Accepted Manuscript

Experimental and Numerical Investigation of Combustion in a Hydrocarbon and

Gaseous Oxygen fuelled Rocket

Tausif Shaikh, Lalit Patidar, Arindrajit Chowdhury

PII: S1359-4311(16)31582-4
DOI: http://dx.doi.org/10.1016/j.applthermaleng.2016.08.222
Reference: ATE 9021

To appear in: Applied Thermal Engineering

Received Date: 11 June 2016

Revised Date: 29 August 2016
Accepted Date: 31 August 2016

Please cite this article as: T. Shaikh, L. Patidar, A. Chowdhury, Experimental and Numerical Investigation of
Combustion in a Hydrocarbon and Gaseous Oxygen fuelled Rocket, Applied Thermal Engineering (2016), doi:
http://dx.doi.org/10.1016/j.applthermaleng.2016.08.222

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
 Experimental and Numerical Investigation of Combustion in a Hydrocarbon and

Gaseous Oxygen fuelled Rocket

Tausif Shaikh1 , Lalit Patidar 1 , Arindrajit Chowdhury1

1
Department of Mechanical Engineering, Indian Institute of Technology, Bombay

Powai, Mumbai, Maharashtra, 400076, India

Abstract

Turbulent spray combustion in a laboratory scale liquid rocket engine fuelled by a liquid

hydrocarbon, gasoline, and gaseous oxygen was studied under both fuel-rich and fuel-lean

conditions. Various pertinent parameters, such as pressures at pertinent locations, fuel and

oxygen mass flow rates, and thrust were measured during the experimental runs. The steady

state values of these parameters were further utilised to calculate the specific impulses and

the c* efficiencies for the test cases. The experimental values of chamber pressure and the

generated thrust were verified by modelling the combustion process in Ansys Fluent by using

n-heptane as the surrogate fuel. The model was utilised to evaluate the capability of four

turbulence models – standard k-ε (SKE), realizable k-ε (RKE), renormalisation group (RNG)

k-ε, and Reynold’s stress model (RSM), combined with either a probability density function

model using equilibrium chemistry or a steady diffusion flamelet model with detailed

chemical kinetic mechanism. Although the RSM model combined with the flamelet model

was found to be the most realistic in prediction of the essential features of turbulent spray

combustion in a liquid rocket engine, the SKE model combined with the flamelet model was

found to predict the experimental pressures most accurately.

Keywords: Spray, Combustion, rocket, model, specific impulse, n-heptane

1
1. Introduction

Spray combustion is the preferred mode of utilization of liquid fuels in power generation

systems and propulsion systems. Liquid fuels are preferred over gaseous fuels in rocket

propulsion systems as they are easier to handle in terms of transportation and storage. Boost

vehicles are primarily comprised of cryogenic or semi-cryogenic liquid rocket engines

(LREs). In the past few decades, significant efforts have been expended to achieve better

efficiency, faster response, and multiple restarts in LREs by implementing innovative

technologies, such as introduction of nanoparticle enriched fuels, increasing turbulence

within the combustion chamber, improving atomization by better injector designs, using high

energy density (HED) fuels etc. [1-5]

Although RP1 & RP2 [6] have been almost exclusively used in semi-cryogenic engines, with

the exception of Syntin during the Soviet Russian era [7], the propellant community strives to

synthesize novel HED compounds with improved specific impulses with liquid oxygen

(LOX) as the oxidizer. Cage hydrocarbons have been envisioned as fuels for semi-cryogenic

LREs in the past and present century with various advantages, such as high density, high

specific impulse, ease of handling in terms of transportation and storage, low vulnerability to

detonation, etc. There is considerable interest within the propellant synthesis community to

design new propellants based on cage hydrocarbons. The objectives of such endeavors are to

primarily synthesize liquid propellants with high energy densities to be utilized directly or as

additives to liquid fuels.

In previous studies conducted by the authors’ group, a range of potential high energy density

(HED) strained polycyclic liquids, such as bis(nitratomethyl)-1,3-bishomocubane

(DNMBHC)[8], nitromethyl-l,3-bishomocubane (NMBHC), nitromethylene-1,3-

bishomocubane (NMyBHC), dinitro-dimethyl-1,3-bishomocubane(DNTMBHC) [9], diazido-

dimethyl-bishomocubane (DADMBHC), ditetrazolo-bishomocubane (DTetzBHC), and

2
diphenyltriazolo-dimethyl-bishomocubane (DPTrizDMBHC) [10] have been synthesized.

Although all of these compounds were found to provide specific impulses that were

comparable to RP1, the density specific impulse of all of these compounds were calculated to

be significantly higher than that of RP1, due to their higher densities. Among these, the

calculated density specific impulse of DADMBHC was found to be approximately 100 s

higher than that of RP1. However, the predicted performance of this compound needs to be

verified by combusting it under realistic conditions in an LRE, and comparing the measured

performance with RP1 or a suitable surrogate of RP1. As a first step towards comprehending

the combustion characteristics of cage hydrocarbon propellants, the present study is focused

on the experimental investigation of spray combustion of gasoline and numerical

investigation of a reference fuel n-heptane to provide an insight into liquid fuel spray

combustion in LREs.

Various laboratory-scale LREs with suitable diagnostic tools have been fabricated and

commissioned to comprehend the nature of turbulent combustion both inside the combustion

chamber and in the expansion plume. Santos et al. [11] developed a LRE with ethanol and

GOX as propellants with variable thrust capacity of 50-100 kg. Tests conducted with a

mixture ratio of 1.6, at four values of chamber pressure (8, 10, 12, and 15 bar), produced

thrusts of 56, 70, 84, and 106 kg respectively. Flame stabilization near the injector face in a

subscale optically accessible rocket, 50 mm in diameter and 430 mm long, was studied by

Lux and Haidn [12]. This 100 bar capacity rocket engine, with a throat diameter of 17.3 mm,

was fuelled by methane and liquid oxygen (LOX). Flame emission spectra at different axial

locations were recorded through an optical window cooled by hydrogen film and the

concentrations of OH and CH radicals were obtained. Navarro et al. [13] had developed a

multipurpose optically accessible rocket chamber designed for producing 11.3 kg of thrust at

20 bar operating pressure using LOX and CH4 as propellants. Caisso et al. [14] provided a

3
brief review of the historical development, current observed trends, and possible areas of

improvements in liquid propulsion systems with a roadmap of future developments.

