Overview of flaring efficiency studies

Several studies over the past 20 years have evaluated flare efficiencies. These include major flare research programs by the U.S. EPA, the Alberta Research Council, and the University of Alberta. Other work includes a CMA study; surveys of operating flares by BP/Statoil, the German Aerospace Centre and Shell in Europe, and Shell Nigeria; and a multi-year PERF study of external combustion. These studies have consistently shown that appropriately designed and operated flares are highly efficient, converting 98-99.5% or more of the hydrocarbon feed to carbon dioxide and water. However, in certain situations, particularly cases of incomplete liquid knock-out, low BTU flare gas, or high wind conditions, the efficiency can be lower. The Alberta Research Council found efficiencies as low as 62% for unpiloted field pipe flares, with no flame stabilization or wind protection, subjected to liquid hydrocarbon carryover.

Overview of Flaring Efficiency Studies John Cain, Jim Seebold, Lyman Young ChevronTexaco Energy Research and Technology Company November 5, 2002 Overview Several studies over the past 20 years have evaluated flare efficiencies. These include major flare research programs by the U.S. EPA, the Alberta Research Council, and the University of Alberta. Other work includes a CMA study; surveys of operating flares by BP/Statoil, the German Aerospace Centre and Shell in Europe, and Shell Nigeria; and a multi-year PERF study of external combustion. These studies have consistently shown that appropriately designed and operated flares are highly efficient, converting 98­ 99.5% or more of the hydrocarbon feed to carbon dioxide and water. However, in certain situations, particularly cases of incomplete liquid knock-out, low BTU flare gas, or high wind conditions, the efficiency can be lower. The Alberta Research Council found efficiencies as low as 62% for unpiloted field pipe flares, with no flame stabilization or wind protection, subjected to liquid hydrocarbon carryover. Based on the body of flare efficiency work reported in the literature, it can be concluded that flares operate with high efficiency (98% and higher depending upon flare gas composition and other operating parameters) as long as certain good operating practices are followed. These practices1 are: 1. Assure that flares are lit when in use. Continuous presence of a pilot will help achieve this. According to 40CFR60.18, "the presence of a flare pilot flame shall be monitored using a thermocouple or any other equivalent device." 2. Provide flares with adequate knock-out / liquids separation to remove all of the hydrocarbon droplets from the gas. Liquid carry-over can significantly reduce flare inefficiency. 3. Give special consideration to low-Btu gases. Gases with heating values below about 300 Btu/ft3 are prone to inefficient combustion. If possible, these gases should be combined with higher heating value streams before flaring. 40CFR60.18 sets minimum gas heat contents of 300 3 3 Btu/ft for steam or air assisted flares, and 200 Btu/ft for unassisted flares Gases with heating values from about 300 Btu/ft3 to 800 Btu/ft3 are susceptible to inefficient combustion. High wind conditions can reduce efficiency, so flares for these gases should be located so to minimize wind impacts and/or equipped with wind guards to protect the flare tip, and should be designed to assure that the stack exit velocity is sufficient to provide stabilization against the wind. Additionally, too high a stack exit velocity can potentially cause flame instability and reduce combustion efficiency. 40CFR60.18 addresses these considerations with stack 1 The 40CFR60.18 General Requirements for Flares address several of these practices. The EPA test data discussed below were the basis for these regulations. 1 exit velocity requirements which vary according to gas heat content and the presence or absence of assist steam/air. However, these velocity restrictions do not apply to modern high-pressure "sonic" flares. 4. Gas streams containing primarily hydrocarbons, with heat contents of about 800 Btu/ft or greater, will usually burn efficiently, even under high wind conditions. Nonetheless, if possible, location to avoid high wind exposure is preferable. 