V = Velocity
X = Mole fraction
Bi = Biot number
h = Convective heat transfer coefficient
k = Thermal conductivity
r = Radius
A = Area
T = Temperature
ρ = Density
Cp = Specific heat at constant pressure
Nu = Nusselt number
M = Mach number
D,d = Diameter
Re = Reynolds number
u = Flow speed
gc = Gravitational constant
R = Gas constant
P = Pressure
xii
m& = Mass flowrate
W = Weight flowrate
X = X function
Y = Y function
Z = Z function
F = Specific stream thrust
η = Efficiency
F = Thrust
θ = Nozzle half angle
φ = Equivalence ratio
S A* = Air specific stream thrust
Subscripts
diff = Diffusion
g = Gas
s = Surface
vap = Vaporization
eff = Effective
PS = Polystyrene
∞ = Infinity
0 = Total, initial
d = Diameter
T = Total
CBE = Choke Block Entrance
CB* = Choke Block Throat
1.5 = Diffuser
2 = Dump plane
3 = Fwd Combustor
xiii
4 = Aft Combustor
5 = Ramjet Nozzle Throat
6 = Ramjet Nozzle Exit
D = Discharge
ΔT = Temperature rise
corr = Corrected
meas = Measured
a = Ambient
x = Axial
atm = Atmospheric
n = Nozzle
a = Air
f = Fuel
xiv
CHAPTER 1 - INTRODUCTION
1.1 Baseline Ducted Rocket Combustor Flowfield

The ducted rocket is a ramjet engine cycle in which the fuel is delivered by
combustion of a primary fuel propellant grain. The propellant contains mainly fuel
ingredients, with only enough oxidizer as required to initiate combustion and expel fuel-
rich products into the ramjet combustion chamber. The result of this marginal primary
combustion is a large amount of particulate carbon and mildly warm gas phase
combustion products, which must be efficiently burned in the ramburner.
Practical examples of the ducted rocket can be found in missile propulsion around
the world, with the earliest implementation being the Soviet SA-6 in the late 1950’s.
Recent focus has been on a type of ducted rocket in which a variable-area fuel exhaust
valve is used to vary the fuel output. This engine cycle has become known as the
Variable Flow Ducted Rocket, or VFDR. The world’s first flying example of this engine
is the Supersonic Sea Skimming Target (SSST), with the ramjet supplied by Aerojet. The
SSST propellant consists of a binder (hydroxyl-terminated polybutadiene) and oxidizer
(ammonium perchlorate), with fuel additives including polystyrene, boron, and
magnesium.
The SSST missile is intended to simulate threat anti-ship cruise missiles by operating
at Mach 2.5 at near sea level for over 50 miles, while executing significant all-aspect
maneuvers. The missile configuration is shown in Figure 1.1-1. The GQM-163A uses a
first stage booster that provides sufficient impulse to achieve ramjet takeover speed, at
which point it separates and ramjet operation commences. The significance of this staged
operation is that there is no internal booster housed in the combustion chamber, as would
be the case for an integral rocket ramjet weapon engine. This makes the addition of
mixing and flame-holding devices in the combustor interior possible.
The ramjet engine was designated the MARC R282, and is shown schematically in
Figure 1.1-2.
The MARC-R282 engine development activity took place from 2000 to 2003, with
an emphasis on rapid development and low technical risk. As a result there was little
1
time available for combustor design optimization. Initial CFD work was performed to
help determine the fuel injector geometry. The combustor inlet head end configuration
was based on water flow visualization studies performed on a similar 4-inlet combustor
by Virginia Research Associates39. Direct-connect testing demonstrated adequate
delivered combustion efficiencies and adequate thermal margins during full duration
mission simulations.
Although the delivered performance was consistent with the propulsion system
specification, a large amount of unburned carbon was produced, resulting in a very
visible exhaust trail (evidenced in Figure 1.1-1). This observation reinforces the need to
develop a better understanding of the combustion process with an eventual aim of
identifying hardware changes to be implemented.
2
Image property of Orbital Sciences Corporation
Boost Phase Sustain Phase
Image property of US Navy

Image property of US Navy
Figure 1.1-1 SSST Missile Configuration
3
Inlets
- Four 2-D Inlets
Combustor
Gas Generator - DC93-104 Insulation
- Steel Case/Closures - Ramjet Nozzle
- ARCADENE 428J Fuel
Control Valve
- EM Linear Actuator
Image property of Aerojet Corporation
Figure 1.1-2 MARC-R282 Ramjet Engine
A second issue to be addressed in the SSST combustor is a region adjacent

to the inlet dump station in which there is evidence of localized heating. This presents a
risk of late flight case burn-throughs and heating of wiring harnesses that run through this
region. Figure 1.1-3 shows the external combustor discoloration following a full-
duration direct-connect test. Again, a suitable design tool is required to manage the
4
internal flow to reduce localized heating, yet maintain acceptable levels of engine
performance.
Figure 1.1-3 MARC-R282 Engine Heating
Figure 1.1-4 SSST Combustor Layout
5
The MARC-R282 combustor layout is shown in Figure 1.1-4. The length to
diameter ratio, L/D, is only 2.5, which presents a challenge to reach a fully mixed state
prior to the exit nozzle. The exit nozzle geometry and inlet dump location, angle, and
area must remain fixed in order to avoid modifying qualified hardware. In attempting to
improve the combustor performance the addition of flameholders, vanes, or other flow
management devices is permissible. The injector diameter is fixed but the exit area and
hole pattern may be modified. The injector length can be increased; however, structural
considerations may prevent changes.
Preliminary CFD studies and post-test insulator mapping have provided a general
indication of the internal combustor flowfield. Recirculation regions are established in
the combustor head end, where higher fuel/air ratios are achieved and a favorable
combustion environment is formed, as illustrated schematically in Figure 1.1-5. Outside
of these regions the flow is much cooler and any carbon or boron not combusted in the
primary recirculation zone is likely quenched and expelled as soot, contributing to the
exhaust signature
Figure 1.1-5 Combustor Flow Features
6
1.2 Research Objectives
The objective of this project is to develop an improved understanding of the
combustion process in the ducted rocket combustor, and to develop a suitable engineering
tool to guide future efforts. The ducted rocket engine cycle introduces multiple
challenges to the engine designer, including particulate combustion in addition to a gas
phase diffusion- and kinetically-controlled combusting flowfield. The experimental and
analytical approach for the research includes:
• Modeling the primary equilibrium thermochemistry in the fuel gas generator and
the associated fuel products injected into the combustor
• Experimentation to better quantify the composition and particle size of the
particulates composing the solid fraction of the fuel exhaust
• The development of a CFD model of the current MARC-R282 internal fuel and
air flowfield
• Model validation using ramjet direct-connect ground test data
1.3 Literature Review

A review of prior combustion research was performed to help devise a strategy for
the current problem. The applicable prior work consisted primarily of carbon soot and
boron particle combustion, and ducted rocket analytical modeling. An overview is
provided here with detailed discussion to follow.
A large base of applicable combustion research was performed by G.M. Faeth21,15,18
and associates. The majority concerned premixed laminar flames, with much focus on
carbon soot formation, and to a lesser on oxidation of the resulting particles. Very little
prior research exists on soot combustion in turbulent diffusion flames, as in all likelihood
these problems are highly specific to individual geometries.
The diesel engine research area has also explored soot combustion; however, the
bulk of work is primarily concentrated on soot formation and catalyst effects. The work
is associated with very specific engine geometries and dynamics (timing).
7
The majority of rocket plume analysis and characterization has been aimed at Al2O3
formation, with a secondary focus on soot formation.
Applicable turbine engine research has been identified dealing with carbon slurry
combustion models. The work describes two combustion stages: fuel vaporization and
combustion followed by carbon particle heating and oxidation (which is of direct interest
here).
Prior research on coal furnace combustion has yielded the most generally applicable
results to the current problem and is discussed in detail in subsequent sections. In
addition to combustion modeling, the broader subject of ramjet computational modeling
is reviewed first, as this is the ultimate aim of the present research. Boron particle
combustion is then considered.
1.3.1 Ramjet Combustor Modeling

Numerical modeling of ramjet combustion chambers can be a formidable task. High
velocity flows are typically encountered, with complex flame-holding zones and high
turbulence and strong flow gradients. In addition, fuel injection and chemical reactions
must be treated. The ducted rocket adds additional degrees of complexity in which the
fuel stream itself is a high temperature reacting multi-phase flow. For these reasons
researchers would be well advised to seek mathematically straightforward models for
each aspect of the problem. The benefit of an accurate numerical combustor tool is
brought to focus by Ogerby27, in which he states that “465 combustor modifications and
140 full-scale engine tests were carried out on the Pratt and Whitney JT8D fan engine to
produce a smokeless combustor.” This is a luxury unavailable to a tactical missile engine
contractor.
Although the paper must be calibrated by significant recent advances in
computational abilities, Edelman et al7 discuss ramjet modeling techniques, and the
conclusion is reached that it is difficult to impossible to provide sufficient resolution to
adequately describe processes such as fuel injection that occurs on a much finer scale
than overall the combustion chamber. They propose a modular model to break up the
combustor into manageable regions. Similar views are expressed by Lilley8, and doubt is
cast that 3-D full simulations will prove practical to the designer. Special procedures
8
were devised for multi-path flows by Rodriguez23 to model complete combustor
configurations, which may be beyond the reach of even sophisticated CFD tools.
Harsha and Edelman10 present a thorough treatment of combustor modeling, and put
forth a modeling “hierarchy” from simple to detailed. Chemical reaction modeling types
are also categorized from global to detailed. Again, a calibration must be made for
present day capabilities and advances in CFD software packages.
Specific work undertaken in ducted rocket modeling is described by Stowe et al13. A
one-stream approach was developed in which all of the fuel components were injected
together and assumed to react instantaneously in the presence of the oxidizer. This
method over-predicted combustion efficiency as one would expect. A more complicated
two-stream model was also used in which the fuel was separated into gas components and
75 nanometer carbon particles, which exhibited better agreement with test data. Soot
from the hydrocarbon fuel combustor was collected and measured from sub-micron to
over 200 microns. The majority of the particles were near the 75 nm value. The average
difference between test data and predicted temperature-rise combustion efficiency was
27% for the one-stream method, and 16% for the two-stream method.
Numerical predictions of ducted rocket performance were made by Lewis17 using
FLUENT gas phase and particle combustion algorithms. The model was exercised to
evaluate trends in equivalence ratio, injector design, port covers, and inlet air conditions.
Lacking experimental data, a 3 micron soot particle size was assumed for the studies, and
the coal particle combustion models contained in FLUENT were utilized. Based on the
present research it was found that larger particle sizes must be treated, and the reaction
rates must be tailored to match the ducted rocket combustion environment.
Although the work was conducted for an Air-Turbo-Rocket, Thomas and Leonard23
present work in which soot combustion is considered in a numerical model. The soot was
modeled as solid carbon spheres of uniform diameter that react with the gas stream,
converting carbon to CO and other products. Minimizing particle size to less than 100
microns results in increased combustion efficiency and reduced residence time
requirements. The behavior of 50 micron and 100 micron particles was nearly identical.
A numerical simulation was performed by Ristori and Defour25 for a hydrocarbon
ducted rocket using a methodology based on a multi-species Reynolds Averaged Navier-
9
Stokes (RANS) approach associated with a turbulent combustion model. No particles
were considered. Various combustion models are evaluated, with strengths and
weaknesses of each identified. They were able to predict a global combustion efficiency
with a precision better than 7 % and a global pressure recovery with a precision better
than 5 %.
1.3.2 Boron Particle Combustion

As the ducted rocket engine cycle utilizes solid propellant combustion products as a
fuel, it is relatively straightforward to add metals to the propellant composition. This is
in direct contrast to liquid fuels, in which a slurry or gel must be prepared, along with the
associated issues concerning settling, pumping, and atomization. Of the metals studied
for application to the ducted rocket, boron is generally seen to be the most attractive fuel
ingredient. Both the gravimetric and volumetric heating values of selected fuels are
shown in Table 1.3.2-1. For volume limited applications like tactical missiles, the
volumetric heating value proves the most attractive attribute, and illustrates why boron is
the preferred choice.
Table 1.3.2-1 Metallic Ducted Rocket Fuel Additives
Gravimetric Heating Volumetric Heating Stoichiometric