Owing to the inherent complexities in design and manufacturing components of LREs and

testing all possible iterations, significant efforts have also been expended in simulating the

processes in the combustion chamber and the nozzle-diffuser assembly. An extensive review

of spray combustion modelling and its current status is given by Faeth [15], Sirignano [16],

and Law [17]. Turbulent spray combustion is a combination of complex physico-chemical

phenomena consisting of spray characteristics, heat and mass transfer between liquid and gas

phase through heating and vaporization, turbulence in gas phase, and its effect on chemical

reactions and droplet dispersion [17]. Individual models accounting for these processes and

their coupling dictate the accuracy with which the behaviour of the system can be predicted.

Among various models that have been developed, some are well-established and are

extensively used. For example the Reynolds Averaged Navier Stokes (RANS) based models

are commonly used to solve fluid flow equations due to their low requirement of

computational resources [18].

For turbulence modelling, different forms of the k–ε models, such as standard k–ε (SKE),

renormalization group (RNG) k–ε and realizable k–ε (RKE) models are commonly employed

along with RANS based models [19-21]. Even though the accuracy of k–ε models is lower

compared to the Reynolds stress model (RSM), they are predominantly used in most of the

turbulent flow simulations because of their simplicity and their ability to accurately predict

the flow in a number of configurations.

Transport equations for one or two conserved scalars i.e. mixture fractions are typically

solved in non-premixed combustion modelling. A Probability Density Function (PDF) is used

to model the turbulence-chemistry interaction. Look-up tables are used during the flow-field

calculation, created based on pre-processed thermo-chemistry calculations. Two well-known

4
combustion models typically utilize the above mentioned approach– Equilibrium PDF model

and Steady Laminar Flamelet model [22], the former focussing on equilibrium chemistry, and

the latter focussing on non-equilibrium chemical kinetics.

The interactions between the continuous and dispersed phases are often captured using the

Discrete Phase Model (DPM) in the Eulerian–Lagrangian formulation [23] during modelling

spray combustion. Modelling of turbulent combustion of liquid oxygen-gaseous hydrogen

provided flame shapes and stabilization points in good agreement with experimental

observations in a ONERA Mascotte A10 test bench [24]. A simulation with quasi global

kinetic mechanism for a surrogate fuel predicted combustion efficiencies with reasonable

accuracy in an uni-element tri-propellant combustor of a kerosene-fuelled thruster plume

[25]. Prediction of ignition delays using a multi-step quasi global mechanism of kerosene

showed a reasonable match with experimental data by tuning the pre-exponential factor and

activation energy [26].

In the current study, a laboratory scale experimental LRE was constructed to analyse the

combustion characteristics of gasoline and GOX under fuel lean and fuel rich conditions.

Additionally, the experimentally derived pressure and thrust for various conditions were

verified by using numerical techniques, with n-heptane as the surrogate fuel. As mentioned

earlier, such a study was aimed at establishing the capabilities for analysing blends of

hydrocarbons with HED fuels.

2. Experimental studies

2.1. Experimental setup

The experimental setup consists of a laboratory-scale LRE of 10 kg thrust capacity. Various

subsystems, such as a thrust stand, a feed system for injection of fuel and oxidizer, a spark

ignition system, and a cooling system were devised and attached to the LRE. The design

pressure of the LRE was 20 bar. The physical dimensions of the LRE were as follows, a

5
diameter of 30 mm, a length of 57 mm, and a throat diameter of 6 mm. The converging

section of the nozzle-diffuser assembly was provided with an angle of 60o while the diverging

section was provided with an angle of 15o.

5.

4.

3. 6.
2.

7.
8.

1. 9.
10.

Figure 1. Sectional view of the rocket engine (1. fuel inlet, 2. oxygen inlet, 3. holes for
oxygen injection, 4. ’O’ring groove, 5. cooling water inlet, 6. pressure tap pipe, 7. cooling
water out, 8. combustion chamber, 9. cooling jacket, 10. injector flange)

Figure 1 shows a sectional view of the LRE to describe various pertinent components. A

commercial solid cone spray nozzle with a spray angle of 60 degrees (B1/8GG-1 from

Spraytech) was used for injection of fuel. Oxygen was injected through four 1.8 mm equally

spaced holes drilled on the injector plate, circumscribed on a circle of 22 mm diameter. In

order to enhance cooling, the combustion chamber was constructed using copper, surrounded

by an SS316 cooling jacket. Additionally, water was circulated through a 1 mm gap between

the cooling jacket and the chamber to provide cooling from a storage container using a

standard pump.

6
 SV PRV PT(Ptank)
NRV Purge

PT(PFi) MCV
Fuel
SV MFM FF Tank N2
NRV

Rocket
Engine Fuel
PT(POi)
SV MFM NV BV
NRV
Water in
Oxygen
Water out
O2

PT(PC)

Figure 2. Schematic of the experimental setup (NRV: non return valve, NV: needle valve,
PT: pressure transducer, SV: solenoid valve, MFM: mass flow meter, MCV: motorized
control valve, PRV: pressure relief valve, FF: fuel filter, BV: ball valve)

Pressure transducers (Omega PX309) were installed to measure pressures in fuel tank, fuel

line, oxygen line and the converging section of the chamber. A thermal mass flow meter

(Bronkhorst F-113AC) and a turbine flow meter (Omega FLR1011BR) were installed in the

oxygen and the fuel line respectively. Pressures at various locations and the fuel flow rates

were recorded using a National Instruments USB6210 data acquisition system. Thrust was

obtained using an S type load cell (Omega LCR150) and recorded by an Agilent 34970A data

acquisition system. A schematic of the experimental setup is shown in Fig. 2.

All pressure transducers were calibrated using a dead weight tester. Various precautions

were undertaken to ensure the safety of the operators as well as that of the expensive

measurement devices. The engine was enclosed by an SS304 confining chamber with a wall

thickness of 15 mm to prevent projectiles from affecting the measurement devices in case of

accidental detonation of the chamber. The experiments were performed in a separate room

and the combustion process was controlled remotely. To automatically shut the engine down

in case of power failure, normally closed solenoid valves were used in the feed systems.

