Basis for Emission Calculation from Flare Systems

In 1983, the Chemical Manufacturer's Association (CMA) sponsored a study in an attempt to define emission factors for flare systems. The study encompassed several variables including the lower heating value of the gas, relief gas flow rate, air assist rate, and steam assist rate. Emissions were continuously monitored by a sample probe located above the flare flame to analyze the concentrations of carbon dioxide, carbon monoxide, total hydrocarbons, sulfur dioxide, nitrogen oxides, and oxygen. The US Environmental Protection Agency (EPA) published the AP-42 guideline for flare system emissions, citing values that are essentially averages considering all of the CMA testing data. Unfortunately, companies using the AP-42 data as their guideline for permitting elevated flares could be citing values that are far from accurate, considering the wide range of conditions tested in the CMA study. This paper will examine the original CMA test data, discuss possible variations between the testing from 1983 and modern flare system designs and operation, and identify the applicable conditions for employing the various published factors for flare emissions.

Basis for Emission Calculation from Flare Systems Scot Smith Director - Flare Division Zeeco, Inc. Ben Pettys Design Engineer - Flare Division Zeeco, Inc. Abstract: In 1983, the Chemical Manufacturer's Association (CMA) sponsored a study in an attempt to define emission factors for flare systems. The study encompassed several variables including the lower heating value of the gas, relief gas flow rate, air assist rate, and steam assist rate. Emissions were continuously monitored by a sample probe located above the flare flame to analyze the concentrations of carbon dioxide, carbon monoxide, total hydrocarbons, sulfur dioxide, nitrogen oxides, and oxygen. The US Environmental Protection Agency (EPA) published the AP-42 guideline for flare system emissions, citing values that are essentially averages considering all of the CMA testing data. Unfortunately, companies using the AP-42 data as their guideline for permitting elevated flares could be citing values that are far from accurate, considering the wide range of conditions tested in the CMA study. This paper will examine the original CMA test data, discuss possible variations between the testing from 1983 and modern flare system designs and operation, and identify the applicable conditions for employing the various published factors for flare emissions. Overview and Background The concern for industrial emissions of pollutants and their effect on the environment has been ever increasing since the 1950's. The emissions of carbon monoxide (CO), nitrogen oxides (NOx), sulfur dioxide (SO2), particulate, and smoke have been considered to have the greatest impact on air quality. Beginning in the 1970's and through the early 1980's, elevated flare stacks were viewed as environmentally problematic due to the highly visible flame and smoke. The EPA proposed to eliminate flares as a viable device to control volatile organic compound (VOC) emissions from refineries and chemical plants. To prevent flare systems from being excluded as an option for destroying industrial emissions, the Chemical Manufacturers Association (CMA) sponsored a study in 1983, primarily to determine the combustion efficiency of flares. Several tests were performed for non-assisted, steam-assisted and air-assisted flares in an effort to encompass the wide range of operating conditions defined by normal industrial applications. Variables involved for both air assisted and steam assisted flares included the relief gas flow rate, lower heating value (LHV) of the flare gas, and air or steam assist rate. Testing was completed utilizing 100% propylene to model high LHV flare gas scenarios and propylene diluted with nitrogen to represent low LHV flare gas scenarios. The emission concentration of CO, NOx, SO2, carbon dioxide (CO2), oxygen (O2), and total hydrocarbons (THC) was continuously monitored by a sample probe located above the flare flame. The testing results concluded that an elevated flare with an open flame had a combustion efficiency of at least 98% when operated correctly. The EPA included all CMA testing to determine elevated flares have a combustion efficiency of 98% however, if 40 CFR 60.18 parameters are followed, the combustion efficiency will increase above 99.5%. These results were published and utilized as a basis by the US Environmental Protection Agency (EPA) to develop the 40 CFR 60.11 to 60.18 regulations as well as the AP 42 Chapter 