Density Value - Hg Value - Hv A/F
(kg/m3) (MJ/Kg) (MJ/m3)
Fuel
Aluminum 2.70 31.07 83.88 3.84
Beryllium 1.85 66.49 122.97 7.67
Boron 2.22 59.28 131.60 9.59
Lithium 0.53 42.96 22.77 4.98
Magnesium 1.74 24.74 43.05 2.84
Titanium 4.50 19.74 88.85 2.88
As a result of the benefits offered by boron, a great deal of research on single particle
boron combustion has been performed. This has lead to a reasonable understanding of
the conditions required to efficiently burn them in a ramjet combustor. Although the
understanding exists, it is difficult in practice to achieve efficient combustion. One of the
most effective methods is to employ combustion aids such as a propellant additive or
10
boron particle coatings. The aim of the present research, however, is not to promote
efficient combustion through propellant tailoring, but rather through an improved
understanding of the combustor flowfield and boron particle history. The following
historical research is presented in order to summarize the body of knowledge on boron
combustion.
The boron used in the validation motor firings uses a small percentage of amorphous
boron with a maximum average particle size of 1 micron, and a boron purity of 95%, with
the major impurity being oxygen. The accepted practice has been to model the particles
for combustion studies as 95% B and 5% B2O3 using the as-received particle size.
Mellor12 studied the particulate exhaust properties of sample boron propellants and found
that the particle sizes were relatively unchanged, however an allowance must be made for
the formation of ammonium pentaborate (NH4B5O8*4H2O) on the particle surface by
reaction with ammonium perchlorate or its decomposition products. The
recommendation is made to characterize the particulate exhaust in cases where motor
data are compared to theory.
A landmark paper was contributed by Komar and King28 for single particle boron
combustion. Laser based diagnostics were used for the determination of particle-size,
velocity, temperature, and times-to-ignition for complete combustion of single boron
particles. The particle combustion process is described in general terms as initially the
heating and vaporization of an oxide layer which inhibits combustion and must be
removed before "full-fledged" combustion can begin; considerable time may be
consumed during this "ignition" stage. In addition, boron has a very high boiling point,
necessitating relatively slow surface burning even after the oxide has been removed.
Also pointed out is that the use of a particle burning law based on diffusion-limited
combustion must become erroneous for sufficiently small particle diameters. Since the
diffusional transport rates are predicted to approach infinity as particle size approaches
zero, the kinetic process rates do not increase with decreasing particle size. Thus, at some
finite diameter, the rate-limited particle burning processes must become kinetic rather
than diffusional, with a resultant shift from the burning mass flux being inversely
proportional to being independent of particle radius. This was found to be in the 30-40
micron particle size range. Thus, the one micron boron utilized in the verification case
11
places it well in the range of kinetic rate control. Although a detailed combustion model
is alluded to, only limited data are presented for the diameters of interest here. Data are
provided for one micron particle combustion times of less than one millisecond for a gas
temperature of 4050°R and an oxygen mole fraction of 0.19, closely resembling
stoichiometric conditions for the present research case.
Ignition delay was studied for small particles at 8.5 atm and 4050°R by Rood30 for
pure oxygen. At 1 micron particle sizes the ignition delays approach 10 microseconds.
Odawara et al31 present data showing ignition delay times account for roughly 80% of the
combustion time, and that delay times drop dramatically above 2160 °R. Heating of the
particles in the primary gas generator of a ducted rocket was shown to also reduce
ignition delay times.
Schadow33 also was able to show the effect of primary gas temperature on the
combustion efficiency of boron. A non-equilibrium chamber temperature T* was
calculated by treating the boron as a heat sink only, and not contributing to the reaction.
A gradual increase in secondary combustion efficiency was seen over a T* range of
1260-3600°R, with no apparent jump near the 2160°R mark discussed in Reference 31.
Additional research by Schadow38 showed that efficient boron combustion is dependent
on ignition and combustion of the gas phase, regardless of combustor operating pressure.
This is a significant finding with regard to high altitude engine operation where
combustor pressures below one atmosphere can be encountered. Mitani37 conducted
strand bomb experiments where it was shown that burning of boron is relatively easy in
the primary chambers of ducted rockets if fine B powders are used. However, this result
would be anticipated based on other research presented herein, as the equilibrium
temperatures for most propellants tested exceeded 3780°R.
Abbott, et al20 investigated boron combustion in a combustor. It was stated that a gas
temperature of 3510°R is required for rapid boron combustion. Based on experiments,
the statement is made that “When ignition of the gases occurs in farther aft regions of the
duct, the mixture is too diluted in air to reach 3510°R. This result implies that rapid,
early mixing can be disadvantageous to promotion of high combustion efficiency”. This
fact must be considered as the SSST combustor is examined. Vigot29 demonstrated a
technique using impinging injector jets to “scrub” the oxide layer in preparation for
12
injection into the combustion chamber, as well as dispersing agglomerates. A
compromise was necessary in practice between injector efficiency and durability in
withstanding the boron particle stream. Up to 50% improvement in combustion
efficiency was shown, but this was limited to the range of 25% for practical injectors for
missile applications.
A somewhat skeptical review of boron combustion is presented by Gany32. He also
gives an ignition temperature of 3420°R-3600°R to begin removal of the oxide layer and
initiation of boron surface reactions. Also presented is a novel theory in which the
gaseous B2O3 partial pressure may reach a point in which further vaporization of the
oxide layer is inhibited, termed “Thermodynamic Blocking”. This effect was stated to
prevent combustion of boron propellants at certain flight conditions. However, it is felt
that through proper flow management encouraging near stoichiometric conditions in
recirculation zones and the appropriate use of combustion aids this effect may not be a
significant effect in practice. Natan and Gany36 also show through modeling that at
3420°R the oxide removal rate increases and ignition times drop.
The preceding discussion illustrates that efficient burning of boron is rather difficult
to accomplish in a ramjet engine. The ducted rocket has the advantage that the boron can
be heated in the primary gas generator, thus beginning the process of oxide removal even
before injection into the combustion chamber. Perhaps the most usefully elegant
statement is made by Schadow34: “The present results indicate that, so long as the
gaseous fuel component from the primary combustor can be made to react with the air in
the secondary chamber before excessive mixing occurs, the resulting temperatures will be
high enough to initiate high boron reaction rate.”
King’s recommendations for future boron combustion modeling remain largely
unaddressed to date. These consist of:
- Development of a kinetics-limited model for the "full-fledged“ combustion stage of
boron burning. For such modeling to be successful, experimental identification of
reaction paths and determination of kinetic data for limiting reaction steps are required.
- Definition of the nature of boron particles entering combustors should be addressed,
(single unit particles, loosely bound porous conglomerates, tightly bound low porosity
13
conglomerates) and ignition/combustion models should be modified as needed to treat
conglomerates as well as single particles.
1.3.3 Carbon Particle Combustion

A great deal of prior work has been undertaken to understand the formation of
carbon soot in combustors, the heating, oxidation mechanisms, and reaction rates. Much
of this work has been directed at gas turbine combustors, which has provided a solid basis
for the chemical processes; however, the particle sizes are in the nanometer range, far
smaller than those encountered in the present research. Another general area of
fundamental research has been coal particle combustion. This research has proven to be
most applicable to the current ducted rocket combustion problem, as the emphasis is
primarily on decomposition verses formation, and the particle sizes are much more in line
with those seen in the ducted rocket. The following discussion provides an overview of
important carbon particle research, and summarizes the pertinent data employed in the
current research.
1.3.3.1 Diffusion Flame Soot Combustion

Several sources exist in which soot formation and reaction rates are presented (Kim
et. al.14, Rodionov, et. al.19, and Blevins26). An excellent review of the subject with
comments on current understanding is provided by Urban and Faeth21. However, their
usefulness for this research is limited to a broader view of the particle combustion
mechanism rather than insight into applicable reaction rates.
1.3.3.2 Carbon Particle Combustion

Turns and Faeth18 present results of combustion research aimed at carbon black
slurries for use in airbreathing engines. The carbon black particle diameters range from
10 to 100 μm, and are thus of interest for the ducted rocket. The observation is made that
particle burning rates are quite low until the particle temperature reaches 2700-3060 °R.
This is slightly higher than the stated values for onset of combustion for coal particles but
still in reasonable agreement.
14
Dempsey, et al41 present data for carbon black particle agglomerates in the 45-66 μm
size range, and states ignition times on the order of 1.2 to 1.6 milliseconds, and
combustion times of nominally 0.9 milliseconds in a methane flat flame burner.
1.3.3.3 Coal Combustion Research

Baum and Street1 investigated the burning of pulverized coal particles in the range of
60 μm in diameter. They relied on a large base of research that observed the process of
coal particle combustion in which hollow spheres are formed upon rapid heating and the
combustion process takes place at nearly constant diameter as oxidizing gases permeate
the sphere and burning takes place internal to the shell. Fragmentation takes place as the
shell is eroded away at the completion of the reaction. A mathematical model was
developed based on observations in a laboratory furnace burning size-graded coal. A
“swelling coefficient” is used to account for the increase seen in diameter as volatiles
begin reacting and a hollow sphere is formed with large holes in its surface. In this state
the remaining carbon burns at a rate determined by the diffusion of oxygen to the surface
of the char (physical control), and the reaction of carbon and oxygen to CO and CO2
(chemical control). Typical swelling coefficients were stated to be from 1.0 to 4.0.
Cangialosi et al16 reports maximum oxidation rates for coal fly ash to be at approximately
1620 °R through thermo-gravimetric analysis.
Peters and Weber6 constructed a mathematical model for nominally 45 μm coal and
compared predictions to actual data from a coal furnace. In the model they found that
good accuracies were seen when a standard k-epsilon model for turbulence was
employed. Magnussen and Hjertager4 also were able to demonstrate good agreement
between their combustion model and experiment by using the k-epsilon turbulence
model. Soot combustion rates for a C2H2 diffusion flame were stated to be approximately
200 g/(m3s) at stoichiometric conditions, where the rate was maximized. For a spherical
particle of 20 μm with an assumed O2 partial pressure of 0.2 in air, this value can be
expressed in terms of surface area as .667 g/cm2s atm for comparison with the much
lower reaction rates shown below. This effect is likely due to the much smaller soot
particle sizes for the acetylene flame.
15
Reaction rates for size-graded coal are reported by Field5 for gas temperatures
between 2160 °R and 3600 °R, and particle size diameters from 20 μm to 100 μm.
Oxygen concentrations were tested at 0,5, and 10 percent. For large particle sizes the
overall reaction rate was determined by diffusion of CO from the particle surface. For
particles smaller than 80 μm it was found that the surface reaction rate coefficient was
smaller, and independent of oxygen concentration and particle size. Photographs of the
raw coal particles are shown, exhibiting irregular angular surfaces closely similar to the
ducted rocket particulate samples obtained in the present research.
The overall reaction rate K (g/cm2s atm) is defined as the rate of removal of carbon
per unit external surface q (g/cm2s) per unit atmosphere partial pressure of oxygen in the
gas pg (atm), and
K = q / pg
The maximum limiting value for diffusion, Kdiff assuming CO transport is

0.75
⎛ T ⎞
K Diff = 0.00128 ⎜ g ⎟ / x (g/cm2s atm)
⎝ 1600 ⎠
where Tg is the gas temperature (°K), and x is the particle diameter (cm).
A surface reaction rate Ks is calculated from the relation

K ∗ K Diff
Ks = (g/cm2s atm)
K Diff − K
Values of Ks were found to be independent of oxygen concentration in the range

from 0 to 10%, and depends only on the particle surface temperature. No effect of size
was detected in the range of 28 μm to 105μm. The relationship given for coal particles
within the surface temperature range of 1400 °K to 2000 °K is
K S = −0.49 + 3.85 ×10−4 TS
16
A method is presented to numerically calculate particle surface temperature by
performing a heat balance assuming combustion of CO.
The data presented by Field are tabulated in Table 1.3.3-1 for 28 μm and 38 μm
particles. These data were used to construct a simplified relationship neglecting the
effects of oxygen concentration and particle size to approximate the reaction rates as
shown in Figure 1.3.3-1. A simplified relationship based on a linear curve fit of the data
presented is
K = .0003 ∗ Tg − 0.3993 where Tg is the gas temperature (°K), and K has units g/cm2s atm
The data clearly illustrate a critical gas temperature to be 1200°K (2160°R), below
which reaction rates approach zero.
Table 1.3.3-1 Coal Particle Reaction Rates (Field Data)
28 Micron Sample 1 28 Micron Sample 2 38 Micron Sample
Gas temp °K K (5% O2) K (10% O2) gas temp K (5% O2) K (10% O2) gas temp K (5% O2) K (10% O2)
1420 0.037 0.074 1320 0.019 0.053 1220 0.000 0.017
1520 0.087 0.144 1420 0.042 0.098 1320 0.026 0.025
1620 0.097 0.165 1520 0.106 0.126 1420 0.053 0.068
1720 0.140 0.167 1620 0.145 0.150 1520 0.103 0.120
1620 0.139 0.145
1720 0.158 0.165
17
0.180
Overall Reaction Rate Coefficient - K 0.160
y = 0.0003x - 0.3993
0.140
0.120
(g/cm 2s atm)
0.100
0.080
0.060
0.040
0.020
0.000
1000 1200 1400 1600 1800
Gas Temperature (°K)
Figure 1.3.3-1 Coal Particle Reaction Rate Curve Fit
The preceding research forms the basis of the pulverized coal combustion
modeling implemented in the FLUENT code42. FLUENT provides a choice of four
heterogeneous surface reaction rate models for combusting particles:
– the diffusion-limited rate model (the default model)
– the kinetics/diffusion-limited rate model
– the intrinsic model (first-order kinetics/diffusion-limited rate model more
effectively modeling pore diffusion)
– the multiple surface reactions model
Both the diffusion-limited rate model and the kinetics/diffusion-limited rate model
were used in the present analysis.
18
CHAPTER 2 – Fuel Exhaust Characterization
2.1 Thermochemistry
2.1.1 Gas generator Exhaust Composition
2.1.1.1 Equilibrium Calculations

Table 2.1.1-1 provides the theoretical equilibrium thermochemical calculations using
the ARCISP equilibrium thermochemistry program for the gas generator exhaust – prior
to mixing with air in the combustion chamber. In practice, fuel rich propellants typically
rarely reach the theoretical characteristic exhaust velocity, c*, or total temperature so
these values must be used only as a guide. Further, the non-equilibrium combustion is
time-dependent as the grain burns back and the free volume and fuel residence time
increase.
Of primary interest in the predicted products is the value of particulate carbon
produced, on the order of 40% by mass. Since combustion efficiencies up to 95% are
achieved in operation, it is apparent that the majority of the carbon is being combusted.
Any remaining carbon is believed to be responsible for the sooty exhaust. The other
major constituents of the fuel output are combustible gases that provide a heat source in
the initial phases of secondary combustion for particle burning.
The thermochemistry analysis also shows that the boron has been reacted to solid
boron nitride and liquid boron oxide. Based on numerous models developed for boron
combustion as discussed in Section 1.3.2, complete combustion in the primary chamber is
not likely, and elemental boron particles will be present in the combustor as well. The
following section discusses these non-equilibrium combustion situations.
19
Table 2.1.1-1 ARCADENE 428J Gas Generator Exhaust Properties
MASS FRACTIONS AT EQUILIBRIUM CONDITIONS ($=solid, *=liquid)
CHAMBER THROAT
CH4 .01817 .01555

CO .20932 .19988
FECL2 .01017 .01229
H2 .05381 .05427
HCL .06038 .06833
MGCL2 .04823 .03728
BN$ .06220 .05867
C$ .39085 .39696
MGO$ .02826 .03337
B2O3* .08645 .09270
FE3C* .01017 .00917
TOTAL MOLES: 7.52482 7.57053
MOLES GAS: 3.82018 3.80817
MOLES $,*: 3.70463 3.76237
PRESSURE TEMP. HT. CAP MOLS GAS VELOC C*

PSIA DEG R CAL MOLES FT FT
_______ _____ ___ ___
K*100GM 100GM SEC SEC
CHAMBER 1000.0000 3134.3 60.282 3.8202
THROAT 580.1835 2963.2 59.802 3.8082 2508.6 3854.7
AIR/FUEL RATIO FOR STOICHIOMETRIC MIXTURE = 7.4488
ACOUS VEL GAMMA PROCESS

FT/SEC (BSD ON HT CAP) GAMMA
CHAMBER 2585.4 1.1441
THROAT 2508.6 1.1449 1.1151
2.1.1.2 Non-Equilibrium Calculations

The SSST propellant uses a significant amount of polystyrene beads as a fuel
ingredient (34% of the propellant mass). They are nearly spherical beads with a nominal
diameter of 220 microns, as shown in Figure 2.1.1-1.
20
Figure 2.1.1-1 Polystyrene Beads (136X)
As the polystyrene beads are fairly large particles that require a finite time to heat up
and begin contributing to the combustion process, it has been postulated that they are a
primary source for the delay in gas generator characteristic velocity for the first several
seconds of the motor firing. For this reason, a series of thermochemistry calculations
were made to investigate the effects of non-reacting polystyrene on the gas generator
exhaust. For these calculations the boron was modeled as non-reacting particles heated to
the gas generator chamber temperature.
21
Figure 2.1.1-2 shows the effect on combustion products as the level of non-reacting
polystyrene increases. It is apparent that the majority of carbon produced is derived from
the combustion of the polystyrene beads, as this is the primary effect seen.
Fuel Products With Non-Reacting Polystyrene
40.00
Carbon
35.00
HCL
CO
30.00
CO2
H2
Mass Fraction (%)
25.00 N2
CH4
20.00
15.00
10.00
5.00
0.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00
Mass of Polystyrene Non-Reacting (%)
Figure 2.1.1-2 Fuel Products With Non-Reacting Polystyrene
This is an important fact when considering solutions for improving ramjet

combustion efficiency and reducing signature, especially in light of the conclusions
reached in Chapter 4 computational modeling. Figure 2.1.1-3 presents the effects of non-
reacting polystyrene on chamber temperature and characteristic exhaust velocity c*.
Although not as dramatic as the effect on carbon produced, the net result seen is a
decrease in temperature and c* with an increased fraction of polystyrene unreacted, as
one might expect.
The preceding results are not entirely rigorous, in that the polystyrene in actuality is
actually only heated to the decomposition temperature before it begins reacting.
Therefore a manual temperature and c* correction is required to reverse a portion of the
enthalpy change occurring from the decomposition temperature to chamber temperature
for the non-reacted polystyrene. This effect is bracketed in Figure 2.1.1-4, where
calculations are shown for the two cases of zero heating of the beads, and the beads
allowed to be heated beyond the vaporization temperature to the gas temperature.
22
Chamber Temperature and c* With Non-Reacting Polystyrene
3500.00
3300.00
c*
3100.00
2900.00
Temperature (°R), c* (ft/sec)