5. Air or steam injection are frequently used to control soot. 40CFR60.18 requires that flares "be designed for and operated with no visible emissions." Even in sooty flames, however, soot typically represents a very small fraction of the hydrocarbon feed, and does not significantly influence overall flare efficiency. Also, excessive steam or air injection can dilute the BTU value of the flare gas to the point where inefficiency becomes a concern (see #3, above). Alternatively, flares can be designed to operate smokelessly with high pressure, high exit velocity " sonic" flare tips. Brief Overviews of Experimental Programs EPA - Major program, 1981-86. Bench scale and commercial scale tests. Found 9899%+ efficiency as long as flame is stable. Stability primarily tied to gas heating value and stack exit velocity. ARC - Initiated in 1990. The one study, along with its follow-up reports, claiming flares to be inefficient. Two operating flares, both which had liquid carry-over (as reported in the study summary), were sampled. As could be expected, the efficiencies in these cases were low (62%-84%). University o f Alberta - Ongoing program, begun in 1996 to resolve the apparent discrepancy between the EPA studies, which found high flare efficiencies, and the ARC studies, which found low efficiencies in two operating flares. Results to date show that there is no discrepancy. Flares are generally 98%+ efficient, but can be inefficient under certain conditions such as improper liquids removal or low BTU value coupled with high wind. German Ph.D. Dissertation - 1980 evaluation using a refinery gas slip stream and a test flare. Found 99%+ combustion efficiency when the manner of flare operation was representative of plant elevated flares. CMA - June, 1982. Commercial-size flares. 98-99.5%+ efficiency when the flares were operated under conditions representative of industrial operating practices. British Petroleum/Statoil - 1992-1994 field tests of flares at three European refineries, using Differential Absorption Lidar. Various steam rates tested. Efficiencies consistently 98%+. German Aerospace Center/Shell Research (UK) - Field testing in 1997 at three natural gas production sites. Used FTIR. 99.5% efficiencies. 2 Shell Nigeria - Field testing in 1998 on "eight representative flares of varying designs and flow rates." Used open-path IR to analyze emissions. No unconverted hydrocarbons were detected through the entire course of the program. Based on the sensitivity of the measurements, flare efficiencies of 98%+ were estimated for seven of the flares, and 95.7% for the eighth. PERF - 1992-97. Major experimental and computational study of combustion in fired heaters. Demonstrated that combustion is equally effective (i.e., extremely efficient) over a wide range of operating conditions, regardless of whether the feed is natural gas or refinery gas. Although this was not a flare study, the lessons learned provide insight into the chemistry and physics of external combustion in general. Natural Resources Canada - New laboratory work, reported at the AFRC meeting in May, 2002, suggesting that turbulent crosswind can more significantly reduce flare efficiency than laminar crosswind. Flare Efficiency Studies - Additional Details EPA In 1981, EPA began working with Energy and Environmental Research (EER, in Irvine, California) to quantify the emissions from and efficiency of industrial flares. The project continued for five years, and produced a number of EPA reports, conference presentations, and articles in the scientific literature, some of which are listed in the "References" section of this report. ChevronTexaco's Jim Seebold served as Technical Advisor to the project. Most of the work, particularly the initial testing, used propane in nitrogen; however, other gases including natural gas, ammonia, butadiene, ethylene oxide and hydrogen sulfide were also tested. Experimental work was conducted both in small scale apparatus, which allowed all of the combustion products to be collected and analyzed, and in a large test facility built specifically for the flare efficiency test program. Experimental and commercial flare tips up to 12-inch diameter were used. A wide range of experimental conditions were examined. Gas heating values ranged from less than 150 Btu/ft3 to greater than 3000 Btu/ft3, and flare tip exit velocities from less than 1 ft/sec to greater than 400 ft/sec. The effects of soot suppression by steam or air injection were evaluated. Stream-to-gas weight ratios, when steam was used, ranged from 0.069 to 3.7. Air-to-gas momentum ratios, when air assist was used, ranged from 0.1 to 100, and the corresponding stoichiometric ratios from 0.117 to 79. The main conclusion of the program was that flare