Lefebvre A.H., Ballal D.R. Gas Turbine Combustion: Alternative Fuels and Emissions

Подождите немного. Документ загружается.

300 Gas Turbine Combustion: Alternative Fuels and Emissions, Third Edition

the total combustion time have increased enough to offset the decrease in

reaction time caused by the increase in inlet air temperature.

These considerations suggest that an increase in combustion pressure,

which also enhances chemical reaction rates, should again serve to alleviate

acoustic oscillations. However, this effect tends to be more than offset by

the corresponding increase in combustion energy that sustains the oscilla-

tions (note that for a constant fuel/air ratio, the combustor heat release rate is

directly proportional to pressure), so the net result is that combustion noise

usually increases with an increase in pressure [15].

From the above discussion, it would appear that any change that reduces

the overall combustion time, such as an increase in fuel volatility that reduces

the evaporation time, or an increase in inlet air temperature that reduces the

reaction time, will tend to alleviate instabilities, and more often than not

this is the case. However, this should not be regarded as a general result,

and many exceptions have been observed in laboratory tests and on engine

hardware. Much depends on the frequency of the oscillation. For the excep-

tionally high frequencies associated with small combustors, an increase in

inlet air temperature tends to promote instabilities. As the key factor is the

ratio of the overall combustion time to the relevant acoustic time, it is hard to

generalize the effect of inlet air temperature or, in fact, any other operational

parameter on combustion oscillations. Any change in the characteristic com-

bustion time could possibly move the combustor from a stable to an unstable

region, and vice versa [21].

7.3.5 influence of Ambient Conditions

The laboratory tests carried out by Janus et al. [17] generally conrmed the

benecial effect of an increase in ambient air temperature in suppressing

noise and also showed that increased ambient humidity decreases combus-

tor pressure oscillations. This latter effect was attributed to the additional

heat capacity of the water molecules, which lowers the peak ame tempera-

ture, thereby reducing the reaction rate. However, the main conclusion to be

drawn from their investigation is that gas turbine manufacturers and users

need to be aware of the effects of ambient conditions (and fuel composition)

on combustion oscillations. According to these workers, signicant variations

in combustor stability performance could occur due to changes in geographic

climate and seasonal weather conditions.

7.3.6 Aerodynamic instabilities

Experimental and numerical studies on combustor aerodynamics have

revealed that the shear layers created by counter-rotating swirling airows

and the breakup regions of air jets are characterized by the presence of

coherent large-scale structures. These structures play an important role in

promoting the high mixing rates that are an essential prerequisite for high

Combustion Noise 301

volumetric heat release rates. However, they can also give rise to unsteady

heat release that could, in turn, induce combustion instability. An important

factor in this instability mechanism is the time delay between the formation

of a coherent vortex structure and the instant that energy is released due to

combustion in the vortex. This delay may provide the proper phase relation-

ship between the oscillating pressure eld and unsteady heat release to drive

the instability [22].

Another, and more prevalent cause of pressure oscillations in gas turbine

combustors, is when the inherent aerodynamic stability of the combustor

is too low. The primary-zone airow pattern is of major importance to both

ame and aerodynamic stability. Many different types of airow patterns

are employed in gas turbine combustors, but one feature common to all is

the creation at the upstream end of the liner of a toroidal ow reversal that

entrains and recirculates a portion of the hot combustion products to mix

with the incoming air and fuel. These vortices are continually replenished

by air owing through the dome swirler and holes pierced in the liner walls.

Additional air is supplied through are-cooling slots and from air employed

in airblast atomization. A satisfactory airow pattern can only be achieved

by good matching of these various modes of air admission. In particular,

it is important to ensure that the large-scale ow reversals induced by the

swirler air and the primary air jets merge and blend in such a manner that

each one complements the other to produce an aerodynamically strong and

stable recirculation zone. Any signicant mismatch between these two main

sources of primary air can lead to the creation of localized regions within

the primary zone in which the ow is sluggish and disorganized. Rates of

combustion and heat release in such regions are highly susceptible to pertur-

bations in ow and pressure.