7
Initially a few experiments were carried out at lower pressures to ensure proper functioning

and operation of the LRE. An inert gas, nitrogen was used for pressurizing the fuel as well as

for purging the chamber. Initial pressures were set to 25 bar in the fuel tank and in the oxygen

line. A spark ignitor was pushed through the nozzle into the combustion chamber, and a low

flow rate of oxygen was introduced in the chamber. Subsequently, a spark was provided and

the fuel supply was turned on by opening the solenoid valve on the fuel line. As soon as a

visible flame was observed at the exit of the chamber, the oxygen flow rate was increased

rapidly until the desired chamber pressure was achieved. After the parameters reached a

steady state and data acquisition was completed, the solenoid valves on the fuel and oxygen

supply lines were turned off and chamber was purged with nitrogen to terminate the test. In

case of ignition failure, fuel supply was turned off immediately, followed by purging of the

chamber to prevent a subsequent hard start. The experiments were conducted with

commercially available gasoline in the absence of RP-1. Specific impulse was calculated as

total thrust generated divided by total propellant mass flow rate. Theoretical characteristic

velocity was calculated using the NASA CEA code for each test run using the actual

operating conditions. The actual characteristic velocity was calculated as a product of

chamber pressure and throat area divided by total propellant mass flow rate.

2.2. Results and discussion

After ensuring safe operation of the engine, the diagnostic instruments were installed and

pressures as well as flow rates were recorded at a sampling rate of 1 kHz. The outputs from

the pressure transducer and the liquid flow meter were treated by a Butterworth’s low pass

filter in MATLAB [27] to reduce random noise. The load cell output was recorded at 23 Hz.

Typical results from the successful experiments are discussed henceforth.

8
 15 15
PFi
POi
Pc
wF
10 wO 10

Flow rate (g/s)

Pressure (bar)

5 5

0 0
0 10 20 30 40 50
Time (s)
Figure 3. Variation of pressure in the fuel line, oxygen line and converging section of
chamber, and fuel and oxidizer flow rates at ϕ = 1

Figure 3 shows the variation of pressure at different locations in the setup for the experiment

carried out at an equivalence ratio (ϕ) of 1. PFi, POi, and Pc denoted the pressures at the fuel

injection line, the oxygen injection line, and the converging section of the combustion

chamber. The combustion event was controlled by controlling P Oi. It was interesting to

observe that Pc and PFi followed the nature of variation of POi. The oxygen mass flow rate

was also found to follow a similar trend, as expected. However, as POi, Pc, and PFi reached the

steady state, the difference between PFi and Pc was found to be reduced, leading to a

reduction in the mass flow rate of the fuel. It was also noticeable that the value of POi was

approximately double that of P Fi. All values, except the fuel tank pressure, were found to be

reduced drastically as purging was initiated.

9
 30
PFi
POi 15
25 Pc
wF
20 wO

Flow rate (g/s)

Pressure (bar)
10
15

10
5

0 0
0 10 20 30 40 50
Time (s)
Figure 4. Variation of pressure in the fuel line, oxygen line and converging section of
chamber at, and fuel and oxidizer flow rates ϕ = 0.8

Experimental results with an equivalence ratio of 0.8 are shown in Fig. 4. The variation of

pressure at all locations was found to be similar to the trends observed in the previous

experiment. The oxygen mass flow rate was varied to change the equivalence ratio while

maintaining the fuel mass flow rate constant.

3 3
 = 0.8


2 2
Thrust (kg)

1 1

0 0
0 10 20 30 40 50
Time (s)
Figure 5. Variation of thrust with time for ϕ = 1 and ϕ = 0.8

10
As seen in Fig. 5, the thrust produced with ϕ = 0.8 was higher than that with ϕ = 1. Steady

state conditions were assumed to have been reached when the values of pressure at all

locations were reasonably invariant with time. The values of pressures and flow rates were

averaged during the steady state operation to calculate the equivalence ratio, thrust, and

specific impulse. Table 1 shows the steady state values of various parameters.

Table 1: Steady state values of various parameters for ϕ = 1.14, 1, 0.9, and 0.8

Parameter ϕ = 1.14 ϕ=1 ϕ = 0.9 ϕ = 0.8

Fuel injection pressure (bar) 5.25 5.58 5.76 7.86

Oxygen injection pressure (bar) 9.63 10.5 11.20 15
Pressure at converging section (bar) 4.95 5.28 5.51 7.55
Fuel flow rate (g/s) 2.48 2.39 2.28 2.71
Oxygen flow rate (g/s) 7.59 8.27 8.91 11.97
Thrust force (kg) 1.70 1.75 1.76 2.53
Specific impulse (s) 169 164 157 172
c* efficiency (%) 80.03 82.38 83.64 89.04

Figure 6. Mach discs at ϕ = 1.14

11
 100 250

80 200

c* efficiency (%) 60 150

Isp (s)
40 100

20 50
c* efficiency
Isp
0 0
0.7 0.8 0.9 1 1.1 1.2 1.3 1.4

Figure 7. Variation of c* efficiency and Specific impulse with equivalence ratio

In the initial experiments, the mass flow rate of fuel was maintained at a value that was

approximately a quarter of the designed flow rate. Hence, the chamber pressure and the thrust

were also around 25% of the design conditions. The specific impulse at a chamber pressure of

approximately 7 bar was calculated to be 180 s whereas the expected specific impulse at a

chamber pressure of 20 bar was 242 s. Thus, it may be concluded that the chamber operated

with reasonable accuracy and is expected to provide better performance when operated close

to the design conditions. A numbers of tests were performed with the range of equivalence

ratios being varied between 0.8 and 1.3, by varying the oxidizer mass flow rate. Figure 6

shows the Mach discs visible in the exhaust plume at ϕ = 1.14. The experimental

characteristic velocities and specific impulses at different equivalence ratios are shown in the

Fig. 7, which were found to be higher under leaner and stoichiometric conditions, owing to

the higher chamber pressures.

The predicted values of chamber pressure and thrust calculated numerically were compared

against the experimental data. In the absence of the experimental values of temperature,

velocity, and species profiles within the combustion chamber and post-combustion region,

12
the present study is focused on the capabilities of various numerical approaches to predict the

chamber pressure and the thrust. The ultimate objective is to establish a numerical technique

to predict the performance of various blends of hydrocarbon fuels with HED compounds

without conducting full scale experiments with each and every combination.

