Presented by

Yousheng Zeng, PhD, PE and Jon Morris

Providence

at

A&WMA Louisiana Section 2012 Fall Conference

October 30-31, 2012

A NEW METHOD TO MEASURE FLARE COMBUSTION

EFFICIENCY IN REAL-TIME

1

INTRODUCTION

2

2010 TULSA FLARE STUDY

Captured flare plume gases, analyzed composition, and determined Combustion Efficiency (CE) and Destruction and Removal Efficiency (DRE)

Yielded important findings regarding steam and air assist and their impact on CE and DRE

Demonstrated high variability in CE and DRE – potential benefits if CE could be monitored and fed back to operator in real time

Extremely large effort; not suitable for routine monitoring

FTIR-BASED HYPER-SPECTRAL IMAGER

Advantages – Imaging the flare plume (2-D compared to a path

measurement; 3-D data cube)

– High spectral resolution

Disadvantages – Low frame rate (~1 scan/sec) – flare plume may

have changed significantly during the same measurement cycle

– Specialist required for data reduction and analysis

– Not suitable for unmanned operations or long-term monitoring

4

PASSIVE FTIR

Advantages

– High spectral resolution

Disadvantages

– Path measurement – representativeness and

aiming issues

– Low scan rate with respect to the rate of

change in flare plume

– Specialist required for data reduction and

analysis

5

THE CONCEPT OF FLARE CE MONITOR

Patented and patent-pending technologies using a 4-band MWIR imager that can – Image the flare plume, and

– Measure flare CE at pixel level

– Determine overall flare CE

Major difference from FTIR based measurement – high frame rate: ~30Hz. As a result, temporal and spatial changes of flare plume within each CE measurement cycle become negligible

Design objective: real time CE output suitable for integration into plant data systems for – Flare operators, or/and

– Process control

6

SETUP OF THE NEW FLARE CE MONITOR

7

Sky

Flare

Flare CE

monitor

Not a path measurement

No scanning; high frame rate

No operator required

THEORETICAL BASIS FOR THE NEW FLARE

CE MONITOR

8

𝐶𝐸(%) =𝐶 𝐶𝑂2

𝑛𝑖 𝐶 𝐻𝐶𝑖𝑖 + 𝐶 𝐶𝑂2+ 𝐶 𝐶𝑂 Eq. (1)

Typical Flare Combustion:

Fuel = hydrocarbon (HC); generically expressed as CnHy

CnHy + n𝑦

2O2 → nCO2 +

𝑦

2H2O

plus some CO if combustion is incomplete

Flare Combustion Efficiency (CE):

GENERAL IDEA

Use a 4-band, high frame rate (~30 fps) Infrared (IR) imager to measure CO2, CO, and HC and calculate CE – Band 1 (Ch1) for HC

– Band 2 (Ch2) for CO2

– Band 3 (Ch3) for CO

– Band 4 (Ch4) for reference

Each pixel in the image represents a region in the flare plume; CE measured at pixel level

Relative strength of signals are measured per Eq. (1); calibrations are achieved through the Ref. Channel (Ch4)

Overall CE is determined by averaging pixel level CE on pixels that form the flare plume “envelope” (pattern recognition algorithms are used to determine the envelope)

10

98 98 98 98 98

98 95

95

92 92

92

95 95 95

98

98

70

SELECTION OF FOUR SPECTRAL BANDS

11

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8 5.0

α

λ(µ)

CO2

CO

Propane

Methane

H2O

Ch1

Ch4

Ch2

Ch3

SIMPLIFIED RADIATIVE TRANSFER EQUATION (RTE)

12

𝐼 = 𝜀 λ 𝐵 𝑇𝑏, λ 𝑒𝑥𝑝 −𝛼 λ 𝐶𝐿 + 𝐵 𝑇𝑔, λ − 𝐵 𝑇𝑔, λ 𝑒𝑥𝑝 −𝛼 λ 𝐶𝐿

When Tg >> Tb,

𝐼 ≈ 𝐵 𝑇𝑔, λ − 𝐵 𝑇𝑔, λ 𝑒𝑥𝑝 −𝛼 λ 𝐶𝐿

𝑒𝑥 = 1 +𝑥

1!+𝑥2

2!+𝑥3

3!+ ⋯ , −∞ < 𝑥 < ∞ Taylor Expansion:

