Lefebvre A.H., Ballal D.R. Gas Turbine Combustion: Alternative Fuels and Emissions
Подождите немного. Документ загружается.
290 Gas Turbine Combustion: Alternative Fuels and Emissions, Third Edition
88. Narasimhan, M. V., Sankaran Kutty, P., and Narayanaswamy, K., “Prediction
of the Air Core Diameter in Swirl Chamber Atomizers,” unpublished report,
1978.
89. Abramovich, G. N., Theory of Turbulent Jets, MIT Press, Cambridge, MA,
1963.
90. Reitz, R. D., and Bracco, F. V, “On the Dependence of Spray Angle and Other
Spray Parameters on Nozzle Design and Operating Conditions,” SAE Paper
790494, 1979.
91. Yokota, K., and Matsuoka, S., “An Experimental Study of Fuel Spray in a Diesel
Engine,” Transactions of the Japanese Society of Mechanical Engineers, Vol. 43, No.
373, pp. 3455–64, 1973.
92. Hiroyasu, H., and Arai, M., “Fuel Spray Penetration and Spray Angle in Diesel
Engines,” Transactions of the Japanese Society of Mechanical Engineers, Vol. 21, pp.
5–11, 1980.
93. Ohrn, T. R., Senser, D. W., and Lefebvre, A. H., “Geometrical Effects On Spray
Cone Angle for Plain-Orice Atomizers,” Atomization and Sprays, Vol. 1, No. 3,
pp. 253–68, 1991.
94. Taylor, G. I., “The Mechanics of Swirl Atomizers,” Seventh International Congress
of Applied Mechanics, Vol. 2, Pt. 1, pp. 280–5, 1948.
95. Rizk, N. K., and Lefebvre, A. H., “Prediction of Velocity Coefcient and Spray
Cone Angle for Simplex Swirl Atomizers,” Proceedings of the Third International
Conference on Liquid Atomization and Spray Systems, London, 11 lC/2/1–16,
1985.
96. De Corso, S. M., and Kemeny, G. Α., “Effect of Ambient and Fuel Pressure on
Nozzle Spray Angle,” Transactions of the American Society of Mechanical Engineers,
Vol. 79, No. 3, pp. 607–15, 1957.
97. Chen, S. K., Lefebvre, A. H., and Rollbuhler, J., “Factors Inuencing the Effective
Spray Cone Angle of Pressure-Swirl Atomizers,” Journal of Engineering for Gas
Turbines and Power, Vol. 114, pp. 97–103, 1992.
98. McVey, J. Β., Russell, S., and Kennedy, J. Β., “High-Resolution Patternator for the
Characterization of Fuel Sprays,” Journal of Propulsion and Power, Vol. 3, No. 3,
pp. 202–9, 1987.
99. McVey, J. Β., Kennedy, J. Β., and Russell, S., “Application of Advanced Diagnostics
to Airblast Injector Flows,” Journal of Engineering for Gas Turbines and Power, Vol.
Ill, No. 1, pp. 53–62, 1989.
100. Chen, S. K., Lefebvre, A. H., and Rollbuhler, J., “Factors Inuencing the
Circumferential Liquid Distribution from Pressure-Swirl Atomizers,” Journal of
Engineering for Gas Turbines and Power, Vol. 115, pp. 447–52, 1993.
101. Chen, S. Κ., Rollbuhler, J., and Lefebvre, A. H., “Inuence of Liquid Viscosity on
Pressure-Swirl Atomizer Performance,” Atomization and Sprays, Vol. 1, Νο. 1, pp.
1–22, 1991.
102. Tate, R. W., “Spray Patternation,” Journal of Industrial and Chemical Engineering,
Vol. 52, pp. 49–52, 1960.
103. Rosfjord, T. J., and Eckerle, W. Α., “Aerating Fuel Nozzle Design Inuences
on Airow Features,” Journal of Propulsion and Power, Vol. 7, No. 6, pp. 849–56,
1991.
104. Rosfjord, T. J., and Russel, S., “Nozzle Design and Manufacturing Inuences on
Fuel Spray Circumferential Uniformity,” Journal of Propulsion and Power, Vol. 5,
No. 2, pp. 144–50, 1989.
