Anda di halaman 1dari 10

R=20050203867 2019-06-26T00:13:55+00:00Z

On the Interaction of a Premixed etc. Combustion instabilities are a

Flame with an Acoustic Disturbance concern in many engineering devices as
they lead to malfunctioning of the device
and a reduction in its efficiency. The
sources of these instabilities are diverse
Caroline Hood and Abdelkader Frendi’ and are dependent on fuekoxygen
University of Alabama in Huntsville mixture, chamber geometry, injection
Huntsville, AL, 35806 mechanism, etc. It is believed that,
combustion instabilities occur due to
sustained pressure fluctuations of
acoustic nature in a chamber where
combustion takes place. These
Abstract fluctuations alter the flame, which in
turn drives sound through unsteady heat
The main objective of this research is to release.
analyze the effect of acoustic It is well known that the
disturbances on a premixed flame and resonance between a fluctuation in the
determine their role in the onset of heat release rate and an acoustic mode of
combustion instabilities. Computations the chamber is one of the main reasons
for the one-dimensional, unsteady for combustion instabilities. Prasad [I]
combustion of a lean premixed methane- investigated the effect of an external
air mixture are performed. An acoustic pressure perturbation on “premixed
excitation is introduced in the chamber flames” using detailed chemistry. He
and interacts with the flame front. Our solved the one-dimensional, time -
results indicate that as the amplitude of dependent, nonlinear conservation
the acoustic excitation is increased, the equations. His results showed that in
flame front position fluctuates rapidly. response to the externally imposed
This phenomenon is even more intense pressure waves, the flame moves to a
when the frequency of the acoustic location where heat release and heat loss
disturbance matches the fundamental are in balance.
frequency of the chamber. Our results Peters and Ludford [2]
suggest that the interactions between the concentrated on small variations of
flame and the acoustic excitation may pressure and found that it was not
result in flame extinguishment. In affecting the flame significantly and
addition various passive control devices hence concluded that substantially large
are tested and we found that the variation in pressure (of the order of
Helmholtz resonator with rounded inlet ambient pressure) might be needed to
corners is the most efficient. quench the flame. On the other hand,
Ledder and Kapila [3] studied the
Introduction response of a premixed laminar flame to
pressure perturbations with emphasis on
Combustion instabilities arise flame extinction. It was found that a very
in many practical devices, such as rocket small variation in dynamic pressure has
engines, aircraft engines, gas turbines, an influence on the burning rate, which
could eventually lead to flame
Corresponding author, Associate Professor, extinction.
Senior Member of AIAA.

This material is a work of the U S . Government and is not subject to copyright pmtedion in the United Stater
Due to the importance of The work performed describes the
combustion instabilities for different interaction of an acoustic excitation with
combustors, many studies have a premixed laminar flame in a closed,
addressed the modeling and control of one-dimensional chamber. Depending on
combustion instabilities in lean premixed the excitation frequency and amplitude,
mixtures. Paxson [4] modeled this interaction leads to the growth of the
combustion instabilities in an LPP pressure fluctuations in the system. At
combustor using a sectored one- substantially high amplitude, the growth
dimensional code and studied the effect becomes nonlinear and results in flame
of an abrupt change of area in the extinction.
combustor rig and how it expedites the
phenomena of combustion instabilities Mathematical Formulation
using single-step kinetics. The author
concludes that the position at which fuel The Mathematical model consist
is injected has a significant effect on of the Nuvier-Stokes equations combined
acoustic instabilities. with a one-step irreversible exothermic
Frendi [5] concentrated on the chemical reaction of the form
coupling between the unsteady heat
release rate from combustion and the v,F +v,O +v,P (1)
acoustic energy of the combustion
chamber. Furthermore, he also studied where “F” is the fuel, “0”the oxygen
the effect of adding a Helmholtz and “P” the products with vi the
resonator and/or baffles to the system.
He found that the use of a Helmholtz corresponding stoichiometric
coefficients. All species are assumed to
resonator and/or baffles shifts the
frequency of the chamber away from the be in the gas phase.
unstable mode.
In addition to the above studies, The full set of conservation
several other studies on different aspects equations in one dimension are:
of combustion instabilities have been
carried out [6-141. The basis for all the aw +-=
- aF
previous studies is due to Lord Rayleigh at ax
whose criteria states that the acoustic
waves in the chamber will amplifL if the where,
acoustic pressure and heat release rate
are “in phase”.
In the present study, we
concentrate on excitation of acoustic
oscillations in combustion chambers, as W=
it is an undesired phenomenon which
results in inefficient burning and can
lead to possible engine failure. However,
accurate predictions of these types of
acoustic instabilities are difficult to
make as it primarily depends on the
geometry of the combustor in question.
The source terms in the fuel and oxygen
pu species conservation equations are