3. Numerical studies

3.1. Model

A numerical model for spray combustion was developed by combining appropriate sub-

models for the gas and liquid phases within the ambit of a commercial CFD software,

ANSYS FLUENT 14.5 [28]. The finite volume method was used to discretize the transport

equations of the compressible reacting turbulent flow model. A segregated approach with

double precision accuracy was used to solve the steady-state equations. The convective terms

were discretized using a second order upwind scheme while the SIMPLE algorithm was

employed for velocity-pressure coupling. Four turbulence models i.e. Standard k–ε (SKE),

Realizable k–ε (RKE), Renormalized grid (RNG) k–ε and Reynold stress model (RSM) were

chosen to evaluate their ability to predict complex turbulent reacting flows. Scalable wall

function was utilized during the simulations such that the y+ at the wall is displaced to a

value of 11.225 irrespective of the level of refinement of the mesh near the wall, triggering

the application of the log law near the wall. This was expected to provide realistic solutions

without increasing computational time by the utilization of enhanced wall functions. The

values of y+ were found to very between 4 and 11 along the wall for all models. Two non-

premixed combustion models were considered equilibrium probability density function

(PDF) model and steady diffusion laminar flamelet model. In the equilibrium PDF model, ten

chemical species (O2, n-C7H16, CO2, CO, H2O, H2, OH, C(s), H2O(l), and CH4) were included

in the equilibrium product mixture. In the flamelet model, a chemical kinetic mechanism

comprising of 41 species and 124 reactions for n-heptane was used [29].

13
3.2. Flow configuration and grid generation

The numerical simulations were performed using a three dimensional domain as shown in

Fig. 8 and the unstructured tetrahedral mesh superimposed over the domain is shown in Fig.

9. Rotational periodicity of the flow configuration was exploited and only a quarter of the

domain was simulated. A buffer zone further downstream of the exit plane of converging-

diverging nozzle was included in the computational domain to facilitate the numerical

treatment of exhaust gases mixing with ambient air. The axial locations chosen to display the

values of various parameters at z = 30, 50, 78, and 90 mm are shown in Fig. 10. The

numerical studies were conducted with a grid comprising of 77,841 nodes and 349,246

elements, which was found to provide a balance between accuracy and the computational

time required.

Figure 8. Computational domain

Figure 9. Unstructured mesh for the rocket engine

14
 Figure 10. Axial locations chosen for displaying profiles of velocity, temperature, and
species.

3.3. Boundary conditions

 Inlets: The inlet temperatures of the fuel and gaseous oxygen were provided as 300 K. The

mass flow rates given as inputs during the simulations were one-fourth of the values used in

the experimental studies owing to the symmetric quarter geometry utilized.

 Chamber walls: The “no slip” condition was specified at the chamber wall and in the

absence of heat transfer data from experiments, “adiabatic” condition was used. For the DPM

model, the “reflect” boundary condition was set to allow unevaporated droplets to return to

the chamber from the walls.

 Interfaces with the atmosphere: The domain downstream of the exit plane models the

ambient atmosphere where the rocket plume expands to atmospheric pressure. Hence, the

“pressure inlet” boundary condition with static pressure of 1 atm was specified for the

boundaries on the top right in Fig. 8. For the outlet, a “pressure outlet” condition with zero

gauge pressure was provided.

 Bounding planes within the chamber: Two planes which constrict the domain into a quarter

were specified as a pair of planes with periodic boundary condition.

 DPM model: A full cone spray of n-heptane with 30o spray half-angle was specified using

the DPM model. The spray was specified as monodisperse with a droplet diameter of 100

μm, with an initial velocity of 35 m/s. The total mass flow rate of fuel was specified as a

15
quarter of 2.48 g/s and 2.28 g/s for rich and lean conditions respectively. Turbulent dispersion

of the particles was accounted for via stochastic tracking. Droplet breakup and coalescence

were not implemented during these studies.

3.4. Post-processing methodology

The important features of the flow and the flame structure was discussed at four axial

locations i.e. within the combustion chamber (z = 30 mm and z = 50 mm), the midsection of

the converging section (z = 78 mm) and the midsection of the diverging section (z = 90 mm)

as shown in Fig. 10. The chamber pressure was found to remain reasonably constant for all

models utilized in this study. Therefore the mean value of the pressure at z = 50 mm was

taken as the chamber pressure. The area-averaged velocity and pressure at the exit plane was

taken as the exit velocity and exit pressure respectively. The thrust force F was calculated

using total propellant flow rate ( ), exit velocity ( ), exit pressure ( ), and nozzle exit

area ( ) as shown by equation (1).

(1)

where, Pamb is the ambient pressure.

16
3.5. Grid independence study

To ensure that the results are independent of the grid size, a grid independence study was

initially performed for four incremental grid sizes, with 34340, 56455, 77841, 103573 nodes

and 147589, 240205, 349256, 483635 elements, which were termed as Grid_1, Grid_2,

Grid_3 and Grid_4 respectively.

(a) (b)

(c) (d)
Figure 11. Velocity profiles at different axial locations for four grid sizes

Figure 11 shows the comparison of velocity profiles at various axial locations for these grids,

demonstrating that the variation between the profiles diminished as the number of elements

were increased, especially the profiles within the combustion chamber (i.e. z = 30 mm and z

= 50 mm). Further downstream of the combustion chamber i.e. in the converging and

diverging sections, the gases accelerate and the velocity profiles were found to be identical.

On further refinement of the grid from Grid_3 to Grid_4, the improvement in velocity

profiles were found to be marginal at the cost of a significant increase in computational time.

17
Therefore computations were henceforth performed with Grid_3 (77841 nodes and 349256

elements) to reduce computational time without compromising accuracy.

3.6. Comparison of turbulence models

As mentioned earlier, three forms of k–ε models i.e. SKE, RNG, and RKE as well as the

RSM model were used to simulate spray combustion in the rocket chamber and gas dynamics

of the converging diverging nozzle. Since experimental data on velocity, temperature, and

species profiles within liquid rocket chambers is relatively sparse in the literature, static wall

pressure and wall temperature profiles along the axial direction are typically used by

investigators to validate their numerical models [30-32].

Figure 12. Contours of velocity superimposed by streamlines near the injector determined
using four turbulence models on the plane passing through the centre of an oxygen inlet and
the central axis.

The flow of oxygen emanating from a circular jet in the given configuration represents a

turbulent flow past a backward facing step as shown by the propellant injection configuration

in Fig. 10. In such configurations, a recirculation zone is typically observed at the corner

created by the step owing to the high momentum of the jet [33]. As seen in Fig. 12, which

18
shows the streamlines for the four turbulence models superimposed over the velocity

contours, the SKE and RSM models were able to predict the recirculation zone, as opposed to

the RKE and RNG models. The difference arises because of the varying rates of diffusion of

oxygen from the inlet stream to the surrounding environment. This leads to substantial

differences in the prediction of the flow features near the injector face as well as in the

vicinity of the oxygen stream.