2700.00
2500.00
2300.00
Temperature
2100.00
1900.00
1700.00
1500.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00
Figure 2.1.1-3 Chamber Temperature and c* With Non-Reacting Polystyrene
Chamber Temperature With Non-Reacting Polystyrene
2900.00
2800.00
Gas Temperature with NR PS mass removed
2700.00
2600.00
Temperature (°R)
2500.00
2400.00
Manual Calculation for PS at Tg
2300.00
Gas Temperature with NR PS at Tc
2200.00
2100.00
2000.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00
Figure 2.1.1-4 Chamber Temperature and c* With Non-Reacting Polystyrene and

Different Assumptions for Polystyrene Heating
23
The polystyrene decomposition process was investigated in 1981 by Dr. Merril
King2, and modeled for a boron/polystyrene fuel by the author in 1994. Early in the
firing, residence time is short in the gas generator, and there is little time for heat-transfer
limited vaporization of spherical polystyrene to occur. Particles remain in the flow
through nozzle with resultant two-phase flow losses. A model was developed to describe
the process but only limited comparisons between theory and data were made. For the
present research initial calculations were made in the event that the phenomenon was
shown to be significant, and worth implementing in the CFD model. The polystyrene
decomposition process is shown schematically in Figure 2.1.1-5.
r r r
V V V
Tgas Tvap Tvap
q&
Particle heating Vaporization PS decomposition

phase T reached and reaction
– mass addition
determined by
regression rate
- Mixture of reacted
and unreacted PS
Figure 2.1.1-5 Polystyrene Decomposition Process
The King model assumes 100% thermal and velocity lags of particles in the nozzle,
i.e. c*eff = c*gas Xg. Any unvaporized polystyrene passes through the nozzle at the
vaporization temperature. Carbon formation is either allowed or suppressed from PS
(allowed for fuel binder) to form limits for the analysis. A heat-up time (theta) is
calculated, and if theta is greater than the particle residence time, the fraction of
24
polystyrene burned is zero, otherwise the particle size is determined from a burning rate
equation.
The Biot number compares the relative magnitudes of surface convection and
internal conduction resistances to heat transfer. When the Biot number is small, Bi<0.1,
T(R,t)=T(t) and the sphere responds as a lumped mass.
h ( VA ) h ( 3r )
Bi = =
k k
For the polystyrene used in the SSST fuel and representative chamber conditions, h
is likely less than 200 W/m^2°C, R ~ 150E-06 m, k (PS) = .08 W/m°K, the Bi is less than
0.125, so lumped heat capacity is assumed for the polystyrene beads.
The change in internal energy of bead is then set equal to the heat transfer from
convection and integrated to determine the time to heat to the de-polymerization
temperature:
dT
q = hA(T − T∞ ) = −CpPS ρ PSV
dt
t −hA t 1
∫
0 ρ CpV
dt = ∫
0 T − T∞
⎛ − hA ⎞
⎜ ⎟t
⎝ ρ CpV ⎠ T − T∞
e =
T0 − T∞
− ρ PS CpPS r0PS ⎡ Tvap − T∞ ⎤

theatup = ln ⎢ ⎥
3h ⎣ T0 − T∞ ⎦
25
Expressing the equation in terms of Nusselt number, the expression becomes;
hD
Nu =
k gas
− ρ PS CpPS r02PS
⎡ Tvap − T∞ ⎤
theatup = ln ⎢ ⎥
1.5k gas Nu ⎣ T0 − T∞ ⎦
Using the expression provided by McAdams43 for Nusselt number:
0.6
⎛u d ⎞
Nu = 0.37 ⎜ ∞ ⎟ ;17 < Re D < 70, 000
⎜ νf ⎟
⎝ ⎠
Nu ≈ 1.5
The vaporization temperature for polystyrene derived from thermo-gravimetric
analysis results presented in Figure 2.1.1-6 was taken as 425°C (1256°R). Additional
Figure 2.1.1-6 Thermo-Gravimetric Analysis Results For Polystyrene
26
required parameters were approximated as:
cal
CpPS = 750
kg ° K
r0 = 220 x10−6 m
cal
k gas = 20000
sec m ° K
T0 = 25°C
T∞ = 1500°C
The resulting heat-up time is found to be less than 0.2 milliseconds to raise the
polystyrene particle to 425 °C (1256°R). Since gas generator residence times are much
longer than this, even in the initial transient period for SSST, this is too quick to have an
effect on ballistics. Particle collection results also later verified no unreacted polystyrene
was found at the exit for the sample collection times. For these reasons all subsequent
calculations assumed that the polystyrene beads are reacted completely in the gas
generator. Supporting this conclusion, sub-micron soot and no evidence of partially
decomposed polystyrene was found as reported in a workshop summary prepared by
Waesche22.
2.1.2 Gas Generator Exhaust Temperature

In order to set initial conditions for the FLUENT fuel inflow condition, the gas phase
temperature must be estimated. Since the RT product is proportional to the characteristic
velocity squared;
γ +1
1 ⎛ γ + 1 ⎞ γ −1
R ∗ T ~ ( c *) taken from
2 c *theo = gc ⎜ RT
γ ⎝ 2 ⎟⎠
the characteristic velocity was determined for the DC-1 gas generator. The inherent
difficulty associated with this calculation, assuming the fuel flow determination is
accurate, is that the gas generator throat area must be known. As the firing proceeds this
value becomes less precise due to deposits that form on the interior surfaces, and thermal
growth of the valve components. By restricting the analysis to the early portion of the
27
test, errors associated with deposits can be minimized. In order to estimate geometrical
effects due to thermal growth, an analysis was performed to estimate the valve plunger
dimensional variations with time. These data were used to calculate the area reduction
experienced with time for the first twenty seconds in which the valve throttle position is
fixed. With the area now known, c* can be determined from;
Pt ∗ At ∗ g c
c* =
m&
And the RT product can be determined from the relationship
( γ +1)
1 ⎛ γ + 1 ⎞ (γ −1) RT0
c* =
γ ⎜⎝ 2 ⎟⎠ M
As seen in Figure 2.1.2-1, it is important to account for thermal growth as the valve
heats and reduces the effective flow area. Using an approximate value of c* ratio equal
to 0.89 for the first twenty seconds of the test results in a ratio of 0.79 for the RT product.
This indicates a situation in which the degree of reaction is less complete than the
equilibrium calculation, with a corresponding decrease in gas temperature, and the
presence of higher molecular weight hydrocarbon compounds.
As a limiting case, if the molecular weight of the exhaust products is taken to be
close to the theoretical value, the gas temperature is reduced to 1944 °R from 2462 °R
(non-reacting boron). In reality, the gas temperature is somewhere between these values;
however, to remain consistent with the gas phase chemistry chosen for the FLUENT
model, the conservative value of 1944 °R was used.
The fact that the c* remains essentially constant during a time when the gas
generator residence time is increasing considerably supports the assumption that
molecular weight is not changing, otherwise the additional combustion time should result
in a more complete reaction and an increase in calculated c*.
28
1.1
1.05
Based on Measured Valve Position

c* Ratio to Theoretical
0.95
Valve Throat Area Adjusted for Thermal Growth

0.9
0.85
0.8
0 2 4 6 8 10 12 14 16 18 20
Test Time (sec)
Figure 2.1.2-1 Calculated Characteristic Velocity, Test DC-1
2.1.3 Combustor Thermochemistry

The previous section dealt with the fuel gas generator prior to mixing with air. For
ramjet combustor analysis, a second set of calculations must be performed using the fuel
ingredients and air to determine theoretical combustor flow conditions. Air is modeled
using standard atmospheric standards for analysis of flight operating conditions. For
ground testing, the Aerojet test facility uses propane combustion to heat the air, with
make-up oxygen added to return the mixture to the correct mole fraction of atmospheric
oxygen. As a result, the air mixture, referred to as vitiated air, is modeled as nitrogen,
oxygen, carbon dioxide, and water. A table of equilibrium values for process gamma,
combustor total temperature, and gas molecular weight are developed as a function of
inlet air temperature, operating pressure, and fuel/air ratio. The input parameters used to
create the test data thermochemistry matrix are tabulated in Table 2.1.3-1.
29
Table 2.1.3-1 Equilibrium Thermochemistry Input Parameters
Inlet Air Temp Combustion

(°R) Pressure (psia) Air/Fuel Ratio
1140.0 50.0 4000.0
1150.0 100.0 400.0
1160.0 125.0 100.0
1170.0 140.0 50.0
1190.0 160.0 27.0
25.0
23.0
20.0
15.0
In the test data analysis process if any individual data point falls outside the values
used to create the thermochemistry matrix the calculations are halted to allow the matrix
to be expanded.
The theoretical combustor total temperature for operation at near nominal flight
conditions (TT0=1200R, PT4=150 psia) is shown in Figure 2.1.3-1. The engine typically
operates at a fuel/air ratio of .05 to .06 during acceleration and cruise. Thus a well-mixed
combustor equilibrium total temperature should be less than 3000 °R at 93%
temperature-rise combustion efficiency. At this temperature the combustion gases are
below the generally accepted 3500°R required to burn boron, and marginally above the
2160°R required to burn carbon. Therefore, a hotter flame-holding recirculation zone
with combustion temperatures closer to stoichiometric values is required to achieve
acceptable levels of combustion efficiency.
30
5000
4500
4000
Combustor TT4 (R)
3500
Engine operating range

3000
2500
2000
1500
0 0.05 0.1 0.15 0.2 0.25 0.3
Fuel/Air Ratio
Figure 2.1.3-1 Typical Engine Operating Temperature
2.2 Particle Collection

Examination of the equilibrium thermochemical exhaust species for the SSST gas
generator fuel shows that a significant fraction is predicted to be composed of solids,
primarily carbon. Additionally, lacking conclusive evidence, the assumption is made that
the boron does not react in the primary combustor and enters the combustion chamber as
a solid. Therefore, when contemplating a method for numerical simulation of the
process, it becomes of primary importance to understand the composition and particle
sizes being injected with the fuel gas phase. Attempts were made experimentally to
characterize the particulate exhaust at each step in the process:
31
– At the propellant surface; the flame front and products ejected from propellant
surface in gas generator
– As the fuel is injected into the combustion chamber; gas / particulate flow exiting
fuel injector
– At the ramjet exhaust after mixing with air
2.2.1 Propellant Surface Particle Testing

Two methods were used to examine the flow properties at the propellant surface -
window bomb testing and particle bomb testing. Although not directly applicable for the
numerical modeling effort, it is important to understand the fuel formation process in
order to help guide future work to improve either exhaust signature or ramjet combustion
efficiency.
2.2.1.1 Window bomb testing

The window bomb apparatus was constructed to examine the propellant surface,
flame front, and ejected solids using a transparent window to allow visual access to
propellant samples burning in an inert argon atmosphere. The test apparatus is shown in
Figure 2.2.1-1. A propellant strand 0.25 in. by 0.25 in. in cross section is ignited and a
video camera is used to record the propellant burning surface while visible through the
window. A single frame of a representative video is shown in Figure 2.2.1-2. A hot
ammonium perchlorate flame zone is seen as well as particles combusting, which are
likely magnesium. The polystyrene beads are clearly identifiable in the photo, both on
the propellant surface as well as in the flame front as they begin to react. The video
clearly shows long strings of beads forming and reacting as they are carried from the
propellant surface. This was a significant finding, as the clumping of beads is not in the
direction of assuring complete reaction prior to introduction to the combustion chamber,
and may be in part responsible for the unreacted carbon seen in the ramjet exhaust.
32
Figure 2.2.1-1 Window Bomb Apparatus
Burning Metal -
Likely AP flame
Magnesium Area
String of beads
Polystyrene
Beads
220 micron
0.25” PS bead diameter
Figure 2.2.1-2 Window Bomb Image of Flame Front
33
2.2.1.1 Particle Collection tests
The window bomb collection tests provide a qualitative indication of the surface
reactions, but little other data. For this reason, a series of particle collection tests was
conducted.
The particle collection bomb shown in Figure 2.2.1-3 consists of an internal stainless
steel cylinder which spins with a layer of alcohol coating the internal surface. The
propellant sample is mounted just above the alcohol layer. The particle bomb is filled
with argon and the propellant is ignited. Particles coming off the propellant are quenched
in the cold alcohol, and the alcohol containing the sample is sent to the analytical lab for
analysis.
Figure 2.2.1-3 Particle Collection Bomb
34
PS beads added for reference
Figure 2.2.1-4 Particle Collection Bomb Results
The particle collection bomb procedure results in particles collected very close to the
burning surface of the propellant. For many standard propellant problems this may
provide valuable data, but for these fuel rich propellants under marginal combustion
conditions such a low residence time prior to collection, it provides only basic knowledge
that the combustion process is underway. The preliminary surface reactions result in
particles that are far from the state at which the fuel exhaust is injected, as illustrated in
the following sections. The most useful information from these tests is that the
polystyrene has begun to react very close to the surface, as seen visually in the window
bomb experiments. Figure 2.2.1-4 shows the particle bomb residue along with neat
polystyrene beads to illustrate the particle sizes obtained. Data collected from laboratory
analysis indicated that 40-50% of the insoluble material collected was charred
polystyrene beads. These findings also support the analytical conclusion that the
polystyrene reacts quickly in the gas generator under sustained heating, and can be
excluded from the particle fraction used to represent the fuel exhaust for the numerical
modeling effort.
35
2.2.2 Gas Generator Exhaust Particle Testing
There exists a wealth of prior research in particle collection from rocket exhausts as
described aptly by Sambamurthi11. The primary obstacle encountered in particle
collection is reliably obtaining measurements over a wide range of particle diameters.
Some methods lend themselves to fine fractions and misrepresent the larger fraction, and
some demonstrate the opposite trend. After reviewing past research in particle collection
and size characterization including optical diagnostics, the copper tape method developed
by Sambamurthi was adapted to the test hardware employed for this project.
The method described in that paper was developed to collect aluminum oxide
particles, but appeared readily usable for boron and carbon collection. In personal
correspondence with the author on 2/23/06, Dr. Sambamurthi (NASA Marshall) endorsed
the method for collection of other rocket motor particulate exhaust, and indicated it was
currently being employed to collect turbine exhaust particulates.
For the SSST fuel collection, double-sided adhesive copper tape (3M® Code 1182)
was placed on three one-inch diameter steel rods. These rods were placed downstream of
the exhaust of a gas generator firing using SSST propellant (no mixing with air). The test
setup is shown schematically in Figure 2.2.2-1.
Propellant Motor Case Closure