The equations are solved in a

nondimensional form using the
(4) following reference variables:

In addition, the equation of state is given

p=pT7T Method of Solution

with p being the pressure, p the mixture

density, T the temperature, R the The set of PDE's are solved
universal gas constant and a the using an explicit finite difference
scheme. The scheme, which is a
mixture molecular weight.
generalization of the original
MacCormack scheme obtained by
The source terms in the energy equations
Gottlieb and Turkel [15], is fourth order
are given by
accurate in space and second order
accurate in time. The numerical scheme
applied to a one-dimensional equation of
the form

at ax '

(7) consists of a predictor step given by

ui**= U,"+ where, ii is the nondimensional

($1- 74 + 84+1- 4+2)

velocity at the inflow boundary, f the
frequency, E the amplitude and @ the
(12) The spark ignition parameters
used for the present system are
followed by a corrector step of the form
rIg= 0.004mI z,, = 2 . 7 5 ~ and
E: = 2.95. 1 0 " J / m 3 s . The domain

q n + l =-
[[g7F,'*+ 4-2.')
- S F , - , .b

In the above equations, the

I length is 0.2 m. Gas properties for a lean
mixture are different. We obtain the
properties for the lean side from Coffee
and Kotlar [17] and values
corresponding to (I= 0.662 are given in
Table 1. Two different cases are
simulated based on changing frequency,
subscript i denotes the spatial grid point a resonant case, f = 1940Hz, and an
and the superscript n denotes the time off-resonance case, f = 1730Hz.
level. Superscript (* * ) corresponds to
an intermediate time level (between n
and ( n + 1)). The fourth order accuracy 1) Resonant Case
is obtained by alternating the scheme By changing the amplitude,&, three
given above with its symmetric variant. cases were studied, with the use of the
Furthermore, Jameson's [161 artificial
resonant frequency f = 194OHz
dissipation terms were explicitly added
to the finite difference scheme, which
captured sudden jumps in pressure and (i) E = 0 - No Excitation
other variables near shock waves and (ii) E = 0.03 - Low Excitation
stagnation points. Adding these terms (iii) E = 0.3 - High Excitation
helped prevent the appearance of the
well-known Gibbs phenomenon in Several observations have been made
regions containing severe pressure from the following figures. Figures 1, 2,
gradients. Both the inflow and outflow 3 and show the temperature profiles at
boundaries were rigid, u = 0 . different times for E = 0 , E = 0.03
and E = 0.3, respectively. In Figure 1 we
Results and Discussions observe that the temperature distribution
is symmetrical and the flame fronts are
Consider a lean mixture of traveling towards both boundaries at
equivalence ratio @ = 0.662 in a closed equal speed and the temperature of the
chamber. Applying an external acoustic flame front is settling down to the
excitation at the inflow boundary of the adiabatic temperature of the lean
form mixture, which is Tb = 2142.2K. Unlike
the profile in Figure 1, the profiles in
Figures 2 and 3 are different. The
ii = E sin( 2@t+ @) acoustic excitation results in increasing
the speed of the positive propagating
flame and slowing the negative 3) Passive Control
propagating flame. Pressure time traces
for different excitation amplitudes are Several passive control devices were
shown in Figure 4, which shows the tested numerically using a resonant
growth of the pressure oscillations with acoustic disturbance alone, without
time and by increasing the amplitude. combustion. The devices tested were, a
baffle, a quatenvave and a Helmholtz
2) Off-Resonance Case resonator. In the later two cases, both
smooth and sharp corners were
By changing the amplitude, E , three considered. Figure 10 shows a
cases were simulated, with the use of comparison of the OASPL measured in
off-resonance frequency of f = 1730Hz the chamber the controlled and
uncontrolled cases. It is clear from the
(i) E = 0 - No Excitation figure that the Helmholtz resonator
(ii) E = 1.2 - Low Excitation results in better control. Figure 11 shows
(iii) E = 2.4 - High Excitation the effect of rounding the various
corners of the resonator. Our results
Figures 5, 6 and 7 show the show that rounding the inlet corners
profiles of temperature at different alone gives the best results. Figure 12
excitation amplitudes and it can be shows that the Helmholtz resonator is
observed that the effect on the flame less effective at high sound pressure
front location as the amplitude is levels.
increased is very significant. Flame
fronts are observed to move with
unequal speeds towards both boundaries. Concluding Remarks
Increasing the excitation amplitude, the
inflow flame front is moving with a low Based on the results presented in this
speed. paper, the following conclusions are
Figures 8 compare the negative made:
and positive propagating flame radius 1. The effect of the acoustic
time histories for E = 2.4 with E = 0 . By excitation is to displace the flame
increasing the amplitude ( E ) , the flame front. Depending upon the
location fluctuates significantly which amplitude and frequency of the
could lead to possible extinction. excitation, the flame front
Figure 9 shows the comparison oscillations can be severe and
of the time traces of the pressure at the may lead to flame extinction.
center of the chamber at different times 2. When the frequency of the
for the same amplitude ( E = 1.2) but acoustic excitation is at
different frequencies ( f = 1730Hz and resonance with the fundamental
frequency of the chamber, we
f = 1940Hz ). It is observed that found that the combustion
pressure is growing much faster in the acoustic interactions in the
chamber when resonance frequency is chamber resulted in rapid growth
imposed compare to the off-resonance of the pressure oscillations. It can
frequency case. also be said that these nonlinear
interaction between the heat
release fiom the flame and the
acoustic excitation may lead to a [6] Wu, X., Wang, M., and Moin, P.,
possible flame extinguishment. “Combustion instability due to the
3. The Helmholtz resonator with nonlinear interaction between sound and
rounded inlet corners is the best flame,” Centerfor Turbulence Research,
passive control device tested. Annual Research Briefs 200 1.

