See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228537444

Combustion Dynamics: Application of Active Instability Control to Heavy Duty

Gas Turbines

Conference Paper · May 2001

CITATIONS READS

14 1,198

2 authors, including:

Jakob Hermann
IfTA Ingenieurbuero fuer Thermoakustik GmbH
39 PUBLICATIONS   378 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Jakob Hermann on 13 February 2016.

The user has requested enhancement of the downloaded file.

 6-1

Combustion Dynamics: Application of Active Instability Control to Heavy

Duty Gas Turbines

J. Hermann, A. Orthmann
IfTA Ingenieurbüro für Thermoakustik GmbH
Industriestr. 33
D-82194 Gröbenzell/Munich, Germany
E-mail: [email protected], [email protected]

Summary
Self-excited combustion oscillations constitute an important problem for the development of modern
combustion systems. Methods to avoid these oscillations by active instability control (AIC) to provide
safe operations for the corresponding combustion system are the subject matter of this paper. In addition
to applications of this technology for a land-based gas turbine, basic actuation possibilities, methods of
measuring oscillation quantities and requirements to be met by control strategies are described.
Following this general section, the installation of AIC systems for gas turbines of the Siemens Vx4.3A
family will be explained. This type of turbine features an annular combustion chamber with a total of 24
burners. In order to be able to damp combustion oscillations arising within this type of combustion
chamber - appearing as azimuthal modes spreading along the circumference of the combustion chamber -
every burner was fitted with a direct drive valve. This type of valve creates mass flow modulations within
the pilot gas supply of the burner which are anti-cyclical to the oscillations characterising the heat release
rate within the flame, thus extinguishing them. Input signals for the feedback control system are obtained
by measuring sound pressures within the combustion chamber indirectly at the burner flanges. Finally,
this type of gas turbine was fitted with a 12-channel controller and 12 sensors in order to allow a damping
of the azimuthal modes excited around the circumference of its annular combustion chamber. Exploiting
the symmetry characterising azimuthal modes, two actuators are driven by every feedback loop. In tests
run on various type V94.3A gas turbines delivering up to 267 MW of electric power according to ISO,
AIC systems were used to damp successfully a great variety of combustion oscillations for several burner
variants and operating points. Thus, it was possible to obtain stable gas turbine operations over their full
power ranges. In addition to the high AIC flexibility in damping various oscillation problems at different
gas turbine operating points, this technology has proved to provide a high degree of fault tolerance and
good long-term characteristics.

Introduction
Primary development aims for all modern combustion systems are minimising pollutant emission values,
enhancing efficiency and increasing power density in order to achieve design dimensions as compact as
possible. This imposes strict operating limits for efficiency and power density; moreover, in order to
achieve low NOx emissions, lean premixed combustion is favoured in most cases. One unwanted side-
effect is the appearance of a special form of combustion instability, so-called self-excited combustion
oscillations. For low thermal power combustion systems, these instabilities primarily lead to increased
noise levels. For combustion systems delivering high heat release rates, such as process gas heaters, or
for combustion systems working under pressure, such as gas turbines, the sound pressure generated will
reach very high levels. Owing to the large surfaces of such systems, high mechanical loads on the

Paper presented at the RTO AVT Course on “Active Control of Engine Dynamics”,
held in Brussels, Belgium, 14-18 May 2001, and published in RTO-EN-020.
6-2

2
combustion chamber as well as on upstream and downstream components will arise. Also the thermal
load on the chamber walls will rise considerably. Depending on sound pressure amplitudes, components
will fail sooner or later; thus, this kind of oscillation must be avoided by all means if those combustion
systems are to be operated safely. In view of the high requirements to be met by up-to-date, highly
optimised combustion systems and their low degree of flexibility as regards operating parameters, the
scope available for avoiding oscillations by modifying burner operation is decreasing all the time. New
methods to prevent oscillations will have to be found.
Basically, the possibilities available to prevent self-excited combustion oscillations can be subdivided
into passive and active measures. For instance, increasing acoustical attenuation, acoustically detuning
systems by design modifications, and operational modifications are considered passive measures. 1-4 By
contrast, active measures imply creating an external feedback loop using an actuator to influence
combustion oscillations so as to damp them down. With this type of "Active Instability Control" (AIC),
the combustion system, properly speaking, does not have to be redesigned, and operations can continue
as usual.
The basic idea to suppress combustion oscillations by an active feedback loop was published in a
theoretical paper on rocket engines by Tsien as early as 1952.5 However, it took until the eighties to
convert this idea into a practical device. Various authors described successful tests based on laboratory-
scaled burners with a thermal power of between 1 kW and 250 kW.6-8 For all these publications,
attenuation of combustion oscillations is achieved by anti-sound signals generated via loudspeakers. In
addition to this method, other types of intervention and control strategies were researched for various
combustion systems; however, all tests were performed at lab scales. 9-16 The first industrial-size
application was realised by Seume et al. 17 in 1996 based on a land-based gas turbine delivering 160 MW
of electrical power. For this gas turbine, active control was achieved by means of anti-cyclical fuel
injection, with direct drive valves serving as actuators. This technique was then also extended to the
largest type of this family of gas turbines, the V94.3A with an electric power output of 267 MW.18, 19 Full
scale tests on afterburners have been published by Moren et al. 20, and on a 67.5° sector of a liquid-fuelled
lean premixed combustor by Hibsman et al. 21
The way this technology was implemented and the problems encountered in doing so constitute the
subject matter of chapter 3. To begin with, however, chapter 2 explains the necessary fundamentals, the
main possibilities for actuation, as well as the components available to do so, such as sensors and
actuators. Moreover, practical requirements to be met by this technology will be illustrated. Results
achieved with AIC are discussed in chapter 4. Chapter 5 and 6 deal with long-term experience and a short
evaluation of the advantages offered by AIC as compared to passive measures.