efficiencies are 98-99% or greater as long as the flame is stable: "Stable flare flames and high (>98-99 percent) combustion and destruction efficiencies are attained when the flares are operated within 3 operating envelopes specific to each flare head and gas mixture tested. Operation beyond the edge of the operating envelope results in rapid flame de-stabilization and a decrease in combustion and destruction efficiencies." The studies identified the heating value and exit velocity of the gas as the two key variables defining the flame stability (and combustion efficiency) operating envelope. The general trend identified was one of increased stability with higher heating value and lower exit velocity.2 For example, based on experiments with propane/nitrogen mixtures, an exit velocity of 1 ft/sec requires a heating value of 350 Btu/ft or greater to guarantee stability, while an exit velocity of 10 ft/sec requires 570 Btu/ft3 or greater. Flare head design and size were found to be of secondary importance. The EPA studies also found that in smoky flares soot typically represented less than 0.5% of the flared hydrocarbons. Submicron carbon particles are very strong light scatterers and absorbers, so a very small quantity produces significant opacity. Steam suppressed soot, but did not significantly influence flare efficiency unless the steam rate was so great (greater than about 0.5 pound steam per pound of fuel) as to begin to quench the flame, in which case combustion efficiency could decrease. The EPA studies did not address the issue of liquid carry-over, which was found to be a very important variable in the subsequent Canadian work, because industrial lowpressure pipe flares are neither designed nor intended to handle liquids. Alberta Research Council In 1990 the Alberta Research Council (ARC), with funding from Environment Canada and other Canadian organizations, began its own study of flare efficiencies and emissions. Over the following years they performed an array of laboratory and pilot plant tests, and measured emissions from two field flares. A summary report was issued in 1996. Consistent with the EPA work, the ARC work found that "pure gas streams such as methane, propane, and commercial natural gas...burn with a high degree of efficiency (98% or greater) under most conditions employed in laboratory or pilot plant tests." The greatest negative effect on efficiency resulted from the addition of liquid hydrocarbon droplets to the gas. The pilot plant tests, which used mixtures of gas and condensate (primarily heptane), also identified crosswinds as a potentially negative influence on flare efficiency. The most contentious p a rt of the ARC investigation was its field testing. According to the 1996 report, "field testing of flares was carried out at two oilfield battery sites, one containing sweet gas and the other sour gas. The sweet gas site had considerable liquid The experiments were conducted in essentially stagnant air. The large-scale flare test rig, in particular, was carefully protected from the wind for safety reasons. As discussed below, later University of Alberta studies found that low exit velocities can lead to instability and inefficient combustion in cases of high wind speed. 2 4 hydrocarbon carryover both to and from the liquid knock-out drum en-route to the flare while the sour site by comparison was considerably drier." Samples of the combustion gases were collected and analyzed by GC. The sweet flare was 12 meters tall, with a 20 cm tip diameter, and burned 8000 m per day of associated gas. Efficiencies varied from 62% to 71%, depending on such variables as the fluid level in the knock-out drum. The sour flare was 15 meters tall, with a 7.6 cm tip diameter, and burned 650 m3/day of associated gas containing 22.8% H2S. The hydrocarbon destruction efficiency was 84%, and the H2S destruction efficiency was 82%. As well as unburned hydrocarbon feed, about 150 VOCs, PAHs, and other compounds, including benzene, styrene, ethylbenzene, naphthalene, toluene, and xylenes, were detected among the combustion products.3 The products from the sour flare also included a variety of sulfur-containing compounds. Despite the low flare efficiencies, the report stated that: "modeling studies...have determined that predicted ground-level concentrations at these two sites would be low in relation to ambient air criteria or observed values." University o f Alberta In 1996 a second Canadian study was begun, this time by the University of Alberta, with