Smart Sensors for Advanced Combustion

Ron HansonHigh Temperature Gasdynamics Laboratory

Department of Mechanical EngineeringStanford University

GCEP presentation June 14, 2005

MotivationSensor strategySensor for IC-engine research*Sensor for gas-turbine exhaust*Sensor for T in coal-fired power plantReal-time T for combustion control*Opportunities

*see student poster

2

Why Research on Smart Sensors?Stanford has pioneered laser-based sensing for combustion systemsSmart sensors can reduce greenhouse gases in multiple ways

Expeditedevelopment of

emerging energy conversion

technologies

Real-Time Control

Internal Combustion Engines

Facilitate research on new energy conversion

technologies

Improve energy efficiency of existing

technologies

Smart Sensors

Coal-Fired Power Plants

Gas Turbine Engines

Research plan: GCEP-designed smart sensors w/collaborations for practical demonstrations

3

Sensor Strategy: Tunable Diode Laser (TDL) Absorption for Temperature and Gas Composition

Sensing Principle:Absorption of laser light by molecular transitions in combustion gases

Beer’s law: Transmission = I/Io = exp(-kνL)Absorption coefficient kν = f(temperature, pressure, gas composition)

Ratio of absorbance on two molecular transitions yields gas temperature

Measurement of absorbance with temperature yields mole fraction Use of additional lasers can yield more species (wavelength-multiplexing)

λ

Abs

orba

nce

λ2

λ1Absorbance ∝ kν

Tempk ν

2/kν

1

Ratio of peak heightyields gas temperature

4

Wavelength-Multiplexing to Extend Sensing Strategy

Wavelength-multiplexed, diode-laser sensors for simultaneous, in situ measurement of multiple quantities

T, H2O, CO, CO2 demonstrated at StanfordPotentially NO, UHC’s, fuel

Fiber-optic technology enables multiple measurement locationsFiber optics allows remote location of sensorsFiber switches allow multiple measurements paths for one sensor

Multiplexed absorption concept pioneered at Stanford

Detect light

Multiplexed lasers

Disperse transmitted

lightλ1

λ2

λ3λ4

λ1 λ3λ2 λ4

5

Smart Sensor Research at Stanford University

Opportunity:New smart sensors offer the potential for reduced GHG emissions

Enabling new innovative energy conversion conceptsExpediting development of emerging energy-conversion conceptsImproving efficiency and pollutant emissions from existing technologies

Strategy:Develop non-intrusive sensors using NIR absorption and fiber technologyTest sensor concepts with collaborations that enable access to practical engine environments, examples:

IC-engines, gas turbines, coal combustors, and new burner concepts

6

EGR(CO2) fuel

CO, NO, UHCs

Diode Lasers Fiber optics

T, H2O, fuel, EGR

Diode-Laser Sensors for IC-Engine Applications

Examples:HCCI

Ignition controlled by chemical kinetics (not spark or fuel injection)Careful control of temperature and fuel/air mixture composition needed

Hydrogen fueled IC-engines

Motivation: Reduce emissions from a major source of GHGChallenges:

IC-engines have limited optical access and time-varying T and P with soot and droplet scatteringRapid time resolution is crucial

Strategy:Develop novel sensors tailored for IC-engine research

7

Initial Sensor Design for In-cylinder T

7306.75 cm-1

7203.90 cm-1

7606 cm-1Non-resonant

3x 1Multiplexer

1200 g/mm gold grating

Detector

N2 Purged Region

aspheric lens

Single mode fiber

Pitch

5-axis stage

N2 purge tube

Engine(top view)

Catch

valves

flat windows

Multi-mode fiber

Intake humid air

Wavelength-multiplexed laser-based sensor (2 colors for H2O and 1 color off-resonance) Perform measurements in an optical research engine to prove sensing concept

8

Proof-of-Principle Tests @ University of Michigan

0.0

0.2

0.4

0.6

0.8

1.0

Crank Angle Position [deg]

intakeN

orm

aliz

ed d

etec

otr s

igna

l

7203.90 cm-1, water line 7306.75 cm-1, water line Off line

compression

intake valve occlusion

scattering from water droplet formation

360 420 480 540 600 660 7200

3

6

9

12

15

18

600 RPM motored engine

Pre

ssur

e

Initial measurement campaign shows promising resultsDemonstrated feasibility of TDL sensing in IC enginesNoise near TDC produced by excess loading of water vapor to intake air

