PM emissions from flares: chasms among current emission factors,fundamental studies, and field observations

Current Research Initiatives PM Emissions from Flares: Chasms among Current Emission Factors, Fundamental Studies, and Field Observations Economic analysis of flaring and venting mitigation Fugitive emissions • Optical diagnostic development • Innovative detection schemes Matthew Johnson, Ph.D., P.Eng. Canada Research Chair in Energy & Combustion Generated Air Emissions Mechanical & Aerospace Engineering Carleton, University Mine face emissions from oil sands Ottawa, ON Canada Emissions from liquid storage tanks Propagation dynamics of premixed and partially premixed flames © M. Johnson, 2010 Invited presentation to IFRF TOTeM Maui, HI, Sept 30, 2010 2 Inst it ut e f or Chemical Process and Environment al Technology Emissions from flaring & venting Motivation - Quantifying PM Emissions Particulate matter (PM) emission from flares is a global issue "Flare efficiency" (Carbon conversion efficiency): η= Emissions from flaring • Soot (PM) emissions • Quantitative Field diagnostics for soot • Greenhouse gas emission models / analysis for Alberta Mass of Carbon Converted to CO 2 Mass of Carbon Originally as Fuel • Global gas flaring exceeds 135 billion m3/year • No quantitative approaches exist to measure these emissions • Current "emission factor" models are flawed Speciated emissions: • Key greenhouse gases More than serious heath effects, PM / soot is a key climate forcer (IPCC, AR4, 2007) - CH4, CO2 • Priority pollutants • +0.2 ± 0.15 W/m2 for fossil fuel black carbon • -0.05 ± 0.05 W/m2 for fossil fuel organic carbon • One prominent study suggests, climate forcing of BC could be +0.9 W/m2 (55% that of CO2) (Ramanthan & Carmichael, 2008) - Soot (carbon based PM), SO2, H2S, NOx - Soot has recently been implicated as a key climate forcer (e.g. Ramanathan & Carmichael, 2008; IPCC AR4, 2007) • Minor species - Volatile organic compounds (VOCs) 3 © Prof. Matthew Johnson, Carleton University, 2010 4 1 So what do we actually know about soot from flares? Brief Notes on Soot / PM Emissions To date, not much… Formation exceedingly complex; entails: • Chemical composition of fuel • Turbulent mixing & diffusion of air and fuel species • Rate of heat transfer from flame • Residence time / temperature history through flame One main set of studies on soot from flares via USEPA: • McDaniel (1983) - while focusing on efficiency measurements, measured plume concentrations of soot for flares with smoke suppression disabled • Pohl et al. (1986) - did not report direct emission rate data but concluded for his test conditions, soot accounts for "less than 0.5% of the combustion inefficiencies" Fundamental work on soot emissions from turbulent jet flames No existing practical approaches for quantifying PM in plumes of flares Measuring PM in general is a challenge no matter the source • Faeth et al. (1990s): studies of "overfire soot" for strongly sooting fuels • Becker and Liang (1982): measurements of soot from high momentum jet flames from millimeter scale burners Extremely limited work on small-scale reacting jets in crossflow • Ellzey et al., 1990; University of Alberta, 2002 5 So how are soot emissions from flares currently reported? 6 Introduction: Current Emission Factors In Canada, PM emissions above threshold amount must be reported to the National Pollutant Release Inventory (NPRI) Source • Problematic since PM from flares is not readily measured CAPP Guide The Canadian Association of Petroleum Producers (CAPP) has developed a guide for reporting • Simple emission factors based on volume of gas flared USEPA FIRE v.6.25 - 2.5632 kg soot per 103 m3 fuel flared • Assumed constant under all conditions regardless of flare size, fuel composition, wind effects, flowrates, etc. USEPA AP-42 Vol. I, sect. 13.5 Some obvious questions: • Where does this number come from? Is this approach reasonable? Is there any other alternative? What liabilities might this