Combustion efficiency of full scale flares measured using DIAL technology

Flares are used in Alberta's oil and gas industry to safely treat H2S containing gas releases during emergencies, and gas well flow tests and as a means of disposing of gases that cannot be recovered economically. Despite the wide use of these flares, little data exists on the combustion efficiency of full-scale flares. Differential Absorption Light Detection and Ranging (DIAL) is a laser-based optical method that can remotely measure the concentration of hydrocarbons in the plume downwind of a flare. By combining DIAL measured 2D concentration maps with wind speed, the mass emissions of hydrocarbons in a flare plume can be calculated. The hydrocarbon combustion efficiency of the flare can then be determined by comparing hydrocarbon mass flux in the plume to that in the gas going to flare. During the spring of 2003, DIAL was used in Alberta to measure the combustion efficiency of one sour gas well test flare and two solution gas flares during normal operation. These flares covered tip exit velocities from 1 to 23 m/s and H2S concentrations in the flared gas from 0 to 11%. Measured combustion efficiencies based on hydrocarbons remaining in the plume ranged from 74 to 98%.

C o m b u s tio n E ffic ie n c y o f F u ll S c a le F la r e s M e a s u r e d u s in g D I A L T e c h n o lo g y Allan K. Chambers, Alberta Research Council Inc., 250 Karl Clark Rd., Edmonton, Alberta, Canada Tony Wootton, Jan Moncrieff and Philip McCready, Spectrasyne Ltd., 3 The Ringway Centre, Edison Road, Basingstoke, Hampshire, RG21 6YH, U.K. ABSTRACT Flares are used in Alberta's oil and gas industry to safely treat H2S containing gas releases during emergencies, and gas well flow tests and as a means of disposing of gases that cannot be recovered economically. Despite the wide use of these flares, little data exists on the combustion efficiency of full-scale flares. Differential Absorption Light Detection and Ranging (DIAL) is a laser-based optical method that can remotely measure the concentration of hydrocarbons in the plume downwind of a flare. By combining DIAL measured 2D concentration maps with wind speed, the mass emissions of hydrocarbons in a flare plume can be calculated. The hydrocarbon combustion efficiency of the flare can then be determined by comparing hydrocarbon mass flux in the plume to that in the gas going to flare. During the spring of 2003, DIAL was used in Alberta to measure the combustion efficiency of one sour gas well test flare and two solution gas flares during normal operation. These flares covered tip exit velocities from 1 to 23 m/s and H2S concentrations in the flared gas from 0 to 11%. Measured combustion efficiencies based on hydrocarbons remaining in the plume ranged from 74 to 98%. Acknowledgements The authors gratefully acknowledge the funding for this project supplied by Environment Canada, the Canadian Association of Petroleum Producers and the British Columbia Oil and Gas Commission. 1 Introduction The province of Alberta has a large oil and gas industry producing and processing oil and natural gas. Alberta contains both sweet and sour natural gas, with sour gas often containing hydrogen sulfide (H2S) at percentage levels. Solution gas flares are used to safely dispose of gas when the gas cannot be economically recovered for sale. These flares are relatively small flares (average flow rate of 1,000 m3/day) and are typically installed near oil wells to dispose of gas associated with the produced oil. Well test flares are much larger (>10,000 m3/day) and are usually a temporary installation used to dispose of gas produced during well servicing and flow testing. Wind tunnel testing of scaled flares was recently completed in Alberta (Kostiuk et al., 2000). From this testing, correlations were developed to predict combustion inefficiency (100 minus combustion efficiency) for a range of wind speeds, flare tip exit velocities and fuel gas compositions. The flare tip diameter of the tested flares in this and related studies ranged up to 4 inches in diameter. Due to the hazardous nature of H2S, these wind tunnel tests were restricted to sweet gas flares. Flare combustion efficiency has also been measured for larger scale flares using gas sampling probe methods. (Pohl et al., 1984; Strosher, 1996). Strosher (1996) measured the combustion efficiency of a sweet and a sour solution gas flare in Alberta and performed detailed analysis of the hydrocarbons present in the flare plume downwind of the visible flame. Calculated combustion efficiencies were based on the gas sampling and analysis from a probe placed near the