Fundamentals

Origin of self-excited combustion oscillations

In numerous cases, self-excited combustion oscillations are due to the feedback mechanism shown in a
simplified diagram in Fig. 1. The injection of a turbulent air and fuel spray will generate broadband
sound. Excitation of acoustic resonant frequencies within the combustion chamber may thus
particularly amplify specific discrete frequencies. These sound pressure perturbations will create
periodical flow modulations, i.e. periodically change the inflows into the combustion chamber, if
appropriate preconditions at the combustion chamber inlet are given. Within the flame, these periodical
flow fluctuations will cause the heat release rate to fluctuate with the same period. If its oscillations are
in phase with sound pressure oscillations within the combustion chamber, further excitation will result
according to Rayleigh’s criterion,22 so that oscillations will keep building up. Within this feedback
cycle, the decisive link between combustion and flow is constituted by the acoustics.
 6-3

3
& = m& 0 ,AIR + m
m & 0,FUEL + m& AIR + m
& FUEL Q& = Q& 0 + Q&
m& 0 ,AIR = const .
m& 0,FUEL = const. Flame holder Flow field Flame

p Q&
Acoustics
m& 0 const . mass flow
~
m& mass flow fluctuation
m& total mass flow
Q& 0 const . heat release rate of the flame
~
Q& fluctuation of the heat release rate of the flame
Q& total heat release rate of the flame
~p sound pressure

Fig. 1: Simplified model of a self-excited combustion oscillation. The feedback loop is determined by
combustor acoustics.

Active control of combustion oscillations

When actively suppressing self-excited combustion oscillations, some physical quantity such as the
sound pressure characterising the oscillation is measured; the signal thus obtained is then fed into a
controller. Using an appropriate control strategy, the controller generates an output signal designed to
counteract the oscillation via an actuator. Thus, the sound pressure signal is attenuated. For this purpose,
the actuator superimposes some separately generated oscillation on some quantity contributing to
combustion oscillations so that self-excited oscillations will be extinguished. Fig. 2 shows this principle
schematically. In this instance, sound pressure is used as controller input signal driving the actuator to
modulate the fuel mass flow m & 0 ,Fuel injected.

& = m& 0 ,AIR + m

m & 0,FUEL + m& AIR + m
& FUEL Q& = Q& 0 + Q&
m& 0 ,AIR = const .
m& 0,FUEL = const. Flame holder Flow field Flame
+
p Q&
Acoustics

Actuator Sensor

Controller

AIC-System

Fig. 2: Schematic of the AIC system. Combustion instabilities are neutralised by modulation of the fuel
flow rate anti-cyclically to the heat release rate of the flame.

POINTS OF ACTUATION AND ACTUATORS AVAILABLE

According to the mechanism generating self-excited combustion oscillations, schematically shown in Fig.
1, there are two basic possibilities for active control of combustion instabilities via actuators: by
influencing the combustion system acoustics or by controlling the flame itself. For actively intervening
via acoustical means, sound pressure fluctuations within the combustion chamber are damped by means
6-4

4
of an anti-sound signal. Owing to the direct coupling between sound pressure fluctuation and related
oscillations of the heat release rate within the flame, the combustion zone itself will likewise be smoothed
out. By contrast, when actively controlling via the flame, fluctuations of the heat release rate are
attenuated by anti-cyclically injecting air and/or fuel mass flows. Seeing that the unsteady flame with its
unsteady heat release rate constitutes the source of sound, the sound field within the combustion system
will likewise be damped.
When implementing various possible means of controlling combustion instabilities, different problems
will occur depending upon the type of intervention or the physical quantities influenced. More
particularly, acting upon acoustics by superimposing an anti-sound signal, say, by means of a
loudspeaker, soon runs up against practical limits. For combustion systems operating under pressure or
characterised by high volume flow rates, the acoustic power that can be generated by loudspeakers is
insufficient to appropriately damp any self-excited sound field. Moreover, there will be problems with
installing this type of actuator. For instance, loudspeakers have to be built in as flush as possible with the
chamber walls in order to guarantee good sound emission. In view of the high wall temperatures
characterising combustion systems, doing so is normally impractical. Installing acoustic actuators within
the air/fuel supply system likewise creates practical problems if mass flows are pre-heated or flammable.
Compared to acoustics, influencing the flows introduced into combustion systems offers a better control
potential. Fuel supplies are normally particularly suitable because, in industrial combustion systems,
volume flow rates of fuel are often substantially lower than those of air. This is particularly true for
gaseous fuels supplied under high pressure, for highly caloric gaseous fuels, or when burning liquid fuels.
Fig. 2 shows this type of control schematically. Given sufficiently reduced volume flow rates,
correspondingly small valves can be used as actuators for high-frequency opening and closing. Various
authors have presented and/or tested this type of valve. Most of these valves are actuated by means of
electromagnetic linear motors or by piezo or magneto-restrictive drives. 16, 15, 23 In addition to such valves,
other devices were successfully used in the past to modulate fuel flows, such as an actuator featuring a
piezo driven piston designed to superimpose a sound field upon a liquid fuel flow that generates mass
flow fluctuations at the injection nozzle.15
In addition to actuation through modulation of the mean fuel supply, there is the possibility of damping
the oscillations prevailing within the flame by injecting additional fuel directly into it. Certain papers
have shown that, depending on the actuator type used, the fuel injection point, or the spray
characteristics, no more than 3-4% of additional fuel will be required to control combustion
oscillations. 10, 14
No matter which type of actuator is employed, its correct installation is highly important. This includes
actuator positioning, taking due account of functional principles and the sound field prevailing in any
specific case. For instance with loudspeakers, care has to be taken to install them in the sound pressure
anti-node area of the eigenmode for active control to be effective. In addition to this requirement,
actuators for industrial AIC systems need high reliability; in some cases, it must be possible to use them
even at high ambient and media temperatures.

