Efficiency of flare flames in turbulent crosswind

Flares are used for the safe, clean and economical disposal of waste gases, whether in upstream oil production (solution gas flares), refineries, gas plants or other chemical processing facilities. Elevated flares are exposed to all the weather patterns, perhaps the most important is the crosswind. An important parameter in the flare flame behaviour is the ratio of the fuel jet momentum flux to the crosswind momentum flux. When this ratio is less than unity, the flame is stabilised by a recirculating eddy downwind of the flare pipe. This paper is concerned with these wake-stabilised flames. The CETC Flare Test Facility (FTF) was constructed to address the questions of performance of solution gas flares and in response to limited field trials that indicated the possibility of very low combustion efficiencies under certain conditions. The FTF consists of a high capacity fan feeding through flow straighteners into the working section, 1.2m wide and 8.2m long. The variable ceiling height is adjustable to 1.5m, 1.8m, 2.1m and 2.6m, permitting a very wide range of crosswind speeds up to 45 km/h. The flare pipe is situated near the front of the working section. Model solution gases are produced from natural gas, propane, carbon dioxide and nitrogen. Liquid droplets can be injected into the fuel gas to simulate the entrainment of liquids. The basic airflow (wind) in the working section has been designed to have very low turbulence intensity. Atmospheric wind is a turbulent shear flow with intensity around 7%. While it is not possible to exactly reproduce atmospheric turbulence in the FTF, using different grids can produce a range of turbulence intensities and integral length scales. The paper will discuss how this grid-generated turbulence relates to the turbulence properties of atmospheric wind. The flare flame efficiency is measured by the conversion of carbon in the fuel to carbon dioxide. Fuels tested are natural gas, propane, and mixtures of these. Conversion efficiency is lowest for pure natural gas, and increases with the amount of propane. However, conversion efficiency decreases significantly with the increase of turbulence intensity of the crosswind, which has implications for existing and novel strategies to improve flare performance. These results are explained by the lower flammability limits of the gas mixtures.

E f f i c i e n c y o f F l a r e F l a m e s in T u r b u l e n t C r o s s w i n d P.E.G. Gogolek, A.C.S. Hayden Advanced Combustion Technologies Natural Resources Canada ABSTRACT Flares are used for the safe, clean and economical disposal of waste gases, whether in upstream oil production (solution gas flares), refineries, gas plants or other chemical processing facilities. Elevated flares are exposed to all the weather patterns, perhaps the most important is the crosswind. A n important parameter in the flare flame behaviour is the ratio of the fuel jet m o m e n t u m flux to the crosswind m o m e n t u m flux. W h e n this ratio is less than unity, the flame is stabilised by a recirculating eddy downwind of the flare pipe. This paper is concerned with these wake-stabilised flames. The C E T C Flare Test Facility (FTF) was constructed to address the questions of performance of solution gas flares and in response to limited field trials that indicated the possibility of very low combustion efficiencies under certain conditions. The F T F consists of a high capacity fan feeding through flow straighteners into the working section, 1.2m wide and 8.2m long. The variable ceiling height is adjustable to 1.5m, 1.8m, 2.1m and 2.6m, permitting a very wide range of crosswind speeds up to 45 km/h. The flare pipe is situated near the front of the working section. Model solution gases are produced from natural gas, propane, carbon dioxide and nitrogen. Liquid droplets can be injected into the fuel gas to simulate the entrainment of liquids. The basic airflow (wind) in the working section has been designed to have very low turbulence intensity. Atmospheric wind is a turbulent shear flow with intensity around 7%. While it is not possible to exactly reproduce atmospheric turbulence in the FTF, using different grids can produce a range of turbulence intensities and integral length scales. The paper will discuss h o w this grid-generated turbulence relates to the turbulence properties of atmospheric wind. The flare flame efficiency is measured by the conversion of carbon in the fuel to carbon dioxide. Fuels tested are natural gas, propane, and mixtures of these. Conversion efficiency is lowest for pure natural gas, and increases with the amount of propane. However, conversion efficiency decreases significantly with the increase of turbulence intensity of the crosswind, which