create? Original Factor as reported Emission Factor (kg PM per 103 m3 fuel) Gas 2.5632 kg PM per 103 m3 fuel 2.5632 "Adjusted" for HHV=45MJ/m3 53 lb PM per 106 ft3 fuel 0.85 Landfill Gas1 17 lb PM per 106 ft3 fuel 0.27 Methane2 0-274 lb PM per 106 BTU 0 - 5301 80% Propylene, 20% Propane 0-274 µg PM per 10-3 m3 exhaust gas3 Not Convertible 80% Propylene, 20% Propane 1.Typ. landfill gas composition: 56% CH4, 37% CO2, 1% O2, and trace amounts of other gases 2.Predominantly enclosed flare measurements at landfill sites 3.The range of 0-274 is based on the "smoking level": non-smoking flares, 0 µg/L; lightly smoking flares, 40 µg/L; average smoking flares, 177 µg/L; and heavily smoking flares, 274 µg/L 7 © Prof. Matthew Johnson, Carleton University, 2010 8 2 Introduction: Current Emission Factors Some Forensics... Source CAPP Guide USEPA FIRE v.6.25 USEPA AP-42 Vol. I, sect. 13.5 Original Factor as reported Emission Factor (kg PM per 103 m3 fuel) Gas 2.5632 kg PM per 103 m3 fuel 2.5632 "Adjusted" for HHV=45MJ/m3 53 lb PM per 106 ft3 fuel 0.85 Landfill Gas1 17 lb PM per 106 ft3 fuel 0.27 Methane2 0-274 lb PM per 106 BTU 0- - 5301 6.28 80% Propylene, 20% Propane 0-274 µg PM per 10-3 m3 exhaust gas3 Not Convertible 80% Propylene, 20% Propane 1.Typ. landfill gas composition: 56% CH4, 37% CO2, 1% O2, and trace amounts of other gases 2.Predominantly enclosed flare measurements at landfill sites 3.The range of 0-274 is based on the "smoking level": non-smoking flares, 0 µg/L; lightly smoking flares, 40 µg/L; average smoking flares, 177 µg/L; and heavily smoking flares, 274 µg/L 9 Current Status of Emission Factors 10 Soot / PM: Current Research Initiatives Two main areas of focus: Error in US EPA emission factor data base has now been formally reported to the correct authorities who are reviewing the appropriate documents 1. Direct measurement of soot emissions from flares in controlled lab setting Even with correction, there are still only three main sources of emission factor data: • Two are based on landfill gas flares (and are likely enclosed flares) • One is based on a propylene/propane flare and is reported in original form as a plume concentration only • Sampling protocol development • Elemental / Organic Carbon measurements • Emission factors development 2. Novel diagnostic to measure soot from flares in the field Single emission factor approach is also overly simplified • Desire simple tool to improve upon qualitative "opacity" • Related work on measuring optical properties of soot aggregates One motivation of current work is to seek better, measurement-based emission factors for PM in flares 11 © Prof. Matthew Johnson, Carleton University, 2010 12 3 Introduction: Flame & Flare "Regimes" Experimental: Burner & Enclosure Regime map of Delichatsios [1993] Sampling Protocol Developed based on Diesel sampling Soot Yield Ys = ρ soot f v , sampleQDT Tsample m& fuelTDT Frg = u e f s3 / 2 ( gd e )1 / 2  ρ∞   ρe    1/ 4 13 Lab-based Experiments 60 50 4-Component Light Mixture 70 4-Component Mean Mixture 4-Component Heavy Mixture 60 50 4-Component Light Mixture 40 30 10 20 20 0 10 10 0 0 N2 C4 CO2 C3 C2 Gas Species C1 H2 He N2 C7 CO2 C6 H2S C5 C4 C3 C2 C1 20 Gas Species 4-Component Heavy Mixture 50 30 30 4-Component Mean Mixture 60 40 40 6-Component Heavy Mixture 70 N2 70 6-Component Heavy Mixture 80 6-Component Mean Mixture 80 CO2 10th, 90th Percentiles • Laser induced incandescence • Gravimetric sampling • EC/OC (NIOSH 5040) 6-Component Mean Mixture 6-Component Light Mixture 90 C4 80 6-Component Light Mixture 100 C3 Data from 2908 distinct oil production sites in Alberta Mean Concentration Controlled experiments using 3 main techniques: C2 90 90 Species Concentration of Surrogate Mixtures (%) 100 C1 Species Concentration in Solution Gas at Sampled Sites (%) Based on 2007 PTAC data from 2908 production sites Species Concentration of Surrogate Mixtures (%) Experimental: Fuel Mixtures 100 14 Gas Species Fuel Mixture Smoke-point (mg/s) Average Raw Mixture 31.2 