AIC INPUT SIGNALS

Basically, every parameter that is part of the oscillation at issue can be used as input signal for an active
feedback control. In practice, however, only those will be used that can be measured easily by suitable
sensors, i.e. mainly sound pressure and heat release rate of the flame. Sound pressure can be measured
very easily, e.g. by microphones or piezo pressure transducers, while photomultipliers or photodiodes are
particularly useful to measure the unsteady heat release rate of the flame. In order to obtain a resolution
sufficient for the time scales of the oscillations, the emissions of short-lived intermediate reaction
products are measured, such as the formation of free OH radicals. In order to be able to capture only the
radiation emitted by them at 306.5 nm, appropriate optical bandpass filters will be used.
AIC sensors have to be positioned in spots where the parameter to be measured correlates sufficiently
with combustion oscillations and is reasonably proportional. Optical sensors to measure heat release rates
for the AIC must cover no less than that area of the combustion zone where fluctuations of the reaction
 6-5

5
rate occur, the entire combustion zone being preferable. If only a small portion of the combustion zone is
covered, any displacement of this zone may blind the sensor to prevailing oscillations. A further problem
with optical sensors is fouling, e.g. by soot.

REQUIREMENTS TO BE MET BY A CONTROLLER

The least complicated controller for an AIC system is an amplifier/phase shifter combination
appropriately amplifying input signals and shifting their phases so as to ensure the required anti-cyclical
operation of the actuator. However, this type of controller will only work satisfactorily if the oscillation
frequency excited is constant and if there is only one oscillation frequency. Further requirements
imposing more complex control strategies are entailed by combustion systems having several burners
distributed over the modes excited. Under certain circumstances, they may require multi-channel control
signal processing. Moreover, in industrial applications, a self-adapting control strategy and/or control
parameter compensation depending on operating points are indispensable. Thus, in the past, a great
variety of control strategies such as model-based controllers, self-adapting controllers based on LMS
algorithms or self-tuning controllers were researched. A summary of the work published on this topic is
to be found in a recent publication of Dowling. 24

Installing AIC on Siemens type Vx4.3A land-based gas turbines

The following section describes the installation of an AIC system on type Vx4.3A land-based Siemens gas
turbines. Fig. 3 shows a longitudinal section through the upper half of this gas turbine. In contrast to former
types of Siemens gas turbines that were fitted with silo combustion chambers, this family of gas turbines
features ring combustion chambers (see left-hand illustration in Fig. 4). In total, this gas turbine comprises
24 Siemens hybrid burners spread uniformly over the circumference of its annular combustion chamber.

One of the 24 Ring combustion Turbine Exhaust

Hybrid Burners chamber channel

Inlet
channel Compressor

Fig. 3: Half-section drawing of a Vx4.3A series gas turbine.

6-6

6
2nd harmonic with a frequency of 170 Hz
Premix
Sound
On of the 24 flame Pressure
burners t = 0.75·T

+
Circumferential
coordinate
Coordinate
Combustion Burner

-
chamber

t = 0·T, 0.5·T and T

Ring Combustion
Chamber

t = 0.25·T

Oszillation: T = 1 / f
Period of oscillation:

Fig. 4: Left: Three-dimensional drawing of the annular combustion chamber of the Vx4.3A with a total
of 24 burners. Right: Excited azimuthal modes for the second harmonic in the V94.3A.

Any Siemens hybrid burner may be operated both on liquid and gaseous fuels. At start-up and with
gaseous fuel, the burner is operated in diffusion mode. When a specific turbine power is reached, there
will be a switch-over from pure diffusion to so-called mixed operation. For this purpose, a combination
of diffusion and premix flames is used. When the temperature within the combustion chamber is
sufficiently high to stabilise a pure premix flame, operation is switched from this mixed mode to purely
premixed operation. To stabilise the premix flame, every Siemens hybrid burner features an additional
pilot burner. This is a diffusion burner delivering approx. 10% of the thermal power provided by the
complete burner unit.

