Numerical Study of Two Fluid Model for Space

and Commercial Applications

Report Submitted to
Propulsion Research Group(PRG)
Vikram Sarabhai Space Centre(VSSC)

Project Supervisors
Dr. T. Jayachandran
Deputy Director, PRSO
VSSC.
1|Page

Pedda Perriah
Engineer/Scientist
VSSC.

Acknowledgement
I would like to take this opportunity to thank my mentor, Dr. T. Jayachandran, for mentoring
and guiding me throughout the course of this project work. It was due to his constant support
and motivation that this project could be successfully completed.
I would also like to acknowledge the efforts put by Pedda Perriah sir in helping me with this
this project. He guided me with each and every aspect of this study and always emphasized to
understand the physics behind every physical phenomenon. The discussions with him proved
very fruitful and always brought up something new.
Last, but not the least, I would also like to expresses my gratitude towards M.V Dekhne Sir
and H.B Hablani Sir, for giving me the opportunity to work among the best Scientists of India
in presumably the best Research Organization of India.

2|Page

Chapter 1: Space Application

Numerical Study of bulk Solid Rocket Motor

Performance Parameters during ejection of an
Inert Particle

Abstract:
Measured time histories for bulk Solid Rocket Motor parameters like thrust, chamber pressure,
have been observed to have frequent perturbations or excursions about some nominal value.
One cause of these perturbations and the subject of this study is the expulsion of an inert mass,
e.g., an igniter that has come loose. The objective of this work was to numerically generate
accurate, high time resolution thrust and head end pressure curves for various idealized cases
of mass discharge to improve the understanding of this phenomenon.

3|Page

Table of Contents
1. Nomenclature

2. Introduction

3. Numerical Procedure

3.1 Geometry of the Solid Rocket Motor

3.2 Governing Equations
3.2.1
a.
b.
c.
d.

Without the introduction of the particle

Source Terms
Input Parameters Used
Initial Conditions
Boundary Conditions

3.2.2 After the introduction of the particle

a. Drag force model
b. Input Parameters for particle introduction

6
7
7
8
8
8
8
9
9
9

4. Numerical Scheme

10

5. Numerical Results and Discussions

11

5.1 Single Phase Flow Results

5.2 Two Phase Flow Results

11
14

6. Conclusion

21

7. References

21

4|Page

1. Nomenclature

5|Page

2. Introduction
In this study, numerical model for solving one dimensional two phase compressible flow
in integrated solid rocket motor nozzle is developed and solved to account for the thrust
and pressure perturbations during condensed phase discharge. In order to make this
problem tractable, several assumptions have been made. The problem is approximated to
unsteady, axisymmetric flow. The condensed phase ejected body is restricted to spherical
in shape, thus completely specifying it by its radius and density. The flow in the rocket
motor is assumed to be that of an inviscid perfect gas with constant ratio of specific heats
(Isentropic flow)

3. Numerical Procedure
To study the thrust and pressure perturbations during mass ejection, firstly, steady state
flow is reached without considering the effect of condensed phase in the governing
equations. Once the steady state or nominal values of thrust and head end pressure are
computed, the particle is introduced in the motor and subsequently time history of thrust,
head end pressure, particle trajectory and particle velocity is monitored.

3.1

Geometry of the Solid Rocket Motor Nozzle (JPL Nozzle [1])

6|Page

Motor Geometrical Parameters

3.2

Governing equations:
A set of four nonlinear coupled Euler equations for two fluid model [6] with source
terms are solved. The two phase flow is assumed to be isentropic. To study the thrust
and pressure perturbations, these Euler equations are solved in two stages. First without
considering the condensed phase intervention. And then introducing the condensed
phase into the motor and taking into account the Drag force source terms. The solution
is then allowed to reach steady state and correspondingly time history of motor
performance parameters are monitored.

3.2.1 Without the introduction of the particle

where U is the matrix of conservative variables, F is the matrix of Flux terms, S is the
source terms. The nomenclature is as usual and can be referred to at the beginning.

7|Page

Here there are 5 unknowns

but only four governing equations
as stated above. So to provide a closure to the above set of equations, energy equation
for gas phase is considered. Considering stagnation enthalpy to be constant (isentropic
flow), dynamic gas temperature can be computed and then through equation of state, gas
pressure is computed.

a. Source terms:
To predict the solid propellant burning rate, APN model is used, which is a semiempirical model. APN model considers grain burning rate as quasi static term as
dependent on mean local pressure, given by the equation

Typically a and n are empirically coefficients and are considered constant for wide
pressure ranges. Pref is the reference pressure usually taken as 0.98 bars if P is taken in
bars.
So, the propellant mass flow rate is
b. Input Parameters Used

c. Initial Conditions
The motor is initialized with a Mac number =0.0001 and corresponding dynamic
temperature and pressure are computed using Isentropic flow conditions from
stagnation state parameters specified above. To account for the solid phase governing
equations, a tolerance value is given to particle fraction (10-7).

d. Boundary Conditions
Wall Boundary condition is imposed on the left side of the motor and supersonic
outflow boundary conditions are given at the right end of the motor.