the primary objective to "bridge the gap between the ARC and EPA studies." Funding was again provided by Environment Canada and other Canadian organizations. The focus has been on "scaled-down, generic pipe flares under well controlled conditions." This program has found that flares burn very efficiently in stagnant air. As crosswind increases, however, the efficiency can decline. To maintain high efficiency, the gas exit velocity from the stack must be sufficient to provide stabilization against the wind. If the stack exit velocity is too low, strong wind can disperse the gas before it has a chance to burn. Combustion efficiency also improves with increasing BTU content of the gas and with increasing diameter of the stack. For the natural gas case, the study found that natural gas flares burn quite efficiently even in fairly strong winds. For example, a 100 mm (4 inch) diameter flare with an 8 m/sec (26 ft/sec) exit velocity would burn efficiently in winds up to 22 m/sec (73 ft/sec, or 49 mph). Larger diameter flares, or flares with higher exit velocities, would burn efficiently in even stronger winds, because of the stabilizing effects of increased exit velocity and stack diameter. Propane was found to burn efficiently at even higher wind speeds than natural gas. 3 Trace quantities of a wide variety products are expected from all hydrocarbon combustion processes. The PERF project discussed below demonstrated that even for "complete" combustion, sensitive enough analytical techniques will find small amounts of virtually all conceivable products. 5 Fuels with lower heat content, however, were found to be much more susceptible to wind. Experiments with natural gas/CO 2 mixtures, for example, showed that adding 41% CO 2 decreased the allowable wind speed (to maintain 98% efficiency) by about a factor of five. Because of the exponential nature of the relationships, for any given wind speed this would increase the required flare diameter by a factor of 25, or the exit velocity by a factor of 125. Based on this finding, the investigators recommend that flaring of gases 3 3 with heat contents any lower than this (20 MJ/m , or 537 Btu/ft ) be prohibited. In tests of the effect of liquid droplets, entrained water droplets did not affect flare efficiency. The maximum amount of water tested was 42 wt% of the gas rate. Entrained droplets of octane (20% of the total hydrocarbon mass flow to the flare) reduced efficiency from 99% to 93%. The University of Alberta findings are consistent with those of both the EPA and ARC. All three studies found that in the absence of cross-wind and liquid droplets, experimental flares are extremely efficient. Both ARC and the University of Alberta performed further tests on experimental flares with hydrocarbon droplets in the gas, and found that combustion efficiency declines. ARC followed this with field tests of flares with hydrocarbon droplets in the gas, and again found reduced efficiencies. All three studies identified the heating value of the gas as a stabilizing factor. One apparent difference between the EPA and University of Alberta studies concerns the effect of stack exit velocity. EPA reported reduced stability and efficiency if stack exit velocity got too high, while the University of Alberta reported that increasing exit velocity helps stabilize the flame and assure high efficiency. So is high stack exit velocity good or bad? It depends on the situation. Flares, particularly with low-BTU gases, can experience flame-out if the stack exit velocity gets too high. This was the EPA conclusion. However, the University of Alberta found that flares, particularly with low BTU gases, can also experience reduced efficiency if the wind is too strong and the stack exit velocity is too low. Considering both of these effects, flares the handle low heating value gases are susceptible to inefficient combustion unless the stack exit velocity is within the proper range. Flares that handle higher heating value gases, such as gases consisting primarily of hydrocarbons, are more robust. Fridericiana Technical University, Karlsruhe, Germany, Ph.D. Dissertation Published in 1980, the year before the beginning of the EPA study, this K.D. Siegel's dissertation reported on the combustion efficiency of a test flare burning a refinery relief gas slip stream.4 The work was conducted using a 8-inch Flaregas steam smokeless flare tip located at 5 meters above grade. The tip had 6 Flaregas coanda type steam injectors. It was designed for a rated load of 10 t/h of flare gas with a density of 0.75 kg/m . Test flows ranged from 0.13 to 2.9 t/h and density ranged from 0.54 to 1.86 kg/m3. 