Instability problem
During the test phase of the prototype and the commissioning of the type of gas turbine described above,
there were problems with self-excited combustion oscillations in various power ranges depending on the
modifications performed on the hybrid burner used. Due to the low frequency of these instabilities, this
effect is sometimes called "humming".
Research on this oscillatory phenomenon described by Seume et al. 17 showed that standing sound waves
are generated within the annular combustion chamber by self-excited combustion oscillations. According
to the direction of propagation of these waves along the circumference of the annular combustor they are
designated "azimuthal modes". They are characterised by regions where sound pressure amplitudes are
high, called "sound pressure anti-nodes", and regions where these amplitudes are very low, called "sound
pressure nodes". The number of nodes and antinodes varies according to the prevailing frequencies. The
right-hand side of Fig. 4 shows the azimuthal modes of the second harmonic, characterised by a total of 4
nodes and 4 anti-nodes. For the largest version of this gas turbine type, V94.3A, this eigenmode is
excited at a frequency of approx. 170 Hz.
The equation f=n⋅c/(π⋅d) provides a theoretical estimate for the characteristic acoustical frequencies of
any annular combustion chamber. At an average combustion chamber diameter of d=3 m and a speed of
sound of c=844 m/s (assuming an average combustion chamber temperature of 1,500°C), the resulting
frequency for a V94.3A is 179 Hz for the second harmonic, a value agreeing well with experimentally
determined characteristic frequencies. For smaller versions of this turbine family, the above equation
returns higher frequencies for the occurrence of combustion oscillations, due to the reduced diameter of
their combustion chambers (see Seume et al. 17).
 6-7

7
AIC installation
In order to avoid the oscillation problem described, an AIC system for this type of gas turbine was
developed in addition to passive measures such as burner design modifications. 4 Fig. 5 provides a
simplified schematic diagram of the basic design of this AIC.
Plenum Piezo pressure
Controller
chamber transducer
Ring
combustor

DDV valve Burner

ne on
Pilot gas Pilot gas

zo usti
ring line feed tube

mb
Co
Turbine
Compressor inlet
discharge

Fig. 5: Schematic AIC diagram of the Siemens model Vx4.3A heavy duty gas turbine

SENSOR
By way of input quantity, this AIC system uses the sound pressure measured at the burner flanges. There,
wall temperatures are substantially lower than within the combustion chamber so that high-temperature
piezo pressure transducers can be used without requiring any additional cooling . It was verified in a
number of tests that the sound pressure signals measured at the burner flanges coincide sufficiently well
in amplitude and phase with the pressure signals within the combustion chamber.
During some initial tests, potential uses of optical probes for measuring AIC input signals were
researched in addition to sound pressure measurements at the burner flanges. However, owing to the
limited field of view and due to thermal problems with probe installation, this method was soon given up.

ACTUATOR
The crucial problem for implementing any AIC system of the type described here is actively influencing
the combustion. Owing to the elevated air and fuel volume flows through Vx4.3A gas turbines, these
flows cannot be sufficiently modulated in full, neither by means of acoustical actuators nor by valves.
Detailed research showed that the main premix flame of the Siemens hybrid burner – controlled by much
smaller pilot diffusion flames – will respond very precisely to fluctuations in the conversion rates of
those pilot flames. Accordingly, it is possible, by modulating the pilot gas mass flow, which comprises
no more than approx. 10% of the entire mass flow, to control not only the pilot flames themselves but
also the premixed main flame to a significant extent. Pilot flames are supplied via their own fuel line. The
actuator, a direct drive valve (DDV) – shown in Fig. 6 - developed specifically for this application by
Moog Germany, is integrated in this fuel line. In order to obtain maximum control levels for main flames
and, accordingly, for the combustion oscillations arising within the gas turbine, every burner of the
turbine was fitted with its own valve. Thus, a total of 24 valves was installed around the annular
combustor of the gas turbine.
In the absence of any AIC signal, DDV valves will be 50% open, so that the pilot gas mass flows
required for normal premixed operation will reach the burners. In order to modulate individual pilot gas
mass flows and thus the pilot flames themselves, valve spools will be moved around their static opening
value of 50%, i.e. opened further or closed down at the frequency of the combustion oscillation to be
damped. The level of fuel flow modulation achieved can be determined by valve spool stroke. According
6-8

8
to the frequency response of the used valve type – shown in Fig. 7 – the valves allow the control of
combustion oscillations with frequencies of up to 400 Hz.

Connector

Coil Centring
spring

Position Bushing
transducer Spool
Linear force
motor

Fig. 6: Schematic diagram of the DDV valve used

4
amplitude [d B]

-4

-8

-12

0
phase [deg ]

-45

-90

-135

-180
5 10 50 100 500 1000
frequency [Hz]

Fig. 7: Frequency response of the DDV valve used

By moving the valve spools at frequencies within this range, sound waves will be generated within the
pilot gas systems. The resulting effects on the modulation of the pilot gas mass flow must be taken into
account and the pilot gas system must be tuned as described in Hermann et al. 15 and Hantschk et al. 16

CONTROLLER
As described in Seume et al. 17, azimuthal modes are excited within ring combustion chambers by
combustion oscillations. This means that the pressure fluctuations excited along the combustion chamber
circumference are characterised by both different amplitudes and different phases, and that the burners
placed evenly along the circumference of the annular combustion chamber are accordingly located at
positions characterised by differing amplitude and phase values. Seeing that fluctuations of the heat
release rate are coupled characteristically with pressure oscillations (self-excitation), any heat released at
 6-9

9
the various burners will fluctuate at differing amplitudes and phases. Since AIC requires both anti-phase
and in-amplitude influencing of the unsteady flame, this means that it must be possible to control every
burner individually. The least complicated case would require a sensor and a feedback loop for every
burner, so that a 24-channel AIC system would have to be installed.