Fuel Injection 291
105. Wang, H. Y., McDonell, V. G., and Samuelsen, G. S., “The Two-Phase Flow
Downstream of a Production Engine Combustor Swirl Cup,” Twenty-Fourth
Symposium (International) on Combustion, pp. 1457–63, The Combustion
Institute, Pittsburgh, PA, 1992.
106. Wang, H. Y., McDonell, V. G., and Samuelsen, G. S., “Inuence of Hardware
Design on the Flow Field Structures and the Patterns of Droplet Dispersion: Part
I—Mean Quantities,” Journal of Engineering for Gas Turbines and Power, Vol. 117,
pp. 282–9, 1995.
293
7
Combustion Noise
7.1 Introduction
Combustion in a owing turbulent fuel–air mixture is always accompanied
by noise. In a gas turbine, the total noise resulting from normal combus-
tion is often referred to as core noise. It comprises two components, (1) direct
combustion noise that is generated solely by the combustion process itself,
and (2) indirect combustion noise that is produced by the ow of hot combus-
tion products through the turbine and exhaust nozzle [1]. In this chapter, we
shall concentrate on direct combustion noise and also on the more insidious
forms of noise that arise when combustion instabilities become coupled to
acoustic modes in a combustion chamber.
Every combustor has natural frequencies that may be excited by the com-
bustion process to produce oscillations, and the gas turbine combustor is no
exception. The oscillations may be longitudinal, radial, or circumferential, or
a combination of these modes. In the longitudinal, or so-called “organ pipe”
mode, the resonant pressure waves are along the length of the combustor,
whereas in the circumferential or “sloshing” mode, the waves are tangential
to the main ow direction.
The topic of combustion instability has gured fairly prominently in the com-
bustion literature, but usually in the context of rocket engines and jet-engine
afterburners. Comparatively little has been published on acoustic oscillations
in main combustors. Usually, the problem only arises during full-scale combus-
tor testing, and sometimes instabilities do not show up until the combustor is
tted to an engine. Typically, the problem has been to eliminate or control an
instability in a combustor at a comparatively late stage in its development, when
relatively few changes can be made except at large expense and loss of time.
In recent years, the problem of sustained oscillating combustion, which
gives rise to noise and engine vibrations, has become of increasing concern.
The main reason for this is the continuing trend toward higher degrees of
fuel–air premixing prior to combustion. The motivation is to further reduce
pollutant emissions, in particular oxides of nitrogen, which can only be done
by improving the mixedness of the fuel–air mixture entering the combustion
zone. Unfortunately, developments in premixed combustion are generally
294 Gas Turbine Combustion: Alternative Fuels and Emissions, Third Edition
accompanied by an increase in the occurrence of oscillating combustion. This
has led to a resurgence of interest in all aspects of combustion noise, includ-
ing its cause, the mechanisms involved in noise generation and suppression,
methods of noise control, both passive and active, and computational models
to provide guidance for developing control strategies. These topics form the
basis of the material presented in this chapter.
7.2 Direct Combustion Noise
Direct combustion noise arises when a volume of gas expands at constant
pressure as it is heated by combustion. The resulting expansion of the sur-
rounding gas produces a sound wave that propagates outside the boundary
of the ame. The pressure in the sound wave, and hence the sound intensity,
depends on the rate of volume generation by the source. Such sources are
described as acoustic monopoles. Since the size distribution of the eddies in
a turbulent ow is governed by the statistical distribution of the turbulent
mixing lengths, it follows that the sound generation in a highly turbulent
ame is equivalent to a statistical distribution of monopole sources through-
out the combustion zone [2,3].
Two parameters are of prime importance in the description of direct com-
bustion noise. One is the radiated sound power and the other is the thermoacoustic
efciency, which is dened as the ratio of the radiated sound power to the heat
released in combustion. Due to the destructive interference that occurs within
the source region, the actual acoustic power emitted in gas turbine combus-
tion tends to be only a minute fraction of the total thermal power [2,3].
The radiated sound power covers a broad spectrum of frequencies from
around 100 to 2000 Hz, reaching a blunt peak between 300 and 500 Hz. This
general shape of the sound power/frequency curve appears to be largely
independent of combustor size, engine power, and ame temperature,
although these factors strongly affect the radiated sound power level [1].