ARCADENE 428J Valve Actuator
Valve
Assembly
Figure 2.2.2-1 Gas Generator Exhaust Sampling Test Setup
36
During the motor firing, the rods were extended using a hydraulic ram and left
exposed to the motor exhaust for a period of one second. Sampling was conducted at 0-1
sec (retract at 1 sec), 2 sec, and 10 sec.
The three copper tape tubes immersed during firing at time intervals were submitted
to the lab for analysis. Scanning Electron Microscopy and Energy Dispersive
Spectrometer (SEM/EDS), was used to determine inorganic material, and Fourier
Transform Infrared Spectroscopy (FTIR) was used to identify any organic material
present. The SEM/EDS samples were prepared by collecting residue that fell off the
tubes onto a stub then coated with palladium. The FTIR samples were prepared by
cutting a piece of the copper tape away from the metal rod then extracting in methylene
chloride. The samples were then filtered and spotted onto a prepared KBr plate for
analysis.
FTIRs indicated only the presence of the adhesive material used on the copper tape.
This verifies the assumption that the polystyrene is reacting in the gas generator.
SEM/EDS detected the presence of magnesium, silicon, chlorine, titanium, and iron.
Scanned photomicrographs at 100x, 200x, and 500x (top to bottom) are shown in Figure
2.2.2-2 for 0, 2, and 10 second samples. The photos exhibit very irregular non-spherical
particle geometry.
Image property of Aerojet

Corporation
Figure 2.2.2-2 Gas Generator Exhaust Particle Photomicrographs
37
The laboratory analysis of the particles collected attempted to determine the
chemical composition and particle sizes. Table 2.2.2-1 presents equilibrium composition
of solid and liquid exhaust products for SSST propellant as a percentage of the total
exhaust and as a percentage of the particulates.
Table 2.2.2-1 SSST Propellant Equilibrium Exhaust Solids
Compound Mass fraction of Mass fraction of particulates

exhaust (%)
Solid Boron Nitride 0.5 0.7
Solid Magnesium
4.5 6.7
Oxide
Solid Carbon 45. 67.
Solid Boron Oxide 17. 25.
The laboratory analysis technique used does not distinguish between elemental boron
and boron oxide, only the total elemental amount of boron present. Table 2.2.2-2
presents the analysis results for boron. The question to be answered is whether the boron
is reacting in the gas generator, or is injected without reacting, as assumed based on
experience expressed in the literature.
Table 2.2.2-2 Boron Analysis Results
Sample Time (sec) Boron % if B2O3 Boron % if B

T + 0.1 to T + 1.1 37.7 5.85
T + 2 to T + 3 17.5 2.71
T + 10 to T + 11 20.9 3.24
38
Since the 38% boron oxide level calculated exceeds even the equilibrium value, it is
apparent that early in the firing all the boron is likely not in oxide form. However, as the
values decrease for later samples it is possible that the boron is reacting more than
thought in the gas generator. Future testing and analysis would be required to better
characterize the form of boron particles, but for the SSST propellant at low levels of
boron, the assumption that the boron is unreacted when injected appears a reasonable
simplification.
Probably more pertinent to the present research are the particle sizes obtained from
the photomicrographs. Each photograph section was used to determine individual
particle volumes and express them as an effective spherical volume, assuming the particle
width is equal to the depth. The results are shown in Figures 2.2.2-3 through 2.2.2-5.
0-1 sec
Mean 50.16350435
Standard Error 5. 149346911
Median 33.31184589
Mode 12.0828555
Standard Deviation 43.99603929
Sample Variance 1935.651473
Kurtosis 2. 485395842
Skewness 1. 689180561
Range 192.9836471
Minimum 7. 702820382
Maximum 200.6864675
Sum 3661.935817
Count 73
Confidence Level(95.0%) 10.26503532
Gas Generator Exhaust Sampling Data
Figure 2.2.2-3 Gas Generator Particle Sizes, 0-1 Second Sample
39
2 sec
Mean 18.96073349
Stan dard Error 1.278773804
Median 13.03698637
Mode 6.796606219
Stan dard Deviatio n 18.70684802
Sample Variance 349.9461628
Kurtosis 8.588922736
Skewness 2.681164836
Range 115.9225476
Minimum 4.8331422
Maximum 120.7556898
Sum 4057.596967
Count 214
Confidence Level(95.0%) 2.520672678
10 sec
x500
40
The 10-11 second sample does not exhibit distinct particles, but represents more of a
“caking”. This may be a result of fine particles coating the tape during the 10 sec period
prior to insertion, preventing a representative sample to be obtained. There is also a
significant drop in mean particle diameter from the first to second samples, indicating
that there is a time-dependent component to the exhaust properties. The 2-3 second
results however would be most representative of the 1 to 20 second period used to
calibrate the CFD model with ramjet ground test data.
2.2.3 Combustor Exit Exhaust Characterization

As a final experimental method, the exhaust of the ramjet combustor was sampled
during a ramjet ground test. The test setup shown in Figures 2.2.3-1 and 2.2.3-2 is
similar to the arrangement used for the fuel exhaust testing discussed in Section 2.2.2.
1” Rod Covered with Copper Tape
12 inch stroke
5 feet
Remotely
Controlled
Linear
actuator
Figure 2.2.3-1 Ramjet Exhaust Sampling Test Schematic
41
Image property of Aerojet Corporation Image property of Aerojet Corporation
Figure 2.2.3-2 Ramjet Exhaust Sampling Test Setup
For this testing, the sampling probes were located 5 feet downstream due to the
higher temperatures experienced at the nozzle exit. Video coverage verified that when
extended, the rods were immersed in the motor exhaust. In this case, the rods were
inserted at 2, 10, and 60 seconds, with a 1 second dwell time for each case. The rod ends
following the test are shown in Figure 2.2.3-3.
1 sec. immersion
at t=2 sec.
1 sec. immersion
at t=10 sec.
1 sec. immersion
at t=60 sec.
Figure 2.2.3-3 Ramjet Exhaust Samples
42
An extraction was performed to remove particles from some strips in a solvent.
They were easily removed and the copper base foil was left clean. The mixture was
sonicated to separate agglomerated particles and a FTIR was used to identify organic
materials. Again, the adhesive was detected, but there was no unreacted polystyrene
present. An SEM/EDS was used to identify metals. Magnesium and copper were
detected, likely copper from the tape, and Mg from the fuel. No boron was detected.
Carbon was not detected by SEM/EDS due to the lab’s type of detector. As was the case
with the propellant surface combustion particles collected, the ramjet exhaust particles
serve more to help with a general understanding of the combustion process. The most
directly pertinent data were obtained with the gas generator exhaust sampling, as these
data address the condition of the fuel particles injected in the CFD model discussed in
subsequent sections.
43
CHAPTER 3 – Ramjet Verification Tests
3.1 SSST Program Direct Connect Tests
3.1.1 Validation Ramjet Test Description

The first three ramjet direct-connect tests conducted under the Supersonic Sea-
Skimming Target (SSST) program at Aerojet were conducted in heavy-weight hardware
with the intention of optimizing the fuel injector configuration and identifying any
thermal issues with the combustor design. A convergent-only exit nozzle was used in
order to obtain more accurate thrust data without considering the losses associated with
flow divergence or separation in the expansion cone.
For the tests, vitiated air was supplied by a specially constructed facility designed to
deliver air at a desired mass flow rate and temperature. The air is heated through
combustion of oxygen and propane, with excess oxygen added to ensure 21% mole
fraction when delivered to the ramjet combustor. The result of vitiated heating becomes
higher amounts of water and CO2 in the “air” mixture, which is then accounted for in the
equilibrium thermochemistry calculations. The ramjet engine is mounted on a stand
allowed to freely translate axially for measurement of thrust. The Aerojet facility is
shown schematically in Figure 3.1.1-1. An ejector is shown downstream of the
combustor exit for tests where low base pressures simulating high altitude flight are
required. The SSST direct-connect hardware is shown installed in the facility in Figure
3.1.1-2. An on-stand vitiator was fabricated for SSST to minimize the run-up time
required to achieve the desired test condition. The curved feed lines supply required
gases and liquids to the vitiator without inducing excessive thrust tares.
44
Test Article Air Vitiator
Steam Ejector
Thrust Stand Image property of Aerojet Corporation
Figure 3.1.1-1 Aerojet Ramjet Test Facility
Combustion Chamber Air Plenum
Inlet Arms (4)

Exit Nozzle

Figure 3.1.1-2 SSST Heavywall Direct-Connect Installation
45
The SSST heavywall test hardware is shown in Figure 3.1.1-3. The inlet arms
include a sonic point to isolate upstream flow influences and to act as a metering point to
determine air flowrate for the tests. The internal gas generator fuel delivery system
flowpath and combustor geometry was simulated by the test hardware.
Figure 3.1.1-3 SSST Heavywall Test Hardware
A different fuel injector geometry was evaluated for each of the tests denoted
DC-1, DC-2, and DC-3. The three injectors are shown in Figure 3.1.1-4. The first test
DC-1 used a split of axial injection and radial fuel injection. Based on the observance of
unburned carbon the in exhaust, test DC-2 eliminated the axial fuel port in an attempt to
improve fuel residence time in the combustor. As the exhaust signature remained
unchanged, a move to a showerhead type injector was made for DC-3.
46
DC-1 DC-2 DC-3
• Injector • Injector • Injector
– 8 Radial Ports: Ø .344 – 8 Radial Ports: Ø .398 – 24 Radial Ports: Ø .0.213
– 1 Axial Port: Ø .566 – No Axial Port – 4 45 Deg Aft Ports: Ø
.0.213
Figure 3.1.1-4 SSST Heavywall Fuel Injectors
These three fuel injectors and the associated direct-connect test data were used to
validate the numerical model developed in the present research.
Pertinent hardware flow areas are presented in Table 3.1.1-1
Table 3.1.1-1 Heavywall Test Hardware Flow Areas
Parameter Value Drawing or

Measured
ACBE Choke Block Entrance Area 120.57 in2 D
ACB* Choke Block Throat Area 26.796 in2 M
A1.5 Diffuser Area 69.806 in2 D
A3 Fwd Combustor Area 132.73 in2 D
A4 Aft Combustor Area 132.73 in2 D
A5 Ramjet Nozzle Throat Area 40.42 in2 M
A6 Ramjet Nozzle Exit Area 40.42 in2 M
47
3.1.2 Test Data
The raw test data presented in this section were used to calculate the combustion
efficiency of the ducted rocket for each test. In each case, the raw data were smoothed
with a 10-point rolling averaging technique. Low speed data obtained using a PSI, Inc.
multiplexed pressure scanner measurement system are reported, at nominally a 10 Hz
data rate.
The test facility air flow rate actually consists of the component flows of air, oxygen,
and propane. Each quantity is redundantly measured at the metering point and in the
process building. In order to accommodate transients in air delivery during a test, the
total flow is also measured just upstream of the test article at a sonic flow point, referred
to as the inlet “choke block”. This flow point is calibrated using steady-state facility flow
data from immediately prior to the actual test. The facility air flow rates for the three
tests are shown in Figure 3.1.2-1, with time zero being onset of facility air flow. All
three tests were run at constant air conditions corresponding to Mach 2.5 at sea level,
with a desired value of 83.4 lbm/sec.
100.0
90.0
80.0
Mass Flowrate (lbm/sec)
70.0
60.0
50.0
40.0
30.0 DC-1
DC-2
20.0
DC-3
10.0
0.0
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0
Test Time (sec)
Figure 3.1.2-1 Direct-Connect Vitiated Air Mass Flowrates
48
The air temperature is computer controlled to a given set point by choosing the flows
of propane and oxygen to the air vitiator. The ratio of oxygen to propane is kept constant
in order to deliver 21% oxygen mole fraction in the inlet air. The desired inlet operating
point for simulating sea level air is 1172°R. The facility was able to hold this value
closely for all three tests as shown in Figure 3.1.2-2.
Air Temperature
1400.0
1200.0
1000.0
Temperature (°R)
800.0
600.0
DC-1
400.0
DC-2
DC-3
200.0
0.0
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0
Test Tim e (sec)
Figure 3.1.2-2 Direct-Connect Air Temperature
49
140.0
130.0
120.0
Pressure (psia)
110.0
100.0 DC-1
DC-2
90.0 DC-3
80.0
0.0 10.0 20.0 30.0 40.0 50.0
Test Tim e (sec)
Figure 3.1.2-3 Direct-Connect Ramburner Static Pressures
The combustor static pressure is measured from several radial positions and axial
stations for use in calculating the combustor total pressure. Averaged data from a station
upstream of the ramburner exit nozzle are shown in Figure 3.1.2-3. In general, the three
tests demonstrate good agreement. The time scale is shown in which zero is gas
generator ignition. The fuel flowrates were varied in tests DC-1 and DC-2, and a
corresponding change in combustor pressure is seen in each case. Figure 3.1.2-4 shows
the gas generator internal operating pressure and Figure 3.1.2-5 shows the corresponding
change in calculated fuel flowrate. Figure 3.1.2-6 shows the delivered gross thrust.
50
1400.0
1200.0
1000.0
Pressure (psia )
800.0
600.0
DC-1
400.0
DC-2
DC-3
200.0
0.0
0.0 10.0 20.0 30.0 40.0 50.0
Test Tim e (sec)
Figure 3.1.2-4 Direct-Connect Gas Generator Chamber Pressures
51
SSST DC-1, -2, and -3 Fuel Flow Profiles
4.5
DC-1
4
DC-2
3.5 DC-3
Wfuel (lbm/s)
2.5
1.5
0.5
0
0 10 20 30 40 50
Tim e (sec)
Figure 3.1.2-5 Direct-Connect Fuel Flowrates
Thrust
12000.0
10000.0
8000.0
Thrust (lb f)
6000.0
4000.0 DC-1
DC-2
2000.0 DC-3
0.0
0.0 10.0 20.0 30.0 40.0 50.0
Test Tim e (sec)
Figure 3.1.2-6 Direct-Connect Gross Thrust
52
3.2 Data Reduction Procedures
In order to calculate engine performance from ground test data, a procedure was
developed as outlined below. These steps were programmed in FORTRAN in a routine
called ARCEFF. This program was used to reduce the test data from the three SSST
validation cases as described in Section 3.3. The ramjet station nomenclature used is
illustrated in Figure 3.2-1. For engine ground testing, there is no capture station c, and
inlet arm total temperature is used as station 0. Some features are not applicable to the
test data examined, but the entire process is described for posterity and future reference.
3
(after dump, before heat/fuel addition)
P (primary)
0 C 2 4 5 6
(inlet dump plane)
Figure 3.2-1 Ramjet Engine Station Nomenclature
The following flow functions are defined here for reference, as they are utilized in
the data reduction procedures.
The X function calculates flow Mach / area relations:
53
γM ⎛A ⎞
X (γ , M ) = γ +1 where X 2 = X 1 ⎜ 1 ⎟
(1 + γ 2−1 M 2 ) 2 ( γ −1)
⎝ A2 ⎠
The Y function is used for mixing and heat addition:
M γ (1 + γ 2−1 M 2 )
Y (γ , M ) =
1+ γ M 2
The Z function is used in stream thrust relations:
1+ γ M 2
Z (γ , M ) = γ where F1 = PT1 A1 Z1
γ −1
(1 + 2
2
M ) γ −1
3.2.1 Thermochemistry
The theoretical equilibrium values for the process gamma, gas constant, total
temperature, and vacuum specific impulse are determined using vitiated air (combustion
of propane, air and make-up oxygen in these cases). Process gamma is that calculated by
equilibrium thermochemistry for isentropic flow between the stations indicated. The
table of properties is generated accounting for the correct amounts of propane and oxygen
added to reach the desired inlet total temperature. The analysis includes a constant
temperature drop reflecting the difference between measured vitiator temperature and
inlet arm temperature to determine the proper vitiator products; however, the reported
properties are at the inlet arm temperature. The data are calculated as a function of inlet
arm total temperature, combustor total pressure, and Air/Fuel ratio.
Vitiated air properties are also determined for use in pre- and post-test air-only
calculations.
54
3.2.2 Air Mass Flow Measurement
The values used for air mass flow in post-test analysis for all tests are calculated
from measured inlet arm temperatures and pressures. The discharge coefficient of the
sonic area of the inlet was measured during the period of constant airflow immediately
preceding the test using facility-based air flow measurements. All subsequent data
calculations used the inlet-based air flowrate to account for air flow transients during the
test.
3.2.3 Flow Areas

