Two

dimensional Model
for
Liquid

Rocket
Transverse
Combustion Instability
by W. A. Sirignano and P. Popov
Mechanical and Aerospace Engineering
University of California, Irvine
Supported by Air Force Office of Scientific Research
Dr.
Mitat
Birkan
, Program Manager
Goals of Current UCI Combustion Instability
Research
•
Develop “simplified” liquid

rocket numerical models of
combustion dynamics to test stochastic approaches and
demonstrate the triggering of combustion instabilities.
•
Examine nonlinear stability with initial conditions matching
linear modes. Identify parameter domains allowing triggering.
•
Examine nonlinear stability with initial conditions
of local
Gaussian

shaped disturbance.
Identify parameter domains
allowing triggering
.
•
Future:
Extend the model with the inclusion of stochastic terms
representing combustion noise and large

amplitude random
perturbations.
•
Future:
Investigate the statistical data for markers of complex
system behavior, e.g. power laws.
•
Future:
Extend the work to more detailed combustion dynamics
models
i
n collaboration with Georgia Tech and
HyPerComp
.
Nonlinear Combustion Instability and Triggering Action
Instability initiates in various ways, depending upon the operational
parameter domain: (1) oscillations that grow from normal combustion
noise are named linear or spontaneous instabilities; (2) oscillations
initiated only by disturbances larger than noise are named nonlinear or
triggered instabilities. All limit

cycle oscillations
are nonlinear;
“linear” and “nonlinear” refer to initiation.
Princeton
experiment,
c
irca 1961.
A “bomb”
(gunpowder
contained
with burst disk)
is used as the
trigger.
EARLY THEORIES
Nonlinear
l
imit cycles were first
predicted by
Crocco
& students
in
the 1960s using perturbation
methods: Sirignano,
Zinn
, and Mitchell
dissertations. Triggering and stable and
unstable limit cycles were predicted.
Later,
Zinn
& Powell followed by
Culick
and co

workers used a
Galerkin
method to predict transient behavior as
well.
Except for one portion of
Sirignano’s
work, all the models used heuristic
representations of combustion: e.g.,
n,
τ
model.
Model Equation System
•
The model equations retain essential physics for the combustion
dynamics but eliminate much of the secondary physics which
could be added in later studies. This model should allow the
testing of our statistical approaches before we engage in a full
analysis.
•
The focus is on transverse oscillations in a cylindrical chamber
allowing averaging over the axial direction and reduction to an
unsteady, two

dimensional problem in the transverse polar
coordinates.
•
Kinematic waves are neglected leaving only the longer acoustic
waves. These kinematic waves travel primarily in the axial
direction and because of larger gradients (i.e., shorter
wavelengths) are more likely to be vitiated by turbulent mixing.
•
Viscosity, heat

conductive, and turbulent

mixing effects on the
longer acoustic waves are neglected.
•
A model for co

axial stream turbulent mixing and reaction is
developed and employed for a multi

injector configuration
.
•
A simplified “short” multi

orifice nozzle boundary condition is
used.
Nonlinear Wave Equation for Pressure
Momentum equations for radial and azimuthal velocities
E
is the energy per unit volume
per unit time released by the
combustion process.
Modelling
of
E
is required.
Triggering disturbance could
appear in several ways:

Introduction through
reacting, mixing flow

field condition

Introduction through injector

face boundary condition.

An intermittent blockage in nozzle
flow.
W
e test wave dynamics with
P = (p

p
0
)/p
0
and
u
sing a
polynomial function
E(p) = a P
4
–
bP
3
+ cP
2
+
dP
Triggering can occur; stable
and unstable limit cycles and
transients are captured.
Co

axial Mixing Model
Energy and Species Equations
Uniform pressure over jet
One

step reaction
Le = 1
Use eddy diffusivity for
D
Ambient gas oscillates
isentropically
.
P and T are collapsed to one
function of entropy.
Schvab

Zel’dovich
Variables
S

Z formulation plus
Oseen
approximation reduces three nonlinear
PDEs to one nonlinear PDE and two linear, homogeneous PDEs.
We may use Green’s function for two equations or numerical
integration for all three equations.
Frequency Response of Single Injector
•
Sinusoidal
pressure of
frequency
f .
•
Two characteristic combustion times
appear:
τ
M
for turbulent mixing,
τ
R
for
chemical reaction.
•
A time

lag results: the energy release
rate
E
lags the pressure
p
oscillation.
•
Reaction rate pre

exponental
factor is
varied from experimental
value
to
explore frequency response
for two
time ratios:
f
τ
M
and
f
τ
R
.
•
E
maximizes in a certain parameter
domain for the two time ratios.
•
The black line shows the path as
frequency varies for the given co

axial
injector design, chamber conditions,
and propellants.
CONFIGURATION:
Ten oxygen