The RKE and the RNG models predicted slower diffusion of the oxygen jet, leading to higher

velocities compared to the SKE and RSM models as the jet expanded past the backward

facing step created by the injector. Although a separation point was created on the injector

face, the small recirculation zones could not be distinguished for these two models.

However, the recirculation zones between two contiguous oxygen jets, although not shown,

were predicted by all the models,

The inability of the RNG and the RKE models to predict swirling flows in complex

geometries has been demonstrated by Menzies [34], and the SKE model was found to be best

suited for the purpose. Additionally, Maele et al. [21] showed that the SKE and RKE models

were capable of predicting swirling reacting flows, with the SKE model providing better

prediction of some flow features, and the RKE model predicting other features better. The

RNG model was found to be the worst of the three models in terms of prediction. Other

numerical studies [22, 24] on combustion also relied on the SKE model to predict

characteristics of turbulent reacting flows.

Since it was expected that the lack of the clear formation of the recirculation zone at the

injector face would not significantly affect the chamber pressure distribution and the

predicted thrust, the use of enhanced wall functions was not further pursued.

19
 (a) (b)

(c) (d)
Figure 13. Velocity profiles at different axial locations determined using four turbulence
models

Velocity profiles at different axial locations are shown in Fig. 13. The oxygen jet was found

to diffuse rapidly and lose its initial momentum as predicted by the SKE model, and as

observed by comparing the velocity profiles predicted by the four models at z = 30 mm and z

= 50 mm, which shows that the oxygen jet centreline velocity decreases significantly as

compared to other models and the velocity profile flattens out. At z = 30 mm, the other three

models predicted similar velocity profiles, with the RKE model predicting the highest

velocities throughout the cross section. It was not possible to detect the presence of the

recirculation zone from Fig. 13 (a) explicitly. As we move downstream to z = 50 mm, the

RKE model was found to retain the highest velocity of the oxygen jet as observed in Fig. 12

as well as in Fig. 13(b). Although the velocity of the oxygen stream predicted by the RNG

model decreased at z = 50 mm compared to z = 30 mm, radial diffusion of oxygen was slow,

allowing the stream to extend up to the end of the combustion chamber and even reach the

20
converging section. As seen in Fig. 13(c), the flow sped up considerably in the converging

section, and the predictions for all the models were similar, with the SKE model predicting

slightly higher values of velocity in the vicinity of the centreline, owing to a wider

combustion zone in the converging section, as will be observed later. A similar observation

was made for z = 90 mm in Fig. 13(d), where the flow was supersonic.

(a) (b)

(c) (d)
Figure 14. Temperature profiles at different axial locations determined using four turbulence
models

The contours of vorticity normal to the plane through an injector hole, demonstrating the

velocity contours in Fig. 12, are shown in Fig. 15. The contours depict trends that are similar

to those shown by the velocity contours.

21
 x-vorticity SKE
6000

5000

-6000
4000
-5500

-5000
3000
-4500

-4000
2000
-3500
-3000
1000
-2500

0
-2000 RKE
-1500
-1000
RNG
-1000
-500

0
-2000
500

1000
-3000
1500

2000

-4000
2500
3000

3500
-5000
4000
4500
-6000
5000
(1/s) 5500
RSM
6000

Figure 15. Contours of vorticity near the injector determined using four turbulence models on

the plane passing through the centre of an oxygen inlet and the central axis.

Temperatures profiles at z =30 mm and z = 50 mm, as shown in Fig. 14(a) and Fig. 14 (b),

also confirmed the similarity between the results obtained using the SKE and RSM models,

and that between those obtained using the RKE and RNG models. Under-prediction of the

oxygen jet spread rate by the RKE and RNG models led to determination of higher wall

temperatures while overdiffusion of the low temperature oxygen jet reduced the wall

temperatures predicted by the SKE and RSM models. Due to this phenomenon, the location

of peak temperature at z = 30 mm was also shifted radially outward from the centreline by the

RKE and RNG models. Figures 14(c) and 14(d) followed the trends set by the velocity

profiles.

22
 (a) (b)

(c) (d)
Figure 16. Contours of (a) temperature [K] and (b) OH mass fraction (c) n-heptane mass
fraction (d) O2 mass fraction determined using four turbulence models on the plane passing
through the centre of an oxygen inlet and the central axis.

Figure 16(a) shows the contours of temperature along with contours of mass fraction of OH

(Fig.16 (b)), n-heptane (Fig. 16(c)), and oxygen (Fig. 16(d)). The temperature, OH mass

fraction, n-heptane mass fraction, and oxygen mass fraction contours determined by all

models were mutually consistent throughout Fig.16. Due to the discrepancy in the radial

spread of the oxygen stream, higher temperatures predicted (Fig. 16(a)) by both RKE and

RNG models in the vicinity of the oxygen jet near the inlet holes were corroborated. For both

of these models, fuel vapours were entrained due to higher jet velocities and lower radial

spread leading to a flame zone surrounding the oxygen jet. Although OH mass fractions were

23
predicted to be high in the vicinity of the oxygen jet by RKE and RNG models, those values

were smaller for the RSM model, while it was negligible for the SKE model.

High mass fractions of OH were also observed near the injector face due to entrainment of

fuel vapour by both the RKE and RNG models, indicating a vigorous reaction zone in the

vicinity of the injector face with the OH concentration and temperatures being higher for the

RNG model than the RKE model. The SKE and RSM models display a recirculation zone in

that region and both the models predicted a less harsh thermal environment for the injector.

Combustion was not completed within the reaction chamber, owing to the larger droplet

sizes emanating from the injector, as well as the length of the combustor. Hence, the

combustion zone was found to extend beyond the converging-diverging section of the rocket

chamber, and into the ambient gases beyond the exit plane. However, discrepancies were

obtained in the nature of the flame zone as predicted by the SKE model as opposed to those

predicted by the other three models. The high temperature zones were found to be the most

diffuse, also borne out by the profiles in Fig. 16(c) and Fig. 16(d). The RNG model predicted

the thinnest reaction zone in the converging-diverging section, aided by the penetration of the

oxygen jet till the converging section.

24
 Figure 17. Pathlines of particles coloured by particle diameter (m) determined using four
turbulence models on the plane passing through the centre of an oxygen inlet and the central
axis.

Spray characteristics for the four models are shown in Fig. 17 by particle tracks coloured by

particle diameter. Particles possessing larger diameters are reflected from the chamber walls

in both the cylindrical and converging section. These results reiterate the need of better

injectors to produce sprays with smaller Sauter mean diameters, allowing combustion to be

completed within the combustion chamber.