Pre- and post-test measurements of ramburner I.D. (sta. 4), nozzle throat (sta. 5),
nozzle exit (sta. 6), and base area were taken and the associated flow areas are normally
varied linearly between the ignition time and gas generator burnout. For the data
reduction performed herein, constant values were used, as determined from pre-test
measurements.
3.2.4 Ramburner Nozzle Discharge Coefficient:

In order to support the calculation of pressure-based combustion efficiency, the exit
nozzle discharge coefficient was determined during a time of constant airflow and
temperature (nozzle sonic) preceding the test. The choke-block entrance total
temperature, station 4 static pressures, and choke block airflow were used in the
calculations with vitiated air properties.
Discharge Coefficient Equations:

⎛C ∗A ⎞
X 4 = ⎜ D 5 ⎟ ∗ X 5 ; Defines M 4
⎝ A4 ⎠
gc
PT5 CD5 A5 X 5 ∗
w& a = R
TT0
γ
⎛ ( γ − 1) 2 ⎞ γ −1
PT4 = P4 ∗ ⎜1 + M4 ⎟
⎝ 2 ⎠
55
3.2.5 Pressure-Based Combustion Efficiency
The data used in the calculation are measured P4, choke-block total temperature,
choke-block air flowrate, and post-test fuel flow calculations resulting from a separate
analysis. Theoretical TT4, Process Gamma, and gas constant are looked up as a function
of TT0, fuel/air ratio, and Combustor Total Pressure.
⎛C ∗A ⎞
X 4 = ⎜ D 5 ⎟ ∗ X 5 ; Defines M 4
⎝ A4 ⎠
gc
PT5 CD5 A5 X 5 ∗
R4
w& a + w& f =
TT4
γ 4 −5
⎛ ( γ − 1) 2 ⎞ γ 4−5 −1
PT4,P = P4 ∗ ⎜1 + 4−5 M4 ⎟
⎝ 2 ⎠
TT4,P − TT0
ηc ,ΔT , P = ∗100.
TT4,Th − TT0
3.2.6 Thrust-Based Combustion Efficiency
The following procedure was followed:
1) Correct the measured thrust for the thrust stand tares. For the Aerojet test stand a
linear spring constant is used:
Fmeas
Fcorr =
0.9865
56
2) Correct for base pressure forces
For tests using an ejector to lower the exit pressure, a base pressure correction should
be made –
Fx = Fcorr + Abase ( Pa − Pbase ) ; where Abase = Acase − A6
For these tests there was no ejector installed, so Fx = Fcorr
3) Correct For Nozzle Divergence
For tests using convergent-divergent nozzles, a correction should be made to the

momentum term in the thrust equation to account for measurement of only the axial
component of thrust. The SSST tests utilized a convergent-only nozzle so no divergence
losses were seen and Fg = Fx.
⎡ Fx − ( P6 − Patm ) A6 ⎤⎦
Fg = ⎣ + ( P6 − Patm ) A6
(1 + cos θ )
2
where theta is the nozzle half-angle.
γ 4−5
⎛ ( γ − 1) 2 ⎞ γ 4−5 −1
P6 = PT6,F / ⎜ 1 + 4 −5 M6 ⎟ ; PT6 F determined below
⎝ 2 ⎠
⎛ Cd A5 ⎞
X 6 = ⎜ 5 ⎟ ∗ X 5 ; Defines M 6 , Z 6
⎝ A6 ⎠
4) Determine the nozzle stream thrust efficiency before the test during air flow only,
after reaching equilibrium. Combustor Mach number, total pressure, and total
57
temperature were determined in the Cd calculations. For the SSST tests the nozzle
efficiency was calculated for the 10 seconds immediately preceding gas generator
ignition.
Stream thrust based on measured thrust; F6 = Fg + Patm A6
F6
PT6 ,F =
A6 Z 6
PT6 ,F
ηn =
PT4 ,P
5) Calculate Thrust-Based Efficiency
The nozzle stream thrust efficiency can be varied linearly between pre- and post-test
values, however the constant pre-test value was used. The efficiency calculation
procedure is the same except PT4F and TT4F are not known.
F6 = Fg + Patm A6
PT6,F = F6 A6 Z 6
PT6,F
PT4 ,F =
ηn
PT6F
PT4F = η
n
gc
PT5,F CD5 A5 X 5 ∗
R4
w& a + w& f =
TT4,F
58
TT4,F − TT0
ηc ,ΔT , F = ∗100.
TT4,Th − TT0
6) Burner Pressure Recovery
Pressure recovery is reported as the ratio of ramburner total pressure ( PT4,P, based on
pressure) to inlet diffuser total pressure based on the inlet static pressure measurements at
station 1.5. The ratio is plotted as a function of modified heat addition parameter MHAP
as defined below.
S A* PT A5 Z 5
MHAP = ∗ φ = 4,P ∗φ
TT0 w& a TT0
3.3 Baseline Verification Test Results
3.3.1 Equilibrium Combustion

The ARCEFF program was used to process the test data for tests DC-1, 2, and 3. As
the results are normalized at ignition with pre-test cold flow coefficients, the most
accurate time for obtaining accurate ramburner performance is soon after gas generator
ignition. For this reason, only the first twenty seconds of data are presented here. All
three tests are near the same throttle setting and fuel flowrate during this time so the
effects of equivalence ratio are removed and the differences in combustion efficiency
reduce to injector design.
Base pressures were measured for each test; however, the change in pressure from
zero flow to ignition was 0.09 psi, or roughly 10 lbf out of 9500 lbf. Since this is 0.1%,
and the effect is normalized by utilizing a cold flow stream thrust coefficient, base
pressures were not included in the analysis.
59
Input files for ARCEFF were created for the fuel thermochemistry, test data, and fuel
flow rate which is separately calculated. The coefficients determined for each test during
the pre-ignition cold flow averaging are shown in Table 3.3.1-1.
Table 3.3.1-1 Ramjet Test Data Coefficients
Test Number Cd 0 Cd 5 ηn
DC-1 1.0058 .9929 .9770
DC-2 1.0269 1.014 .9767
DC-3 1.0096 .9914 .9747
Tests one and three are consistent in terms of coefficients obtained; however, the test
2 discharge coefficients appear out of family. This discrepancy could be due to errors in
instrumentation channels or facility parameters affecting airflow measurements. The
calculated temperature rise combustion efficiencies for each test for the first 20 seconds
are shown in Figures 3.3.1-1 through 3.3.1-3. The red line is thrust-based in all cases.
110
100
Combustion Efficiency (temp rise) - %
90
80
70
60
50
0 2 4 6 8 10 12 14 16 18 20
Test Time (sec)
Figure 3.3.1-1 DC-1 Temperature Rise Combustion Efficiency
60
110
100

90
80
70
60
50
0 2 4 6 8 10 12 14 16 18 20
Test Tim e (sec)
110
100
90
80
70
60
50
0 5 10 15 20
Test Tim e (sec)
61
Both DC-1 and DC-3 exhibit extremely good agreement between pressure-based and
thrust-based combustion efficiencies. The disagreement between the two on test DC-2
casts some doubt on the absolute levels for either. The pressure-based efficiencies for the
three tests are shown together in Figure 3.3.1-4 for comparison. The effects of injector
geometry for the three tests is clearly defined, with the DC-1 injector yielding the highest
efficiency. These results were used to gauge the effectiveness of the numerical FLUENT
model and the ability to correctly predict injector effects.
110
100
90
80
DC-1
70
DC-2
DC-3
60
50
0 5 10 15 20
Test Time (sec)
Figure 3.3.1-4 Pressure-Based Temperature Rise Combustion Efficiency
An alternate method of calculating combustion efficiency uses the fuel/air definition

shown in Figure 3.3.1-5. As in the case of temperature rise efficiency, the temperature at
station 4 is calculated; however, for this definition an “effective” fuel/air ratio is found
corresponding to the calculated temperature. The ratio of effective to operating
62
efficiency forms the fuel/air efficiency. Although not entirely accurate since non-
equilibrium intermediate reactions can be the source of reduced total temperature, the
value provides a rough indication of unburned fuel at the exit. The data based on a
temperature calculated from the measured pressure are provided in Figure 3.3.1-6, again
for comparison to the FLUENT predictions.
f
ηc f = a eff ective
• 100
f Stoichio metric f/a
Comb ustor Tota l Tempe ra ture
a
a ope rating
TT 4 theoretical
Calculated TT 4
Operating f/a
Effective f/a
Figure 3.3.1-5 Fuel/Air Combustion Efficiency Definition
63
110
100
Combustion Efficiency - %
90
80
DC-1
70 DC-2
DC-3
60
50
0 5 10 15 20
Test Time (sec)
Figure 3.3.1-6 Fuel/Air Combustion Efficiency – Based on Pressure
The burner pressure loss for each test is shown in Figure 3.3.1-7 as a function of
Modified Heat Addition Parameter (MHAP). As there is little change in test conditions
and MHAP between tests, the data are centered at a value of approximately 1.18. The
pressure loss for tests DC-1 and DC-3 is consistent at 0.875. Test DC-2 is again out of
family at over 0.92.
64
0.96
0.94
0.92
DC-1
PT4/PT2
0.9 DC-2
DC-3
0.88
0.86
0.84
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Modified Heat Addition Parameter (MHAP)
Figure 3.3.1-7 Ramburner Pressure Losses
3.3.2 Test Data Used for CFD Analysis

Since the DC-2 test rendered results that appear corrupted, the tests DC-1 and DC-3
were used to validate the FLUENT combustor model. These tests represent two
distinctly different injector configurations and present the opportunity to evaluate the
model capability to distinguish between the two flowfields.
A nineteen second average was calculated for the tests for use in correlating the CFD
models. Transient effects are apparent during the first second and, later in the test,
uncertainties are introduced such as flow area changes, instrumentation drift, and fuel
flowrate accuracy as the grain face may deviate from a flat surface. The temperature rise
efficiencies and corresponding total combustion temperatures were determined as:
65
Test DC-1
ηc = 93.6%, TT4 = 1352.6 °K (2434.7 °R)
Test DC-3
ηc = 87.9%, TT4 = 1309.7 °K (2357.5 °R)
This data analysis determines total combustion temperatures assuming that boron
contributes to the combustion process. The final FLUENT modeling neglected boron in
the modeling; therefore, these temperature results require modification to serve as a basis
for modeling. Figure 3.3.2-1 illustrates the effect of boron combustion on the equilibrium
total temperature as a function of fuel/air ratio.
4000
3500
Equilibrium
3000
Combustion TT4 (°R)
2500
2000
No Boron
1500
1000
500
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07
Fuel / Air Ratio
Figure 3.3.2-1 Effect of Boron Combustion on Ramjet Flame Temperature
66
7
Temperature Reduction (%) 6
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07
Fuel / Air Ratio
Figure 3.3.2-2 Temperature Decrease by Neglecting Boron
In the region of interest for the FLUENT modeling, the fuel/air ratio is .042, and the
corresponding decrease in combustion temperature by neglecting boron is 5.7%. The
corresponding total temperature basis for the FLUENT model then becomes:
Test DC-1, TT4 = 1279.7 °K (2303.4 °R)

Test DC-3, TT4 = 1239.1 °K (2230.3 °R)
Perhaps a better indication of model accuracy would be that under a given set of
assumptions, the DC-1 case shows higher combustion efficiency, on the order of a 6%
improvement. This is the case in which a significant portion of the fuel is injected axially
through a center port in the injector. This distinguishing feature proved significant in the
development of the FLUENT model in the following chapter.
67
CHAPTER 4 – Computational Fluid Dynamics Modeling
With the fuel exhaust characterization experiments, theoretical thermochemical
calculations, and validation test data reduction complete, a numerical CFD model of the
SSST heavywall combustor was constructed using GAMBIT for grid generation and
FLUENT Version 6.2 to solve the steady flow equations for continuity, momentum, and
energy. Models were developed with increasing complexity, starting with reacting gas
phase only, then adding particle combustion using built-in FLUENT algorithms
following the theory described in Section 1.3.3.
4.1 Grid Development

A solid model of the SSST heavywall combustor test hardware was used to create a
¼-slice and imported into GAMBIT to form the vertex and line structure for the
combustor inlet, wall, and nozzle. The initial nominal grid spacing was .05 inches, which
corresponds to 0.6% of the inlet flow area minimum dimension, 0.8% of the combustor
flow area radius, and 1.4% of the nozzle internal radius. A baseline grid was created in
this geometry as shown in Figure 4.1-1. Details for a mapped grid are shown in Figures
4.1-2, 4.1-3, and 4.1-4. The models include a total of 783,116 cells for the DC1
configuration, and 730,775 cells for the DC3 configuration, with the differences being
mostly in the fuel injector region. The maximum cell skewness is 0.968 for both the DC1
and DC3 cases. Edge boundary conditions were defined for the inlet and combustor
walls, inlet flow, and outlet flow as shown in Figure 4.1-1.
68
Air mass flow 20.75 lbm/s
Total temperature 1167oR
No-slip and adiabatic wall
Extrapolated
Fuel gaseous mass flow 0.52 lbm/s
Fuel particle (Carbon) mass flow 0.31 lbm/s
Particle and gas total temperature 1944oR
Symmetry
Figure 4.1-1 Baseline Grid and Edge Boundary Condition Designations
69
Figure 4.1-2 Baseline .05 in. Nominal Grid – Dump Region
Figure 4.1-3 Baseline .05 in. Nominal Grid – Dome Region
70
Figure 4.1-4- Baseline .05 in. Nominal Grid – Nozzle Region
4.2 Turbulence Modeling