8|Page

3.2.2 After introduction of the particle

Having achieved a convergence of order 10-8 for all the four conservative variables,
condensed phase particle is introduced in the motor chamber and correspondingly the
governing equations are modified to incorporate the effect of the particle. The
conservative and flux terms remain same, but the source terms are modified for the gas
and solid momentum equations. A drag force component is added to the gas momentum
equation and is subtracted from the solid momentum equation.

a. Single Particle Drag force model [1]

According to the above drag model, drag coefficient is calculated based upon the
particle Reynolds number.

b. Input Parameters for Particle introduction

In order to introduction to introduce the particle in the motor, two parameters are to be
specified. First is the particle radius and the other is its initial grid cell location. Also
for the numerical scheme requirements, no. of cells in which the particle is distributed
has to be specified. For the current grid, 2 cells of 10 mm thickness are specified to
share the particle volume. Then based upon the specified particle radius and the cell
location, particle fraction values of those cells are modified, which is calculated by
volume of the particle that it contains divided by that cell volume. Also the initial
particle velocity is given as 99% of the gas velocity of that cell.

9|Page

Having computed the particle fraction values for the effected cell, the conservative
variables of those cells are modified accordingly to incorporate the particle with
conditions as specified above.
To study the motor bulk performance parameters after the particle introduction, four
cases corresponding to four different input conditions of particle introduction are
considered.

1.
2.
3.
4.

Rp =5mm
Rp =10mm
Rp=5mm
Rp=10mm

released at 500 mm from the head end

released at 500 mm from the head end
released at 700 mm from the head end
released at 700 mm from the head end

4. Numerical Scheme
The four coupled non-linear hyperbolic system of equations are solved numerically using
JST Scheme (Jameson-Schmidt-Turkel). The entire motor geometry is divided into cells
and computational grid is generated. The system of equations is discretised at every cell
centre location and is solved iteratively. Now to compute the flux crossing the cell faces,
various schemes are proposed which are efficient in capturing the shocks and
discontinuities without violating the physics and at the same time stabilizing the
numerical solution. JST is one of the schemes which is extensively used. It computes the
cell face flux based upon the arithmetic average of the fluxes of the neighbouring cells
and an artificial dissipation term based upon some sensor functions. The scheme is
explained as follows:

10 | P a g e

5. Numerical Results and Discussions

With the above mentioned JST scheme, computational iterations were done using
FORTRAN language to attain numerical steady state results for both single phase (prior
to particle introduction) and two phase models (after particle introduction). The CFL
value used was 1.

5.1 Single Phase Flow Results

The single phase simulation was done with CFL value of 1 and was run for 497 ms for
steady state value to be reached. Pressure and Mac number variation along the length
of the motor and time history of the head end pressure was monitored. The results are
shown in Fig 2 and Fig 3.

11 | P a g e

Fig. 2

The mac number plot shows a value of 1 around the throat region, which follows the observed
physical phenomenon of choking.

Fig. 3

12 | P a g e

Fig.3 shows a steady state Head End Pressure value of 93.87 bars. This result is validated using
the analytical expression for steady state chamber pressure derived from the basic mass
conservation principle.

C* is the characteristic flow speed at the nozzle exit. It depends upon the stagnation
temperature, ratio specific heats of the gas and also its characteristic constant ( R).
Kn is the area ratio of the total propellant burning area to throat area.
the propellant burning rate.

rb as already defined is

Substituting the motor input parameters into this analytical expression gives a steady state
Pressure value of 89.9 bars. So the numerically computed pressure value lies within 4% of
the analytically calculated value. The small difference between the two values of the thrust is
assumed to be caused by the fact that the real nozzle flow is not one-dimensional. A second
cause of the difference is the artificial viscosity that stabilizes the numeric and is implicit in
JST numerical scheme. Because of the artificial viscosity, the numerically computed flow is
not exactly isentropic.
As far as present scope of study is involved this error would not impact the final perturbations
in motor parameters. The aim of this study is to analyse the perturbations about some nominal
value, so a little variation in nominal value would not impact the final results
.

After head-end pressure, steady state thrust was calculated using the formula

The steady state thrust was found to be equal to 20.121KN

13 | P a g e

5.2 Two phase Flow Results

As mentioned earlier, four different cases of condensed mass ejection are considered.

