Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
Propulsion Controls and Diagnostic WorkshopDec. 8-10, 2009
Detection and Control of Instabilities and Blowoff for Low Emissions Combustors
J. Seitzman, T. LieuwenJ. Crawford, R. Kiran
Georgia Institute of Technology
SUP-06-0030 NNX07AC92A
Technical Monitor: J. DeLaat NASA GRC
Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
MotivationMotivation
Adiabatic Flame Temperature (°F)
NO
x (p
pm) a
t 15%
O2
2400 2600 2800 3000 3200 3400 36000
20
40
60
80
100
Fuel Lean
McVey et al., ASME 92-GT-133
• Combustor control can provide optimal tradeoff between low emissions and engine reliability/ operability– static and dynamic instability
(blowout and oscillations)
-5-4-3-2-1012345
0.1 0.15 0.2 0.25 0.3
Time (s)
Aco
ustic
Pre
ssur
e
0
0.050.1
0.15
0.2
0.250.3
0.35
0.4
Che
milu
min
esce
nce
PressureChemiluminescence
OperatingLines
φ
Flow
rate
Safe
LBO
LBOlimit
Margin=φop- φLBO
Heat Release
Hea
t Rel
ease
Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
Combustion Control Combustion Control • Control Needs: Control systems that can 1) provide low emissions
(lean operation) and 2) prevent or recover from loss of dynamic stabilityand 3) prevent loss of static stability
Engine Power Control(Fuel Rate and Placement)
Emissions DynamicStability
StaticStability
Prevention Recovery Prevention Recovery
MarginSensing
MarginSensing
RelightSensing
ω,ϕSensing
Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
NeedsNeeds• Goal: fundamental understanding needed to convert signals
from simple sensors to stability margin estimates and active control inputs
• Tasks
Active DynamicStability Control
Static StabilityMargin Sensing
Local Fuel-AirSensing
Reliable Combustor Operation at Low
Emissions Conditions
Dynamic StabilityMargin SensingAvoiding
Instability
ReducingInstabilityManaging
Fuel Placement
Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
Active Control of Combustion DynamicsActive Control of Combustion Dynamics• State of the art
– widely demonstrated that active control can reduce instability amplitude
– poorly understood variability in effectiveness
• SPL reductionsrange from 3 to 40 dB
– controlled combustors can exhibit significant amplitude modulation
• Goal– model factors that limit
control effectiveness
Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
Modeling ApproachModeling Approach( ) ( ) ( ) ( ) ( ) ( )ttptptptptp ccvv σξτετεωζω +−′+−=+′+′′ 22
• Use Culick’s formulation to describe time evolution of combustor modes with time delays and noise• include both “internal” and control induced delays
• Solve for statistical and spectral characteristics of pressure as function of control time delays and gain
Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
AccomplishmentsAccomplishments
• Developed statistical theory of controlled combustion dynamics including time delays and noise– closed form solutions for statistics and spectra of combustor pressure
• Reproduced common experimental observations of controlled combustor dynamics– e.g., one controller can have significantly different influences on instability,
depending upon “nominal” combustor dynamics
Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
Interplay between Interplay between ““NominalNominal”” Combustor Combustor Dynamics and ControlDynamics and Control
• Increasing control time delays decreases control effectiveness and window of controllability
• Controller impact varies with internal time delays
•Plots show inverse limit cycle amplitude of controlled combustor
Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
Quantifying Controller PerformanceQuantifying Controller Performance• Model captures experimentally
observed– increase of minimum limit cycle
amplitude with controller time delay
– variation of controller performance with internal delays
• Future work - add physics– model predicts much smaller
amplitude modulation than experimentally observed
– currently working on incorporating observer dynamics
Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
Static Stability : LBO Precursors
0 100 200 300 400 500 600 700 800
-0.5
0
0.5
Ban
dpas
sou
tput
0 100 200 300 400 500 600 700 8000
0.5
1
time [msec]
OH
sig
nal
Aco
ustic
Se
nsor
Opt
ical
(OH
*)
Sens
or
Precursor Events
• Goal: Estimate LBO margin in-situ• Approach: Occurrence of local
extinction/re-ignition events = LBO precursors– optical/acoustic detection– e.g., double thresholding of signals
127
mm
70 mm Dia.