The most important single factor governing combustion noise is engine
power. This is because both the mass ow rate through the combustor and
the temperature level in the combustor increase with engine power. If the
thermoacoustic efciency remained constant, the radiated sound power
would be proportional to the power developed by the engine. In practice,
sound levels are higher than this simple relationship would predict because
the thermoacoustic efciency also increases with engine power [1].
7.2.1 Theory
In an early study by Bragg [2], the turbulent ame zone is assumed to com-
prise a region of uncorrelated amelets that produce monopole-type sound
Combustion Noise 295
on burning. His model shows that the radiated sound power should vary
with the fuel reactivity and is proportional to the square of the mixture ow
velocity. It also predicts a peak frequency in the region of 500 Hz and a ther-
moacoustic efciency of around 10
−6
for a typical hydrocarbon fuel.
The results obtained by Thomas and Williams [3] show that the normal
burning velocity of a fuel has a substantial inuence on both sound power
and thermoacoustic efciency. For example, it was found that increasing the
burning velocity from 50 to 100 cm/s raised the thermoacoustic efciency
from 10
−6
to 10
−5
. These workers suggest that for practical ames, the ther-
moacoustic efciency will be lower than these values and will probably lie
in the range from 10
−8
to 10
−7
. This result would appear to have important
implications for combustion noise in gas turbines.
As described by Ballal and Lefebvre [4,5], under conditions of high turbu-
lence the combustion zone comprises a fairly thick region of burned gases
interspersed with multitudinous small eddies of unburned mixture. Within
each eddy, the burning velocity is greatly enhanced by the ow of heat and
active species into the unburned mixture from the enveloping ame front.
Sometimes, the acceleration of chemical reactions and ame speeds within
an individual eddy may proceed to such an extent that combustion occurs
almost instantaneously throughout its volume. It would, therefore, be a con-
siderable oversimplication to regard the combustion process in a highly tur-
bulent primary zone, where turbulent intensities can exceed 30%, as simply a
collection of assorted amelets in which the rate of expansion is determined
primarily by the normal burning velocity of the unburned mixture. Thus, in
gas turbine combustion, the thermoacoustic efciency may be appreciably
higher than the values indicated by Thomas and Williams. However, normal
burning velocity may still provide a useful yardstick for the relative noise
levels emitted by the turbulent combustion of different fuels.
In their review of theoretical work relating to direct combustion noise,
Mahan and Karchmer [1] note that several parameters are common to nearly
all theoretical developments: the total mass ow rate through the burner, the
burner length and cross-sectional area, the fuel/air ratio, and some measure
of the fuel reactivity. Although some theories present their ndings in terms
of burner pressure drop and/or temperature rise, these terms are directly
relatable to the above parameters.
Among the most highly developed combustion noise theories is that of
Strahle [6]. In its most practical form, this theory predicts that the sound
power radiated by a can-type combustor is given by
SPUT qN ALα
33
23
20115 1
ref
2
to
to
e
()
().
,
−−
−−
( 7.1)
where S is the sound power in watts, P
3
and T
3
are the combustor inlet pres-
sure and temperature, respectively, U
ref
is the combustor reference velocity,
q is the fuel/air ratio, N is the number of fuel injectors, A
e
is the combustor
296 Gas Turbine Combustion: Alternative Fuels and Emissions, Third Edition
outlet area, and L is the combustor length. It is of interest to compare this
theoretical prediction of sound power with the following equation in which
the various exponents denote the experimental values obtained by Strahle
and Muthukrishnan [7].
SCPUTqNAL=
−− −
3
19
3
25 13 078151.....
.
ref
3.4
e
( 7. 2)
The constant C has the value of 0.047 if the various quantities in Equation 7.2
are expressed in S.I. units.
Comparison of Equations 7.1 and 7.2 shows reasonably good agreement
between theory and experiment, except for P
3
and U
ref
. The higher exponents
for these two quantities in Equation 7.2 are attributed in Reference [7] to the
intervention of jet noise in the experiments. Since jet noise varies as velocity
to the eighth power, it is generally found that although theory tends to pre-
dict a velocity exponent of around 2, the experimentally determined values
are usually close to 3 [8–11].