The turbulence modeling options available in FLUENT 6.244 are summarized below:
Spalart-Allmaras model
This is a one equation model that solves a modeled transport equation for the
kinematic eddy (turbulent) viscosity. It is not recommended where flow changes
abruptly from a wall-bounded to a free shear flow. Turbulent dispersion of particles
cannot be included.
k-ε models
Standard k-ε model
This is a two-equation model in which the solution of two separate transport
equations allows the turbulent velocity and length scales to be considered. It is known
for its robustness, economy, and reasonable accuracy for a wide range of turbulent flows.
Renormalization-group (RNG) k-ε model
71
The analytical derivation results in a model with constants different from those in the
standard k- ε model, and additional terms and functions in the transport equations for k
and ε. The RNG model is more responsive to the effects of rapid strain and streamline
curvature than the standard k- ε model, which explains the superior performance of the
RNG model for certain classes of flows.
Realizable k-ε model
The term “realizable“ means that the model satisfies certain mathematical constraints
on the normal stresses, consistent with the physics of turbulent flows. A weakness of the
standard k- ε model or other traditional k- ε models lies with the modeled equation for the
dissipation rate (ε). The well-known round-jet anomaly (named based on the finding that
the spreading rate in planar jets is predicted reasonably well, but prediction of the
spreading rate for axi-symmetric jets is unexpectedly poor) is considered to be mainly
due to the modeled dissipation equation. The realizable k-ε model was intended to
address these deficiencies of traditional k-ε models, and was extensively validated for a
wide range of flows, including rotating homogeneous shear flows, free flows including
jets and mixing layers, channel and boundary layer flows, and separated flows. For all
these cases, the performance of the model has been found to be substantially better than
that of the standard k- ε model.
k-ω models
Standard k-ω model
This is an empirical model based on model transport equations for the turbulence
kinetic energy (k) and the specific dissipation rate (ω). A gradual change from the
standard k-ω model in the inner region of the boundary layer to a high-Reynolds-number
version of the k- ε model in the outer part of the boundary layer is made. A modified
turbulent viscosity formulation is used to account for the transport effects of the principal
turbulent shear stress.
Shear-stress transport (SST) k-ω model
Turbulent viscosity is modified to account for the transport of the principal turbulent
shear stress.
Reynolds stress model (RSM)
72
This method solves transport equations for each of the terms in the Reynolds stress
tensor. It is good for highly swirling flows and stress driven secondary flows, but 7
additional transport equations are introduced.
Detached eddy simulation (DES) model
RANS modeling with LES for applications in which classical LES is not affordable.
Large eddy simulation (LES) model
The CPU time is prohibitive for complicated flows.
The standard k-ε model was selected here for the baseline calculations. Sensitivity to
turbulence model selection was also considered.
4.3 Gas Phase Model

As a first step in the CFD analysis, a gas-phase only FLUENT model was developed
for the DC-1 and DC-3 test cases. In order to define the fuel constituents, a simplified
representation of the fuel was created. Equilibrium thermochemistry calculations were
performed with non-reacting boron to identify the major gas phase fuel compounds to be
methane, hydrogen, carbon monoxide, hydrochloric acid, nitrogen, and water as shown in
Table 4.3-1.
73
Table 4.3-1 Gas Generator Throat Products for Non-Reacting Boron
THROAT MASS FRACTIONS AT EQUILIBRIUM CONDITIONS (non-reacting boron)

________________________________________
(BOF)3 0.00001 (FECL2)2 0.00149 (HOBO)3 0.00269

(MGCL2)2 0.00000 B2O3 0.00000 BF3 0.00003
BHF2 0.00000 BOCL 0.00000 BOF 0.00003
BOHF 0.00000 BOHF2 0.00009 CH4 0.05566
CL 0.00000 CL2 0.00000 CO 0.21625
CO2 0.01725 FE 0.00000 FECL 0.00000
FECL2 0.01878 FECL3 0.00001 FEF3 0.00000
FEH2O2 0.00000 FEOH 0.00000 H 0.00000
H2 0.04015 H2O 0.03115 H3BO3 0.00184
HBO 0.00000 HOBO 0.00015 HCL 0.09360
HCN 0.00003 HF 0.00168 MG(OH)2 0.00000
MG<NR> 0.00023 MGCL 0.00000 MGCL2 0.00032
MGCLF 0.00000 MGF2 0.00000 N2 0.04129
NH3 0.00026 OH 0.00000 B$<NR> 0.05003
B4C$ 0.00000 BN$ 0.00000 C$ 0.35567
FE$ 0.00000 FE2O3$ 0.00000 FE2O3$ 0.00000
FE3C$ 0.00541 FE3O4$ 0.00000 FE3O4$ 0.00000
MG$<NR> 0.00000 MGF2$ 0.00426 MGO$ 0.04685
B*<NR> 0.00000 B2O3* 0.01479 FE3C* 0.00000
FEO* 0.00000 MGO* 0.00000
These major gaseous products, as well as the solid carbon and boron were extracted
and the major inert species were adjusted to account for the neglected products as shown
in Table 4.3-2.
Table 4.3-2 Adjusted Throat Mass Fractions
Simplified "Fuel" Mass Fractions Unaccounted Lumped into Major Gaseous Inerts (CO, N2, HCL)
Gas Phase Gas Phase Total Gas Mass Flow (.875 lb/sec total)
(lbm/sec)
CH4 0.05566 CH4 0.05566
H2 0.04015 H2 0.04015
CO 0.21625 CO 0.21625 0.5944 0.52
N2 0.04129 N2 0.06774
HCL 0.09360 HCL 0.15355
H2O 0.03115 H2O 0.06100
Solid Phase Solid Phase
C 0.35567 C 0.35567 0.4057 0.35

B 0.05003 B 0.05003
Unaccounted 0.11620
74
Thus, with gaseous fuel used as the only injectant, the resulting mass fractions for
the major products are shown in Table 4.3-3.
Table 4.3-3 Gaseous Fuel Only Mass Fractions
Gaseous Fuel Mass Fractions
CH4 0.09365
H2 0.06755
CO 0.36384
N2 0.11397
HCL 0.25835
H2O 0.10263
These fuel ingredients were used as the injectant in a FLUENT model, in addition to
vitiated air including appropriate amounts of CO2 and water. For air heated to 1167°R
with propane and makeup oxygen added to reach 21% mole fraction the oxidizer mass
fractions are: N2 – 0.737, O2 – 0.233, CO2 – 0.02, H2O – 0.01.
In order to check the results of the FLUENT case, special thermochemistry
calculations were required to arrive at a predicted equilibrium combustion temperature
with these fuel ingredients and vitiated air ingredients. In order to accomplish this, the
heats of formation were calculated using NIST data44 for each ingredient at the
temperature that they are introduced (1944°R for the fuel, and 1166°R for the air). The
resulting equilibrium combustion total temperature is shown in Figure 4.3-1, and is
expressed in terms of temperature rise combustion efficiency in Figure 4.3-2.
75
Equilibrium Combustion Temperature
2900
2400
Chamber TT (°R)
1900
1400
900
400
0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080
f/a Ratio
Figure 4.3-1 Combustion Total Temperature for Simplified Gaseous Fuel
76
Gas Phase Temperature Rise Combustion Efficiency
120
100
Combustion Efficiency (%)
80
60
40
20
0
800 850 900 950 1000 1050
Combustion Total Temperature (°K)
Figure 4.3-2 Combustion Efficiency for Simplified Gaseous Fuel
For the baseline fuel/air ratio of 0.025, the corresponding combustion temperature is
1853 °R.
Non – premixed combustion FLUENT cases were run for test DC-3 with air as the
inlet gas and the fuel ingredients as shown in Table 4.3-3. The mass flowrates of 20.75
lbm/sec of air and 0.52 lbm/sec of fuel represent approximately one-fourth of the full
scale test mass flowrates. Important options selected for the solver and controls are
provided below for the baseline case:
DC3 Case (-102 Injector Configuration):

Mass flow inlet – 20.75 lbm/s
Fuel Inlet boundary condition temperature: 1944R
77
Fuel Mass Flow: 0.52 lbm/sec (gaseous portion of 0.875 lbm/sec GG fuel flow)
3-D model & grid
Pressure based implicit
K-epsilon turbulence model
Combustion modeling:
Non-premixed Combustion (Mixture Fractions)
Radiation Model – P1
Operating Pressure 120psi
Pressure outlet Boundary Condition @ throat (-119 psi)
Solver Options
Coupled solver
Initial Courant number of 1, increased to 10, then 50
Under relaxation factors: Energy 0.5, Mean mixture fraction 0.9, mixture fraction
variation 0.9
First order upwind
Predicted contours of static temperatures are shown in Figure 4.3-3. The contours
illustrate that for the gas phase reactions, the majority of the combustion takes place in
hot zones in the combustor head end and immediately downstream of the inlet arms. The
temperature profile has nearly stabilized by the time the nozzle is reached. This should
point to a reasonably high value for combustion efficiency.
Contours of mean fuel mixture fraction are shown in Figure 4.3-4. These contours
illustrate again that there is very little unreacted gaseous fuel at the nozzle exit.
The mass-weighted average total temperature calculated for this case at the nozzle
exit was 1018.7 °K. Using the thermochemical calculations above, this equates to a
combustion efficiency of 97.2%. This indicates that at least the gaseous phase of the fuel
injectant should be readily combusting in the ramjet engine. This case also served to
validate the thermochemical inputs to the FLUENT model, as they closely matched an
independent analysis method.
78
Figure 4.3-3 Predicted Contours of Static Temperatures for Case DC-3 – Gas Phase
Only
79
Figure 4.3-4 Predicted Contours of Fuel Mixture Fraction for Case DC-3 – Gas
Phase Only
80
4.4 Particle Combustion Modeling
The next step in the modeling effort was to introduce solid particles as part of the
fuel stream. A module is built into the FLUENT code based on pulverized coal particle
combustion. The solid phase carbon fuel exhaust was added to the CFD model using this
option, using a 20 micron particle diameter based on the collection data, and a mass
flowrate based on equilibrium thermochemistry calculations. The small amount of boron
present in the fuel was neglected for these studies. Important options selected for the
solver and controls are provided below for the particle combustion cases:
Pure C(s) added as a secondary fuel stream to the pdf

Temperature = 1944R
Rich Flammability Limit
0.8 for primary stream (gaseous fuels phi=~3)
0.2 for secondary stream (C<s>, stoich = 0.086)
Set up Discrete Phase Model:
Interaction with Continuous Phase
10 Continuous Phase Iterations per DPM iteration
Max # Steps = 10,000, Step Length Factor = 5
Spherical Drag Law
Particle Radiation Interaction
Injection Properties:
Injection Type = Surface, released from “fuel_inlet”
Particle Type = Combusting
Material = “carbon-kin/dif-limit” (Fluent Database carbon, with kinetic/diffusion-
limited combustion enabled)
Diameter Distribution = uniform
Oxidizing Species = O2
Devolatilizing Stream = secondary
Char Stream = secondary
Point Properties
Diameter = 0.0007874 inch = 20 um
Temperature = 1944R
Velocity Magnitude = 1500 ft/s (matches velocity at “fuel_inlet” surface; this
was important to prevent incomplete particle tracking)
Total Flow Rate = 0.3112 lbm/s (mdot of solid carbon; solid boron mdot not
included)
Injected using face normal direction
Turbulent Dispersion
Stochastic Tracking (in the stochastic tracking approach, FLUENT predicts the
turbulent dispersion of particles by integrating the trajectory equations for
individual particles, using the instantaneous fluid velocity along the particle path
81
during the integration. By computing the trajectory in this manner for a
sufficient number of representative particles (termed the number of “tries"), the
random effects of turbulence on the particle dispersion may be accounted for. In
FLUENT, the Discrete Random Walk (DRW) model is used. In this model, the
fluctuating velocity components are discrete piecewise constant functions of
time. Their random value is kept constant over an interval of time given by the
characteristic lifetime of the eddies)
Discrete Random Walk Model
# of tries = 10 (default)
Time Scale Constant = 0.15 (default)
Material Properties for Combusting Particle:
Selected “carbon” from the Fluent Database
Density = 124.8517 lbm/ft3
Cp (Btu/lb-R) = 0.2913997
Thermal Conductivity (btu/h-ft-R) = 0.02623823
Latent Heat (btu/lbm) = 0
Vaporization Temperature = 1944R
Volatile Fraction = 0%
Binary Diffusivity (ft2/s) = 0.0004305564
Particle emissivity = 0.9
Particle Scattering Factor = 0
Swelling Coefficient = 1 (stays constant diameter)
Burnout Stoichiometric Ratio = 2.67
Combustible Fraction = 100%
Devolatilization Model (1/s) = 0
Combustion Model = kinetics/diffusion-limited
Additional manual equilibrium calculations were again made for the fuel mixture
including particulate carbon and boron as shown in Figure 4.4-1. For the case of gaseous
fuel and carbon (no boron), at a fuel/air ratio of .04, the total temperature is 1401°K.
Also shown is a traditional equilibrium thermochemistry calculation performed the
traditional way using fuel raw ingredients as the ingredients. The results are very close to
the piece-wise technique required due to the simplified fuel chemical modeling.
82
4000
Equilibrium Using Propellant
3500
air 20.75 lb/sec a/f
fuel 0.52 lb/sec 24.9639 Carbon, Boron, Gas Phase Reacting
boron 0 lb/sec 0.044
3000
carbon 0.3112 lb/sec
Carbon, Gas Phase Reacting
total fuel 0.8312 lb/sec
2500
MF mass tot MF
Chamber TT (°R)
fuel (1080KCO 0.36387 0.189212 0.2276
CH4 0.093656 0.048701 0.0586
2000
N2 0.11398 0.05927 0.0713
H2 0.067558 0.03513 0.0423
HCL 0.25839 0.134363 0.1616 Gas Phase Reacting
1500
H2O 0.10256 0.053331 0.0642
B 1 0 0.0000
DC-2 Case, Gas Only
C 1 0.3112 0.3744
1000
500
total 1.0000
0
0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080
f/a Ratio
Figure 4.4-1 Thermochemistry for Gas Phase and Carbon Particles
The initial results indicated that any axial particles in this size range injected axially
would bypass the significant combustion zones and exit the combustor unreacted. Figure
4.4-2 shows the carbon particle mixture fraction for a representative case, and Figure 4.4-
3 shows the particle traces for test DC-1, in which the axially injected particles are
predicted to essentially pass through most of the combustor unreacted.
83
Figure 4.4-2 Predicted Mixture Fraction for Carbon Particle Fuel for Case DC-3
84
Figure 4.4-3 Predicted Case DC-1 Particle Traces
Another concern with the initial particle combustion runs was that the DC-3
combustor configuration (with all radial fuel injection) showed higher mean total
temperatures at the nozzle exit. This is in direct contradiction to the ground test results.
Representative temperature profiles for tests DC-1 and DC-3 are shown in Figure 4.4-4,
illustrating the higher temperatures at the nozzle exit.
DC-1 DC-3
Figure 4.4-4 Predicted Temperature Profiles for Tests DC-1 and DC-3
85
The conclusions drawn from the initial FLUENT runs for both DC-1 and DC-3 were
that:
1) If using a pulverized coal carbon combustion model and particle diameters

obtained from ground testing for the solid phase, insufficient particle combustion
would take place to match test temperature rise results
2) Axial carbon particle injection such as DC-1 results in lower combustion

efficiency as the particles pass through unreacted, in direct contrast to the test
results, wherein DC-1 presented higher performance
Options available to numerically increase solid phase particle combustion consist of:
– Arbitrarily decrease mean particle size
– Utilize a particle size distribution (test data exist), as the smaller particles react
and provide a heat source to accelerate the combustion of larger particles
– Arbitrarily increase the reaction rate coefficient. No data exist for solid ducted
rocket carbon reaction rates
– Decrease the injected particle velocity relative to the fuel gas phase to reduce
axial momentum and enhance mixing. The baseline cases were run with particle
velocity matching the gas phase.
A series of FLUENT cases were run to help determine trends for these various
assumptions. The results are shown in Table 4.4-1.
86
Table 4.4-1 Carbon Combustion Parametric Results
Fraction Total
of Carbon Temperature
Burned at Exit
lbm/s oR
DC-3 with nozzle
Gas only, Tf=1944oR n/a
Baseline reaction rate 0.143 2107
Double reaction rate 0.196 2231
10X reaction rate 0.311 2374
Injection velocity 10 fps 0.136 2013
Particle size 2 microns 0.311 2288
Test data - 2358 R
The parametric FLUENT particle results show that simply adding a particle size
distribution or increasing reaction rate will improve combustion, but not show improved
DC-1 combustion over DC-3. The demonstrated temperature rise can be achieved in the
FLUENT model if all carbon is forced to react (10X rate case). The conclusions drawn
from these parametric runs were:
Carbon reaction rate -> strong effect