Ring combustion
Ring combustion
chamber
chamber (cutaway
(cutaway
view across ^
view acrossthe
the Valve 11
Valve p
gasgas
turbine axis)
turbineaxis)

sound pressure p
Sensor11
Sensor time t
Control
Control
Loop11
loop

Burner
Burner sound pressure at valve 1
-1 sound pressure at valve 13

Valve 13
Valve

Fig. 8: Left: Exploiting the symmetry of azimuthal modes, e.g. for the first harmonic. One sensor and
one controller provide the input signals for two DDVs. Right: Sound pressure at the two valve
positions.

For minimising this high number of feedback loops, the symmetry of azimuthal modes arising within the
combustion chamber was exploited. As shown by the azimuthal mode corresponding with the first
harmonic in Fig. 8, the sound pressure fluctuations for two precisely opposite burners are characterised
by identical amplitudes p̂ and a phase shift ϕ amounting to 180° with respect to each other. As indicated
in Fig. 8, for an AIC system this means that a signal measured at a specific position by one sensor can be
used to control not only the actuator for this position but also the one located at the precisely opposite
point of the combustion chamber by merely inverting the controller output signal for the second actuator.
In the event of second or even higher harmonics, the number of feedback loops can be reduced even
further, as shown in Fig. 9. Here, a total of 4 actuators are controlled via one input signal and one
feedback loop. For the control system of V94.3A burners, an installation as in Fig. 8 was chosen since,
for this type of gas turbine, the first harmonic will likewise appear due to self-excited oscillations. In
total, this type of gas turbine was fitted with 12 sensors and 12 feedback loops.

Valve 1
Burner

Sensor 1
Control
Control
-1
Valve 19 loop 1 1
System

Valve 7

Ring combustion
chamber
(cutaway view
across gas Valve 13
turbine axis)

Fig. 9: Exploiting the symmetry of azimuthal modes, e.g. for the second harmonic. One sensor and one
controller provide the input signals for four DDVs.
6-10

10
Those 12 loops were realised by means of 6 signal processors, every processor handling two input signals
and generating four output signals. The algorithm used works within the frequency range; it is suitable
for controlling two frequencies, a restriction entailed by the limited computing capacity of the type of
signal processor used. The necessary control parameters depend on the operation parameters of the gas
turbine and are set automatically during continuous operation. Hermann et al. 19 provides a detailed
description of this control system.

Results
The AIC system described above was implemented in two differently sized gas turbines within the
Vx4.3A family. The first tests were run using a type V84.3A machine delivering 170 MW of electric
power at Siemens test facilities in Berlin. Detailed specifications of the AIC structure used and the tests
performed are to be found in Seume et al. 17 Owing to these successful tests , AIC systems were
subsequently built into type V94.3A machines for field testing purposes. Depending on the version used,
these machines deliver between 233 MW and 267 MW of electrical power under ISO conditions. For this
purpose, the electronic system and the control strategy used for V84.3A were completely redesigned and
further developed into an industrial-grade system.
During field tests, AIC systems were used on V94.3A machines together with various burner
configurations showing oscillation problems at various operating points of the corresponding gas turbine.
Below, tests on three different burner configurations are presented.

Burner configuration A: active control at switch-over

Burner configuration A mainly presented problems due to combustion oscillation during switch-over
from mixed to premixed operation, i.e. when transitioning from a combination of diffusion and premixed
flame to a purely premixed flame. To stabilise the premixed flame, the pilot burner is switched on while
switch-over is in progress. Fig. 10 shows this process without (left) and with activated AIC system
(right). From top to bottom, the following quantities are plotted:
− maximum sound pressure spectrum calculated on the basis of the signals delivered by the 12 AIC
sensors at three different moments in time;
− maximum RMS value of the sound pressure versus time, calculated on the basis of the signals
delivered by the 12 AIC sensors;
− opening cross-section versus time for the gas valves actuated during switch-over, as a percentage
of the maximum opening cross-section for the corresponding valve.
As shown in Fig. 10 in the diagrams on the left, strong combustion oscillations occur approx. 3.5 seconds
after opening the pilot gas main valve (= start of switch-over, marked in the bottom diagram by a vertical
arrow) as the pilot gas flows into the combustion chamber. After another 2 seconds, the diffusion gas
valve is closed and the premix gas valve opens, with the thermal power within the combustion chamber
being kept constant during that process. About 10.5 seconds after initiating the switch-over (t=15s), i.e.
as the diffusion gas valve closes, the combustion oscillation amplitude slightly weakens before it returns,
after another 5 seconds, to the value characterising constant operation. At the t=20s moment in time, the
gas turbine is running in its fully stabilised premix mode. Considering the frequency spectrum of the
sound pressure measured at various times during switch-over, it becomes obvious that, in the beginning,
the first harmonic dominates at around 90 Hz (t=9.7s). At t=13.7s, this frequency will have dropped to
approx. 80 Hz. By way of non-linear effect, high amplitudes of the first harmonic will additionally
produce its higher-order harmonics.
The diagrams on the right-hand side of Fig. 10 show the same switch-over process, but subject to an
activated AIC system. As the sound pressure history for the RMS value indicates, this leads to an almost
complete attenuation of combustion oscillations. Only a specific position of the diffusion gas valve
(t=15.2s) produces a short peak lasting approx. 0.5 s, its RMS value already substantially reduced.
Considering this moment in time on the frequency spectrum, it becomes apparent that the harmonic
excited is no longer the first but the third one, at approx. 250 Hz. To achieve even further damping for
 6-11