has implications for existing and novel strategies to improve flare performance. These results are explained by the lower flammability limits of the gas mixtures. INTRODUCTION Flares are used for the safe, clean and economical disposal of waste gases, whether in upstream oil production (solution gas flares), refineries, gas plants or other chemical processing facilities. Elevated flares are exposed to all the weather patterns, perhaps the most important is the crosswind. The flaring of solution gas (also called associated gas) was a common practice in upstream oil extraction. This gas is a mixture of light hydrocarbons, mostly methane but including possibly significant amounts of ethane, propane, butane, hydrogen, perhaps some inert gases (nitrogen and carbon dioxide), and sometimes hydrogen sulphide (sour gas). The composition and amount of solution gas can vary a great deal between production sites. Over 800 million m of gas was flared in Alberta in 2000, which is almost a 50% reduction from 1996 [1]. The current global flaring rate is estimated between 100 and 126 billion m of solution gas each year [2]. Strosher [3] conducted laboratory-scale investigations on an enclosed flare and an openair flare, and on two commercial flares in the field. He found in the field investigations that combustion efficiency is typically about 70%, sometimes as low as 62%. The unburned portion consisted of methane, unsaturated hydrocarbons, light aromatics such as benzene, and polycyclic 2 aromatic hydrocarbons (PAH). This work has led to much of the current interest in flare research by industry, government, academia and the public. An important parameter in the flare flame behaviour is the ratio of the fuel jet momentum flux to the crosswind momentum flux. When this ratio is less than unity, the flame is stabilised by a recirculating eddy downwind of the flare pipe. There have been some studies published on model flares in wind tunnels, producing wake-stabilised flames. Gollahalli and Nanjundappa [4] describe a three-zone flame with a recirculating eddy attached to the lee-side of the pipe, a mixing layer of non-reacting fuel from the pipe carried over the eddy, and an axisymmetric main tail of flame. Huang and coworkers [5, 6] extended the range to high fuel momentum jets, and attempted a more extensive classification of combustion behaviour, using the momentum flux ratio. Kostiuk and coworkers [7-10] have done extensive work on the qualitative and quantitative behaviour of these flames. They correlate their combustion efficiency data with the ratio , (1) U u f1/3 which collapses their data for different pipe size and wind speeds on a single characteristic curve for each fuel. This group has also published a study of the characteristics of solution gas flares in Alberta [11]. The crosswind in these wind tunnels was close to laminar, with turbulence intensity kept below 0.01. The open wind is a turbulent shear flow between the surface of the earth and the upper atmosphere [12,13]. It forms the atmospheric boundary layer (ABL), the height of which is determined by the surface characteristics. For open, flat country, which is the terrain in Alberta, the height of the ABL is around 300 m. The lower fifth of the ABL is a region o f nearly constant shear stress, called the surface layer. A solution gas flare is typically 10 to 12 m high, definitely in the surface layer. At this height, the turbulence intensity is around 0.07 and the integral length scale is around 4 m. The CETC Flare Test Facility (FTF) was constructed to address the questions of performance and emissions of solution gas flares. We have used this facility to corroborate the published results of other workers in a laminar crosswind [14]. In this paper, we present results on the efficiency of flare flames in crosswinds with grid turbulence. The grid turbulence is related to the turbulence of atmospheric wind, though different in several significant features. Two grids were used, and turbulence is shown to greatly increase the inefficiency of flare flames. EXPERIMENTAL The FTF consists of a high capacity fan feeding through flow straighteners into the working section. The fan is capable of delivering 23.6 m /s (50,000 cfm) at 1.5 kPa (6" wc). The volumetric airflow is measured using a differential pressure flow meter. The working section is 1.2m wide and 8.2m long, with variable ceiling height adjustable to 1.5m, 1.8m, 2.1m and 2.6m. This produces a range of crosswind speed up to 45 km/h. The flare pipe is situated near the front of the working section. Flare pipes are 1" Nom., 2" Nom., and 4 " Nom., Sch. 40, carbon steel, usually 1.0 m long. This configuration allows full development of the flare flame, without impingement on the walls, ceiling or floor o f the working section. More complicated flare tips can be used, for example having a windshield over the pipe opening. Windows provide visual access to the flare flame. 