13.5 emission factors. The EPA 40 CFR 60.18 guidelines use variables such as flare type, presence of a continuously lit pilot, LHV of the flare gas, and the exit velocity of the gas to determine if the flare is operating at the minimum 98% destruction efficiency. Once the flare is operating within these constraints, emission factors presented in AP 42 Chapter 13.5 can be used to estimate the emissions of CO, NOx, unburned hydrocarbons (UHC), and soot from the flare. The EPA developed and published the AP 42 document as an emissions guide for flares used to control VOC's. These emission factors are an average of applicable CMA test data. While AP 42 emission factors account for exit velocity, lower heating value of the flare gas, and flare type, these values suggest emission rates from elevated flares do not differ across the wide range of industrial applications. This generalization ultimately causes estimated emission rates to be substantially different for particular applications. As a result, the Texas Natural Resource Conservation Commission (TNRCC) reviewed the original CMA test data in 1994 and published emission factors that apply to certain flare types and lower heating values. This paper will explain how NOx and CO emission factors were calculated by the EPA and TNRCC, as well as discuss possible modifications to emission factors in an effort to better represent actual flare emissions. CMA Test Setup As previously discussed, propylene was utilized as flare gas and diluted with nitrogen to decrease the lower heating value of the gas. The flow rate for testing ranged from as low as normal purge conditions and to as high as 703 SCFM. The study was completed to establish a wide range of tests similar to operating conditions for common industrial applications. Throughout testing, the air-assisted flare had a 4" gas riser and the steam assisted flare used an 8" gas riser. This testing provided additional data to estimate emissions; however, the results may not apply to the wide range of applications that include flare design, tip exit area and relief gas composition. When considering these variables, the actual emission rate may differ from the data obtained by CMA testing. Development of Various Emission Factors Emission factors published by the EPA and TNRCC are based on data obtained from CMA testing performed in 1983. The assumption was made that when complete combustion takes place, every mole of propylene is burned and three moles of carbon dioxide is produced. The chemical reaction assuming complete combustion is shown below: 𝐶!𝐻! + 4.5 𝑂! + 3.76𝑁! 3𝐻!𝑂 + 3𝐶𝑂! + 16.92𝑁! The molar relationship between propylene and carbon dioxide was then converted on a mass basis using the molecular weight of propylene and carbon dioxide as shown below: 𝑙𝑏𝑚𝑜𝑙 𝐶!𝐻! 𝑥 𝑀𝑊!!!! 3 𝑥 𝑙𝑏𝑚𝑜𝑙 𝐶𝑂!𝑥 𝑀𝑊!"! In the Flare Efficiency Study published by the CMA in 1983, an equation was developed to determine an emission factor (ENOx) based on measured concentration levels of NOx and CO2. This relationship is shown below: 𝐸!"! = 𝑀𝑜𝑙𝑒𝑠 𝑁𝑂! 𝑀𝑜𝑙𝑒𝑠 𝐶𝑂! 𝑥 46 𝑙𝑏𝑠 𝑚𝑜𝑙𝑒 𝑁𝑂! 44 𝑙𝑏𝑠 𝑚𝑜𝑙𝑒 𝐶𝑂! 𝑥 132 𝑙𝑏𝑠 𝐶𝑂! 𝑝𝑟𝑜𝑑𝑢𝑐𝑒𝑑 42 𝑙𝑏𝑠 𝑝𝑟𝑜𝑝𝑦𝑙𝑒𝑛𝑒 𝑏𝑢𝑟𝑛𝑒𝑑 𝑥47.2 𝑙𝑏𝑠 𝑝𝑟𝑜𝑝𝑦𝑙𝑒𝑛𝑒 𝑏𝑢𝑟𝑛𝑒𝑑 10! 𝐵𝑡𝑢 Where: 𝑀𝑜𝑙𝑒𝑠 𝑁𝑂! 𝑀𝑜𝑙𝑒𝑠 𝐶𝑂! = 𝑃𝑃𝑀 𝑁𝑂! 𝑀𝑒𝑎𝑠𝑢𝑟𝑒𝑑 𝑃𝑃𝑀 𝐶𝑂! 𝑀𝑒𝑎𝑠𝑢𝑟𝑒𝑑 𝐻𝐻𝑉!!!! = 21,186 𝐵𝑡𝑢/𝑙𝑏 Methods discussed above could be applied to estimate CO emissions using the measured concentration of CO and CO2 from CMA testing. This was accomplished by replacing the concentration level and molecular weight of NOx from the equation shown above with the concentration level and molecular weight of CO. The relationship for the emission rate of carbon monoxide is shown below: 𝐸!" = 𝑀𝑜𝑙𝑒𝑠 𝐶𝑂 𝑀𝑜𝑙𝑒𝑠 𝐶𝑂! 𝑥 28 𝑙𝑏𝑠 𝑚𝑜𝑙𝑒 𝐶𝑂 44 𝑙𝑏𝑠 𝑚𝑜𝑙𝑒 𝐶𝑂! 𝑥 132 𝑙𝑏𝑠 𝐶𝑂! 𝑝𝑟𝑜𝑑𝑢𝑐𝑒𝑑 42 𝑙𝑏𝑠 𝑝𝑟𝑜𝑝𝑦𝑙𝑒𝑛𝑒 𝑏𝑢𝑟𝑛𝑒𝑑 𝑥47.2 𝑙𝑏𝑠 𝑝𝑟𝑜𝑝𝑦𝑙𝑒𝑛𝑒 𝑏𝑢𝑟𝑛𝑒𝑑 10! 𝐵𝑡𝑢 The emission factors for NOx and CO were calculated by applying the relationships above for all tests using data within the statistical summary of the CMA study with results shown in Table A-1 of Appendix A. Development of TNRCC Emission Factors In 1994, the TNRCC published emission factors primarily based on statistical data from the original study performed by the CMA and EPA in 1983. To achieve higher accuracy for estimating actual emissions from elevated flares, the TNRCC emission factors accounted for the flare type and lower heating value of the relief gas. The emission factors developed by the TNRCC are shown below in Table 1. Upon reviewing the statistical data collected during the CMA testing, the emission factors shown in Table 1 reflect an average of the derived NOx and CO emission factors pertaining to the