AVG-6-component 31.8 AVG-4-component 32.8 15 © Prof. Matthew Johnson, Carleton University, 2010 16 4 Soot Emissions from Lab Flares Soot Measurement Results 0.00125 Fuel Composition Order of magnitude change in soot yield for very small changes in fuel composition Composition is critical for meaningful soot factors Not surprisingly, soot emission rate varies with flare conditions • Fuel composition • Flare diameter • Flowrate 2" Diameter Lab-flare Fuel flowrate = 18 SLPM Soot Yield, Ys (kg soot / kg fuel) Data from ongoing experiments 0.001 0.00075 0.0005 0.00025 0 100% CH4 87% CH4 7% C2H6 6% C3H8 70% CH4 20% C2H6 10% C3H8 17 Suggested by Delichatsios [1993], the "global fire Froude number", Frg , appears to show most promise for developing regime based correlations • Flame length Richardson ratio [Becker and Liang, 1982] • Global characteristic residence time [Becker and Liang, 1982] • Buoyant residence time [Canteenwalla, 2007] • Simple residence time [based on Sivathanu and Faeth, 1990] 106 10 5 10 4 10 3 10 2 Turbulent Buoyant (transition shear) nt oy a t Bu yant) ulen n buo b r Tu nsitio (tra Laminar Buoyant 101 • Fire Froude number [Delichatsios, 1993] 10 19 © Prof. Matthew Johnson, Carleton University, 2010 Laminar Momentum Several parameters investigated in an attempt to scale aerodynamic affects Turbulent Momentum Results: Scaling Aerodynamics Source Reynolds Number, Re Results: Scaling Aerodynamics 18 -3 -2 -1 0 10 10 10 10 Global Froude number, Frg 1 10 2 20 5 "Engineering Attempt" EF model "Engineering Attempt" EF model "Assume" constant soot yield at Frf > 0.003 For narrow range of anticipated fuel mixtures, soot yield scales linearly with heating value In the absence of anything better, can form a simple linear EF model based on heating value The chasm remains however, in linking a useful model with fundamental understanding 22 21 PM from flares: Preliminary Conclusions Preliminary Results: EC/OC Results of current work are showing (as expected) that soot emission rate is strongly affected by: • Fuel composition, flare diameter, and flare gas flowrate Current emission factors are oversimplified and are of questionably related origin Best engineering guess based on current data suggests CAPP emission factor for flare generated soot is likely a factor of 2 or more too high • However, we are a long way from having a defensible model based on fundamental understanding of the soot formation process in a flare Measurements of elemental carbon content in flare generated soot Preliminary results suggest that EC is ~95% of total • Good agreement with laser induced incandescence data 23 © Prof. Matthew Johnson, Carleton University, 2010 24 6 Development of a Novel Soot Plume Field Diagnostic Soot / PM: Current Research Initiatives Two main areas of focus: Current standards are based on opacity, which is only qualitatively related to soot emission rate 1. Direct measurement of soot emissions from flares in controlled lab setting • E.g. EPA test method 9, a ‘human observed' standard • Sampling protocol development • Elemental / Organic Carbon measurements • Emission factors development Is there any way to make quantitative measurements in the field? • Would be invaluable not only for corroborating lab-based data but also as a field tool 2. Novel diagnostic to measure soot from flares in the field Ongoing investigation of a novel concept that is now showing considerable promise… • Desire simple tool to improve upon qualitative "opacity" • Related work on measuring optical properties of soot aggregates 25 Novel soot diagnostic: principle 26 New Field Diagnostic for Soot Plumes Idea: Can we use sky-light to make a quantitative, openended, optical measurement of soot in a plume? Mathematical basis: m& soot = Novel camera based technique under development to directly measure strongly sooting flares under field conditions Lab-based development: • Thomson et al., Applied Optics, 2008 • Johnson et al. (1), Environmental Science & Technology, Accepted