Even when a strong and stable recirculation zone has been achieved,

imperfections in fuel spray patternation and/or airow distribution can lead

to the formation of local regions in which the fuel/air ratio is appreciably

lower than the average value. Changes in engine power setting may cause

ame extinction to occur in these regions, but the combustor continues to

function because elsewhere in the primary zone the fuel/air ratio lies inside

the normal stability limits. However, cessation of combustion in part of the

primary zone may cause the recirculating airow to redistribute itself in

such a way that mixture strengths in these local regions now fall within the

normal burning range and reignition occurs. Thus, the conditions for local

ame extinctions are restored, and a cycle of ame extinction and reignition

in local pockets of mixture is established. If the pressure pulses generated by

this sequence of events become coupled with the combustor’s acoustic eld,

the oscillations are strengthened and sustained.

An example that illustrates the importance of a stable primary airow pat-

tern to the attainment of noise-free combustion has been provided by Scalzo

et al. [23]. The eld conversion of two Westinghouse 104 MW W50 1D5 gas

turbines to burn medium Btu synthetic fuel gas resulted in excessive 100 Hz

302 Gas Turbine Combustion: Alternative Fuels and Emissions, Third Edition

airborne sound and unacceptable engine vibration when burning natural gas

with steam injection. The combustors tted in these engines contain no air

swirlers and the basic recirculation ow pattern is established using primary

air scoops. In the original design, the fuel and steam injection holes were

drilled at angles that promoted good mixing with the primary air, but did

not directly oppose the recirculating ow. As a result of the modication to

accommodate the syngas fuel, the angles of natural gas and steam injection

were both reduced, in one case to directly oppose the recirculating air pattern,

and in the other, to provide more opposition to the recirculating airow. This

led to a disruption of the primary recirculation ow pattern and a dramatic

increase in combustion noise.

A number of tests were carried out that demonstrated the importance of

the recirculating primary scoop ow and fuel gas momentum vectors to

combustion noise. Increasing the included angle of the natural gas injection

holes by 40°, and blanking off the central gas injection hole, caused the fuel

and steam momentum vectors to, once again, be “in sympathy” with the pri-

mary air recirculation pattern. This decreased the noise level from 115 dB to

97 dB, an intensity reduction of 64–1.

7.3.7 Fuel-injector instabilities

It is well established that uctuations in fuel ow rate are a major cause of com-

bustion oscillations in many combustion systems, notably rocket engines and

gas turbines. Depending on the noise frequency, it may be called “chugging”

or “rumble,” but in all cases it refers to an interaction between the acoustics of

the combustor and the fuel-injection system.

Suppose one of the combustor’s acoustic modes produces pressure pulsa-

tions at the fuel nozzle(s). If the fuel supply pressure is low, these pressure

pulsations will create oscillations in the fuel ow rate, and these, in turn, will

produce oscillations in the heat-release rate. If these heat-release oscillations

are properly located and in phase with the acoustic mode, they will add energy

to the mode and sustain the oscillating condition [16].

In addition to their inuence on fuel ow rates, uctuations in fuel

pressure can alter various spray characteristics, such as mean drop size,

drop-size distribution, and spray cone angle in ways that also affect the

heat-release rate.

The problems of oscillating combustion in gas turbines began to emerge

and gain signicance in the 1960s and 1970s, during the period when most

engines were tted with dual-orice fuel nozzles. With these injectors, there

is always a range of fuel ows, starting from the point at which the pressuriz-

ing valve opens, over which the secondary fuel delivery pressure is low and

the system is highly susceptible to the onset of combustion oscillations.

Methods of alleviating this source of combustion noise include changing

th e p rimary n oz z l e  o w n um b e r a n d /o r c h a n gi n g t h e p r e s s u r i z i ng v a l v e o pen -

ing pressure [24]. Note that these changes can also affect both the combustor

Combustion Noise 303

pattern factor (via the secondary head effect) and the ability of the fuel nozzle

to provide good atomization over the entire operating range.