4 See References: Siegel, 1980. 6 Waste gas flowing to the elevated flare through the main flare header was diverted to the test flare using a liquid seal. Some tests were conducted in a crosswind condition generated by a blower. Blower "wind" ranged up to 6 m/s (13.4 mph). Siegel's results may be stated by quoting from his own summary in the dissertation (translated from the original German text which is rather complex in its phrasing): " The results of combustion tests of a flare operated in a manner representative of plant elevated flares has determined that: * In the case of soot-free flare flames, at least 99% of the organically bound carbon contained in the waste gas is converted to carbon dioxide; * In the case of a smoking flare flame, the emissions of all combustion products except carbon-dioxide and water vapor is a maximum of 1 %." Chemical Manufacturers Association In June, 1982 (concurrent with the beginning of the EPA studies with EER), CMA and EPA sponsored an evaluation of flare efficiencies using a commercial-size flares provided by the John Zink Company. Other participants in the program included Optimetrics and Engineering-Science. The flare gas consisted of propylene-nitrogen mixtures blended to heating values ranging from 80-2183 Btu/ft . Propylene was selected because it is one of the most difficult hydrocarbons to burn. Steam-to-gas weight ratios ranged from 0 to 123. Air assist was also studied, with resultant stoichiometric ratios up to greater than 1. Sampling was performed using a probe suspended by a crane over the flares. The samples extracted by the probe were analyzed by continuous emission monitors to determine concentrations of carbon dioxide, carbon monoxide, total hydrocarbons, sulfur dioxide, NOx, and oxygen. Additionally, integrated samples of the flare gas were collected for hydrocarbon speciation by GC. A variety of operating conditions were run over a three week test period. The study found that "when the flares were operated under conditions which were representative of industrial operating practices, the com bustion efficiencies at the sam pling probe w ere determ ined to be g reater than 98%." Excessive steam (steam-to-gas ratio greater than 5) could quench the flame and lead to less efficient combustion, as could excessive assist air (stoichiometric ratio greater than 0.7). British Petroleum/Statoil From 1992-1994, BP and Statoil (Norway) tested flares in three European refineries5. The flare at one refinery was tested at two different times, with significantly different feed compositions. Overall, the hydrogen contents of the flared gases ranged from 10.8% to 78.1%, the C2+ contents ranged from 11.1% to 83.3%, the average molecular weights varied from 8.8 to 45.3, and the heating values ranged from 47.8 MJ/kg to 59.8 MJ/kg, or 5 See References: Boger, 1996. 7 520 Btu/ft3 to 2460 Btu/ft3. The flare tips diameters were 42 inches and 48 inches. Gas rates ranged from 0.9 to 11 tonnes/hours, and were determined by ultrasonic flow meter, orifice meters, and estimation based flame length using on API RF 521. Steam-to-gas weight ratios were varied within the range of 0-3. Differential Absorption Lidar was used to determine the amounts of methane and C2-C6 saturated hydrocarbons in the flare plume. The ratio of these hydrocarbons to the total carbon content of the feed gas was never greater than 0.7%. Other potential products of incomplete combustion, such as CO and soot, were not quantified. Based on previous studies, however, the authors estimated that the amount of these other products would probably not be greater than the amount of methane plus C2-C6 paraffins. This appears to be a reasonable assumption. The resulting flare efficiencies, therefore, were estimated to be 98%+ in all cases. German Aerospace Center/Shell Research (UK) Testing was conducted in 1997 at four natural gas production sites in the Netherlands, under low flow and high flow conditions6. Low flow corresponded to the continuous combustion of a small quantity of sales gas, containing about 81% methane. High flow occurred when additional gas which was richer in higher hydrocarbons was directed to the flare. Additional details are not provided. Fourier-transform infrared spectroscopy was used to determine the concentrations of methane, higher hydrocarbons, carbon dioxide, and carbon monoxide in the flare