Particle size -> strong effect
Particle initial velocity -> little or none
Turbulence dispersion -> strong effect with small particles, less effect with large particles
(for large particles on the order of 20 microns there is less interaction with the gas flow
and particle dispersion effects are minimal).
Interestingly enough, runs made with just the gas phase reacting for DC-1 and DC-3
show improved performance for DC-1, in better agreement with the test data.
Representative temperature profiles for the two cases are shown in Figure 4.4-5. Since
the gas phase only FLUENT runs indicate that DC-1 has higher performance, the
87
conclusion can be reached that the fuel solid phase likely contains a large fraction of sub-
micron carbon particles that follow the gas flow and react quickly, or less carbon is
produced than equilibrium calculations would predict. The remaining carbon would
likely be in the form of higher molecular weight gases that do not fully react in the gas
generator and are introduced to the combustion process as part of the fuel gas phase. In
either case, the FLUENT model must be adjusted to account for a large portion of the
fuel particulates that are more closely coupled to gas phase reactions than the particle
combustion model.
DC-1 DC-3
Figure 4.4-5 Predicted Gas Phase Combustion Profiles for Tests DC-1 and DC-3
The next, and final, series of runs in model development were undertaken to replace
a fraction of particulate carbon in the fuel with gaseous fuel constituents. This was done
to test the hypothesis that all the carbon in the fuel was not introduced as 20 micron
particles as the particle collection tests would lead one to believe. Although visible in the
exhaust, they may actually only make up a small percentage of the mass flow, and may in
fact represent agglomerates of much finer soot particles.
Table 4.4-2 shows the matrix of runs made with the fraction of solid carbon in the
fuel varying from 100% down to 3.2%, with the resulting mass fractions of gaseous
species.
88
Table 4.4-2 Fuel Solid Phase Substitution
Percent of C in solid phase 100% C 3.2% 20.0% 40.0% 60.0%

is solid is solid is solid is solid is solid
CH4 0.0937 0.0593 0.0633 0.0689 0.0756
H2 0.0676 0.0428 0.0457 0.0497 0.0545
N2 0.1140 0.0722 0.0771 0.0839 0.0920
H2O 0.1026 0.0649 0.0694 0.0755 0.0828
HCL 0.2584 0.1636 0.1747 0.1901 0.2085
CO 0.3639 0.2304 0.2461 0.2677 0.2936
C 0.0000 0.3668 0.3238 0.2642 0.1931
SUM 1.0000 1.0000 1.0000 1.0000 1.0000
Fraction of C in particle form 0.0321 0.2000 0.4000 0.6000

Fraction of C in "gaseous" form 0.9679 0.8000 0.6000 0.4000
Gaseous fuel flow rate, lbm/s 0.52 0.52 0.52 0.52 0.52
Particle carbon flow rate, lbm/s 0.3112 0.0100 0.0622 0.1245 0.1867
"Gaseous" carbon flow rate, lbm/s 0 0.3012 0.2490 0.1867 0.1245
SUM 0.8312 0.8312 0.8312 0.8312 0.8312
Gaseous fuel and "gaseous" C flow rate, lbm/s 0.5200 0.8212 0.7690 0.7067 0.6445
As the validation ground tests were conducted with fuel containing boron, an
adjustment is required to compare results of the FLUENT calculations in which the boron
is ignored. For the specific case with an air temperature of 1167°R, a combustion
pressure of 125 psia, and fuel/air ratio of .04, the theoretical combustion total temperature
is 2553 °R. The combustion efficiencies obtained from the validation tests as discussed
in Section 3.3.2 are restated below:
Test DC-1 ηc = 93.6%, TT4 = 1279.7 °K (2303.4 °R), boron neglected

Test DC-3 ηc = 87.9%, TT4 = 1239.1 °K (2230.3 °R), boron neglected
The FLUENT model results for cases varying the fraction of fine/large carbon
particles are shown in Figure 4.4-6. The curves labeled DC-1 and DC-2 represent the
89
baseline CFD results for the two configurations, and the curves labeled DC-1 Test and
DC-3 Test are the average values of combustion efficiency reported for the first twenty
seconds of the ramjet direct-connect test. As the 20 micron particles are replaced with
gaseous species, the absolute levels of combustion efficiency increase, and approach the
magnitudes seen in the validation direct-connect testing. Initially DC-3 shows improved
performance over DC-1, but as the gaseous fraction increases, DC-1 is predicted to show
5-6% better combustion efficiency. This behavior is in good agreement with the
performance levels and trends observed in testing. These results confirm the hypothesis
that a large portion of the carbon soot predicted by equilibrium thermochemistry is in fact
sub-micron in size and/or is actually still in the form of gaseous hydrocarbons. With
these results in hand, the baseline model development was considered complete.
Mass Weighted Average Temperature at Combustor Exit
100.00
95.00
90.00
DC-1
DC-3
85.00
DC-1 Test
DC-3 Test
80.00
75.00
70.00
0.0000 0.2000 0.4000 0.6000 0.8000 1.0000
Fine Carbon Fraction
Figure 4.4-6 Combustion Efficiency as a Function of Fine Carbon
90
4.5 Model Parametrics
Additional FLUENT runs were conducted to evaluate the sensitivity of the model
results to select modeling parameters. The results indicated that there is little variation in
solutions such that the baseline case assumptions remained valid.
4.5.1 Turbulence Model

The realizable k-ε model was chosen as the most applicable scheme to compare to
the baseline model results as discussed in Section 4.2. Figure 4.5.1-1 presents the
baseline model data compared to the realizable turbulence model. The curves labeled
DC-1 and DC-2 represent the baseline CFD results for the two configurations, and the
curves labeled DC-1 Realizable and DC-3 Realizable are the values of combustion
efficiency obtained with the realizable turbulence model. All other parameters were left
unchanged in the analysis. As seen, the levels of combustion efficiency for the two
models compare favorably at all levels of gaseous fuel fraction, validating the use of the
standard k-ε model for all other analysis.
91
100.00
95.00
DC-1 Realizable
90.00
DC-3 Realizable
DC-1 Test
85.00
DC-3 Test
DC-1
80.00
DC-3
75.00
70.00
0.0000 0.2000 0.4000 0.6000 0.8000 1.0000
Figure 4.5.1-1 Model Results for Realizable k-ε Model
4.5.2 Higher Order Calculations

As another check on model sensitivity, a second order upwind analysis was made.
While the first-order upwind discretization generally yields better convergence than the
second-order scheme, it generally will yield less accurate results, especially in situations
in which the flow is not aligned with the cell face. A comparison of second order
discretization compared to the first order baseline computations is shown in Figure 4.5.2-
1. Again in this case very good agreement is seen with the baseline analysis.
92
100.00
95.00
DC-1
90.00
DC-3
DC-1 Test
85.00
DC-3 Test
DC-1 2nd order
80.00
DC-3 2nd order
75.00
70.00
0.0000 0.2000 0.4000 0.6000 0.8000 1.0000
Figure 4.5.2-1 Second Order Model Results
4.5.3 Grid Dependency