When x << 1, 𝑒𝑥 = 1 + 𝑥

𝐼 ≈ 𝐵 𝑇𝑔, λ − 𝐵 𝑇𝑔, λ 1 − 𝛼 λ 𝐶𝐿 = 𝐵 𝑇𝑔, λ 𝛼 λ 𝐶𝐿

𝐶 =𝐼

𝐵 𝑇𝑔,λ 𝛼 λ 𝐿 Eq. (2)

𝐵 𝑇 , λ =2ℎ𝑐2

λ51

𝑒ℎ𝑐

λ𝑘𝐵𝑇 − 1

EQUATION FOR FLARE CE MEASUREMENT

13

𝐶𝐸(%) =

𝐼2𝐵 𝑇𝑔,λ2 𝛼 λ2

𝑛 𝐼1

𝐵 𝑇𝑔,λ1 𝛼 λ2+

𝐼2𝐵 𝑇𝑔,λ2 𝛼 λ2

+𝐼3

𝐵 𝑇𝑔,λ3 𝛼 λ3

Eq. (3)

Substitute C in Eq. (1) with Eq. (2);

Use subscripts 1, 2, 3 for HC (Channel 1), CO2 (Channel 2), and CO (Channel 3), respectively;

Cancel out L; and

Use weighted avg. n and α for HC.

Eq. (2) becomes:

VARIOUS MEASUREMENT APPROACHES CONSIDERED

B(Tg,λi): – Method 1 – Assume the 4 λ’s are

close enough and B(Tg,λ) for the three channels are equal (and cancelled out)

– Method 2 – Use the Ref. Band (Ch4) and Planks law to determine Tg and calculate B(Tg,λi) for other 3 channels – Not desirable and not used at this time

– Method 3 – Calculate ratios of B(Tg,λi)/B(Tg,λRef) in the expected temp. range (e.g., 800-1200 OF)

14

𝐶𝐸(%) =

𝐼2𝐵 𝑇𝑔,λ2 𝛼 λ2

𝑛 𝐼1

𝐵 𝑇𝑔,λ1 𝛼 λ2+

𝐼2𝐵 𝑇𝑔,λ2 𝛼 λ2

+𝐼3

𝐵 𝑇𝑔,λ3 𝛼 λ3

Eq. (3)

0.00E+00

5.00E+08

1.00E+09

1.50E+09

2.00E+09

2.50E+09

3.00E+09

3.00 3.50 4.00 4.50 5.00

B(T

g,λ

)

λ (µ)

VARIOUS MEASUREMENT APPROACHES CONSIDERED

α(λi):

– For CO2 and CO, values of α(λ) are calculated

based on their IR spectra

– For HC, use weighted avg. (need some

knowledge of flare gas composition)

– Correction for H2O – see next slide

15

𝐶𝐸(%) =

𝐼2𝐵 𝑇𝑔,λ2 𝛼 λ2

𝑛 𝐼1

𝐵 𝑇𝑔,λ1 𝛼 λ2+

𝐼2𝐵 𝑇𝑔,λ2 𝛼 λ2

+𝐼3

𝐵 𝑇𝑔,λ3 𝛼 λ3

Eq. (3)

EFFECT OF GAS PHASE H2O

Correction factor=α λ1

α λ4

Where α(λ)=absorption coefficient of H2O at wavelength of Ch1 and Ch4

16

0.00E+00

2.00E+06

4.00E+06

6.00E+06

8.00E+06

1.00E+07

1.20E+07

1.40E+07

1.60E+07

2.652.752.852.953.053.153.253.353.453.553.653.753.853.954.054.154.254.354.454.554.654.754.854.95

IR In

ten

sit

y

Mid-point λ(µ) in each 0.1 µ spectral band

CO2

CO

H2O

Aerosol?

Methane

Ethane

Propane

Butane

Pentane

Benzene

Propene

Total

𝐼1𝐼4=

𝐵 𝑇𝑔, λ1 𝛼 λ1 𝐶𝐻2𝑂𝐿

𝐵 𝑇𝑔, λ4 𝛼 λ4 𝐶𝐻2𝑂𝐿

𝐼1𝐼4=

𝐵 𝑇𝑔, λ1

𝐵 𝑇𝑔, λ4

𝛼 λ1𝛼 λ4

Independent

of H2O conc.!