Oscillating combustion is by no means conned to dual-orice pressure

nozzles. It can arise with any type of fuel injector at engine operating condi-

tions where the fuel delivery pressure is so low that fuel ow rates and spray

characteristics are affected by uctuations in combustion pressure. One of the

main advantages of airblast atomizers is their ability to provide good atomi-

zation at low fuel-injection pressures, which makes them especially prone

to this mechanism of noise generation. Fitting a pilot nozzle (see Chapter 6)

should eliminate the problem at low engine speeds, but it could reappear at

higher speeds when the fuel ow rate is again just above the pressurizing

valve opening point.

The discussion so far has focused on liquid fuel injectors, but pulsations in

gaseous fuel delivery pressures and ow rates are equally effective in pro-

moting combustion oscillations. One method of eliminating fuel injector-

induced noise in multinozzle combustors, which is equally efcacious for

both liquid and gaseous fuels, is by fuel staging, or sector burning, as dis-

cussed in Chapter 9. With this technique, fuel is supplied only to selected

combinations of nozzles. This enriches the localized combustion zones,

thereby moving their operating point away from the lean blowout limit

and, at the same time, raises the fuel-injection pressure, which reduces the

sensitivity of the fuel supply to acoustic oscillations. The main drawback to

sector burning is that it creates a circumferentially nonuniform exit tem-

perature distribution, with consequent loss of turbine efciency and length-

ening of engine startup time. In fact, although there are many tools at the

combustion engineer’s disposal for eliminating or alleviating this category

of noise, they all need careful consideration for their impact on other impor-

tant aspects of combustor and engine performance.

As discussed below, nonuniformities in fuel delivery not only trigger com-

bustion oscillations, but may also be used to eradicate them. Thus, the need

to understand and predict the dynamic behavior of fuel nozzles is becom-

ing increasingly important and has provided the motivation for a number

of experimental studies on the inuence of fuel ow perturbations on spray

characteristics (see, e.g., Ibrahim et al. [25]).

7.3.8 Compressor-induced Oscillations

The possibility that combustion noise might be due to pressure uctuations in

the air supply from the compressor should not be overlooked. At or near the

design point, the compressor efux is smooth and continuous, apart from tur-

bulence uctuations that generally have a benecial effect by inhibiting ow

separations within the combustor diffuser (see Chapter 3). However, at cer-

tain “off-design” or transient conditions, such as, for example, during engine

acceleration, the compressor may operate close to its stall line. In consequence,

the air supply to the combustor may contain pressure uctuations that are

304 Gas Turbine Combustion: Alternative Fuels and Emissions, Third Edition

amplied in the combustion process to produce noise. Compressor-induced

oscillations of this type can be a trigger for low-frequency noise (e.g., growl).

7.3.9 LPM Combustor Noise

As pointed out by Richards et al. [26,27], lean premixed (LPM) combustors

are especially susceptible to instabilities because acoustic losses are smaller

due to the absence of liner holes and wall-cooling slots, and very little acous-

tic energy is absorbed by the hard liner walls. Moreover, its basic require-

ment to operate at mixture strengths near the lean blowout limit means that

small perturbations in fuel/air ratio tend to produce disproportionately large

variations in the rate of heat release (see also Keller [28]). However, this does

not mean that instabilities in LPM combustors are conned solely to oper-

ate at very weak mixture strengths. Janus et al. [17] observed that combus-

tion oscillations can occur over the entire burning range and not just near

the lean blowout limit, as in conventional combustors. Presumably, this is

because when the latter are operating close to lean blowout, maldistributions

in fuel–air mixing can give rise to a cycle of extinction and reignition in local-

ized regions of the ame, and hence to oscillating combustion, as discussed

above. However, once the primary-zone fuel/air ratio has risen to a level at

which the mixture strength is within the normal burning limits in all regions

of the combustion zone, regardless of maldistributions in fuel–air mixing,

the main driving force for oscillating combustion is no longer present. By

contrast, LPM combustors are characterized by complete homogeneity in

the fuel–air mixture entering the combustion zone. In this situation, there is

no reason why the mechanism for instability near the weak extinction limit

should not also be operative at higher fuel/air ratios where, although the

response of the heat release rate to uctuations in fuel/air ratio is lower, more

energy is available to sustain the instability.