Based on the results obtained, the RKE and RNG models, which could not predict a

physically realistic recirculation zone on the injector face, were considered inadequate for the

present study and were discarded for further analyses. The SKE and RSM models were

mutually consistent to a certain extent but discrepancies were evident towards the terminal

region of the combustion chamber and in the diverging section. The SKE model was

discarded since its inherent simplicity was deemed unsuitable for such complex reacting

flows, and the RSM model was utilized for further simulations.

25
3.7. Comparison of combustion models

Once the RSM model was finalized, a comparison between the chemical equilibrium based

PDF model and the steady diffusion flamelet model was conducted to ascertain their

suitability to analyze the combustion of n-heptane and gaseous oxygen. The mass fractions of

important species such as OH, O2, CO2, CO, and n-heptane as well as temperature contours

are shown in Fig. 18.

Although the magnitudes of flow velocities were small inside the combustion chamber, the

gases start to accelerate after the converging section and the flow becomes highly strained in

the transonic and supersonic regimes. Thus, significant departure from equilibrium and

correspondingly lower temperatures were expected and the flamelet model was capable of

such predictions as shown in Fig. 18(a).

As shown in Fig. 18(f), the equilibrium model predicted higher concentration of CO 2 in the

diverging section, indicating that equilibrium was established in that zone, which may be

impractical due to highly strained nature of the flow. Similarly, higher concentrations of CO

predicted by the flamelet model within the combustion chamber, as shown in Fig. 18(e), were

indicative of incomplete combustion which would be expected owing to the non-equilibrium

nature of combustion.

26
 (a) (b)

(c) (d)

(e) (f)
Figure 18. Contours of (a) temperature (K) and mass fraction mass fraction of (b) OH (c) n-
heptane (d) O2 (e) CO (f) CO2 determined using two combustion models on the plane passing
through the centre of an oxygen inlet and the central axis.

On analyzing the phenomena near the injector face, it was observed that the equilibrium

model predicted a thin flame zone evident from Fig. 18(b), and high mass fraction of CO2,

indicating an equilibrium flame with low diffusion rate. However significant gradients in

species and temperature were expected, due to the presence of the recirculation zone and

incoming cold stream of oxygen leading to faster diffusion of radicals and energy. The

flamelet model was able to predict this higher diffusion rate as seen by a thicker flame zone

and low CO2 mass fraction.

27
Further downstream away from the recirculation zone, the equilibrium model predicted a

thicker reaction zone with higher temperatures which led to faster diffusion of species and

energy in radial direction inside the chamber. On the other hand, lower radial diffusion and

lower temperatures were predicted by the flamelet model. A liquid core with lower

temperatures and lower evaporation rates was observed using the flamelet model which

further affected the temperature and species profiles. The equilibrium model was deemed

unsuitable for such flows and the flamelet model was used for further simulations to study

fuel lean and fuel rich conditions.

3.8. Comparison of fuel-lean vs. fuel-rich conditions

Simulations were performed for two propellant flow rates i.e. fuel-lean and fuel rich

conditions with equivalence ratios of ϕ = 0.9 and ϕ = 1.14 respectively. Decrease of the

oxygen flow rate from 8.91 g/s to 7.59 g/s and hence the velocity of the oxygen jet caused

variations in liquid phase and gas phase velocities leading to differences in heat and mass

transfer between the two phases. Higher concentrations of CO and n-heptane were observed

in fuel-rich conditions as shown in Fig. 19.

28
 (a) (b)

(c) (d)

(e) (f)
Figure 19. Contours of (a) temperature (K) and mass fraction mass fraction of (b) OH (c) n-
heptane (d) O2 (e) CO (f) CO2 for fuel lean and fuel rich conditions on the plane passing
through the centre of an oxygen inlet and the central axis.

3.9. Comparison of integral performance parameters with experimental results

(a) (b)

29
 Figure 20. Comparison of chamber pressure, determined using different models with
experimental values for (a) fuel lean, and (b) fuel rich conditions

Figure 20 shows the comparison of chamber pressure determined by various combinations of

turbulence and combustion models. The oxidizer-fuel ratio was varied by adjusting the

oxygen flow rate through oxygen injection pressure which indirectly determined the chamber

pressure. For higher oxygen flow rates in fuel lean conditions the chamber pressure was

higher as compared to lower oxygen flow rate in fuel rich conditions. The SKE model

slightly over-predicted the jet spreading rate leading to enhanced mixing and more residence

time for the propellants. This caused combustion to occur to a greater extent within the

combustion chamber and hence the predicted chamber pressure was found to be higher as

compared to other models as well as the experimental values, except for the SKE model

combined with the flamelet model in fuel lean conditions. The RNG model under-predicted

the rate of spread and the oxygen jet was allowed to penetrate up to the end of the combustion

chamber. Lower extent of mixing, and hence combustion, in this case caused the chamber

pressure to be the lowest among all cases considered for both of the combustion models. The

RKE model behaved in a fashion similar to the RNG model, though higher values were

predicted than the RNG model. Predictions by the SKE model combined with flamelet model

were found to be closest to the experimental values. However, for the RSM model combined

with the flamelet model was also found to be satisfactory, except the fuel rich conditions.

However, it was encouraging to observe that the RSM model with equilibrium chemistry

performed appreciably for both the cases. This would prove to be advantageous in the long

run as detailed chemical kinetic model would be unavailable for the new HED compounds to

be tested in the future.

30
 (a) (b)
Figure 21. Comparison of thrust, determined using different models, with experimental
values

The flamelet model predicted slightly lower chamber pressures for all turbulence models as

compared to equilibrium PDF model with respective turbulence models, but the variation was

not significant. The values of the thrust predicted as shown in Fig. 21 follows similar trends

as that of chamber pressure, though the deviations from the experimental values, especially

for fuel rich conditions, were noticeable. The numerical evaluation of thrust was based on

exit velocity which was strongly affected by combustion in supersonic flow regime, leading

to the incongruences in prediction of thrust. It was surprising to observe that the SKE model

with equilibrium combustion predicted values of thrust that were closest to the experimental

values. This aspect of the work would require further analysis.

The analysis conducted in the present work was intended to establish a numerical approach to

simulate the performance of LREs with n-heptane as a baseline fuel, so that further studies

with addition of HED cage hydrocarbons may be performed in the future. Although various

turbulence and combustion models were considered, a rigorous investigation of discrete

phase model parameters and further experimental characterization of the spray is essential.