3 Model solution gases are produced from natural gas, propane, carbon dioxide and nitrogen, or mixtures of these. Liquid droplets can be injected into the fuel gas at the base of the flare pipe to simulate the entrainment of liquids. Fuel flow rate is measured using thermal mass flow meters and controlled automatically. The maximum flowrate achievable is 42 kg/h. However, the tests reported here had flow rates of 10.5 kg/h and 21.0 kg/h of natural gas or a mixture of 70% natural gas with 30% propane (by mass). The stack gases are sampled using a 0.46 m long sintered metal tube placed at the centerline of the stack. The sampled gases are passed to a bank of analyzers, including O2 (paramagnetic), CO and CO2 (IRD), CH4 and NMHC (FID). Because of the extremely high dilution, it is necessary to measure the ambient air composition as well. The airflow in the working section has been designed to have very low turbulence intensity. Grids are used to produce a turbulent wind in the working section. The grids are steel slats, with width and spacing to produce the desired turbulence intensity and integral length scale. Two grids were used in this work. One has 7.6 cm (3 inch) wide slats with 10.2 cm (4 inch) gap; this is called the 3 " grid. The other has 2.5 cm (1 inch) wide slates with 15.2 cm (6 inch) gap; this is called the 1" grid. RESULTS The conversion efficiency of a flare flame as used in this paper is the mass flow rate of carbon as CO2 in the stack gas less the CO2 in the inlet air, divided by the flowrate of carbon into the flame in the fuel. mass rate of C as CO2 in stack gas - mass rate of C as CO2 input n = ----------------------------- 2------------ 5---------------------------------- 2- - (2) mass rate of combustible C in fuel Since this is typically close to 1, it is easier to see the effects by looking at the conversion inefficiency, 1 -n , expressed as a percentage. The methane slip is the corresponding conversion (or rather the failure to convert) to methane of fuel carbon. mass rate of C as CH4 in stack gas - mass rate of C as CH4 in air n f = ----------------------------- 4------------ ----------------------------------- 4-------(3) mass rate of combustible C in fuel This number is particularly important for estimating the Global Warming Potential of the emissions from flares. The performance of flare flames in laminar crosswind was studied extensively. The results are summarised in Fig. 1. Here the inefficiency is correlated with the factor R ew Re - 14, combining the wind speed, pipe size and fuel mass flow rate in a dimensionless form. These data are for wake stabilised flare flames, where the momentum flux ratio of fuel to wind is less than one. The effect of fuel composition is not correlated by this factor. Natural gas flames are the most inefficient, and inefficiency decreases with amount of propane in the fuel gas mixture. The maximum inefficiency is less than 3% with a laminar crosswind. The wake of the flare pipe, however, is strongly turbulent, with R ew > 10,000 . 4 0 500 1000 1500 2000 2500 3000 R e w / R e , 04, - Figure 1 - Correlation of carbon conversion inefficiency for laminar crosswind. Introducing grid turbulence into the working section has a strong visible effect on the flare flame. With a laminar wind, there is a clear recirculation zone in the immediate wake, which holds the flame. In a turbulent flow, the eddy remains but is diffuse without a clear recirculation pattern. The flame is shortened in a turbulent wind. These effects are stronger for the 3 " grid but still quite evident for the 1" grid. The inefficiency curves for natural gas flames against wind speed are given in Fig.2. Note that these data include 1", 2" and 4 " flare pipes, and fuel flow rates of 10.5 kg/h and 21 kg/h. The effects of pipe size and fuel flow rate are dominated by the wind speed effect. The maximum inefficiency is about 18% for natural gas flames in the turbulent wind produced with the 3 " grid. This is more than 6 times the maximum inefficiency observed in a laminar wind. The 1" grid produces a maximum inefficiency of around 6%, more than double the corresponding laminar case. The functional form of these curves is hard to define because of the relatively few data points in the middle range o f wind speed. 5 W i n d Speed, km/h Figure 2 - Conversion inefficiency in turbulent crosswind, for 1" and 3 " grids. The fuel composition still has a strong effect. Fig. 3 compares the inefficiency curves for natural gas and a mixture of 70% natural gas with 30% propane (by mass), with turbulence produced by the 1" grid. The inefficiency of the mixture is about 2/3 that of the natural gas. This is about the same order of effect as seen with the laminar crosswind. This seems to indicate that a similar mechanism is operating in both cases. 