7.3.10 Test rig Simulations

Practical solutions to instability problems are seldom clearly dened and

are often fraught with doubts concerning the specic mechanism(s) driving

a given oscillation. One complicating factor is that the acoustic modes and

inlet conditions may be different between the test rig and the nal engine

design. Furthermore, for any given engine or test rig, a change in ambient

conditions, or a change in fuel of the same type but from a different source,

may serve as a trigger for oscillating combustion.

Many important aspects of combustion performance, such as combustion

efciency, extinction limits, and pollutant emissions, can be studied effectively

on simple test rigs comprising, for example, 60° or 90° sectors of annular com-

bustors that contain only a small fraction of the total number of fuel nozzles.

Much useful experimental data can be acquired from such test rigs, with the

advantage of considerable savings in the cost of fuel and air supplies.

Combustion Noise 305

Unfortunately, this approach has little to offer when dealing with combustion

instabilities. Rig tests on combustor sectors, or even full-scale combustors, can-

not fully reproduce the acoustic properties of the engine combustor, which are

inuenced by the inlet air conditions and by the geometry of the hot sections

downstream. That is why remedial combustor and/or fuel nozzle modica-

tions must often be performed at great cost and inconvenience at a late stage of

engine development.

From their work on the development of LPM combustors, Richards et al.

[29] have suggested that some aspects of engine oscillations can be studied

in single-nozzle test rigs provided that steps are taken to replicate the engine

ame geometry and to minimize acoustic losses in the test device having a

natural frequency corresponding to the oscillating frequency observed (or

expected) on the engine. Their measurements on a single-nozzle test device

showed similar oscillations to those on the engine at comparable operating

conditions. However, these workers recognize that such test devices can-

not reproduce the arbitrary oscillations that occur on the engine, nor can

they simulate oscillations that are controlled by transverse acoustic modes.

Nevertheless, they contend that the simplicity of single-nozzle testing makes

the approach very attractive for a preliminary assessment of how changes

to individual fuel nozzles will affect combustor stability. An interesting

by-product of this research is the nding that two key parameters in the

simulation of oscillating combustion are reference velocity and inlet air

temperature.

7.4 Control of Combustion Instabilities

Sustained combustion oscillations are the result of resonant interaction

between two or more physical processes. A driving process generates the per-

turbations of the ow, whereas a feedback process couples this perturbation

to the driving mechanism and produces the resonant interaction that may

lead to oscillatory combustion [30]. This coupling is shown schematically in

Figure 7.1. Acoustic wave propagation is usually responsible for the feedback

mechanism that relates the downstream ow to the upstream region where

the perturbations are initiated. Methods for alleviating or eliminating com-

bustion instabilities generally fall into the two main categories of “passive”

and “active” control, which are described below.

7.4.1 Passive Control

Passive control techniques have been widely used in industrial burners for

many years [31]. The passive methods employed for stabilizing combustion

oscillations in liquid rocket engines have been described in publications

306 Gas Turbine Combustion: Alternative Fuels and Emissions, Third Edition

by Harrje and Reardon [32] and, more recently, by Yang and Anderson [33].

Passive methods for controlling combustion instabilities in ramjet dump

combustors have been studied by Gutmark et al. [34]. Their application typi-

cally involves modications to the fuel injector or combustor hardware to

eliminate the source of the variation in heat release or to increase the acous-

tic damping in the system, thereby reducing the amplitude of any pressure

oscillations. They include bafes, resonators, and acoustic liners of the type

that have proved highly successful in suppressing screech in afterburner sys-

tems. These devices are much less effective at the low frequencies encoun-

tered in main combustors, where the emphasis changes to the fuel delivery

system, as discussed above, in particular to those aspects that govern fuel

pressure and fuel distribution pattern. Inevitably, passive control techniques

tend to be applied on a costly trial-and-error basis.

In recent years, increasing interest has been shown in the control and sup-

pression of combustion instabilities by actively and continuously perturbing

in real time the processes responsible for coupling the heat release with the

acoustic pressure. The growing importance of these active control techniques

merits their discussion in some detail.