centerline of the plume. Measured combustion efficiency of these flares ranged from 62 to 84%. Profiling a full scale flare plume using the conventional probe gas sampling technique is difficult and time consuming. Low concentrations of the combustion products entails sampling at each point for up to 1 hour. Fluctuations in wind speed and direction cause large movements in the flare plume position and complicate both sampling and data analysis. Thus very little experimental data exists with full scale flares. H2S hazards further complicate probe sampling methods for sour gas flares, which are common in Alberta. Sour gas well test flares are typically 50 m high making physical sampling difficult. Remote optical methods that can quickly scan the whole plume downwind of a flare would be the preferred method to measure flare combustion efficiency and flare emissions. Differential absorption light detection and ranging (DIAL) is one remote optical method that can do this. The results of one set of testing of three large refinery flares are available in the public domain (Boden et al., 1996). All three of the refinery flares tested by Boden et al. (1996) used steam injection to increase flare combustion efficiency. For the three refinery flares, combustion efficiency as measured with the DIAL was greater than 98% over a range of flow rates and steam injection rates. 2 Objective The objective of this project was to demonstrate the DIAL method for measurement of the combustion efficiency of full-scale flares as operated by industry in Alberta. The combustion efficiency of three types of flare were to be measured, a sour well test flare, a sour solution gas flare and a sweet solution gas flare. The measured efficiencies would be compared to predicted efficiencies using correlations developed from previous wind tunnel studies in Canada. 3 3.1 Experimental Description of DIAL Method DIAL is a laser-based optical method capable of measuring the concentration of a gas species at a remote point in the atmosphere. The DIAL method uses a pulsed laser operating at two wavelengths, one strongly absorbed by the gas species of interest and one weakly absorbed. A system of mirrors and lenses is used to direct the laser beam toward the target gas volume. A telescope collects light back-scattered from particles and aerosols in the atmosphere at each of the two wavelengths. From the time taken for the return signal and the relative strength of absorbed and non-absorbed wavelengths, a gas species concentration profile along the light path can be calculated. A steering mirror system can be used to scan an area of interest and develop a 2D or 3D map of gas species concentration in the atmosphere. Combining the gas concentration maps with measured wind speed enables the calculation of mass emissions of a gas species from a source. Figure 1 is an illustration of DIAL as applied to measure fugitive emissions from a tank farm. Concentration Height (m) (mg/m3) Scan plane met stations & sorption tubes Figure 1. Mode of operation employed by the Spectrasyne DIAL system. Spectrasyne Ltd., UK, (www.spectrasyne.ltd.uk) has commercially operated a DIAL system in Europe for over 15 years to measure fugitive emissions of hydrocarbons from oil and gas processing and storage facilities, combustion efficiency of flares, hydrocarbon emissions from airports, benzene emissions from petrochemical facilities and NOx emissions from flares. To perform this project, Spectrasyne Ltd. was subcontracted to bring their DIAL equipment and operating staff to Alberta during May-June of 2003. Over a four week period, the mobile DIAL unit was used to measure fugitive emissions of hydrocarbons from four gas processing plants, the combustion efficiency of three flares and to track the SO2 plume from tail gas incinerator and a well test flare 3.2 Calculation of Combustion Efficiency To measure the combustion efficiency of flares, the DIAL equipment was used to measure the concentrations of hydrocarbons, usually methane, ethylene and C2+, in a vertical slice through the flare plume. For some flares, the concentration of benzene was also measured. By combining the 2D concentration profiles with the measured wind speed and direction data, the mass flow of hydrocarbons in the flare plume was calculated. Several repeat scans were made through the plume to provide a time weighted mean mass flow of hydrocarbons in the plume. The composition and flow rate of flared gas was monitored and a time weighted mean average flare gas flow rate was calculated. The combustion efficiency was then calculated as follows: C a r b o n e m itte d in p lu m e H C s E ffic ie n c y = 1 --------------------------------------------- x 1 0 0 C a r b o n in f l a r e g a s The DIAL equipment does not measure concentrations of soot or carbon monoxide (CO) in the plume. Thus the combustion efficiency calculation assumes no significant quantities of soot or CO in the plume and that large amounts of combustible non-hydrocarbon species are not present in the flared gas. 3.3 Estimated Accuracy of Flare Combustion Efficiency Measurements Potential for measurement errors exist in: - DIAL measurements of hydrocarbon concentration - wind speed and direction - fuel gas flow rate - fuel gas composition - unmeasured products of incomplete combustion (e.g. soot and CO) During the flare combustion efficiency testing in Alberta, soot was not visible from any of the three flares tested, thus significant amounts of soot were likely not present. Carbon monoxide is often used as an indicator of incomplete combustion in flames. Strosher (1996) analyzed gas samples from one sweet and one sour solution gas flare plume and measured CO concentrations that were only 4% of the total carbon concentrations in the plume for a flare with a combustion efficiency of 84%. Gogolek et al. (2001) analyzed combustion products from 2 to 4 inch diameter flares tested in a wind tunnel. The proportion of CO to hydrocarbons in the flare plume varied with fuel composition and flare efficiency. For propane flares, CO accounted for almost all of the ‘inefficiency', with relatively small amount of hydrocarbons in the plume. For natural gas flames, CO accounted for about one half of the inefficiency for flares operating at efficiencies above 99.5% but this decreased to less than 25% of the inefficiency for flares below 97% efficiency. Thus depending on the fuel composition and combustion efficiency, CO may or may not make up a significant amount of the unburned species. An efficiency based on only the hydrocarbons in the plume may overestimate the flare combustion efficiency. As combustion efficiency increases, errors in measuring the hydrocarbons remaining in the plume will have a declining effect on the calculated combustion efficiency. Assuming an error range in measuring the fuel gas flow rate and composition of ±5% and the range of accuracy of the DIAL plume hydrocarbon mass flow of -15% to +5%, the confidence range in the calculated hydrocarbon combustion efficiency was estimated. For a low measured combustion efficiency of 70%, the expected range of actual efficiency, taking into account experimental errors, is quite large, from 64% to 73%. For a combustion efficiency of 95%, the expected range of actual efficiency is 93.8 to 95.5%. For a combustion efficiency of 98%, the expected range is 97.5 to 98.1%. 3.4 Site Descriptions DIAL measurement of combustion efficiency was completed at three flare sites. All of the flares were simple pipes with a continuous pilot for ignition. The following briefly describes each site. 3.4.1 Sweet Solution Gas Flare The sweet solution gas flare site was a single production well site. The oil and gas produced from the field were separated at the site with the gas flared in a 12 m high, 3.5 inch diameter solution gas flare and the oil piped to a refinery via a small on-site storage tank. The site was located in a 120 m2 clearing in the forest surrounded by relatively flat terrain. DIAL measurements were completed in light winds with a light rain falling. The site operator recorded the flare gas flow rate as measured with an orifice plate meter installed in the gas line. No instantaneous gas analysis data was available but a typical gas analysis for the site was provided and used for the efficiency calculations. The flow of flared gas varied widely from about 1,000 Sm3/d to over 6,000 Sm3/d as slugs of gas came up intermittently with the oil. DIAL scans were taken through the flare plume for a period of about 5 hours over a wide range of flared gas flow rates. For the first 1.5 hours of DIAL measurements, the C2+ emissions in the flare plume were quantified. After this period, the infrared laser was switched to methane for about 1 hour followed by C2+ measurements for a second 1.5 hours followed by a final 0.5 hours of methane measurements. Benzene concentrations were measured with the ultraviolet laser during the full 5 hours. 3.4.2 Sour Solution Gas Flare The sour solution gas flare site was a separation and storage facility serving an oil well and also used as a gathering station for production brought in by tank trucks from other oil wells. The separated solution gas from the producing well, after leaving the separator, passed through a demister and was then sent to an 8 inch diameter solution gas flare. Flared gas flow rate varied from 2,100 to 4,500 m3/d. DIAL scans through the plume were collected over a two day period. 