7.2.2 Core Noise Prediction Methods
The General Electric (GE) Company has used combustion noise theory,
supplemented by engine data, to derive an equation that allows the overall
sound power level (direct plus indirect combustion noise) to be calculated
from a knowledge of just a few combustor and turbine operating parameters
[12]. This equation was found to predict the core noise levels of a number of
turboshaft, turbojet, and turbofan engines to an accuracy of within 5 dB. Of
special interest in the GE equation is that it identies combustion pressure
and combustor temperature rise, along with mass ow rate, as the key fac-
tors governing direct combustion noise.
The Pratt and Whitney Company has made extensive use of combus-
tion noise theory to develop a prediction method that yields both the over-
all sound power level and the peak frequency [13]. Since it employs more
parameters than the GE expression, including several operating and geo-
metrical variables, it has a potentially broader range of application. The
method has been applied successfully to several engine types, including
the JT8, JT8D, and JT9D. The data presented in Reference [13], which cover
wide ranges of operating conditions for these engines, show close agree-
ment between the predicted and measured sound levels, with a standard
deviation of 1.7 dB.
A full description of the GE and P&W noise prediction methods is beyond
the scope of this chapter. For further information, reference should be made
to the original publications [12,13] and to the review article on core noise by
Mahan and Karchmer [1].
Combustion Noise 297
7.3 Combustion Instabilities
Combustion instabilities are usually the result of interactions between the
combustion process and the acoustic elds within the combustor. Some com-
bustion instabilities are caused by uctuations in the air supply to the com-
bustor, others by aerodynamic disturbances created within the combustor
itself, and the remainder are due either to variations in the fuel supply to the
nozzles or to maldistributions of fuel in the combustion zone, which give rise
to a cycle of extinction and reignition in localized regions of the ame.
If, for any reason, the combustion heat release process is periodic, the
resulting acoustic pressure waves emanating from the combustion zone are
periodic with the same frequency. The presence of a combustor liner causes
these pressure waves to be returned to the combustion zone with a time delay
that depends on the chamber shape and size and the average speed of sound
in the combustor. Energy is added to the pressure waves at any frequency
for which the instantaneous peak in acoustic pressure in the combustion
zone coincides with the instantaneous peak in heat release. This coupling
between the combustion process and the acoustic eld causes energy to be
added to the system with each cycle, which results in oscillations that grow
in amplitude until dissipative viscous losses arrest further growth. This situ-
ation conforms to the well-known Rayleigh criterion for combustion oscilla-
tions [14]. In a similar manner, energy is removed from a pressure wave at
any frequency for which the wave is 180° out of phase with the periodic heat
release in the ame zone [1].
7.3.1 Descriptions of Acoustic Oscillations
Various terms are used to describe the sound emitted from a combustor
when combustion instabilities are present. There are no general guidelines
laid down; for example, in some engine companies, the term “rumble” is
used to describe all audible acoustic oscillations, regardless of their fre-
quency, whereas in others, the description “rumble” or “growl” is reserved
for the noise emitted in the low frequency range from 50 to 180 Hz, which
usually occurs at sub-idle conditions. For the higher frequencies associated
with engine speeds around and above idle, the term “howl” or “humming”
is generally considered to be more appropriate.
7.3.1.1 Growl
The characteristics of growl vary from engine to engine and between engine
startups, but its onset may occur soon after ignition is accomplished and it
may persist at engine speeds up to idle. Growl is considered to be undesir-
able because it lengthens engine startup time and reduces compressor stall
298 Gas Turbine Combustion: Alternative Fuels and Emissions, Third Edition
margins. According to Seto [15], some compressors are quite tolerant to growl,
whereas others appear to have growl-related stall problems.
Certain engine operating parameters affect growl. An increase in com-
bustor inlet air temperature decreases the speed range and intensity of
growl, whereas an increase in combustion pressure has the opposite effect
of promoting growl [15].
