Understanding and Control of Combustion Understanding and Control of - - PowerPoint PPT Presentation

Understanding and Control of Combustion Understanding and Control of Combustion Dynamics in Gas Turbine Combustors Dynamics in Gas Turbine Combustors

Georgia Institute of Technology Georgia Institute of Technology

Ben T. Zinn, Tim Lieuwen, Yedidia Neumeier, and Ben Bellows

SCIES Project 02-01-SR095 DOE COOPERATIVE AGREEMENT DE-FC26-02NT41431

Tom J. George, Program Manager, DOE/NETL Richard Wenglarz, Manager of Research, SCIES

Project Awarded (05/01/2002, 36 Month Duration) $452,695 Total Contract Value

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Gas Turbine Need Gas Turbine Need

• Need: Gas turbine reliability and availability is

important factor affecting power plant economics

− Problem: Combustion driven oscillations severely reduce part life, requiring substantially more frequent outages

• Ultimately affects consumer through price of electricity
• Need: Maximum gas turbine power output is

needed in order to meet growing demand

− Problem: Combustion driven oscillations often necessitate de-rating of turbine power output

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Project Objectives Project Objectives

• Task 1 - Improved understanding of combustion

driven oscillations

− Will improve capabilities for designing combustors with reduced dynamics problems

• Task 2 - Active control of combustion driven
• scillations

− Will improve capabilities for suppressing detrimental dynamics

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Project Schedule Project Schedule

Month

36

Task 1 Improved Understanding of Combustion Dynamics

Sub-task 1.1 - Turbulent Flame-Acoustic Wave Interactions

• 1. Low frequency turbulent flame-acoustic wave interaction modelling
• 2. Multi-connected flame fronts modelling
• 3. Experimental assessment of model predictions .

Sub-task 1.2 – Measurements and Physics-based Models of Background Noise Effects

• 1. Additive combustor noise source modelling
• 2. Parametric combustor noise source modelling

3.Measure background noise sources

• 4. Experimentally investigate noise effects
• 5. Experimentally investigate noise effects upon instability amplitude
• 6. Identify dominant background noise effects

Sub-task 1.3 Measurements and Modeling of Nonlinear Combustor Characteristics

• 1. Experimental transfer function measurements
• 2. Deterministic flame dynamics modelling
• 3. Stochastic flame dynamics modelling

Sub-task 1.4 - Evaluation of Modeling/Analysis Tools Upon Full Scale Data From Industrial Partner

Task 2 Active Control of Combustion Dynamics

Subtask 2.1 - Experimental Studies of Active Control Authority

• 1. Experimental studies of operating condition affects upon active control

authority 2.Experimental studies of background noise effects upon control authority

• 3. Experimental studies of time delay affects upon control authority

Sub-task 2.2 Modeling and Analysis of Active Control Authority

• 1. PDF modeling of parametric noise effects
• 2. PDF modeling incorporating active control terms
• 3. Statistical modeling incorporating time delays

Sub-task 2.3 - Control Authority Tests on Full Scale System

Write Final Report

24 30 1 6 12 18

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Accomplishments Accomplishments

• High impact accomplishments to date:

− Improved understanding of factors that affect instability amplitude

• Experimental characterization of combustion process

nonlinearities

• Developed and validated theoretical analysis for prediction of

flame nonlinearities − Improved methods for active instability control

• Demonstrated open loop control of instabilities
• Improved understanding of factors influencing open loop

control effectiveness − Developed and validated models of turbulent flame/acoustic wave interactions that occur during screeching instabilities

• Results are improving understanding of

combustion instability physics and methods of suppressing oscillations

Experimental Characterization of Heat Experimental Characterization of Heat Release Nonlinearities Release Nonlinearities

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Motivation: Linear and Nonlinear Processes in Motivation: Linear and Nonlinear Processes in Unstable Combustors Unstable Combustors

• Linear processes

− Cause inherent disturbances to become self excited and grow in amplitude exponentially, A~eαt

• Nonlinear processes

− Saturate amplitude of self-excited

• scillations

− Amplitude prediction capabilities require understanding nonlinearities!