11
this oscillation by means of an AIC system, the system was pre-set for both frequencies (first and third
harmonics). As shown by the spectrum of another switch-over recorded some 10 minutes later, the
amplitude of this oscillation was further attenuated, without influencing, yet damping of the first
harmonic. It has to be stated that the damping achieved by AIC was more than sufficient for safe gas
turbine operation.
AIC system off AIC system on
sound pressure spectrum (maximum of 12 sensors) sound pressure spectrum (maximum of 12 sensors)
200 200
180 9.7 s 180 10.8 s
13.7 s spectrum at switch-over
160 160 15.2 s
arbitrary units [-]

arbitrary units [-]

with AIC system adjusted
140 17.4 s 140 about 10 minutes later
to the 90 Hz oscillation
120 120
spectrum at switch-over
100 100
with AIC system adjusted
80 80 to the 90 and 250 Hz
60 60 oscillation
40 40
20 20
0 0
0 100 200 300 400 500 0 100 200 300 400 500
frequency [Hz] frequency [Hz]

RMS value of the sound pressure (max. of 12 sensors) RMS value of the sound pressure (max. of 12 sensors)
200 200
180 180
case at switch-over
160 160
arbitrary units [-]

arbitrary units [-]

with AIC system

140 140 adjusted to the 90 Hz
120 120 oscillation
100 100
80 80
60 60
40 40
20 20
0 0
0 5 10 15 20 25 0 5 10 15 20 25
time [s] time [s]

valve opening position valve opening position

100 100
valve opening pos. [%]

valve ppening pos. [%]

80 80
start of the
switch-over Mixed operation
mixed premixed operation
60 60

40 40

20 20

0 0
0 5 10 15 20 25 0 5 10 15 20 25
diffusion gas valve (non-premixed) [%] time [s] diffusion gas valve (non-premixed) [%] time [s]
premix gas valve [%] premix gas valve [%]
main pilot gas valve [%] main pilot gas valve [%]

Fig. 10: Switch-over from mixed to premixed operation without (left) and with AIC (right). The inlet
guide vane remained permanently closed.

Burner configuration B: active control at load change during premixed operation

During premixed operation over a power range of 60% to 100% of base load, burner configuration B was
characterised by self-excited combustion oscillations. If this gas turbine is operated without AIC,
increased amplitudes for the third harmonic at a frequency of 270 Hz may appear as early as at 60% of
base load. At approx. 80% of base load, this frequency shifted to 170 Hz, corresponding to the second
harmonic of the combustion chamber.
By activating the AIC system, the oscillations were attenuated enough to allow safe turbine operations up
to 100% of base load power. Fig. 11 demonstrates the AIC operation for a load increase (left) and a load
6-12

12
decrease (right). From top to bottom, the following oscillation and operating parameters of the turbine are
shown, all of them plotted versus time:
− maximum RMS sound pressure value calculated on the basis of the signals of 12 AIC sensors;
− maximum sound pressure amplitude of the two characteristic frequencies for which the AIC is tuned;
− electrical power output of the gas turbine.

RMS value of the sound pressure (max. of 12 sensors) RMS value of the sound pressure (max. of 12 sensors)
50 50

40 AIC system on 40 AIC system on

arbitrary units [-]

arbitrary units [-]

control of the control of the control of the control of the
30 270 Hz oscillation 170 Hz oscillation 30 170 Hz oscillation 270 Hz osc.

20 20

10 10

0 0
0 2 4 6 8 10 12 14 16 18 20 22 0 2 4 6 8 10 12 14 16
time [min] time [min]

sound pressure amp. at 170 and 270 Hz (max. of 12 sensors) sound pressure amp. at 170 and 270 Hz (max. of 12 sensors)
50 50
amplitude of the
40 amplitude of the 40 amplitude of the
arbitrary units [-]

170 Hz oscillation arbitrary units [-]

270 Hz oscillation amplitude of the 270 Hz oscillation
30 30 170 Hz oscillation

20 20

10 10

0 0
0 2 4 6 8 10 12 14 16 18 20 22 0 2 4 6 8 10 12 14 16
time [min] time [min]

electric power electric power

100 100
% of base load

% of base load

80 80

60 60

40 40

20 20

0 0
0 2 4 6 8 10 12 14 16 18 20 22 0 2 4 6 8 10 12 14 16
time [min] time [min]

Fig. 11: Load increase to base load (left) and decrease to part load (right) in premixed operation with
activated AIC system. While running up the gas turbine, the system must be activated earlier
than when the gas turbine is being run down. The AIC system was also always activated at base
load to suppress combustion instabilities.
First the AIC is tuned to the third harmonic at approx. 270 Hz, and automatically activated at 60% of
base load power. Immediately after AIC activation, the amplitudes characterising the oscillations already
present are damped by approx. 65% and remain at more or less this level, as the load is further increased
up to 80% of base load power. At about that level, the second harmonic is excited at around 170 Hz,
while the amplitude of the third harmonic continues to decrease further. By converting the AIC target
frequency to the one characterising the second harmonic, its amplitude is reduced immediately by around
30%. Switching over AIC from the third to the second harmonic, as described, was required since, at that
point of the test runs, simultaneously damping even and odd-numbered modes had not yet been
implemented for the AIC system.
The same damping efficiency as shown above for the case of a load increase was likewise achieved when
the load level was decreased even though, in this case, switching from the second to the third harmonic
did not produce any significant amplitudes. The right-hand trend diagrams in Fig. 11 demonstrate this
AIC operation.
 6-13