plumes. Soot could not be measured, but based on previous work its quantity as a fraction of the total feed carbon was expected to be small. Based on the measured concentrations, the flare efficiencies were calculated to have averaged about 99.5%. Shell Nigeria study on operatingflares In 2000, Shell presented the results of testing conducted in Nigeria on "eight representative flares of varying designs and flow rates."7 Two gas plant and six flow station flares were analyzed. Detailed information is not provided on gas compositions or flow rates, or on the use of steam or air assist. Flare combustion efficiency was measured using open-path IR. No unconverted hydrocarbons were detected through the entire course of the program. Based on the sensitivity of the measurements, flare efficiencies of 98%+ were estimated for seven of the flares, and 95.7% for the eighth. PERF Petroleum Environmental Research Forum (PERF) Project No. 92-19 was a five-year effort to develop a sound technical background for the development of heater and boiler 6 7 See References: Haus, 1998. See References: Ozumba, 2000. 8 regulations. Participants included DOE, EPA, GRI, Amoco, Chevron, Mobil, Shell, Texaco, Southern California Gas Company, Sandia National Laboratories (SNL), Lawrence Livermore National Laboratories (LLNL), and UCLA. Although this was not a flare study, the lessons learned provide insight into the chemistry and physics of external combustion in general. The program consisted of three major tracks: The chemical mechanism track, conducted at LLNL, developed computational models of the chemical mechanisms by which combustion products are produced and destroyed in flames. Supporting data were gathered using laboratory scale flames at SNL and UCLA. The burner experimental track was conducted at Sandia's Burner Engineering Research Laboratory. The emissions resulting from the combustion of different gas mixtures under various operating conditions, including extreme "failure modes" which were anticipated to result in poor combustion, were measured. The field data analysis track, performed at SNL, reviewed the data from previous field studies to relate the combined information from the previous two tracks with industrial experience. The key finding of the program was even under non-optimal conditions, the combustion process is extremely efficient at converting hydrocarbons to the intended products carbon dioxide and water. This was true both for natural gas and for process gases of various compositions. O f p articu lar relevance to flares, the program showed that even though higher hydrocarbons are initially produced in significant amounts within a flame, these species are quickly consumed, even before continued mixing provides stoichiometric air. The same chemistry applies to flare flames, the caveat being the possible negative effective of environmental conditions such as wind and rain. Because of the rapidity of the chemical reactions relative to mixing processes, however, environmental factors would probably need to be quite severe to significantly affect the combustion efficiency. The PERF experimental program used gases with heating values of 800-1500 BTU/scf. As discussed above, gases with significantly lower heating values are more subject to inefficient combustion. Natural Resources Canada At the American Flame Research Committee Spring Meeting on May 8-10, 2002, Natural Resources Canada presented experimental results suggesting that turbulent crosswind can more significantly reduce flare efficiency than laminar crosswind.8 Natural gas, propane, and natural gas/propane mixtures were burned using 1", 2", and 4" flare pipes. Air flow 8 See References: G o g olek and Hayden, 2002. 9 was produced with a high capacity fan, and turbulence was created by blowing the air through metal grids with 1" or 3" slats. For natural gas, combustion efficiency in 20 km/hour crosswind was about 98% for laminar flow (no grid between the fan and the flare), but only about 94% for turbulent flow. At 40 km/hour, the combustion efficiencies were about 95% and 83%, respectively. Given the field testing results described above, this study's applicability to real flares operating around the world is unclear. This is a potential area for additional research. Summary An extensive body of knowledge about flare combustion efficiencies has been developed over the past two decades. Test programs have included laboratory and test facility studies as well as field testing on both upstream and downstream