[7] Matalon, M., and Matkowsky, B.J.,

Acknow ledprernents “Flames as gas dynamics
discontinuities,” J. Fluid Mech. (1983),
Support for this research was VOI. 124, pp. 239-259.
provided by a grant from NASA
Marshall Space Flight Center, Grant [8] Landau, L.D., “On the theory of slow
number NCC8-200, with Tom Nesman combustion,” 1944, Acta
as the technical monitor. Physicochimica URSS 19,77.

References [9] Umurhan, O.M., “Exploration of

fundamental matters of acoustic
instabilities in combustion chambers”
[l] Prasad, K., “Interaction of Pressure Center for Turbulence Research, 1999.
Perturbations with Premixed Flames,”
Combustion and Flame, Vol. 97, 1994, [lo] Lee, D.H., and Lieuwen, T.C.,
pp. 173-200. “Premixed Flame Kinematics in a
Longitudinal Acoustic Field”, A I M - 0 1 -
[2] Peters, N. and Ludford, G.S.S., “The 385 1, Joint Propulsion Conference,
effect of pressure variations on premixed 200 1.
flames,” Combustion Science
Technology, 34,33 1. [ 113 Paschereit, C.O., Flohr, P.,
Gutmark, E., and Bockholts, M.,
[3] Ledder, G., and Kapila, A.K., “The “Numerically and Experimental
Response of Premixed Flames to Analysis of Acoustically Excited
Pressure Perturbations,” Combustion Combustion Instabilities in Gas
Science and Technology, 1991, Vo1.76, Turbines” AIAA 2000-2076,
pp. 2 1-44. Aeroacoustics Conference, June 2000.

[4] Paxson, D., “A Sectored One- [12] Culick, F.E.C., “A Note on

Dimensional Model for Simulating Rayleigh’s Criterion,” Combustion and
Combustion Instabilities in Premix Science Technology, Vol. 56, 1987, pp.
Combustors,” Glenn Research Center, 159- 166.
Cleveland, Ohio, 2000, AIAA-2000-
0313. [13] Nguyen, Q. V., “Measurements of
Equivalence Ratio Fluctuations in a
[5] Frendi, A., “On the role of Acoustic Lean Premixed Prevaporized (LPP)
Coupling on Combustion Instabilities,” Combustor and its correlation to
AIAA-2003-3181, Hilton Head, South Combustion Instability” National
Carolina, 2003.
Aeronautics and Space Administration,
Cleveland, OH. Figure 1: Temperature profiles in a “Closed
Boundary System” as a function of time for
[ 141 Harper, J., Johnson, C., Neumeier, a lean (@= 0.662) flame with no excitation
Y., Lieuwen, T., and Zinn, B.T., ( & = O ) , f =1940Hz
“Experimental Investigation of the
Nonlinear Flame Response to Flow
Disturbance in a Gas Turbine
Combustor,” A I A - 0 1-0486.