methane co

axial injectors
are placed in a combustion
chamber. The model equation is
solved with a co

axial mixing and
reaction for the heat

release.
•
Chamber length
is
0.5m
and diameter is 0.28m .
•
Injector
outer diameter,
1.1cm
•
Injector
inner wall
0.898 cm
•
Inner flow of gaseous oxygen,
•
Outer flow of methane gas
•
“Short” multi

orifice nozzle
•
Steady

state
pressure is
200atm
,
temperature
is
2000K
First, an individual injector is examined subject to a prescribed
pressure oscillation; then, the analysis is made with ten injectors
c
oupled to the chamber wave dynamics.
Ten

i
njector
Simulation
An initial condition of a 1T mode with
sufficient amplitude results in triggering.
Below a threshold for initial amplitude,
decay and stability result. Above the
threshold, a stable limit cycle develops.
The frequency spectrum analysis shows that
n
onlinear resonance, in this case, produces
a
1T mode, a 2T mode , and a
sub

harmonic with frequency equal to
difference of 1T and 2T frequencies.
Sub

harmonics and Nearly Periodic Limit Cycles

A sub

harmonic mode often appears in nonlinear resonance
with a frequency equal to the sum or difference of integer multiples
of natural frequencies.

The
presence of the 1T, 2T, and sub

harmonic
modes prevents a
periodic behavior.. If one natural mode dominates, a nearly periodic
behavior results.

Linear theory does not predict the existence of harmonics for
circular cylinders or sub

harmonics for any chamber.

Galerkin
methods require the assumption of the modes present;
they do not predict the presence independently.
Left: disturbance decays
Right: growth occurs
w
ith new modes and
aperiodic behavior
Pure Second Tangential
(2T) Mode is the Initial
Condition
Triggering is possible.
Nonlinear resonance induces
fourth tangential mode plus
second harmonic of 2T.
No 1T or sub

harmonic.
Comparison of Variations in E and p

E is found to lag p in time t and position
θ
for
tangential spinning modes. Example with 2T mode
below.

Energy release is localized in the ten injector streams
but differences from one stream to another occur due to
pressure phasing.
Test with Variation
in Reaction Time

1T mode is initial condition.

Chemical kinetic constant is
changed causing longer reaction
time.

The black line in contour plot
adjusts and 2T position moves to
less sensitive region.

Consequently, nonlinear
resonance does not include 2T
and sub

harmonic.

Nonlinear resonance now
involves energy transfer to
harmonics of 1T mode.
Local Abrupt Disturbance
Local Gaussian

shaped
d
isturbance is provided
i
nitially. A direction of initial
motion
ϴ
and an initial
a
mplitude are chosen.
For
ϴ
< 10
o
, stability occurs.
For
10
o
<
ϴ
<
43
o
,
1T mode
t
riggering occurs.
For
ϴ
>
43
o
, 1T
mode
plus 1
st
Radial mode
triggering
occur with a
resulting larger
limit

cycle amplitude.
Conclusions from model

equation solution with
co

axial injection
, mixing and
reaction
•
A nonlinear acoustics model for transverse modes and a co

axial model
for propellant mixing and reaction are used to study combustion
dynamics with a ten

injector geometry.
•
Triggering of first tangential mode (1T) and the second tangential mode
(2T) is possible; both stable and unstable limit cycles are identified.
The stable limit behavior is not always exactly periodic.
•
Two characteristic combustion times are found and prove to be critical.
Accordingly, a time

lag in the combustion response is found
.
•
Depending on characteristic time and frequency values, nonlinear
resonance can transfer energy to second tangential mode (2T) and to a
sub

harmonic mode. Or energy can be transferred to higher harmonics.
The instability occurs for a frequency where the heat

release response
to pressure variation is very strong.
•
Triggering can result from either disturbances with well defined profiles
corresponding to natural modes or localized disturbances.
•
Directional travel orientation of a local disturbance is consequential for
triggering. Different stability and different limit cycles can be induced.
CURRENT WORK DIRECTIONS
UCI is developing the stochastic framework with the Polynomial
Chaos Expansion method and applying it to the transverse
instability model.
A longitudinal model with co

axial injection is being developed
to benchmark
against Purdue data.
Data results of transverse model are being delivered to
HyperComp
where Reduced Basis
Modelling
(RBM) will be
performed with the data.
A
multi

injector (more than ten) chamber is being configured that
will be used for 2D model (UCI)
,
LES (Georgia Tech), and
RBM (
HyPerComp
). Comparisons will be made.
Thank you.
Enter the password to open this PDF file:
File name:

File size:

Title:

Author:

Subject:

Keywords:

Creation Date:

Modification Date:

Creator:

PDF Producer:

PDF Version:

Page Count:

Preparing document for printing…
0%
Comments 0
Log in to post a comment