flares under a variety of operating conditions. Based on these studies, the following conclusions can be drawn: The combustion efficiency of properly operated flares burning dry, hydrocarbon gases, under low to moderate wind conditions, is very high. Even under high wind conditions, efficiency can be maintained by sufficient stack exit velocity and by the multiple pilots with continuous pilot monitoring, flame retention and wind protection devices that are found on all modern flares. Liquid droplets in the gas stream can reduce efficiency. Low BTU gases can burn inefficiently when exposed to moderate or high cross­ winds. Combustion is stabilized by increasing stack exit velocity, but this can be a balancing act since low BTU gases can also experience blow-out if the exit velocity gets too high. Given these considerations, the recommended practices listed in the overview at the beginning of this document should prove valuable in assuring high (98-99.5%+) flare combustion efficiency. The recent turbulent crosswind work by Natural Resources Canada suggests a potential area for further investigation. 10 References EPA study "Evaluation of the Efficiency of Industrial Flares: Background - Experimental Design Facility," EPA-600/2-83-070, August, 1983. "Evaluation of the Efficiency of Industrial Flares: Test Results," EPA-600/2-84-095, May, 1984. "Evaluation of the Efficiency of Industrial Flares: Flare Head Design and Gas Composition," EPA-600/2-85-106, September, 1985. "Evaluation of the Efficiency of Industrial Flares: H2S Gas Mixtures and Pilot assisted Flares," EPA-600/2-86-080, September, 1986. J. Pohl, et al., "Combustion Efficiency of Flares," presented at the 77th Annual Meeting of the Air Pollution Control Association, San Francisco, California, June 24-29, 1984. J. Pohl, et al., "Combustion Efficiency of Flares," Combustion Science and Technology, Vol. 50, pp. 217-231, 1986. Alberta Research Council study M. Strosher, "Investigations of Flare Gas Emissions in Alberta," Alberta Research Council final report to Environment Canada Conservation and Protection, the Alberta Energy and Utilities Board, and the Canadian Association of Petroleum Producers, November, 1996. D. Leahey et al., "Theoretical and Observational Assessments of Flare Efficiencies," J. Air & Waste Manage. Assoc., Vol 51, pp. 1610-1616, December, 2001. University of Alberta study M. Johnson et al., "The Combustion Efficiency of a Propane Jet Diffusion Flame in Cross Flow," presented at the Fall Meeting of the Western Section of The Combustion Institute, Seattle, Washington, October 26-28, 1998 M. Johnson et al., "The Combustion Efficiency of Jet Diffusion Flames in Cross-flow," presented at the Joint Meeting of the United States Sections - The Combustion Institute, Washington D.C., March 15-17, 1999 M. Johnson et al., "Effects of a Diluent on the Efficiencies of Jet Diffusion Flames in a Crosswind," presented at The Combustion Institute, Canadian Section, Spring Technical Meeting, Edmonton, Alberta, May 16-19, 1999 11 M. Johnson et al., "Efficiency Measurements of Flares in a Cross Flow," presented at Combustion Canada 1999, Calgary, Alberta, May 26-28, 1999 "University of Alberta Flare Research Project," University of Alberta interim report, December, 2000. CMA study M. Keller and R. Noble, "RACT for VOC - A Burning Issue," Pollution Engineering, July 1983 "A Report on a Flare Efficiency Study, Volume 1," report prepared by EngineeringScience for CMA PERF study "The Origin and Fate of Toxic Combustion Byproducts in Refinery Heaters: Research to Enable Efficient Compliance with the Clean Air Act," PERF 92-19 Final Report, August 5, 1997. Other J. Boden, et al., "Elevated Flare Emissions Measured by Remote Sensing," Petroleum Review, November, 1996, pp. 524-528. P. Gogolek and A. Hayden, "Efficiency of Flare Flames in Turbulent Crosswind," presented at the American Flame Research Committee Spring Meeting, Ottawa, Canada, May 8-10, 2002. R. Haus, et al., "Remote Sensing of Gas Emissions on Natural Gas Flares," Pure Appl. Opt., Vol. 7, pp. 853-862, 1998. C.I. Ozumba and I.C. Okoro (Shell), "SPE:61025 - Combustion Efficiency Measurements of Flares Operated By An Operating Company," 2000. Siegel, K. D., Concerning the degree of conversion of flare gas in refinery elevated flares: Pollutant emissions from refinery elevated flares as a function of their operating condition, Ph.D. Dissertation, Chemical Engineering Department, Fridericiana Technical University, Karlsruhe, Germany, 1980. 12