SIMULATION RESULTS

17

SIMULATED FLARE PLUME

Flare gases: typical

refinery fuel gas (ref. John

Zink Combustion

Handbook)

Assumptions for Base

Case (Case 1):

– Plume temp=800 F

– Plume depth=1 m (3.28 ft.)

– Distance from flare to the

CE monitor=300 ft.

Compound Conc. CO2 12.000% CO 0.200% H2O 15.960%

Methane 0.720% Ethane 0.360%

Propane 0.400% Butane 0.040% Pentane 0.000%

Benzene 0.000% Propene 0.160%

Assumed Composition

in flare plume

SIMULATED RESULT – BASE CASE (CASE 1)

19

77.52% 77.01% 81.59% 79.30%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

90.00%

True CE "Measured"CE-Method 1

"Measured"CE-Method 3

"Measured"CE-Avg. 1&3

CE

SIMULATED RESULTS – CASES 2-7

EFFECT OF FLARE GAS COMPOSITION AND H2O

Case 2: Unburned fuel=1/2 of Case 1

Case 3: Unburned fuel=1/10 of Case 1

Case 4: Double the conc. of ethane and add pentane and benzene

Case 5: Add 10% more H2O in the plume

Case 6: Add 30% more H2O in the plume

Case 7: Add 30% more CO2 in the plume

20

0.0%

20.0%

40.0%

60.0%

80.0%

100.0%

120.0%

Case 2 Case 3 Case 4 Case 5 Case 6 Case 7

CE

True CE M1 M3 Avg. M1&M3

A SIMPLE EXPERIMENT

USING A BUTANE BURNER AND LAB MULTI-SPECTRAL IR IMAGER

21

A SIMPLE FIELD EXPERIMENT

USING 2 SPECTRAL BANDS

Dual cooled Mid-Wave IR cameras with different spectral filtering (i.e., 2 of the 4 bands required for CE monitor)

Emission plume with steam and propane thoroughly mixed

Steam emission rate: 1200 lb/hr

Propane emission rate: 3 lb/hr

Plume diameter at release: 3 inches

Distance from cameras: 175 feet

Demonstrated the ability to identify and isolate propane from steam at the pixel level using spectral radiance

22

EFFECT OF PLUME TEMPERATURE

23

50.0%

55.0%

60.0%

65.0%

70.0%

75.0%

80.0%

85.0%

90.0%

0 500 1000 1500 2000

CE

Flare Plume Temerature (F)

M1 M2 M3 True CE Avg. M1&M3

Note:

pixels are

expected

to cover

flare’s

high and

low T

regions

EFFECT OF ERROR IN

HC ABSORPTION COEFFICIENT ESTIMATION

24

50.0%

55.0%

60.0%

65.0%

70.0%

75.0%

80.0%

85.0%

90.0%

-60% -40% -20% 0% 20% 40% 60%

CE

Error in α(λ) Estimation

M1 M3 Avg. M1&M3 True CE

EFFECT OF BACKGROUND EMISSIVITY

AND FLARE PLUME DEPTH

25

50.0%

55.0%

60.0%

65.0%

70.0%

75.0%

80.0%

85.0%

0 0.1 0.2 0.3 0.4 0.5 0.6

CE

Background Emissivity

M1 M3 Avg. M1& M3 True CE

50.0%

55.0%

60.0%

65.0%

70.0%

75.0%

80.0%

85.0%

0 2 4 6 8 10 12C

E

Flare Plume Depth (ft.)

M1 True CE

CONCLUSION

26

CONCLUSION

Theoretical analysis and model simulation results demonstrate the feasibility of the real-time flare CE monitoring device – Influential factors: plume temperature, estimate of HC

absorption coefficients based on knowledge of flare gas composition

– Less significant factors: plume depth, background, flare gas composition, actual CE

– Calibration is accomplished inherently through the Reference Channel in the CE calculation which relies on relative measurements at the pixel level; no external calibration is required

Field experiment with propane and steam demonstrates the feasibility of cancelling out interferences by using two spectral bands

NEXT STEPS

Perform field experiment using a 42-band

laboratory spectral imager to further prove

the concept and narrow the design

parameters

Design and develop the first prototype

Extensive field testing

Launch commercial product