3.4.3 Sour Well Test Flare The sour gas flared at this site contained about 11% H2S. For purposes of flaring, the well site equipment included a portable gas/liquid separator and a temporary test flare that was 50 metres high with a diameter of 10 inches. The pilot gas for the flare was sweet gas piped from a nearby gas plant. During the period that the DIAL equipment was on site, gas flaring was carried out primarily for the purpose of the DIAL measurements. Objectives of testing at this site included measuring the combustion efficiency of the flare, the efficiency of conversion of H2S to SO2 and to track the SO2 plume. DIAL measurements at the site were made over two days. The flare gas flow rate for the first test was 60,000 Sm3/d of sour gas containing 11% H2S. For the second test, the sour gas flow was supplemented with approximately 40,000 Sm3/d of sweet gas to increase the total gas flow to the flare to about 100,000 Sm3/d of gas containing about 6.6% H2S. Flared gas flow rates were held steady during the DIAL measurement periods. Figure 2 shows the DIAL equipment measuring hydrocarbons in the plume from the sour solution gas flare. Figure 2: Spectrasyne DIAL Measuring Sour Solution Gas Flare 4 Results and Discussion Figure 3 gives two examples of cross sections through the plume about 1.5 and 4 flame lengths downwind of the tip of the visible flame for the well test flare. The visible flame length was about 9 m for this flare. These profiles show the variation of SO2 concentration in the flare plume and a typical variation in plume shape from scan to scan. This close to the flare, maximum SO2 concentration were about 10 mg/m3 and, although the plume has already broken up, the full plume was captured during a DIAL scan. The lower edge of the plume was at a similar elevation as the flare tip (50 m) while the upper edge of the plume had risen to between 150 to 170 m. Flare SO 2 Approx 1.5 Flame Lengths DW (3.2) 13:05 10-06-03 Light Rain Flare SO 2 Approx 4 Flame Lengths DW (3.7) 13:55 10-06-03 Light I iao C l EPtcrRiSr-ie:L .1 Range (m} ange (m) Figure 3: SO2 Concentration Profiles in Well Test Flare Plume near the Flare Table 1 summarizes the mass flux of CH4 in the plume as calculated from the profiles measured with DIAL during a set of measurements on the well test flare. Table 1 also lists wind speed and direction at the time of the profiles. During these measurements the flare gas flow rate was 60,000 Sm3/day of gas with 11% H2S. Each cross section profile took form 5 to 10 minutes to complete. The CH4 fluxes for each scan ranged from 59 to 121 kg/h with a time weighted mean average of 94.7 kg/h. Changes in mass flux from scan to scan result from changes in wind direction, wind speed and turbulence during the time taken to fully profile the plume. Several scans are required to average these fluctuations. Table 1: Methane Flux Data from a Set of Well Test Flare DIAL Measurements Scan No. Wind Speed (m/s) Wind Direction (deg) CH4 Flux 1 3.5 320 (kg/h) 121 2 2.6 329 59 3 1.4 322 108 4 1.8 322 86.8 5 2.0 342 118 6 2.7 302 81.0 CH4 and C2+ concentration profiles were measured in the plumes of all three flares. In most of the flares the concentration of ethylene, a partial combustion product, was also measured. In the two solution gas flares profiles of benzene concentration were also measured. Benzene was detected in the solution gas flare plumes but the concentrations were near the detection limit (about 1 ppb), suggesting little or no entrained liquids in the flared gas.. The relative amounts of CH4 and C2+ hydrocarbons (including ethylene) generally mirrored the relative amount of CH 4 and C 2+ in the gas being flaring. Table 2 summarizes the mass ratio of methane to other hydrocarbons in the flared gas and the plume for the three flares. Table 2: Variation in CH4 to C2+ Hydrocarbons in Flared Gas and Flare Plume Flare CH4/C2+ weight ratio flared gas flare plume (DIAL) sweet solution gas flare 2.3 2.1 sour solution gas flare 0.9 2.1 well test flare (60,000 m3/d) 11.8 23.4 well test flare (100,000 m3/d) 15.0 25.7 For both of the solution gas flares, the flare gas flow rate fluctuated during the period of the DIAL measurements. The combustion efficiency calculations have been based on the time weighted mean gas flow rate and the time weighted mean emissions of CH4, C2+ and ethylene. Table 1 summarizes the results of the flare efficiency measurements in Alberta. The solution gas flares were both surrounded by relatively flat terrain while the well test