6 W i n d Speed, km/h Figure 3 - Conversion inefficiency in turbulent wind from 1" grid, for natural gas (NG) and 70% natural gas, 30% propane (70-30) flames. The inefficiency can arise from unreacted fuel or from partial combustion. The former is particularly important because the fuel is mostly methane, a strong GHG. Partial combustion, as indicated by the CO emissions, will eventually oxidise to CO2. The formation of toxic and carcinogenic compounds (aromatics and polyaromatics) was found to be insignificant [14]. However, we have not studied the strongly sooting flames sometimes seen in the field. The methane slip, as a percentage of the total inefficiency, is shown in Fig. 4 for natural gas flare flames in winds from the 1" grid and 3 " grid. Included is the methane slip for the 70-30 mixture with the 1" grid in place. The more intense turbulence from the 3 " grid produces greater methane slip, 70% compared to 50%, at the lowest wind speed of 5 km/h. The methane slip converges as wind speed increases and reaches effectively 100% of the inefficiency at the maximum wind speed of 40 km/h. The 70-30 mixture was not tested at the lower wind speeds, but the methane slip shows a similar increase from 20 km/h to 40 km/h. Note that the methane slip at the highest wind speed is around 80% of the total inefficiency, which is greater than the percentage of methane in the fuel gas. This means that methane fails to ignite even as the propane burns. 7 Uw, km /h Figure 4 - Methane slip, as a percentage of the total inefficiency, for natural gas flames as a function of wind speed for 1" grid, and 3" grid. Included is the methane slip for the mixture of 70% natural gas with 30% propane, with 1" grid. DISCUSSION The grids have identical characteristic length, but the 3 " grid produces more intense turbulence. The turbulent kinetic energy can be estimated using the correlation for grid turbulence =A x - x„ U ,2 } (4) B where B is the characteristic length of the grid, xo is the virtual origin and A depends on grid geometry. A recent evaluation [15, p. 160] of the published data recommends n = 1.3 and = 0 . The integral length scale is derived from (2) as o / A T = A 2B V x - Xo B (5) Since the characteristic lengths of both grids are the same, the turbulent kinetic energy and integral scale for the grid turbulence differ only by a constant multiplier. More significantly, the turbulent dispersion, given by D T = 4 2 k A T , is different only by a constant multiplier for the two flows. Note further that D T is linear in the wind speed. Thus, if wind turbulence is the controlling mechanism, the inefficiency curves in Fig. 2 can be made to coincide by a multiplier. 8 This is roughly true, as shown in Fig. 6. It is not clear if there is a linear relationship between inefficiency and wind speed. 25 20 >; 15 o c (U o % 10 c 5 0 0 10 20 30 40 50 W in d S p e e d , km /h Figure 5 - Conversion inefficiency in turbulent wind; data in Fig. 2 replotted with the inefficiency with 1" grid multiplied by 3. This gives a rough coincidence of the inefficiency curves for the winds with different turbulence properties. The effect of fuel composition is related to the lower (lean) flammability limit (LFL). Methane has an LFL of 5%(vol); propane has LFL of 2.1%(vol) [16]. For this simple mixture, Le Chatelier's principle can be used to calculate the LFL [16], giving 4.2%(vol) for 86.5%(vol) (70% mass) methane and 13.5%(vol) (30% mass) propane. This is significantly below methane but still twice the LFL for propane alone. While the LFL provides a qualitative explanation, a quantitative prediction will have to be more complex in order to explain the preferential slip of methane shown in the data for the 70-30 mixture in Fig.4. CONCLUSION While the inefficiency of flame flames in a laminar crosswind is less than 3% for speeds below 45 km/h, imposing turbulence on the wind produces a large increase in inefficiency. Since the atmospheric boundary layer is a turbulent flow, with intensity around 7% at the height of a typical flare tip (12 m), it is important to study the effect of wind turbulence on flare performance. Two grids were used to produce turbulence in the working section of the Flare Test Facility. The 3 " grid produces turbulence intensity comparable to that of wind at the flare pipe. 9 The turbulence from the 1" grid is much less intense. Both grids produce a large increase in inefficiency, up to 18% at the highest wind speed with the 3 " grid. The maximum inefficiency with the 1" grid is 6%. The effect of wind turbulence dominates the effects of pipe size and fuel