Methods of alleviating growl include improvements in primary-zone ow
patterns and structure, and modications to the fuel-injection system to
raise the fuel delivery pressure. As growl is normally most prevalent when
the primary-zone fuel/air ratio is near the weak extinction limit, any change
to the acceleration schedule that raises the fuel/air ratio in the primary zone,
or any change in fuel spray characteristics that lowers the lean blowout limit,
such as a reduction in spray cone angle, will tend to suppress growl.
7.3.1.2 Howl
The phenomenon of howl or humming is closely related to growl, but it
occurs at higher engine speeds. Its frequency is usually in the range from 200
to 500 Hz. As with growl, its intensity is dependent on ambient air tempera-
ture and falls off rapidly as engine inlet air temperatures rise above normal
atmospheric values (around 288 K). It is sensitive to fuel type and diminishes
in severity with an increase in fuel volatility.
With growl, engine compressor instabilities play an important role and
may even be the trigger for growl, but with howl, the compressor is much
less aerodynamically involved. The primary cause of howl appears to be
from fuel pressure perturbations. Isolating the feedback mechanism tends
to eliminate howl.
7.3.2 Characteristic Times
The heat release in a gas turbine combustor does not take place immediately.
The fuel–air mixture travels at least part way around the recirculation zone
in the dome region of the combustor before releasing the major portion of its
heat [16]. The time that elapses between the injection of fuel and the region of
maximum heat release represents the characteristic combustion time, which
is obtained as the sum of the characteristic times for fuel evaporation, mixing
of fuel vapor with air and combustion products to reach a critical reaction
temperature, and chemical reaction.
tttt
combustion mixing evaporationreaction
=+ + .
( 7. 3)
The system becomes prone to instabilities when the overall combustion time
becomes equal to a characteristic acoustic time of the combustor.
Very detailed studies would be required to quantify time scales in Equation
7.3 for specic gas turbine applications but, even without this information,
Combustion Noise 299
Equation 7.3 does allow some qualitative assessments to be made of the
impact on combustion oscillations of changes in various relevant fuel and
combustion parameters.
7.3.3 influence of Fuel Type
The arguments presented above suggest that a change in fuel type or com-
position may either increase or reduce noise amplitudes, depending on
whether the resulting change in chemical reaction time moves the overall
combustion time closer or further away from the characteristic acoustic time.
Tests carried out by Janus et al. [17] on a subscale combustor burning varying
proportions of natural gas, propane, and hydrogen showed that instability
regimes are markedly affected by changes in fuel composition. Keller et al.
[18] also observed changes in both oscillation frequency and amplitude with
change in gaseous fuel composition. As mixing times are largely unaffected
by variations in fuel chemistry, these ndings are clearly due to the change
in chemical reaction time resulting from the change in fuel composition.
With liquid fuels, the inuence of evaporation time must also be consid-
ered in addition to reaction time. The amount of available information on
this effect is not large, but Vandsburger et al. [19] observed that the oscillat-
ing behavior of step-stabilized ames was appreciably different for various
combinations of spray vs. gaseous combustion, whereas Mehta et al. [20] have
demonstrated the inuence of fuel volatility on oscillating combustion in aero
engines. If the difference in evaporation time between one fuel and another is
signicant (say, more than half the acoustic period), then the overall combus-
tion time of the two fuels will also be quite different. In consequence, with
one fuel the coupling between the combustion and acoustic elds may be
weak or nonexistent, resulting in relatively noise-free combustion, whereas
the other fuel may produce strong coupling and large acoustic amplitudes.
7.3.4 influence of Combustor Operating Conditions
Janus et al. [17] have conducted a number of experiments to determine the
inuence of inlet air temperature on combustion stability characteristics.
Their tests were conducted on a subscale combustor burning a variety of
gaseous fuels. The results of this study are presented in the form of stabil-
ity maps. Examination of these maps indicates that the combustor is highly
unstable over a wide operating range at 273 Κ inlet temperature, but relatively
stable over the same operating range at 394 Κ inlet temperature. This result
can readily be explained by reference to Equation 7.3. An increase in inlet air
temperature decreases the chemical reaction portion of the total combustion
time. The new combustion time represents a condition at which acoustic loss
exceeds acoustic gain. It is of interest to note on the stability maps presented
by Janus et al. [17] that the instability region has not disappeared. It has sim-
ply moved to a new location on the map where one or more components of