flare was situated in the foothills. The well test flare was operated at two steady controlled flow rates while the flow rate to both of the solution gas flares varied during normal operation. The measured combustion efficiency ranged from 74 to 98%. Table 1: Summary of Flare Combustion Efficiency Measured by DIAL Flare sweet solution gas sour solution gas1 well test sour gas well test sour gas 1 corroded flare tip Gas Flow Rate (x103 Sm3/d) 1 to 6 4 60 100 H2S Content (%) 0 1.15 11 6.6 DIAL Combustion Efficiency (%) 98 74 to 92 91 91 The combustion efficiency of the sour solution gas flare was strongly dependent on wind direction, with 74% during a NW wind and 92% during a SE wind. Close observation of the flare tip revealed a large hole corroded in the south side of the flare. This hole likely caused the large difference in combustion efficiency with wind direction, with a partially premixed flame during a SE wind. Table 2 compares the combustion efficiencies measured with the DIAL to predicted combustion efficiency based on correlations developed at the University of Alberta from wind tunnel studies of sweet gas flares. In general the U of A correlation predicted combustion efficiencies that were higher than those measured in the field. The tip exit velocity for the well test flare was well outside the range of wind tunnel experimental data. The correlation does predict a trend of increasing combustion efficiency with increasing tip exit velocity, thus a value of >99% combustion efficiency is shown for the prediction. As mentioned above, the sour solution gas flare tip was badly corroded, with combustion efficiency dependent on wind direction. Table 2: Comparison of Measured and Predicted Combustion Efficiency Flare Exit Velocity (m/s) sweet solution 2.8 - 14.8 gas sour solution gas 0.8 - 1.6 13.2 well test sour gas 23.1 well test sour gas 1 corroded flare tip 2 exit velocity well above the Wind Range (m/s) U of A Correlation Prediction (%) 0.3 - 2.1 99.5 - 99.7 DIAL Combustion Efficiency (%) 98 2.9 - 9.2 1.3 - 4.2 1.5 - 3.9 93.1 - 99.4 >992 >992 74 to 921 91 91 correlated data range The mass flux of SO2 in the flare plume was measured for the well test flare. The SO2 mass flux was 46 to 72% of the flux expected if all of the H2S in the flared gas was converted to SO2. The presence of a secondary plume and the lack of significant H2S from ambient air monitoring suggests that sulphur compounds other than SO2 may be formed during flaring of sour gas. 5 Conclusions and Recommendations The ability of DIAL to measure combustion efficiency of full-scale flares under actual atmospheric conditions of wind and turbulence was demonstrated in Alberta. The combustion efficiencies based on DIAL measurement of hydrocarbons in the plume downwind of the three flares tested ranged from 74 to 98%. The measured efficiencies of these three flares were lower than those predicted by correlations based on wind tunnel tests. Although only one sweet and two sour flares were measured in this experiment, the sour gas flares both had lower combustion efficiencies than the sweet gas flare. DIAL measurement of SO2 in the well test flare could account for less than 75% of the sulphur contained as H2S in the flared gas. Other sulphur species were likely formed during sour gas flaring. Further DIAL measurements are recommended to measure the efficiency of a solution gas flare under a range of controlled conditions of fuel flow rate and H2S content. This work would complement the existing data set from the wind tunnel experiments with sweet gas flares and would improve the understanding of sour gas flaring. The proposed program should also investigate products of H2S combustion other than SO2 that might be present in sour gas flare plumes. References Boden, J.C., K. Tjessem, A. Wootton and J. Moncrieff, ‘Elevated fare emissions measured by remote sensing', Petroleum Review Vol. 50:598. p 524-528, Nov 1996. Gogolek, P., S. Madrali and A.C.S. Hayden, ‘Performance and Speciation of Solution Gas Flares in the CANMET Flare Test Facility', Final Report, 2001. Kostiuk, L.W., Majeski, A.J., Poudenx, P., Johnson, M.R., and Wilson, D.J., ‘Scaling of WakeStabilized Jet Diffusion Flames in a Transverse Air Stream', Proceedings o f the Combustion Institute, 28:553-559, 2000. Pohl, J.H., R. Payne and J. Lee, ‘Evaluation of the Efficiency of Industrial Flares: Test Results', EPA-600/2-84-095 (NTIS PB84-199371), 1984. Strosher, M., ‘Investigations of Flare Gas Emissions in Alberta', Alberta Research Council final report to Environment Canada, Alberta Energy and Utilities Board and the Canadian Association of Petroleum Producers, November, 1996.