Optical fiber
AcousticSensor
CH4
Air
CH4/air mixture
Pilot fuel
Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
LBO Control
0 0.01 0.02 0.03 0.04 0.05 0.06 0.070
0.5
1
1.5
2
2.5
3
φOp -φ LBO
# of
eve
nts
/ sec
Premixed gas combustor - OpticPremixed gas combustor - AcousticLiquid fueled combustor - OpticJet-fuel aviation burner
Num
ber E
vent
s/se
c
φop-φLBO
0 0.01 0.02 0.03 0.04 0.05 0.06 0.070
0.5
1
1.5
2
2.5
3
φOp -φ LBO
# of
eve
nts
/ sec
Premixed gas combustor - OpticPremixed gas combustor - AcousticLiquid fueled combustor - OpticJet-fuel aviation burner
0 0.01 0.02 0.03 0.04 0.05 0.06 0.070
0.5
1
1.5
2
2.5
3
φOp -φ LBO
# of
eve
nts
/ sec
Premixed gas combustor - OpticPremixed gas combustor - AcousticLiquid fueled combustor - OpticJet-fuel aviation burner
Num
ber E
vent
s/se
c
φop-φLBO
• LBO proximity parameter : event rate (avg number of events/sec)– more frequent events as LBO is approached
• Can use as input to LBO avoidance controller– e.g., fuel distribution control
• Static stability can be linked to dynamic instabilityLBO margin sensing in the presence of combustion dynamics?
0 20 40 60 80 100 120 1400.75
0.8
φ
0 20 40 60 80 100 120 1400
5Ev
ent c
ount
0 20 40 60 80 100 120 1400
10
20
time [sec]
Pilo
t %
Unpiloted LBO Limit
Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
Precursor Events: Combustion Dynamics• Two kinds of instability mechanisms
– Equivalence ratio oscillations (w/ Φ' )– Flame-acoustic interactions (w/o Φ')
• Low pass filtering (~50Hz) for improved precursor detection
• Same event signatures as dynamically stable conditions
Fuel Plenum
Choked Valve
Fuel
Optical Fiber
Fuel
AcousticSensor
Time (ms)
0 50 100 150 200 250 3000
0.4
0.8
1.2
1.6Un-filtered
0 50 100 150 200 250 300
0.2
0.4
Low Pass Filtered
Sign
al (a
.u.)
Sign
al (a
.u.)
Low pass filtered
Un-filteredw/ Φ'
Sign
al (a
.u.)
Sign
al (
a.u.
)
0 50 100 150 200 250 3000
0.5
1
1.5
Un-filtered OH*
0 50 100 150 200 250 3000
0.20.40.60.8
Filtered OH*
Time (ms)
Un-filtered
Low pass filtered
w/o Φ'
Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
Precursor Events: Acoustic Signal• Precursors not evident in raw acoustic signal• LP filtering reveals precursor event signature• Same event characteristic shape as dynamically
stable conditions– can use same event sensing algorithm (e.g.,
identification with wavelet filter)
0 50 100 150 200 250 300-4
-2
0
2
4
Unfiltered
0 50 100 150 200 250 300-0.2
-0.1
0
0.1
0.2Low Pass Filtered
Time (ms)
Sig
nal (
a.u.
)S
igna
l (a.
u.)
Un-filtered
Low pass filtered0 50 100 150 200 250 300
-5
0
5Un-filtered Acoustic
0 50 100 150 200 250 300-0.2
0
0.2Filtered Acoustic
Time (ms)
Sig
nal (
a.u.
)S
igna
l (a.
u.)
Low pass filtered
Un-filteredw/o Φ'w/ Φ'
Wavelet
Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
LBO Sensing• Robust optical signal
stability parameter : Stability Index (SI)
SI
0 0.02 0.04 0.06 0.080
0.5
1
1.5
2
WithoutWith Φ'
Φ'
Num
ber o
f eve
nts/
sec
Stability Margin (Φ- ΦLBO)
Acoustic
0 0.02 0.04 0.06 0.080
1
2
3
4
5
6
7
8
WithoutWith Φ'
Φ'
Stability Margin (Φ- ΦLBO)
Stab
ility
Inde
x (S
I) %
Optical
Num
ber o
f eve
nts/
sec
Stability Margin (Φ- ΦLBO)
0 0.02 0.04 0.06 0.080
0.5
1
1.5
2
2.5
WithoutWith
Φ'Φ'
Optical
Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
Single Nozzle LDI System
Inlet
Single nozzle Injector Quartz flame
tube
Quartz optical windowsCooling flow
Water coolingPressure Vessel
Exhaust
• Objective: Examine LBO margin sensing in advanced, low NOx, elevated pressure combustor
Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
LDI Tests – Jet A• Initial shakedown at
atmospheric pressure• Flame zone has lean
signature (blue flame)
T3=740 F (667K), Φ=0.46,Vref = 46 ft/s (14 m/s)
T3=751 F (672K), Φ=0.7, Vref=28 ft/s (8.5 m/s)
Movie
T3=740 F, Vref = 46 ft/s, Φ=0.51-0.37
Detection and Control of InstabilitiesPropul. Controls & Diag. Workshop, 8 Dec 2009
Summary
• Developed statistical theory of controlled combustion dynamics including time delays and noise– closed form solutions for statistics and spectra of
combustor pressure• Reproduced common experimental observations of
controlled combustor dynamics– can use to study limits of control effectiveness and
improve controller design• Demonstrated robust LBO precursor sensing
– robust event sensing algorithms: multiple combustors, with and without dynamics instability
• Currently extending work to NASA single-element LDI combustor