Radiation was neglected in the present study and needs to be included in the simulation along

with realistic wall heat transfer characteristics from the combustion chamber. Only integral

performance parameters of the engine were validated experimentally. Detailed experimental

31
data on velocity, temperature, and species profiles are required to establish the validity of

these models.

4. Conclusions

The study was aimed at establishing an experimental laboratory-scale facility to analyse the

performance of solutions of cage hydrocarbon-based HED fuels with standard hydrocarbon

fuels, as well as to determine whether numerical models could be utilised to predict their

combustion behaviour within the chamber and to determine their performance parameters

using a few experimentally determined parameters. The experimental facility was operated by

injecting GOX through circular orifices on the injector plate and liquid gasoline through a full

cone commercial injector to establish steady state combustion. The chamber pressures

achieved were approximately 5 bar, with a thrust of approximately 17 N, which yielded

specific impulses of 160 s with a c* efficiency of 80%. The numerical study was conducted

using a 3D RANS solver in ANSYS Fluent. A qualitative comparison of various turbulence

models showed that the RKE and RNG k-ε models were unable to physically predict a

recirculating zone at the injector plate. The SKE and RSM models were mutually consistent

to a certain extent, with the presence of the recirculating zone and the spread of the oxygen

jet being similar for both. The encouraging result was the ability of the equilibrium PDF

model to predict the features of combustion as well as the experimental pressures within the

combustion chamber, since the chemical kinetic mechanisms of the new HED fuels would

not be readily available for simulation. However, the departure from equilibrium conditions

were evident in the transonic and the supersonic flow regimes at the throat and the diverging

section. This affected the prediction of the exit velocity and hence the thrust of the rocket

engine was under-predicted by approximately 20 to 50%.

32
Acknowledgements

We are grateful to the Department of Science and Technology, India for providing funds for

the experimental studies.

References

[1] A. Sergienko, Liquid rocket engines for large thrust: Present and future, Acta

Astronautica, 29 (1993) 905-909.

[2] T.P. Lebas J., Ohara J., Turbulence in a gaseous hydrogen-liquid oxygen rocket

combustion chamber, in: NASA (ed.), United States 1975.

[3] R.F. Neu, Injection Principles for Liquid Oxygen and Heptane Using Nine-element

Injectors in an 1800-pound-thrust Rocket Engine, National Advisory Committee for

Aeronautics, 1957.

[4] R.J. Priem, Propellant Vaporization as a Criterion for Rocket Engine Design: Calculations

of Chamber Length to Vaporize a Single N-heptane Drop, National Advisory Committee for

Aeronautics, 1957.

[5] S. Sonawane, U. Bhandarkar, B. Puranik, S.S. Kumar, Effect of particle size in aviation

turbine fuel-Al2O3 nanofluids for heat transfer applications, World Academy of Science,

Engineering and Technology, 81 (2011).

[6] G.P. Sutton, Oscar Biblarz, Rocket Propulsion Elements, JOHN WILEY & SONS.,

Nexdv York, 1992.

[7] C. Andrew, Considerations for the Implementation of Privately-Built Manned Orbiting

Vehicles, AIAA SPACE 2007 Conference & Exposition, (2007).

[8] S. Rajkumar, R.S. Choudhari, A. Chowdhury, I.N.N. Namboothiri, Synthesis and

pyrolysis studies of bis (nitratomethyl)-1, 3-bishomocubane—A high-energy high-density

liquid, Thermochimica Acta, 563 (2013) 38-45.

33
[9] S. Lal, S. Rajkumar, A. Tare, S. Reshmi, A. Chowdhury, I.N.N. Namboothiri,

Nitro‐Substituted Bishomocubanes: Synthesis, Characterization, and Application as Energetic

Materials, Chemistry–An Asian Journal, 9 (2014) 3533-3541.

[10] S. Lal, L. Mallick, S. Rajkumar, O.P. Oommen, S. Reshmi, N. Kumbhakarna, A.

Chowdhury, I.N.N. Namboothiri, Synthesis and energetic properties of high-nitrogen

substituted bishomocubanes, Journal of Materials Chemistry A, 3 (2015) 22118-22128.

[11] E.A. Santos, W.F. Alves, A.N.A. Prado, C.A. Martins, Development of test stand for

experimental investigation of chemical and physical phenomena in liquid rocket engine,

Journal of Aerospace Technology and Management, 3 (2011) 159-170.

[12] J. Lux, O. Haidn, Flame Stabilization in High-Pressure Liquid Oxygen/Methane Rocket

Engine Combustion, Journal of Propulsion and Power, 25 (2009) 15-23.

[13] C. Navarro, J. Betancourt-Roque, L. Sanchez, N. Robinson, A. Choudhuri, Development

of a multi-purpose optically accessible rocket combustor for liquid oxygen and hydrocarbons,

in: 47th AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit 2011, 2011.

[14] P. Caisso, A. Souchier, C. Rothmund, P. Alliot, C. Bonhomme, W. Zinner, R. Parsley,

T. Neill, S. Forde, R. Starke, W. Wang, M. Takahashi, M. Atsumi, D. Valentian, A liquid

propulsion panorama, Acta Astronautica, 65 (2009) 1723-1737.

[15] G.M. Faeth, Evaporation and combustion of sprays, Progress in Energy and Combustion

Science, 9 (1983) 1-76.

[16] W.A. Sirignano, Fluid Dynamics of Sprays—1992 Freeman Scholar Lecture, Journal of

Fluids Engineering, 115 (1993) 345-378.

[17] C.K. Law, Recent advances in droplet vaporization and combustion, Progress in Energy

and Combustion Science, 8 (1982) 171-201.

34
[18] P. Ghose, J. Patra, A. Datta, A. Mukhopadhyay, Effect of air flow distribution on soot

formation and radiative heat transfer in a model liquid fuel spray combustor firing kerosene,

International Journal of Heat and Mass Transfer, 74 (2014) 143-155.

[19] C. Hollmann, E. Gutheil, Modeling of turbulent spray diffusion flames including

detailed chemistry, in, Vol. 26, Elsevier, pp. 1731-1738.

[20] A. Datta, S.K. Som, Combustion and emission characteristics in a gas turbine combustor

at different pressure and swirl conditions, Applied Thermal Engineering, 19 (1999) 949-967.

[21] K. Van Maele, B. Merci, E. Dick, Comparative study of k-epsilon turbulence models in

inert and reacting swirling flows, in: Proceedings of the 33rd AIAA Fluid Dynamics

Conference, pp. 1-10.