1.
2.
3.
4.

Rp =5mm
Rp =10mm
Rp=5mm
Rp=10mm

released at 500 mm from the head end

released at 500 mm from the head end
released at 700 mm from the head end
released at 700 mm from the head end

For each case, time history of velocity, thrust and head end pressure was monitored.
The trajectory of the particle in the motor from the onset of its introduction in the motor to its
exit from the nozzle is coupled with time history of velocity, thrust and head end pressure to
provide a clearer picture of the perturbations caused by the particle in the motor parameters.
Case 1 & 2:

Fig. 4

Fig. 5

14 | P a g e

Fig. 6

Fig. 7

The points to note from the above two plots is the magnitude of the pressure perturbation for the larger
and the smaller particle, and the lag between the pressure reaching its peak and the entry of the particle
in the nozzle(or throat). The lag is possible because it takes some time for the rarefaction wave to travel
up the length of the motor and hence the disturbance caused by the particle is not immediately affected
at the head end. The larger perturbation in case of bigger particle is quite possibly due to more area
blockage and more blockage time, as the velocity of the particle varies inversely with mass. So the
disturbance caused is more in case of bigger particle.
15 | P a g e

Figure 8

Figure 9

16 | P a g e

Cases 3 & 4: Released at 700mm

Figure 10

Figure 11

17 | P a g e

Figure 12

Figure 13

18 | P a g e

Figure 14

Figure 15

The main observation from the above two plots (fig.14 & fig 15) of thrust time history for two
different masses, is the thrust deviation in case of larger particle is ten times the deviation in
case of smaller particle. One possible explanation of this could be by doubling the particle
radius, mass of the larger particle is increased 8 times, so throat blockage time for the larger
particle is more. The amount of time that the throat is blocked the time duration of the negative
thrust perturbation is expected to vary as the square root of the body mass. (This corresponds
to a body velocity inversely proportional to the square root of the mass.) This is consistent
from the above two plots. In fig.14 the total time particle remains inside the nozzle is lesser
than in fig.15

19 | P a g e

Figure 16

In Fig. 16, Thrust deviations from the nominal value (20.121KN) is presented. These deviations
are caused due to the condensed particle blocking the throat area. As can be seen from the
graphs, larger the particle size, more is the deviation from the nominal value. The minimum
and maximum values of thrust are almost 10 times for the larger particle as compared to
smaller particle, for both the release positions of the particle. Again this trend can be
explained in terms of throat blockage time, which increases with increasing the mass of the
particle, and subsequently more perturbation in thrust and pressure are observed.
The heavier the particle, more momentum it acquires during its traverse through the nozzle
and consequently has a larger exit velocity. As can be seen from the table below.
Now, from Fig.16 total impulse imparted by the particle for all the four cases are calculated
by finding the area under the curve of Thrust deviation time curve. And the results are
tabulated in the table1. As expected the larger particle gives more total impulse and also
higher specific impulse to the motor as compared to smaller particle for both the release
points.
Mass Ejection parameters
Case

Released
at(in mm)

Radius
(mm)

Density
(kg/m3)

Mass
(gms)

Impulse
(N.s)

1.
2.
3.
4.

500
500
700
700

5
10
5
10

1760
1760
1760
1760

0.9219
7.37
0.9219
7.37

0.13238
1.49108
0.11655
1.101622

Table 1

20 | P a g e

Specific
Exit
Impulse(m/s) Velocity
(m/s)
143.60549
59.5082
202.3184
92.42
126.427
71.65
149.4738
104.244

6. Conclusion
It is seen through this study that although the thrust perturbation is usually negative
when the ejected body blocks the nozzle throat, the total impulse associated with mass
ejection is always positive. Also, the positive thrust perturbation associated with the
ejection may be present long after the body has exited the nozzle.

7. References
[1] One and two phase nozzle flows I-Shih Chang. The Aerospace Corporation, El Segundo,
Calif.

[2] Rocket Thrust Perturbation from Discharge of an Inert Body, John W. Murdock .The
Aerospace Corporation, El Segundo, California.
[3] Effects of Slag Ejection on Solid Rocket Motor Performance R. Harold Whitesides and David
C. Purinton, ERC Incorporated Huntsville, AL
[4] One dimensional, Two-phase Flow Modelling toward Interpreting Motor Slag Expulsion
Phenomena, Timothy P. Kibbey, Jacobs Technology, Jacabs ESTS Group, Huntsville, AL.
[5] Finite Volume method For Hyperbolic Problems by Randall J. Leveque
[6] Principles of combustion 2nd Edition by Kenneth K. Kuo.
[7] Computational Gas Dynamics by Culbert B. Laney

21 | P a g e