Potential for significant improvements w/2nd generation sensor design: e.g. line selection to optimize T sensitivity at high P, improved optical engineering to suppress vibration noise,…Sensor can be tailored to investigate various engine operation modes and cycles

e.g. HCCI, Compression Ignition, Spark Ignition, High-EGR, Supercharged, …

540 570 600 630 660 690 720

300

400

500

600

700

800 600 RPM motored engineXH2O

= 0.03

Compression stroke

Tem

pera

ture

[K]

Crank Angle Position [deg]

Measured

Quartz Cylinder Temperature

droplet scattering

9

Tests Planned for 2nd Generation T Sensorin Optical Research Engines @ Sandia National Laboratory

Two state-of-the-art optical engine facilities for sensor validation and testUtilize TDL-based T and XH2O sensor to investigate HCCI engine operationSensor goal: Facilitate development of HCCI engines to reduce GHG emissions

Dual-Engine HCCI LabPI: John Dec

Automotive HCCI LabPI: Richard Steeper

10

Opportunities for Future IC Engine Measurements

Extend T and P sensing ranges to monitor fired-engine operation

Investigate other species (CO, CO2, fuel, UHCs, NO)

Measure non-uniformities in engine using new TDL concepts

Utilize advanced TDL schemes for improved SNR (e.g. WMS)

Monitor at multiple locations: intake, exhaust, in-cylinder

Extend to Diesel engines

Impact: Unique TDL sensors will facilitate development and control of new concept IC-engines

For details see student poster (Mattison)

11

Gas Temperature Sensor for Stationary Gas Turbine ExhaustMotivation:

Innovative in situ exhaust sensors could provide engine health data needed to minimize environmental impact

Optimizing maintenance schedule for planned load managementOptimizing burners for minimal pollutants and GHG emissions

2x1 Multiplexer

Gas TurbineExhaust Casing(cross section)

Diode LaserController

InGaAsdetectorDSP

Laptopcomputer

High-temperaturesingle-modeoptical fiber

High-temperaturemulti-modeoptical fiber

Instrumentation Shed

Laser 1

Laser 2

Rotor

Research Plan:Two-color T sensor designed and tested at SUDemonstrated in exhaust of a 20MW power generator (collaboration with GE Research Center):

For details see student poster (Liu)Readily extended to multiple paths (tomography)

12

Gas Temperature Sensor for Coal-Fired Power Plants

Strategy:Sensors developed at Stanford tested in coal combustors in Colorado and TVA with collaboration with Zolo Technologies:

Provides fiber technology to enable practical implementationProvides access to power plants to test measurement concepts

Motivation:In situ sensors can improve combustion efficiency

Efficiency increase of 1% reduces GHG by 1% and saves $1M/year in fuel (coal) costs for a 600 MW boilerPath-integrated sensing allows optimization of over-fire-air addition

Sensors can optimize maintenanceIdentify burner malfunction from temperature profile

Challenges:Long path, nearly opaque, vibration and thermal movement

13

T Sensor Design via H2O AbsorptionT sensor based on water vapor absorption H2O absorption is strong in combustion gases

4-12% H2O present in combustion products (depends on coal H2O content )Nearly 500,000 lines in HITRAN between 1 and 2 μm

10-4

10-3

10-2

10-1

100

101

S [cm

-2

/atm

]

2.01.81.61.41.21.0Wavelength [μm]

ν2+2ν32ν1+ν2ν1+ν2+ν3

2ν3ν1+ν32ν1

ν2+ν3ν1+ν2

10-4

10-3

10-2

10-1

100

101

S [cm

-2

/atm

]

2.01.81.61.41.21.0Wavelength [μm]

ν2+2ν32ν1+ν2ν1+ν2+ν3

2ν3ν1+ν32ν1

ν2+ν3ν1+ν2

ν2+ν3ν1+ν2

Use Stanford design rules to choose multiple water lines for optimal sensor performance in the harsh environment of the coal combustorExploit 1250-1650 nm region with available telecommunications lasers and optical fiber technology

14

T Sensor Design: H2O Line Selection

Coal combustor has long path lengths (10-20 m) and poor transmissionTherefore need weak lines (database is not accurate or complete)

Strategy: Choose 9 suitable lines, scanned by 3 diode lasersNeed to determine spectroscopic parameters in lab to enable measurements

Comparison of HITRAN and SU dataLines selected for sensor (A-D)

Note: A, C missing from HITRAN

-1.0 -0.5 0.0 0.5 1.00.0

0.1

0.2

0.3

0.4

Abs

orba

nce

Relative wavenumber [cm-1]