13
Burner configuration C: active control at start-up and at lower part load
operation of the gas turbine
This type of gas turbine is started - and subsequently operated, at its lower part load range – exclusively
in diffusion mode where flames are easier to stabilise despite the low level of combustion chamber
temperatures still prevailing. Once a certain temperature limit has been reached, there will be switch-over
to mixed operation. In this lower part load range – from start to switch-over into mixed operation –
burner configuration C will be subject to problems caused by self-excited combustion oscillations, with
the second and the fourth harmonics being excited simultaneously within the ring combustion chamber.
Even though the AIC application presented relies on modulating pilot gas mass flow and pilot flames
which are, in principle, provided only for premixed operations, it proved possible to use the AIC system
successfully even during part load operations. For this purpose, the pilot burners were used together with
diffusion flames, and the AIC system was activated. The system was tuned for the two frequencies
excited.
As shown in Fig. 12, which displays the maximum frequency spectrum of the 12 AIC sensors with and
without AIC for this point within the operational range, it was possible to damp both excited frequencies
almost completely. For the second and fourth harmonics, damping levels amounted to 20 dB and 14 dB
respectively.
5

0
sound pressure amplitude [dB]

-5

-10

-15

-20

-25

-30

-35
0 200 400 600 800

frequency [Hz]
AIC o n AIC off

Fig. 12: Suppression of two frequency peaks by AIC during part load operation.

800
reduction of AIC damps only
700 AIC output for the one frequency
4th harmonic (2nd harmonic)
600
AIC off
frequency [Hz]

500
AIC on AIC on
400

300 4t h harm.

200
2n d harm.
100

0
0 50 100 150 200
time t [s]

Fig. 13: Separate damping of two dominant eigenmodes of a combustion instability by AIC.
6-14

14
The separate damping of two eigenmodes by AIC, as described above, is exemplified again in Fig. 13.
The diagram plots sound pressure as a spectrogram (maximum values for the 12 sensors) versus time.
Dark areas indicate high amplitudes, lighter areas the general noise level. The two horizontal bars at
completely deactivated AIC mark the two eigenmodes excited, in this case, at 145 Hz and 290 Hz.
Starting at t = 70 s, by slowly reducing the AIC output signal to suppress the fourth harmonic at 290 Hz
within the 70 s < t < 85 s range, the amplitudes of this harmonic increase. During the next 23 seconds,
AIC only damps oscillations at 145 Hz. By completely switching off the AIC system at t=108 s, the
second harmonic is likewise excited at 145 Hz. By reactivating the AIC for both frequencies at t=127 s,
the two oscillations are damped out completely once again. This example demonstrates that both
oscillations can be self-excited independently of each other and that in general - if combustion
oscillations are to be avoided completely - both will have to be damped by using AIC.

AIC fault tolerance and long-term experiences

In order to research the influence of a failure of any feedback loop on AIC damping characteristics, for
instance, due to defective pressure transducers or valves, specific loops were switched off on purpose
during test runs. It was shown that reducing the number of feedback loops to ten – corresponding to no
more than 20 active valves – had no significant influence on damping characteristics.
In test runs to determine the long-term suitability of the AIC system, it was kept operating for more than
8,000 hours in some gas turbines. No valve failure and no decrease in the damping efficiency occurred
and abrasion of the moving parts of actuators was negligible. During these long-term tests, AIC systems
were operated automatically in conjunction with the corresponding gas turbine control units and had to
control automatically a wide range of operating conditions and situations so as to damp any arising
combustion oscillations safely.

Conclusions
As compared to passive measures, AIC systems have a number of advantages. While the damping effect
of passive measures frequently remains restricted to narrow frequency or operating ranges of the
combustion system concerned, the AIC is distinguished by its high degree of flexibility and its wide
range of operation. The examples described – different types of oscillation problems occurring for a
variety of operational situations being solved successfully – make this obvious. For damping low
frequency oscillations numerous passive measures such as Helmholtz resonators or sound absorbers
require a great deal of space and thus cause much unwanted design complications, while the AIC system
needs little space and can be installed in a comparatively simple manner.
Another AIC advantage is that – thanks to its degree of development reached by now – it takes
significantly less time to install and requires fewer tests. Accordingly, novel combustion systems can be
commissioned faster and take less time to market. By comparison, developing suitable passive measures
is still mostly a trial-and-error process and takes much experimenting, i.e. time and money. Thus, higher
expenses frequently required for AIC systems soon pay off in most cases.
In addition to the already mentioned benefits of AIC, this system may render it possible to operate certain
combustion systems at lower emission levels. In the tests run in premixed mode presented above, it was
possible with an activated AIC to operate the gas turbine at lower levels of pilot gas mass flows than
without the system. Since pilot flames are important for influencing NOx emissions of gas turbines, the
AIC system allowed to reduce them by more than 60% in certain cases.