[22] D. Joung, K.Y. Huh, 3D RANS simulation of turbulent flow and combustion in a 5 MW

reverse-flow type gas turbine combustor, Journal of Engineering for Gas Turbines and Power,

132 (2010) 111504.

[23] G.M. Faeth, Mixing, transport and combustion in sprays, Progress in Energy and

Combustion Science, 13 (1987) 293-345.

[24] L. Gomet, V. Robin, A. Mura, Lagrangian modelling of turbulent spray combustion

under liquid rocket engine conditions, Acta Astronautica, 94 (2014) 184-197.

[25] T.-S. Wang, Thermophysics characterization of kerosene combustion, Journal of

Thermophysics and Heat Transfer, 15 (2001) 140-147.

[26] J.-Y. Choi, A quasi global mechanism of kerosene combustion for propulsion

applications, AIAA Paper, 5853 (2011) 2011.

[27] http://in.mathworks.com/help/signal/ref/butter.html, in.

[28] ANSYS INC., ANSYS Fluent Theory Guide, Release 14.5, USA, 2012.

35
[29] Y.-D. Liu, M. Jia, M.-Z. Xie, B. Pang, Enhancement on a skeletal kinetic model for

primary reference fuel oxidation by using a semidecoupling methodology, Energy & Fuels,

26 (2012) 7069-7083.

[30] A. Balabel, A.M. Hegab, M. Nasr, S.M. El-Behery, Assessment of turbulence modeling

for gas flow in two-dimensional convergent–divergent rocket nozzle, Applied Mathematical

Modelling, 35 (2011) 3408-3422.

[31] P.K. Tucker, S. Menon, C.L. Merkle, J.C. Oefelein, V. Yang, Validation of high-fidelity

CFD simulations for rocket injector design, AIAA Paper, 5226 (2008) 2008.

[32] D. Byun, S.W. Baek, Numerical investigation of combustion with non-gray thermal

radiation and soot formation effect in a liquid rocket engine, International Journal of Heat and

Mass Transfer, 50 (2007) 412-422.

[33] X.D. Yang, H.Y. Ma, Y.N. Huang, Prediction of homogeneous shear flow and a

backward-facing step flow with some linear and non-linear K–ε turbulence models,

Communications in Nonlinear Science and Numerical Simulation, 10 (2005) 315-328.

[34] K. Menzies, An evaluation of turbulence models for the isothermal flow in a gas turbine

combustion system, in: 6th International Symposium on Engineering Turbulence Modeling

and Experiments, Sardinia Italy, 2005.

36
List of figure captions

Figure 1. Sectional view of the rocket engine (1. fuel inlet, 2. oxygen inlet, 3. holes for

oxygen injection, 4. ’O’ring groove, 5. cooling water inlet, 6. pressure tap pipe, 7. cooling

water out, 8. combustion chamber, 9. cooling jacket, 10. injector flange) .............................. 6

Figure 2. Schematic of the experimental setup (NRV: non return valve, NV: needle valve,

PT: pressure transducer, SV: solenoid valve, MFM: mass flow meter, MCV: motorized

control valve, PRV: pressure relief valve, FF: fuel filter, BV: ball valve) .............................. 7

Figure 3. Variation of pressure in the fuel line, oxygen line and converging section of

chamber, and fuel and oxidizer flow rates at ϕ = 1 ................................................................. 9

Figure 4. Variation of pressure in the fuel line, oxygen line and converging section of

chamber at, and fuel and oxidizer flow rates ϕ = 0.8 ............................................................ 10

Figure 5. Variation of thrust with time for ϕ = 1 and ϕ = 0.8 ................................................ 10

Figure 6. Mach discs at ϕ = 1.14.......................................................................................... 11

Figure 7. Variation of c* efficiency and Specific impulse with equivalence ratio ................. 12

Figure 8. Computational domain ......................................................................................... 14

Figure 9. Unstructured mesh for the rocket engine............................................................... 14

Figure 10. Axial locations chosen for displaying profiles of velocity, temperature, and

species. ............................................................................................................................... 15

Figure 11. Velocity profiles at different axial locations for four grid sizes ........................... 17

Figure 12. Contours of velocity superimposed by streamlines near the injector determined

using four turbulence models on the plane passing through the centre of an oxygen inlet and

the central axis. ................................................................................................................... 18

Figure 13. Velocity profiles at different axial locations determined using four turbulence

models ................................................................................................................................ 20

37
Figure 14. Temperature profiles at different axial locations determined using four turbulence

models ................................................................................................................................ 21

Figure 15. Contours of vorticity near the injector determined using four turbulence models on

the plane passing through the centre of an oxygen inlet and the central axis......................... 22

Figure 16. Contours of (a) temperature [K] and (b) OH mass fraction (c) n-heptane mass

fraction (d) O2 mass fraction determined using four turbulence models on the plane passing

through the centre of an oxygen inlet and the central axis. ................................................... 23

Figure 17. Pathlines of particles coloured by particle diameter (m) determined using four

turbulence models on the plane passing through the centre of an oxygen inlet and the central

axis. .................................................................................................................................... 25

Figure 18. Contours of (a) temperature (K) and mass fraction mass fraction of (b) OH (c) n-

heptane (d) O2 (e) CO (f) CO2 determined using two combustion models on the plane passing

through the centre of an oxygen inlet and the central axis. ................................................... 27

Figure 19. Contours of (a) temperature (K) and mass fraction mass fraction of (b) OH (c) n-

heptane (d) O2 (e) CO (f) CO2 for fuel lean and fuel rich conditions on the plane passing

through the centre of an oxygen inlet and the central axis. ................................................... 29

Figure 20. Comparison of chamber pressure, determined using different models with

experimental values for (a) fuel lean, and (b) fuel rich conditions ........................................ 30

Figure 21. Comparison of thrust, determined using different models, with experimental

values .................................................................................................................................. 31

38
List of table captions

Table 1: Steady state values of various parameters for ϕ = 1.14, 1, 0.9, and 0.8

39
Sectional view of the rocket engine Variation of pressure in the fuel

line, oxygen line and converging

section of chamber, and fuel and

xidizer flow rates at ϕ = 1

Velocity and temperature contours for four turbulence models

40
 Turbulent spray combustion in a gaseous oxygen and liquid fuelled rocket engine

 Chamber pressure and thrust were determined experimentally for various parameters

 Simulations conducted using FLUENT showed RSM with flamelet model to be the

best

 Standard k-ε model with equilibrium PDF model was also reliable

 Chamber pressures were accurately predicted, while thrust was under-predicted

41