7.4.2 Active Control

The active control measures employed to dampen or eliminate acoustic oscil-

lations in combustion systems appear to vary widely, both in regard to the

theoretical basis for the control system and the actual hardware employed.

They are described in two important review papers by Candel [30] and

McManus et al. [22]. A recent paper by Richards et al. [35] also outlines some

of the strategies and devices used in the area of active control.

Active control methods were derived from a number of studies on rocket

motor instabilities carried out in the early 1950s. One outcome of this work

was the notion of introducing perturbations into the combustor using an

actuator to decouple the physical processes responsible for the oscillations.

Combustion

Acoustics

Variable

pressure

Variable

heat release

Figure 7.1

Representation of coupling between combustion and acoustic processes.

Combustion Noise 307

However, it is only within the past decade that the practical demonstration

of this concept has been realized [30].

Active control systems fall under the two main headings of “open-loop”

and “closed-loop.”

7.4.2.1 Open-Loop Systems

A key feature of an open-loop system is that the control action is indepen-

dent of the combustor’s response to the control input. Essentially, they pro-

vide a xed stimulus to the combustor in order to disrupt the coupling of

the combustion instability mechanism. Usually, an oscillatory signal with

a xed amplitude and frequency is applied to the combustor, using some

form of control actuator. The objective is to introduce a perturbation in some

physical variable that has a strong inuence on the combustion process, such

as the acoustic pressure eld or the inlet ow velocity prole.

The types of actuators commonly used include acoustic drivers, ow valves

with rotating or oscillating elements to produce periodic ows, and shaking

devices to produce mechanical oscillations. The relative merits of these dif-

ferent types of control actuators for various applications have been discussed

by McManus et al. [22]. According to these workers, the main advantage of

open-loop controllers is that they tend not to suffer from control system insta-

bilities. However, their effectiveness is very dependent on their calibration,

which can be a difcult task. For this reason, the more versatile closed-loop

system is generally preferred.

7.4.2.2 Closed-Loop Systems

The most distinguishing feature of closed-loop controllers is their use of

feedback to control combustion oscillations. The basic idea of feedback in

its application to continuous-ow combustion systems is that, by monitor-

ing both the output and input of a combustor, appropriate adjustment of the

input can be made that will eliminate any instabilities in the output. The

feedback signal is produced by a sensor, usually a pressure transducer or

ame detector, which monitors some time-varying property of the combus-

tion process, and is then passed on to the controller. The output signal from

the controller is applied to the combustor via an appropriate form of actuator,

as described above, in order to eliminate the acoustic oscillations.

7.4.3 examples of Active Control

McManus et al. [36] used an open-loop conguration to control combustion

instabilities in a dump combustor. The control system consisted of an actua-

tor that applied a cross-stream velocity perturbation to the inlet boundary

layer of the combustor test section. The actuator was driven by amplied sine-

wave signals in order to create a periodic oscillation in the boundary layer.

308 Gas Turbine Combustion: Alternative Fuels and Emissions, Third Edition

Application of active control with a 160 Hz sine-wave signal produced a 30%

reduction in the RMS pressure uctuation level, due mainly to a decrease in

amplitude of the acoustic mode associated with the instability.

Choudhury et al. [37] used a similar dump combustor in their active con-

trol experiments. The actuator employed in this investigation was a row of

pulsed gas jets located just upstream of the rearward-facing step. Periodic

forcing at a xed frequency and amplitude was achieved by inserting a rotat-

ing valve into the gas supply line. The disruptive effect of the pulsed jets on

the recirculation zone created by the step produced a signicant decrease in

the combustion oscillations.