flow rate observed under laminar wind conditions. Turbulent dispersion by the wind appears to be the dominant mechanism for inefficiency. The fuel composition effect (natural gas producing the most inefficient flames, efficiency increasing with propone fraction) appears to have the same significance with turbulent wind as for laminar wind. The lower flammability limit (LFL) is a good predictor of the relative efficiencies of the flare flames. Fuel slip is the major source o f inefficiency. It accounts for 50% to 80% o f the inefficiency at low wind (5 km/h) conditions, rising to essentially 100% at high wind (45 km/h) conditions. This means that for real winds, that is turbulent winds, flare flames can be a significant source of methane. Thus the proper accounting of flares has to account for the local wind conditions (distribution of wind speed and terrain) and the composition of the gas flared. These results provide a starting point for such an accounting. ACKNOWLEDGEMENTS This work was funded by the Climate Change Action Plan of PERD (Program of Energy Research and Development) through Environment Canada, the Canadian Association of Petroleum Producers, Petroleum Technology Alliance Canada, and the government of Alberta through AOSTRA. C. Balderson is the technician who conducted the experiments, assisted by J. Tremblay, C. Grimwood, and L. Robichaud. The FTF was constructed and commissioned under the leadership of Dr. S. Madrali. The support of B. Reynen o f Environment Canada is worthy of special mention. NOMENCLATURE A B Dp Coefficient for decay of turbulent kinetic energy downstream from a grid. [-] Length scale for grid. [m] Pipe diameter. [m] DT Coefficient of turbulent dispersion. [m /s] k Uf Turbulent kinetic energy per unit mass of fluid. [m /s ] Superficial velocity of fuel in pipe. [m/s] Uw Superficial velocity of wind. [m/s] x Xo Position downstream from grid. [m] Position of virtual origin of grid. [m] n nf Conversion of fuel carbon to CO2 in flare flame. [-] Fuel slip. [-] AT Integral length scale of turbulence. [m] 2 22 10 REFERENCES 1. Berg, A. [2002], Flaring Management in Alberta, Canada, presented at World Bank Global Gas Flaring Reduction Initiative Conference, Oslo, Norway, April 15-16, 2002. 2. World Bank Group [2002], Global Initiative on Natural Gas Flaring Reductions - Report on Consultations with Stakeholders, presented at World Bank Global Gas Flaring Reduction Initiative Conference, Oslo, Norway, April 15-16, 2002. 3. Strosher, M. [1996], Investigation of Flare Gas Emissions in Alberta, Alberta Research Council report to Environment Canada, Alberta Energy and Utilities Board, and Canadian Association of Petroleum Producers; November 1996. 4. Gollahalli, S.R. and B. Nanjundappa [1995], Burner Wake Stabilized Gas Jet Flames in Cross-Flow, Combustion Science and Technology, 109, pp. 327-346. 5. Huang, R.F. and M.J. Yang [1996], Thermal and Concentration Fields of Burner-Attached Jet Flames in Cross Flow, Combustion and Flame, 105, pp. 211-224. 6. Huang, R.F. and S.M. Wang [1999], Characteristic Flow Modes of Wake-Stabilized Jet Flames in a Transverse Air Stream, Combustion and Flame, 117, pp. 59-77. 7. Bourguignon, E., M.R. Johnson, and L.W. Kostiuk [1999], The Use of a Closed Loop Wind Tunnel for Measuring the Combustion Efficiency o f Flames in a Cross Flow, Combustion and Flame, 119, pp. 319-334. 8. Kostiuk, L.W., A.J. Majeski, P. Poudenx, M.R. Johnson and D.J. Wilson [1999], Scaling of Wake-Stabilized Jet Diffusion Flames in a Transverse Air Stream, Proceedings of the 28th Symposium (International) on Combustion, The Combustion Institute. 9. Johnson, M.R. and L.W. Kostiuk [2000], Efficiencies of Low-Momentum Jet Diffusion Flames in Crosswinds, Combustion and Flame, 123, pp. 189-200. 10. Johnson, M.R., D.J. Wilson and L.W. Kostiuk [2001], A Fuel Stripping Mechanism for Wake-Stabilized Jet Diffusion Flames in Crossflow, submitted to Combustion Science and Technology. 11. Johnson, M.R., J.L. Spangelo, and L.W. Kostiuk [2001], A Characterization of Solution Gas Flaring in Alberta, Journal of the Air and Waste Management Association, in press. 12. Panofsky, H.A. and J.A. Dutton [1984], Atmospheric Turbulence: M odels and Methods fo r Engineering Applications. Wiley-Interscience. 13. Kaimal, J.C. and J.J. Finnigan [1994], Atmospheric Boundary Layer Flows: Their structure and measurement, Oxford University Press. 14. Gogolek, P., S. Madrali, and A.C.S. Hayden [2001], Performance and Speciation of Solution Gas Flares in the CANMET Flare Test Facility, Final Report. 15. Pope, S.B. [2000], Turbulent Flows, Cambridge University Press. 16. Coward, H.F. and G.W. Jones [1952], Limits of Flammability of Gases and Vapors, Bulletin 503, Bureau of Mines, United States Government Printing Office, Washington.