• Objective of this part of work is to

measure shape of “Driving” curve

Driving/Damping Driving α Driving/Damping Driving α Amplitude Driving/Damping Damping Driving α

A 3 A 2

1

A

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Experimental Approach Experimental Approach

• Determine transfer function between chemiluminescence and

flow forcing amplitude − Dependence upon driving frequency, flow rate, equivalence ratio − Reactants premixed ahead of choke point to ensure constant fuel/air ratio − Reynolds Number based on premixer exit diameter: 21000 – 43000 (mean velocity = 20-45 m/s) − Amplitude dependence of transfer function determined at 96 conditions/frequencies

• Key Findings:

− Flame response nonlinearities significantly more complicated and varied than simple saturation − Mechanisms identified:

• Amplitude-dependent flame liftoff
• Vortex roll-up
• Excitation of parametric instability

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Nonlinear Transfer Function Nonlinear Transfer Function

• CH*’-u’

relationship remains linear up to ~ 35% of mean velocity

• CH*’ response

saturates at large amplitudes of driving

0.2 0.4 0.6 0.8 0.1 0.2 0.3 0.4 0.5 u′ / uo CH*′ / CH*o

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Saturation Amplitude Can Vary Saturation Amplitude Can Vary Substantially! Substantially!

• Similar saturation

value as assumed in Dowling nonlinear flame model (temporary global extinction)

• Mechanism is not

instantaneous heat release equaling zero here, but flame liftoff

0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1

u′ / uo

CH*′ / CH*o 130 Hz 140 Hz 150 Hz

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Nonlinear Flame Response More Nonlinear Flame Response More Complicated than Simple Saturation Complicated than Simple Saturation

• Very similar

behavior to recent

• bservations of

Balachandran et

• al. (C&F, 2005)
• Reynolds number

~21000, fdrive = 410 Hz

0.1 0.2 0.3 0.4 0.5 0.6 0.1 0.2 0.3 0.4 0.5 0.6 u′/ uo CH*′ / CH*o

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Even More Complicated Nonlinear Flame Even More Complicated Nonlinear Flame Response Observed as Well Response Observed as Well

• Transfer function

shape changes drastically

− Chemiluminescence initially increases then sharply decreases followed by further increase

• Response of flame

shifted to 1st harmonic

• Reynolds number

~ 30000

0.1 0.2 0.3 0.4 0.5 0.05 0.1 0.15 0.2 0.25 0.3 u′/uo CH*′ / CH*o 160 Hz 170 Hz 180 Hz

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Summary of Nonlinear Flame Characteristics

• Characterization of

flame nonlinearities substantially more complicated than simple saturation amplitude

• Here, we plot

amplitude at which nonlinearity is first

• bserved.
• Results indicate that

variety of behaviors (shape, mechanisms) exist in single combustor

100 150 200 250 300 350 400 450 0.2 0.4 0.6 0.8 1 Frequency (Hz) CH* Nonlinear Amplitude Re = 21000 Re = 30000 Re = 43000

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Mechanisms of Nonlinearity Mechanisms of Nonlinearity

• Performed Large Number of OH-PLIF Imaging

Studies to Elucidate Flame Dynamics at two driving frequencies- 130 and 410 Hz

− 5 driving amplitudes − 8 phases taken during cycle, for total of 4000 images per data set

• Many thanks to D. Santavicca and J.G. Lee for

their assistance and advice!

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Simultaneous OH Simultaneous OH-

• PLIF Imaging to

PLIF Imaging to Elucidate Flame Dynamics Elucidate Flame Dynamics -

• 410 Hz

410 Hz

• Subsequent

images taken at two indicated driving amplitudes

0.1 0.2 0.3 0.4 0.5 0.6 0.1 0.2 0.3 0.4 0.5 0.6 u′ / uo CH*′ / CH*o

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Low Low Amplitude Forcing Amplitude Forcing

• Fdrive = 410 Hz
• Convecting

structures can be seen in some images

• Any

suggestions for good averaging techniques that don’t turn images into mush?