Sept. 23, 2010. − uρ soot λ ln(τ λ ( y )) dy 6π E (m) λ (1 + ρ sa ) ∫ Initial field trial completed to measure emissions from a large sooting flare in Uzbekistan If we can develop a quantitative system to measure transmissivity, we can make field measurements of soot plumes • First ever, quantitative field measurement of soot from a flare • Johnson et al. (2), currently under review with Env. Sci. Technol. as of Sept. 2010. • Need optical properties of soot (paper submitted to Appl. Phys. B) • Need good estimate of plume velocity (see subsequent slides) 27 © Prof. Matthew Johnson, Carleton University, 2010 28 7 Field testing of new soot diagnostic Methodology for Field testing Sky-LOSA Field data collected in Uzbekistan, July 2008, as part of separate World Bank funded project to estimate flare volumes in collaboration with Dave Picard (Clearstone Eng.) Sky-LOSA (plume transmissivity) acquisitions • 16 bit Peltier-cooled CCD camera with 532nm filter • Commercial lens • Laptop control via custom written acquisition software High-speed images • Casio EX-F1 digital camera • No collection filter: acquisition in the visible • 300 frames per second @ 512 x 384 pixel resolution Soot flux calculated as: & soot = 1 1 m N f Nz N z Nf ∑∑ A ∫ U(y, z )ln (τ (y, z ))dy λ j =1 i =1 29 First test of new Sky-LOSA diagnostic 30 Quantification of Plume Velocity Field measurement of large (1.04 m diameter) flare at a gas plant in Karshi, Uzbekistan Flare gas composition and flow rate data not available "Image correlation velocimetry" to quantitatively measure plume velocity from high-speed video of visible soot plume 31 © Prof. Matthew Johnson, Carleton University, 2010 32 8 Sky-LOSA processing Quantification of Plume Velocity Velocity measurements presented based on available 61s of high-speed video only (18,300 frames) z x y • Uncertainties estimated at 25%, although this could be reduced significantly with better optimization • Short time span of velocity measurements also impacts accuracy Sky interpolation 1.03 z y x Transmissivity τλ 0.60 2.10 Instantaneous Ensemble average u ln[τ λ ] 33 2. Field Diagnostic for Soot Plumes -0.06 34 Conclusions Preliminary results of first tests Current emission factors oversimplified and of questionable accuracy • Results for one (1) flare only and should not be generalized • Analysis currently under review in article submitted to Env. Sci. Technol. Best engineering estimate from current data suggests CAPP factor for flare generated soot is likely >2x too high • Because of regime transitions, experiments (and possibly models) at a variety of scales will be required to develop a defensible emission factors • Some empiricism likely unavoidable -- fundamental understanding of soot formation in turbulent flames remains as the most challenging problem in combustion Soot flux quantified at ##-## g/s • Uncertainty currently estimated to be +/-33% • Approximately equivalent to ##-## buses running continuously (assuming New "Sky-LOSA" field measurement technique continues to show promise and is ready for more field testing 299 mg PM2.5/km [Keogh et al., Env. Sci. Poll. Res, 2010]) 35 © Prof. Matthew Johnson, Carleton University, 2010 36 9 Research Partners Current Research Team Principle Investigators: Project Manager: Michael Layer, Natural Resources Canada • Matthew Johnson, Canada Research Chair in Energy & Combustion Generated Pollutant Emissions, Associate Prof., Carleton University • Kevin Thomson, Research Officer, NRC-ICPET Natural Resources Canada Post. Doctoral Fellows / Research Engineers: • Adam Coderre, M.A.Sc. • James McEwen, M.A.Sc. Robin Devillers, Ph.D. Graduate Students: • Brian Crosland, Ph.D. candidate Jan Gorski, M.A.Sc. candidate • Ian Joynes, M.A.Sc. Candidate Clifton Pereira, M.A.Sc. candidate • Stephen Schoonbaert, M.A.Sc. cand. Patrizio Vena, Ph.D. candidate Recent Alumni: • Carol Brereton, M.A.Sc. • Pervez Canteenwalla, M.A.Sc. 37 © Prof. Matthew Johnson, Carleton University, 2010 Chen Yang, M.A.Sc. 38 10