Bloxsidge et al. [38] used a 250 kW model jet-engine afterburner test rig

to investigate the possibility of closed-loop control of instabilities by vary-

ing the inlet ow area to the combustor. The apparatus consisted of a long

duct with a centerbody ameholder located in the downstream portion

of the duct. A premixed ethylene fuel–air mixture entered the test section

through a nozzle whose ow area was determined by the axial position of

a translating plug located at the center of the nozzle. A mechanical shaker

connected to the plug allowed the inlet ow area to be modulated, thereby

serving as a control system actuator. A pressure transducer located down-

stream of the nozzle provided the feedback signal required to control the

instability. Without control, the combustion process was characterized by

high-amplitude oscillations in the frequency range from 80 to 300 Hz. These

oscillations were attributed to the coupling between unsteady heat release

and longitudinal acoustic modes of the duct. Activation of the control system

effectively suppressed the oscillations.

The same apparatus was used by Langhorne et al. [39] to investigate the

possibility of eliminating combustion oscillations by adding fuel out of

phase with the oscillation. For these experiments, the translating plug was

removed to give a xed-area inlet nozzle, and an additional fuel-injection

manifold was tted just upstream of the ameholder to provide a controlled

unsteady injection of fuel in addition to the main ethylene–air mixture ow-

ing through the inlet nozzle. It was found that a closed-loop control, using the

pressure signal to modulate just 3% of the total fuel ow, reduced the noise

level by 12 dB. Sivasegaram and Whitelaw [40] also used periodic fuel injec-

tion, in both open- and closed-loop congurations, to reduce the oscillating

pressure amplitude in a laboratory test rig by values up to 15 dB, depending

on the type of control used.

Schadow et al. [41] have reported on a number of control techniques,

including fuel ow modulation. In a subsequent paper, Schadow et al. [42]

compared fuel modulation with a markedly different approach, whereby a

spark discharge ignites a portion of the fuel before it enters the main ame

of a dump combustor. Finally, Richards et al. [35] have described the applica-

tion of active control to a premixing fuel nozzle, using natural gas fuel. Cyclic

injection of 14% control fuel was found to produce a 30% (10 dB) reduction in

oscillating pressure amplitude at 300 Hz.

Combustion Noise 309

7.4.4 influence of Control Signal Frequency

In their study of combustion instabilities in a dump combustor, McManus

et al. [36] found that maximum reduction in pressure oscillations level

was obtained when using a control signal frequency of 160 Hz, as noted

above, but they also observed that pressure oscillations could be reduced

when active control was applied over a range of frequencies from 100 to

1000 Hz. Several other workers have reported that effective control can be

achieved at frequencies much lower than that of the acoustic oscillation.

For example, Brouwer et al. [43] found that a slow modulation of atomiz-

ing air (less than 2 Hz) reduced combustion oscillation in a liquid-fuelled

gas turbine combustor, whereas Richards et al. [35] were able to stabilize

oscillating combustion using a low-frequency modulation of the natural

gas fuel. Gemmen et al. [44] also observed similar behavior when modu-

lating a pilot ame. Oscillation control was again achieved by changing

the ame conditions at a frequency that was much lower than the acoustic

oscillation. These results are interesting because they do not conform to the

generally held view that uctuations in heat release can only be smoothed

out by repeated perturbations that occur at about the same frequency as

the uctuations.

7.5 Modeling of Combustion Instabilities

Many attempts have been made to develop analytical models for predict-

ing combustion instabilities along with their amplitudes and frequencies,

but these efforts have met with only partial success. The problems involved

are formidable because audible engine noise is usually the result of complex

instability mechanisms that cannot be modeled or described using standard

analytic techniques until many simplications have been made to render the

problem tractable [22]. Clearly, this approach could result in an oversimplied

view of the problem. At the present time, there is no universal model for pre-

dicting combustion instabilities, thus modeling of combustion instabilities is

usually carried out on a case by case basis.

Much of the early work on the modeling of combustion oscillations was

motivated by the instabilities encountered in liquid-fueled rocket engines.

The so-called τ–n analysis of Crocco and Cheng [45], which relates heat

release and pressure oscillations by a time lag, τ, and an interaction index, n,

has proved useful in analyzing rocket engine data. It invokes the Rayleigh cri-

terion that heat release and acoustic uctuations should be in phase to drive

oscillations and out of phase to dampen oscillations [14]. The basis of the τ–n

model is that acoustic disturbances produce a change in heat release rate, but

delayed by a time interval, τ. Given the correct value of τ, and assuming the