Pure H2O, 1095K, 20.75 Torr, L=381 cm measured HITRAN 2004

A

BC

D

Accurate diode laser sensing of temperature requires laboratory measurements of fundamental spectroscopic data

Example of need for accurate spectroscopic data

15

Laboratory Experiments for Quantitative Spectral Data

NEL DFBDiode laser

computer

Detector 1

N2-purged Area

vacuum vacuumH2OThermocouples

30”50”

3-zone furnace

Detector 2

Etalon

Collimating lens

Measure line strength versus T to verify E” assignment and provide quantitative S(T), line broadening, and line shift dataUse multi-pass geometry for accurate measurement of weak linesApproach yields substantial improvements to current spectral database

16

Linestrength Measurements: Example for Line A

Measure absorbance for known H2O in heated cell for S(T)Fit to confirm E” assignment and find S(T0)

500 1000 1500 2000 2500 30000.0000

0.0005

0.0010

0.0015

Line

stre

ngth

[sm

-2/a

tm]

Temperature [K]

Line A measured fitted: E"=3655.53cm-1, S(T0)=6.93 E-8 E"=3655.53cm-1, S(T0) +/- 5%

( ) ( ) ( )( )⎥⎦

⎤⎢⎣

⎡−−−−

×⎥⎥⎦

⎤

⎢⎢⎣

⎡⎟⎟⎠

⎞⎜⎜⎝

⎛−−=

00

00 "exp1

"exp111"exp)(

)(kThcEkThcE

TTkhcE

TQTQ

TSTS

End result: First determination of line strength for this previously unknown hot line

17

Example Results w/ Nine-Line H2O T Sensor: 220MW Plant

Detect light w/ detector/laser

Multiplexed lasers

Disperse transmitted

light

D1 D3D2

Multiplex 3 lasers and scan 9 H2O linesValmont coal-fired 220MW power-plant in Colorado (dry coal)Boltzmann plot fits one temperature(1500K) with XH2O ~ 5.9%Sensor confirms uniform T in combustor, i.e. well-balanced burners with proper over-fire-air flow

1000 2000 3000 4000 5000 6000

4

6

8

10

12

14

16

18

20

22

ln(A

rea i/S

0 i)

E" [cm-1]

measured linear fit: T=1485+/-80 K (5.4%)

A

B

C

D

E

FG

H

I

Laser 1

Laser 2

Laser 3

Example Results w/ 2nd Generation Sensor: 280MW Plant (T vs Time w/Multiple Paths)

TVA: coal-fired power plant280 MW, tangentially firedFiber-coupled sensor (multiple locations)Coarse tomography (4X2) of T field

200 250 300 350 400 450 5001250

1300

1350

1400

1450

1500

1550

12.5

13.0

13.5

14.0

14.5

15.0

15.5

H2O

(%)

Tem

pera

ture

[K]

Time [minutes]

Temperature

H2O (%)

Path5 (SOFA), 10-kHz scan rate, 10-s averaging

2-D temperature map @ 10mSecondary-over-fire air location(Max: 1528K, Min: 1425K)

Thundershowers begin

9.7 m

7.7

m

Note sensitivity of sensor to intake air/fuel conditionsSuccess achieved has led to a proposal for a permanent sensor installation to improve combustor efficiency and reduce GHG emissions

19

Demonstrate Use of T Sensor for Combustion Control

Motivation:Reducing the fuel-air equivalence ratio can improve combustor emissions and combustion efficiencyBut fuel lean systems are notoriously unstableTDL sensors offer the potential for control strategies to extend lean operating limits

Fiber-Coupled Diode Laser

Filtered Detector

Collimator

Jet-A fueled

Challenges:Poor optical access, high noise/vibration, time-varying transmission, and need for high-speed, real-time sensor

Strategy:Develop fast, real-time (2kHz) gas temperature measurement for combustion control

Novel wavelength modulation design to enable real-time data analysisTest Stanford sensors in practical swirl-stabilized burner

Triple annular swirl-stabilized burner at Stanford simulates next-generation gas turbine fuel nozzles

20

HardwareLock-in

Scanned-Wavelength Modulation with 2f Detection:Facilitates Real-Time T Measurement

Use of 2f simplifies data analysis ==> increases measurement bandwidthDevelopment of 2f method for T requires extension of 2f theory

5

4

3

2

1

0

Det

ecto

r si

gnal

(Vol

ts)

1.00.80.60.40.20.0Sampling time (ms)