0° 45° 90° 135° 315° 270° 225° 180°

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Large Amplitude Forcing Large Amplitude Forcing

• Fdrive = 410 Hz
• Large

amplitude driving

− Flame liftoff throughout driving cycle − Stabilization point of flame moves from centerbody to local low velocity location downstream

0° 45° 90° 135° 315° 270° 225° 180°

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

0.1 0.2 0.3 0.4 0.5 0.6 0.1 0.2 0.3 0.4 0.5 0.6 u′/ uo CH*′ / CH*o

Simultaneous OH Simultaneous OH-

• PLIF Imaging to

PLIF Imaging to Elucidate Flame Dynamics Elucidate Flame Dynamics -

• 410 Hz

410 Hz

Initiation of Flame liftoff behavior coincides with saturation point

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Simultaneous OH Simultaneous OH-

• PLIF Imaging to

PLIF Imaging to Elucidate Flame Dynamics Elucidate Flame Dynamics -

• 130 Hz

130 Hz

• Subsequent

images taken at two indicated driving amplitudes

0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 u′ / uo CH*′ / CH*o

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Low Low Amplitude Forcing Amplitude Forcing

• Well-

defined flame position, structure throughout driving cycle

0° 45° 90° 135° 315° 270° 225° 180°

• Fdrive =

130 Hz

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Large Amplitude Forcing Large Amplitude Forcing

• Turbulent flame speed is apparently modulated –

peaks at highest instantaneous velocity

• Vortex rollup with occasional merging (3-D effect)
• Fdrive =

130 Hz

0° 45° 90° 135° 315° 270° 225° 180°

Frequency Locking and Open Loop Control

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Objective Objective

• Investigate nonlinear interaction between

driven acoustic oscillation and natural combustor mode during unstable combustion

• Determine important parameters which are

affected by frequency spacing between driven

• scillation and combustor mode
• Investigate the effectiveness of open-loop

control on reduction in acoustic power in combustor

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Effect of Acoustic Forcing on Instability Effect of Acoustic Forcing on Instability Amplitude Amplitude

200 400 600 800 1000 0.005 0.01 0.015 Frequency (Hz) p′ / po

Driven Oscillation Unstable Mode Reduction in instability amplitude by

• pen-loop forcing

200 400 600 800 1000 0.005 0.01 0.015 Frequency (Hz) p′ / po

Driven Oscillation Unstable Mode Reduction in instability amplitude by

• pen-loop forcing

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Typical Result Typical Result

• Investigated

parameters in this study −Ledge Width, AL −Instability Rolloff, δp −Entrainment Amplitude, AE

0.1 0.2 0.3 0.4 0.005 0.01 0.015 0.02 u′drive / uo p′instability / po Increasing Driving Decreasing Driving RMS Pressure

AE AL δp

0.1 0.2 0.3 0.4 0.005 0.01 0.015 0.02 u′drive / uo p′instability / po Increasing Driving Decreasing Driving RMS Pressure

AE AL δp

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Pressure Entrainment Amplitude Pressure Entrainment Amplitude Characteristics Characteristics

• Pressure Entrainment

exhibits nonlinear frequency dependence

− Is velocity a better parameter?

150 200 250 300 350 400 450 3 4 5 6 7 8 9 10 x 10

• 3

Driving Frequency (Hz) p′entrainment / po

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Entrainment Amplitude Characteristics Entrainment Amplitude Characteristics

• Entrainment

amplitude increases with increasing frequency spacing

• More intuitive

result compared to pressure dependence

150 200 250 300 350 400 450 0.05 0.1 0.15 0.2 0.25 0.3 Frequency (Hz) AE 150 200 250 300 350 400 450 0.05 0.1 0.15 0.2 0.25 0.3 Frequency (Hz) AE

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

150 200 250 300 350 400 450 70 75 80 85 90 95 Frequency (Hz) Maximum Reduction in Acoustic Power (%)

Acoustic Power Reduction Acoustic Power Reduction

• Acoustic power reduced by at least 70%. Best results seen

where pressure entrainment amplitude is minimized.

CLEMSONPRES.PPT, 10/28/2003, B.T. ZINN, T. LIEUWEN, Y. NEUMEIER

Concluding Remarks Concluding Remarks

• Experimental studies of flame nonlinearity

− Nonlinear flame characteristics significantly more complicated than simple saturation − Shape of transfer function is a function of frequency, Reynolds number

• Single combustor can exhibit a variety of behaviors

− Mechanisms identified:

• Amplitude-dependent flame liftoff
• Vortex roll-up
• Excitation of parametric instability
• Nonlinear Entrainment studies

− Study clarifies nonlinear interactions between driven acoustic oscillations and unstable combustor modes

• Velocity entrainment amplitude seen to decrease with decreasing

frequency spacing − Open loop forcing of combustor at frequencies different from unstable mode shown to be quite effective at studied operating condition.

• Reduction in acoustic power up to 90%. Best results occur at pressure

entrainment amplitude minima.