In order to evaluate the effects of grid size on model accuracy, a finer grid was
created by bisecting each edge of the baseline grid for the DC-1 case. DC-1 was used for
the grid dependency analysis as it was felt to present a more complex flow environment
due to the strong axial component of particulate fuel injection. The number of cells
increased from 783,000 to 6,290,000. A dramatic increase in computation time was
observed for each iteration. The mass-averaged exit flow temperature history is shown in
Figure 4.5.3-1 as the iterations proceeded over a period of 10 days using sixteen CPUs.
Although a decrease in exit temperature of over 50 degrees is seen, it was unclear
whether a larger amplitude sinusoidal pattern would result with continued computations,
and where the solution would eventually converge. In order to more aggressively
approach the solution the momentum under-relaxation factor was increased from 0.1 to
0.3, and the Courant Number was increased from 20 to 50. The resulting temperature
convergence history is seen in figure 4.5.3-2.
93
Figure 4.5.3-1 Fine Grid Convergence History – Baseline Under-Relaxation
Figure 4.5.3-2 Fine Grid Convergence History – Aggressive Under-Relaxation
94
With the more aggressive under-relaxation the expected sinusoidal behavior was
observed but the required computation time required to reach convergence again was
prohibitive.
A simpler case using gas phase fuel combustion only for case DC-1 was then used to
examine the effect of a finer grid. This appeared a reasonable approach since the CFD
results had shown that test data matched predicted model temperatures more closely
when the majority of fuel was included in the gaseous phase. In this case the fine grid
converged much more quickly and demonstrated good agreement with the baseline grid.
The mass-averaged total temperature at the throat was 2397°R for the baseline grid, and
2407°R for the fine grid.
Figure 4.5.3-3 Fine Grid Convergence History – Gas Phase Combustion Only
95
The basic flow structure for the fine grid and baseline grid compared well, as shown
in Figure 4.5.3-4. There was little change in internal flow patterns, with the exception of
very near the nozzle exit, where slightly higher Mach numbers are seen. If greater
resolution is desired in this nozzle for other applications it would prove wise to transition
to a finer mesh in the nozzle, while maintaining the baseline grid sizing to speed
computations. For the present research the baseline grid was judged to be sufficiently
accurate, and offered reasonable computation times for convergence.
Mach Number Mach Number
Baseline Mesh Fine Mesh
Figure 4.5.3-4 Fine Grid Convergence History – Gas Phase Combustion Only
96
CHAPTER 5 - Conclusions and Recommendations
The present research was undertaken to numerically model a very complex
combustion flow problem, including multi-phase fuel injection with momentum and
temperature through a ported injector into a highly turbulent combustor flowfield. This
type of flow is found in practice in pulverized coal combustors and ducted rocket ramjets,
for example. The approach taken was to develop an understanding of a baseline ducted
rocket combustion case, primarily focused on characterization of the fuel properties,
followed by the development of a numerical model and comparison to combustion test
data for the baseline case.
The ducted rocket engine used as a basis for computational model development was
the MARC-R282, developed by Aerojet for the SSST missile target. This is a four-inlet
side-dump combustion chamber with no flameholding devices. The solid fuel is partially
combusted to produce a fuel rich mixture containing combustible gases and carbon soot
that are injected into the main airstream in the combustor.
The first step in the investigation was to develop a better understanding of the gas
generator exhaust properties. This process included conducting experiments to
investigate the propellant burning surface (window bomb and particle collection tests),
the gas generator exhaust prior to mixing with air (static motor firings with exhaust
sampling), and the flow exiting the combustor exit nozzle (direct-connect ramjet tests
with exhaust sampling). Important conclusions drawn from these experiments were: 1)
the polystyrene used as a fuel ingredient was shown to begin reacting very close to the
propellant surface and with reactions completed prior to injection into the combustion
chamber, 2) the boron fuel ingredient was shown to be incompletely reacted exiting the
gas generator early in the firing, and 3) the particles collected from the gas generator
exhaust were not spherical, but highly irregular with a wide range of sizes, and were
found to have a mean effective particle diameter of 20 microns.
In order to formulate a set of test data to validate a computation model, a series of
ground test ramjet test data was collected and analyzed for different fuel injector
geometries in the MARC-R282 engine. Two distinct injector geometries formed the
basis for the comparison; one with a large axial component to the injector flow, and one
97
with nearly all injectant directed radially. The combustion efficiency for the version with
axial fuel injection was shown to be roughly 6% higher than the radial injector. This
difference in performance caused by injector geometry was used to test the predictive
capability of the model developed in this project.
A computational model was developed using the FLUENT 6.2 commercial CFD
code. First, a simplified “fuel” was developed for the gas-phase components based on
equilibrium thermochemistry calculations. This fuel was injected in a CFD model of the
MARC-R282 combustor at air flow conditions matching the ground test data. FLUENT
predicted temperature-rise combustion efficiencies approaching 100%. The gas-phase
model served to validate the thermochemical inputs to the CFD computations, and
provided confidence that the combustor flowfield was conducive to efficient combustion
if the kinetic limitations inherent in the fuel are not exceeded.
A second simplified “fuel” was then developed for FLUENT modeling adding
particulate carbon at a mass fraction predicted by equilibrium thermochemistry for the
gas generator exhaust. The built-in routines developed for pulverized coal particles in
FLUENT were used, with a mean particle diameter of 20 microns, based on the data
collected from the ground test experiments. The solid boron particles contained in the
exhaust were neglected for this study due to their low concentration. The model
predictions for this particulate-laden fuel for the two validation test cases showed a trend
opposite to that seen in ground testing, i.e., the axial injector geometry resulted in lower
combustion efficiency than the radial injector. This was shown to be a result of poor
dispersion of the solids and incomplete particle combustion.
It was then theorized that potentially some fraction of the particulates were in fact
much smaller than those collected, and the actual mass fraction of solids in the exhaust
was smaller than equilibrium thermochemistry would predict. To test this hypothesis a
series of FLUENT cases was run in which a varying portion of the particulate exhaust
was replaced with gaseous fuel. The cases showed that as the solid fuel fraction was
reduced, the difference in model predictions narrowed and actually reversed to be
consistent with the ground test data. This indicated that the soot particles collected in
experiments may actually be a small mass fraction of the total fuel exhaust, and may in
fact be agglomerates of much finer soot particles introduced with the fuel. The absolute
98
combustion efficiencies predicted by the revised FLUENT model were also very close to
matching levels seen in ground testing.
These results served to verify the accuracy and predictive capability of the
computational model. The robustness of the computational model was also evaluated
parametrically by examining the effects of turbulence model, order of discretization, and
grid size. Comparable trends and results were seen for all cases examined.
This study clearly demonstrated the usefulness of planning an engineering project
combining CFD with complementary experimental data. By simply relying on the
laboratory data collected for particle sizes with a single ground test configuration, one
could have used the many variables at hand to force the particle combustion to fit the test
data and been misled in the process. By considering multiple sets of ground test data,
guided by CFD interpretations, a much more complete understanding of the process was
reached.
This combustor analysis tool can now be employed for similar flow problems
involving particulate-laden fuel injected with high velocity and temperature. For the
specific problem investigated, the model was able to help understand particle behavior in
a combustion environment, with results correlated to ground test observations. The
results show that for the MARC R282 engine it should be possible to use the model to
identify hardware changes to manage the fuel and air mixing to improve combustion
efficiency, and thereby reduce the amount of visible soot in the exhaust. Indeed, for
initial combustor design efforts to improve the MARC R282 combustor it may be
possible to neglect the particulate fuel phase altogether until more rigorous modeling is
needed. Nonetheless, the framework exists in this modeling approach to investigate
future flow problems in which larger particles are present and a more substantial
influence on the combustion process.
Additionally, the analytical studies and data collected in this project indicated that
the polystyrene is likely the cause of the majority of the soot produced. In addition to
flow management to improve combustion efficiency, this also would suggest chemical
formulation approaches as a possible avenue to reducing soot in the exhaust. Options
could include the use of alternate fuels or the use combustion catalysts in the fuel
formulation.
99
This combustion problem is complex, and there are many assumptions made that
further research may be directed toward. Recommended areas to focus future attention
on include: 1) actual gas temperature measurements, 2) boron combustion modeling, 3)
tests to determine boron combustion levels in the ramjet combustor, and 4) fuel exhaust
gas sampling to determine gaseous species and molecular weight.
100
References
1 Baum, M.M., and Street, P.J., "Predicting the Combustion Behavior of Coal
Particles", Combustion Science Technology, Vol 3 pp. 231-243, 1971.
2 Corporan, E., Oppelt, J.B., and Buckley, P.L., "A Ducted Rocket Fuel Injector
Evaluation with a Gas Generator Simulator", 1992 JANNAF Propulsion Meeting.
3 Wilson, K.J., Shadow, K.C., Gutmark, E., and Smith, R.A., "Passive Combustion In a
Ducted Rocket with Side Mounted Inlets" 1990 JANNAF Propulsion Meeting, Anaheim,
CA 3-5 Oct 1990, pp383-396.
4 Magnussen, B.F., and Hjertager, B.H., On Mathematical Modeling of Turbulent
Combustion with Special Emphasis on Soot Formation and Combustion", 16th
Symposium (International) on Combustion, pp. 719-729, The Combustion Institute,
Pittsburgh, 1976.
5 Field, M.A., "Rate of Combustion of Size-Graded Fractions of Char from a Low-
Rank Coal between 1200K-200K", Combustion and Flame, 13:237-252, 1969.
6 Peters, A.F. and Weber, R. (1997), Mathematical Modeling of a 2.4 MW Swirling,
Pulverized Coal Flame, Combustion Science and Technology, 122, 131.
7 Edelman, R.B., and Harsha, P.T., "Modeling Techniques for the Modeling of Ramjet
Combustion Processes" AIAA Journal, Vol. 19, No. 5, May 1981.
8 Lilley, D.G., "Computer Modeling in Ramjet Combustors", AIAA Journal, Vol. 19,
No. 12, December 1981.
9 King, M., "Modeling of Effects of Various Aids (Halogens, Metal Coatings, LiF
Coatings) on Boron Particle Ignition", JANNAF Paper?
10 Harsha, P.T., and Edelman, R.B., "Computational Hierarchy for the Analysis of
Combustion Phenomena in Ramjet Environments", 1984 JANNAF Propulsion Meeting -
Specialist Session, Computational Methods, 9 Feb, 1984, New Orleans, LA.
11 Sambamurthi, J.K., "Al2O3 Collection and Sizing from Solid Rocket Motor Plumes",
Journal of Propulsion and Power, VOL. 12, NO. 3, May-Jun 1996.
12 Mellor, A.M., "Particulate Matter in the Exhaust of a Boron-Loaded Solid
Propellant", AIAA Journal, Vol 9, No. 10, October 1971.
101
13 Stowe, R.A., et al, "Two Phase Flow Combustion Modelling of a Ducted Rocket"
AIAA 2001-3461, 37th AIAA Joint Propulsion Conference, 8 July 2001, Salt Lake City
Utah.
14 Buckley, P., et al. "The Use of Swirl for Efficient L/D Combustors", USAF
Document 1982.
15 Kim, C.H., El-Leathy, C.M., and Faeth, G.M., "Laminar Soot Processes (LSP)
Experiment: Findings from Ground Based Measurements, NASA/CP-2003-
212376/REV1
16 Cangialosi, F. et al. "Combustion Reactivity of Unbirned Carbon in Coal Combustion
Fly Ashes" Technical University of Bari, Italy.
17 Lewis, D.J, "Numerical Prediction of Ducted Rocket Ramjet Combustor
Performance", JANNAF paper 2003-0010G.
18 Turns, S.R., and Faeth, G.M., "A One-Dimensional Model of Carbon-Black Slurry-
Fueled Combustor", Journal of Propulsion, Vol 1, No.1, Jan-Feb, 1985.
19 Rodionov, A.V., et al. "Soot Oxidation Modeling in Plumes", AIAA 2001-3858, 37th
AIAA Joint Propulsion Conference, 8 July 2001, Salt Lake City Utah.
20 Abbott, S.W., Smoot, L.D., and Schadow, K., "Direct Mixing and Combustion
Efficiency Measurements in Ducted, Particle-Laden Jets", AIAA Journal, Vol 12, No. 3,
March 1974.
21 Urban, D.L., and Faeth, G.M., "Soot Research in Combustion Science: Introduction
and Review of Current Work", AIAA 2001-0322, 39th AIAA Aerospace Sciences
Meeting, January 2001, Reno, NV.
22 Waesche, R.H.W.W., Workshop Report - "Ignition and Combustion in Ducted
Rocket Combustors" JANNAF Workshop held June 13, 1984, Cincinnati, OH.
23 Carlos G. Rodriguez, C.G.,"ONE-DIMENSIONAL, FINITE-RATE MODEL FOR
GAS-TURBINE COMBUSTORS", Dissertation submitted to the Faculty of the Virginia
Polytechnic Institute and State University in partial fulfillment of the requirements for the
degree of DOCTOR OF PHILOSOPHY in Mechanical Engineering, July 1997.
24 Thomas, M.E. and Leonard, A.D., "AIR-TURBO-ROCKET COMBUSTION",
AIAA-95-0813, 33rd Aerospace Sciences Meeting and Exhibit, January 9-12, 1995,
Reno, NV.
102
25 Ristori, A. , and Dufour, E., "Numerical Simulation of Ducted Rocket Motor", AlAA
2001 -3193, 37th AIAA Joint Propulsion Conference, 8 July 2001, Salt Lake City Utah.
26 "Blevins, L.G., ""PARTICULATE MATTER EMITTED FROM AIRCRAFT
ENGINES"", AIAA 2003-2764, AIAA/ICAS International Air and Space Symposium
and Exposition: The Next 100 Years
Dayton, Ohio, 14 - 17 Jul 2003."
27 "Ogerby, T. ""Literature Review of Turbine Combustor
Modeling and Emissions"", AIAA JOURNAL, VOL. 12, NO. 6, JUNE 1974 "
28 Komar, J.J, and King, M.K., "Diagnostics of Single Particle Boron Combustion",
AIAA 80-0070, AlAA 21st Aerospace Sciences Meeting, January 10-13, 1983, Reno,
Nevada
29 Vigot, C., Bardelle, L., and Nadaud, L. “Improvement of Boron Combustion in a
Solid-Fuel Ramrocket”, AIAA-86-1590, AIAA 22nd Joint Propulsion Conference, June
16-18, 1986, Huntsville, Alabama.
30 Rood, T.J., et al, "Ignition dynamics of boron particles in a shock tube", AIAA-1997-
3234, AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, 33rd, Seattle,
WA, July 6-9, 1997.
31 Odawara, T., et al, “Ignition Characteristics of Boron Particles in Ducted Rockets”,
AIAA 2005-4498, 41st AIAA/ASME/SAE/ASEE Joint Propulsion Conference &
Exhibit, 10 - 13 July 2005, Tucson, Arizona
32 Gany, A., “Comprehensive Consideration of Boron Combustion in Airbreathing
Propulsion”, AIAA 2006-4567, 42nd AIAA/ASME/SAE/ASEE Joint Propulsion
Conference & Exhibit, 9 - 12 July 2006, Sacramento, California
33 Schadow, K., "Experimental Investigation of Boron Combustion in Air-Augmented
Rockets", AIAA JOURNAL VOL. 7, NO. 10
34 Schadow, K., Study of Gas-Phase Reactions in Particle Laden, Ducted Flows", AIAA
JOURNAL VOL. 11, NO. 7
35 Natan, B., and Gany, A., "Ignition and Combustion of Boron Particles in the
Flowfield of a Solid Fuel Ramjet", Journal of Propulsion and Power, VOL. 7, NO. 1,
JAN.-FEB. 1991, p. 37.
103
36 Natan, B., and Gany, A., "Combustion Characteristics of a Boron-Fueled Solid Fuel
Ramjet with Aft-Burner", JOURNAL OF PROPULSION AND POWER, Vol. 9, No. 5,
Sept.-Oct. 1993
37 Mitani, T and Izumikawat, M., "Combustion Efficiencies of Aluminum and Boron in
Solid Propellants", Journal of Spacecraft, VOL. 28, NO. 1, JAN.-FEB. 1991, p. 79
38 Schadow, K., "Fuel-Rich Particle-Laden Plume Combustion", AIAA Journal, VOL.
13, NO. 12, DECEMBER 1975, p.1553
39 “Water Flow Tests of a Ducted Rocket Combustor”, Virginia Research Associates
Final Report to Atlantic Research Corporation, December 1985.
40 King, M.K., “An Explanation for Early Progressivity in DR-PTV Gas Generator
Firings”, Atlantic Research Corporation Memorandum, April 1981.
41 Dempsey, J.V., et. al., "Ignition and Combustion of Carbon Particles", AIAA Paper
AIAA-95-2993, 31st AIAA Joint Propulsion Conference, San Diego CA, July 10-12
1995.
42 FLUENT 6.2 User’s Guide, January 2005
43 McAdams, W.H.: “Heat Transmission,” 3rd ed. McGraw-Hill Book Co, New York
1954.
44 National Institute of Standards and Technology web site:
http://webbook.nist.gov/cgi/cbook.cgi?ID=C630080&Units=SI&Mask=1#Thermo-Gas
104

Numerical Modelling and Experimental Validation - Ducted Rocket Motor

Uploaded by

Copyright:

Available Formats

Numerical Modelling and Experimental Validation - Ducted Rocket Motor

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Numerical Modelling and Experimental Validation - Ducted Rocket Motor

Uploaded by

Copyright:

Available Formats

Numerical Modeling of a Ducted Rocket Combustor

With Experimental Validation

Dissertation submitted to the faculty of the Virginia

September 04, 2008

Keywords: Ducted Rocket, Ramjet, Combustion, Boron, Soot, Gas Generator

CHAPTER 1 - INTRODUCTION .............................................................................................................. 1

1.1 BASELINE DUCTED ROCKET COMBUSTOR FLOWFIELD ......................................................................... 1

CHAPTER 2 – FUEL EXHAUST CHARACTERIZATION ................................................................. 19

2.1 THERMOCHEMISTRY ........................................................................................................................... 19

CHAPTER 3 – RAMJET VERIFICATION TESTS............................................................................... 44

3.1 SSST PROGRAM DIRECT CONNECT TESTS.......................................................................................... 44

CHAPTER 4 – COMPUTATIONAL FLUID DYNAMICS MODELING............................................ 68

4.1 GRID DEVELOPMENT .......................................................................................................................... 68

CHAPTER 5 - CONCLUSIONS AND RECOMMENDATIONS.......................................................... 97

REFERENCES ......................................................................................................................................... 101

FIGURE 1.1-1 SSST MISSILE CONFIGURATION............................................................................... 3

FIGURE 1.1-2 MARC-R282 RAMJET ENGINE .................................................................................... 4

FIGURE 1.1-3 MARC-R282 ENGINE HEATING .................................................................................. 5

FIGURE 1.1-4 SSST COMBUSTOR LAYOUT....................................................................................... 5

FIGURE 1.1-5 COMBUSTOR FLOW FEATURES................................................................................. 6

FIGURE 1.3.3-1 COAL PARTICLE REACTION RATE CURVE FIT .............................................. 18

FIGURE 2.1.1-1 POLYSTYRENE BEADS (136X)................................................................................. 21

FIGURE 2.1.1-2 FUEL PRODUCTS WITH NON-REACTING POLYSTYRENE ............................ 22

FIGURE 2.1.1-3 CHAMBER TEMPERATURE AND C* WITH NON-REACTING

FIGURE 2.1.1-4 CHAMBER TEMPERATURE AND C* WITH NON-REACTING

FIGURE 2.1.1-5 POLYSTYRENE DECOMPOSITION PROCESS .................................................... 24

FIGURE 2.1.1-6 THERMO-GRAVIMETRIC ANALYSIS RESULTS FOR POLYSTYRENE........ 26

FIGURE 2.1.2-1 CALCULATED CHARACTERISTIC VELOCITY, TEST DC-1............................ 29

FIGURE 2.1.3-1 TYPICAL ENGINE OPERATING TEMPERATURE.............................................. 31

FIGURE 2.2.1-1 WINDOW BOMB APPARATUS................................................................................. 33

FIGURE 2.2.1-2 WINDOW BOMB IMAGE OF FLAME FRONT ...................................................... 33

FIGURE 2.2.1-3 PARTICLE COLLECTION BOMB............................................................................ 34

FIGURE 2.2.1-4 PARTICLE COLLECTION BOMB RESULTS......................................................... 35

FIGURE 2.2.2-1 GAS GENERATOR EXHAUST SAMPLING TEST SETUP ................................... 36

FIGURE 2.2.2-2 GAS GENERATOR EXHAUST PARTICLE PHOTOMICROGRAPHS............... 37

FIGURE 2.2.3-1 RAMJET EXHAUST SAMPLING TEST SCHEMATIC ......................................... 41

FIGURE 2.2.3-3 RAMJET EXHAUST SAMPLES ................................................................................ 42

FIGURE 3.1.1-1 AEROJET RAMJET TEST FACILITY ..................................................................... 45

FIGURE 3.1.1-2 SSST HEAVYWALL DIRECT-CONNECT INSTALLATION ............................... 45

FIGURE 3.1.1-3 SSST HEAVYWALL TEST HARDWARE ................................................................ 46

FIGURE 3.1.1-4 SSST HEAVYWALL FUEL INJECTORS ................................................................ 47

FIGURE 3.1.2-1 DIRECT-CONNECT VITIATED AIR MASS FLOWRATES ................................ 48

FIGURE 3.1.2-2 DIRECT-CONNECT AIR TEMPERATURE ........................................................... 49

FIGURE 3.1.2-3 DIRECT-CONNECT RAMBURNER STATIC PRESSURES................................. 50

FIGURE 3.1.2-4 DIRECT-CONNECT GAS GENERATOR CHAMBER PRESSURES .................. 51

FIGURE 3.1.2-5 DIRECT-CONNECT FUEL FLOWRATES.............................................................. 52

FIGURE 3.1.2-6 DIRECT-CONNECT GROSS THRUST.................................................................... 52

FIGURE 3.2-1 RAMJET ENGINE STATION NOMENCLATURE .................................................... 53

FIGURE 3.3.1-1 DC-1 TEMPERATURE RISE COMBUSTION EFFICIENCY................................ 60

FIGURE 3.3.1-2 DC-2 TEMPERATURE RISE COMBUSTION EFFICIENCY................................ 61

FIGURE 3.3.1-3 DC-3 TEMPERATURE RISE COMBUSTION EFFICIENCY................................ 61

FIGURE 3.3.1-4 PRESSURE-BASED TEMPERATURE RISE COMBUSTION EFFICIENCY..... 62

FIGURE 3.3.1-5 FUEL/AIR COMBUSTION EFFICIENCY DEFINITION...................................... 63

FIGURE 3.3.1-6 FUEL/AIR COMBUSTION EFFICIENCY – BASED ON PRESSURE ................. 64

FIGURE 3.3.1-7 RAMBURNER PRESSURE LOSSES ........................................................................ 65

FIGURE 3.3.2-1 EFFECT OF BORON COMBUSTION ON RAMJET FLAME TEMPERATURE 66

FIGURE 3.3.2-2 TEMPERATURE DECREASE BY NEGLECTING BORON.................................. 67

FIGURE 4.1-2 BASELINE .05 IN. NOMINAL GRID – DUMP REGION.......................................... 70

FIGURE 4.1-4- BASELINE .05 IN. NOMINAL GRID – NOZZLE REGION..................................... 71