[l5] Gottlieb, D., and Turkel, E.,

“Dissipative Two-Four Methods for
Time-Dependent Problems,” 1976,
Mathematics of Computations, Vol. 30,
NO. 136,703-723.

[16] Jameson, A., Schmidt, W., and

Turkel, E., “Numerical Solutions of the
Euler Equations by Finite Volume
Methods Using Runga-Kutta Time-
Stepping Schemes,” 198 1, A M - 8 1 -

[17] Coffee, T.P., Kotlar, A.J., and Figure 2: Temperature profiles in a “Closed
Miller, M.S., “The Overall Reaction Boundary System” as a function of time for
Concept in Premixed, Laminar, Steady a lean ( @= 0.662) flame with low excitation
State Flames I. Stoichiometries,”
( & = 0 . 0 3 ) , f = 1940Hz
Combustion and Flame 54:155- 169
(1 983).

Figure 3: Temperature profiles in a

“Closed Boundary System” as a function
of time for a lean (4 = 0.662) flame with a lean ($I= 0.662) mixture with no
low excitation ( E = 0.3), f = 1940Hz excitation ( E = 0 ), f = 1730Hz .
Tim (a)
- - - O.oo0968
-0 . m

ta I . , I * I # * II I I I I . , , . I
0 0.05 0.1 0.15 0.2
DiSanee (m)

Figure 6: Temperature profiles in a “Closed

Boundary System” as a function of time for
a lean ( $ = 0.662) mixture with low
Figure 4: Time traces of pressure at the excitation ( E = 1.2),f = 173OHz.
center of the chamber with different
excitations fx a lean mixture in a closed
chamber (a) E = 0 , (b) E = 0.03 , (c) E = 0.3
a00 r
and (d) E = 1.2, f = 194OHz

l hM
--..-..-- - - 0.000968
-0.0115 aom

f 1750
En 1500

g 1250 t I I I t I 1 I 1 I I I I 1 I I I I , I I
1000 0 0.05 0.1 0.15 0.2
Distance (m)


250 7 7 I-FT-1 Figure 7: Temperature profiles in a “Closed

0 0.05 0.1 0.15 0.2 Boundary System” as a function of time for
Disenee(@ a lean ( $ = 0.662) mixture with high
excitation ( E = 2.4), f = 1730Hz .
Figure 5: Temperature profiles in a “Closed
Boundary System” as a function of time for
r .

180 -
N.C. :No Cortrd
B :Baffle
Q.W.: aumtaw.n
H.R. : Helmblh R e w m b r

J 160

' ' ' ' '

' ' ' ' '
' ' ' '
o h Figure 10: Comparisonofoverall sound
Tine (s) pressure level inthe chamber obtained
with the various passive devices.
Figure 8: Flame trajectories as a function of
time for ( E = 2.4) and
( E = O ) , @ = 0.662, f = 1730Hz.
180 -
N.C. : No Control
Sh.C. :Sharp Cornets

Sm.l.C. :Smooth I n l e t C o m r

Sm.T.C. :Smooth Thmat Comers

b -
A.C.Sm. : A # Comers Smooth
j 160

140 I

Figure 11: Effect of smoothing the

resonator corners on the overall sound
pressure level.

Figure 9: Time traces of pressure at the

center of the chamber with different
frequencies for a lean mixture in closed
chamber (a) f = 173OHz, (b) f = 1940Hz
for &=1.2, @ = 0 . 6 6 2 .
110 - N.C. :No Cottrd
Sh.C.: Sharp Cornera
Sm.C.: Smoo(h Cornua

I -
J 160


Figure 12: Effect of a Helmholtz

resonator on the overall sound pressure
level of a combustion chamber at high
noise levels.