Laser Scan with Modulation

0.06

0.04

0.02

0.00

-0.02

-0.04

2f S

igna

l (ar

b. u

nits

)1.00.80.60.40.20.0

Sampling time (ms)

2f Line Shape

Ramp

Measurement path

+

Detector signals to lock-in

Diode laser Controller

Computer

Functiongenerator

Diode Laser

Collimator

Modulate at f

Lock-in amplifier

Reference signal

Functiongenerator

Harmonic Signal

21

The transmission coefficient can be written in terms of the modulation and expanded in a Fourier series: The interpretation of 2f signals is simplified for weak absorbances (kvL << 1) :

The 2f Fourier component simplifies as:

The ratio of 2f peak heights is used to determine temperature :

WMS Sensor for Real-Time T@ 2kHz: Relevant Absorption Theory

))2cos(( fta πντ ⋅+ ∑∞

=0

)2cos(),(k

k ftkaH πν=

2f P

eak

Rat

io

Temp

2f S

igna

l

Wavelength

Ratio of 2f peak heightyields gas temperatureλ1

λ2 2f peak height

θθθνφπ

νπ

πda

LpSaH i ⋅⋅+

⋅⋅−= ∫

+

−2cos)cos(),(2

112 1

2 22 2

( cos )cos 2( )( ) ( cos )cos 2

a dH SSH a d

π

ππ

π

φ ν θ θ θνν φ ν θ θ θ

+

−+

−

+ ⋅ ⋅=

+ ⋅ ⋅

∫∫

( ) ( ) ( ) LPXXTPTSLkLk OHOH 22,,11exp φτ νν −=−≈−=

Impact: New 2f T-sensor should have broad impact on energy conversion research and technology

22

Example Results: Sensor Reveals Thermal Acoustic Instability

First application of TDL sensing in a spray flameQuartz duct used to generate natural flame instabilityPower spectrum yields the Trms at the instabilities near 350 and 700 HzScanned-wavelength 2f sensor shows promise for control applications

Fiber-Coupled Diode Laser

Collimator

Filtered Detector

FilteredPMT

CH* chemiluminescence's signal

Microphone

• 2kHz real-time T measurement in unstable ethanol/air flame

10080604020

0

Trm

s[K

]

10008006004002000Frequency [Hz]

3.0x103

2.5

2.0

1.51.0

T [K

]

0.50.40.30.20.10.0Time [s]

FFT

2000 Hz Real-time

23

Example Results: First TDL Sensor Control of Swirl-Stabilized Flame

First application of TDL sensing to stabilize a spray flameNaturally unstable flame from finite length flame ductActive control of air flow suppresses instabilityResults suggest that the T sensor has good potential as a localized control variable, and could be combined with species concentration sensor

For details see student poster (Li)

0.30

0.25

0.20

0.15

0.10

0.05

0.00

Po

wer

Sp

ect

rum

[a.

u. rm

s2]

8006004002000Frequency [Hz]

Control On (10∼10.5 s)

2f Sensor0.30

0.25

0.20

0.15

0.10

0.05

0.00

Po

wer

Sp

ect

rum

[a.

u. rm

s2]

8006004002000Frequency [Hz]

Control Off (9.5∼10 s)

2f Sensor

Control Off Control On

24

Summary of Key Accomplishments – Future Opportunities

Key Accomplishments:Innovation and design of unique in-situ sensors based on fiber-coupled near-infrared diode lasersContributions to the NIR H2O spectral databaseSensors demonstrated in practical environment with help from collaboratorsSensors can reduce GHG emissions in multiple ways:

Improve fuel efficiencyFacilitate R&D of new energy conversion conceptsReduce combustion-generated pollutants

Future Opportunities:Extend sensing concept to mid-infrared for fuels and trace gasesDevelop sensors for applications to other energy conversion schemes; e.g. fuel cellsContinue Stanford’s unique role of sensor innovation, engineering, and transfer to practical combustion systems

25

Acknowledgements

Students: Dan Mattison, Xiang Liu, Hejie Li, and Xin Zhou

Staff: Dr. Dave Davidson and Dr. Jay Jeffries

Collaborators who enabled sensor tests in practical combustion systemsVolker Sick, University of MichiganDennis Seibers, John Dec, and Dick Steeper, Sandia National LaboratoryMark Woodmansee, GE Global Research CenterAndy Sappey, Zolo TechnologyEphraim Gutmark, University of Cincinnati with thanks to GE Aircraft Engines (fuel spray concept) and Goodrich Corporation (fuel injector)