References
1 Culick, F. E. C., "Combustion Instabilities in Liquid-Fuelled Propulsion Systems - An Overview",
AGARD Conference on Combustion Instabilities in Liquid-Fuelled Propulsion Systems, Bath,
AGARD-CP-450, pp. 1-1 - 1-73, (1988).
 6-15

15
2 Konrad, W., Brehm, B., Kameier, F., Freeman, C., Day, I. J., "Combustion Instability Investigations
On The BR710 Jet Engine", ASME-Paper No. 96-TA-36, presented November 5-7, 1996 at the
ASME Turbo Asia Conference, Jakarta, Indonesia, (1996).
3 Scalzo, A.J., Sharkey, W.T., Emmerling, W. C., "Solution of Combustor Noise in a Coal Gasification
Cogeneration Application of 100-MW-Class Combustion Turbines", Transactions of the ASME
Journal of Engineering for Gas Turbines and Power, Vol. 112/39, (1996).
4 Berenbrink, P., Hoffmann, S.: Suppression of Dynamic Combustion Instabilities by Passive and
Active Means. 2000-GT-0079, Proceedings of: ASME TURBOEXPO 2000, May 8-11, Munich,
Germany, (2000).
5 Tsien, H. S.: Servo-stabilisation of combustion in rocket motors. Am. Rocket Soc. J., Vol. 22, pp.
256-263, (1952).
6 Lang, W., Poinsot, T., and Candel, S., "Active Control of Combustion Instability", Combustion and
Flame, Vol. 70, pp. 281-289, (1987).
7 Gulati, A., Mani, R., “Active Control of Unsteady Combustion-Induced Oscillations,” J. Propulsion
and Power, Vol. 8, No. 5, pp. 1109-1115, (1992).
8 Poinsot, T., Veynante, Bourienne, F., Candel, S., and Esposito, E., "Initiation and Suppression of
Combustion Instabilities by Active Control", 22nd Symposium (International) on Combustion, Seattle,
pp. 1363-1370, (1988).
9 Bloxsidge, G., Dowling, A., Hooper, N., Langhorne, P.: Active control of reheat buzz. AIAA Journal,
Vol. 26 (7), pp. 783-790, (1988).
10 Langhorn, P.J.: “Reheat Buzz an acoustically coupled combustion instability,” J. Fluid Mech. vol.
193, pp. 417-443, (1988).
11 Poinsot, T., Candel, S., Esposito, E., Lang, W., Bourienne, F., “Suppression of combustion
instabilities by active control,” Journal of Propulsion and Power, Vol. 5, pp. 14-20, (1989).
12 Wilson, K. J.; Gutmark, E.; Schadwo, K. C.: Flame-Kernel Pulse Actuator for Active Combustion
Control. ASME 1992, Active Control of Noise and Vibration, DSC-Vol. 38, pp. 75-81, (1992).
13 Yu, K.H., Parr, T.P., Wilson, K.J. Schadow, K.C., Gutmark, E.J., “Active Control of Liquid-Fueled
Combustion Using Periodic Vortex-Droplet Interaction, “Twenty-sixth Symposium (International) on
Combustion, Vol. 2, pp. 2843-2850, (1996).
14 Yu, K., Wilson, K.J., Schadow, K.C.: “Scale-Up Experiments on Liquid-Fueled Active Combustion
Control,” AIAA 98-3211, (1998).
15 Hermann, J., Gleis, S., and Vortmeyer, D., "Active Instability Control (AIC) of Spray Combustors by
Modulation of the Liquid Fuel Flow Rate,", Combust. Sci. and Tech., Vol. 118, pp. 1-25, (1996).
16 Hantschk, C., Hermann, J., and Vortmeyer, D., "Active Instability Control with Direct Drive Servo Valves
in Liquid-Fuelled Combustion Systems," 26. Int. Symp. on Combustion, Naples, (1996).
17 Seume, J. R., Vortmeyer, N., Krause, W., Hermann, J., Hantschk, C.-C., Zangl, P., Gleis, S., Vortmeyer,
D., Orthmann, A., "Application of Active Combustion Instability Control to a Heavy Duty Gas Turbine",
ASME Paper No. 97-AA-119, presented September 30 – October 2, 1997 at the ASME ASIA ´97
Congress & Exhibition, Singapore, (1997).
18 Hoffmann, S., Weber, G., Judith, H., Hermann, J. and Orthmann, A., “Application of Active Combustion
Control to Siemens Heavy Duty Gas Turbines,” Presented at the Symposium of the AVT Panel on Gas
Turbine Engine Combustion, Emissions and Alternative Fuels, Lisbon 12-16, October, RTO Meeting
Proceedings 14, pp. 40-1 – 40-13, (1998).
19 Hermann, J., Orthmann, A., Hoffmann, S., „Application of Active instability control
to a heavy duty gas turbine,“ XIV ISABE, 5-10 Sept., Florence, Italy, A99-34186, (1999).
 6-16

16
20 Moran, A.J., Steele, D., Dowling, A.P., “Active Control and its Application,” NATO/RTO Active Control
Symposium, May 8-12, Braunschweig, Germany, (2000).
21 Hibshman, JR, Cohen, JM, Banaszuk, A, Anderson, TJ, Alholm, HA, “ Active Control of Combustion
Instability in a Liquid-Fueled Sector Combustor,” ASME, 99-GT-215, (1999).
22 Rayleigh, Lord J. W. S., "The Explanation of Certain Acoustical Phenomena", Nature, July 18, pp.
319-321, (1878).
23 Neumeier, Y., Zinn, B. T., “Theoretical and Experimental Investigation of the Performance of an Actively
controlled Fuel Actuator,” 36th Aerospace Science Meeting, Reno, NV, AIAA-98-0355, (1998).
24 Dowling, P. A., “Active Control of Instabilities in Gas Turbines,” NATO/RTO Active Control
Symposium, May 8-12, Braunschweig, Germany, (2000).

View publication stats