flare type and lower heating value of the relief gas. When calculating the emission factors for steam-assisted flares relieving a low Btu waste gas and airassisted flares relieving a high Btu waste gas, all test data collected during the CMA testing was included in the average of the derived emission factors. In order to calculate emission factors for the remaining categories, multiple tests had to be disregarded due to various reasons. For instance, during testing in the high Btu steam-assisted category, the emission probe was placed into the flare flame during test 67, resulting in a substantial increase in the concentration level of both THC and carbon monoxide. During tests 61 and 55, it was noted that the flare was capped by the assist steam, contributing to destruction efficiencies well below 98%. Omitting tests 67, 61 and 55 from the high Btu steam-assisted waste gas data, the average of the derived NOx and CO emission factor of the remaining tests resulted in the values shown above in Table 1. To calculate the emission factor for air and non-assisted flares relieving low Btu waste gas, certain CMA test data were excluded in the calculation performed by TNRCC. Upon detailed review of the CMA test data, one possible method for calculating the emission factors in Table 1 entails disregarding tests 66, 29, 29a, 29b, and 62. Reasoning for this assumption may be due to the lower heating value of the relief gas being below 184 Btu/SCF. However, data from test 33 appears to be included in the overall average Type Waste Gas NOx lb/MM Btu of Waste Gas Steam-­‐Assisted High Btu (>1000/scf 0.0485 Steam-­‐Assisted Low Btu (192-­‐1000/scf) 0.0680 Air and Non-­‐Assisted High Btu (>1000/scf 0.1380 Air and Non-­‐Assisted Low Btu (184-­‐1000/scf) 0.0641 0.2755 0.5496 CO lb/MM Btu of Waste Gas 0.3503 0.3465 Table 1. Emission factors developed by the TNRCC for NOx and CO emission factor and should not be applicable since the lower heating value of the flare gas is 83 Btu/SCF. Once this test point is excluded from the overall average, the emission factors for NOx and CO are 0.0645 and 0.4989 lb/MM Btu, respectively. Development of EPA AP 42 Emission Factors According to EPA AP-42 Chapter 13.5, emission factors were developed based on data obtained during the 1983 CMA testing, which included steam-assisted flare tests with gas exit velocities between 130 and 3,750 ft/min and air-assisted flare tests at gas exit velocities between 617 and 13,087 ft/min. It is also stated that the EPA established that steam-assisted and air-assisted flares could attain destruction efficiencies greater than 98% when the LHV of the flare gas is at least 300 Btu/SCF. The emission factors published by the EPA for NOx and CO are shown below in Table 2. Since the emission rate of NOx is independent of combustion efficiency, all CMA tests were included in the average of the derived NOx factor taken by the EPA. This resulted in the NOx emission factor presented above in Table 2. In order to determine the CO emission factor published by the EPA, an assumption was made that CMA tests having a destruction efficiency less than 98% could be excluded. When this assumption is applied to the CMA test data, the average of the derived CO emission factors for the remaining tests resulted in the value presented above in Table 2. Component Emission Factor, lb/MM Btu Carbon Monoxide 0.37 Nitrogen Oxides 0.068 Table 2. AP 42 Emission Factors for CO and NOx Conclusion To estimate emission rates for carbon monoxide and nitrogen dioxide based on CMA test results, variables such as the LHV, exit velocity and flare type must be considered. These variables were considered by both the EPA and the TNRCC. However, the discrepancy between the published emission factors has caused some companies to permit elevated flares using AP 42 guidelines citing values that are not representative of actual emissions. In order to address this issue, supplemental testing must be performed that considers additional variables to determine emission rates for current industrial applications. References 1. AP 42, Fifth Edition, Compilation of Air Pollutant Emission Factors, Volume 1: Stationary Point and Area Sources, U.S. Environmental Protection Agency, Research Triangle Park, NC, January, 1995. 2. Flare Efficiency Study, EPA-600/2-83-052, U.S. Environmental Protection Agency, Cincinnati, OH, July 1983. 3. Technical Guidance Package for Chemical Sources: Flare Sources, Texas natural Resource Conservation Commission, Austin, Texas, November 1994. Appendix A Type Test No. Heating Value Btu/SCF NOX Concentration ppm CO Concentration ppm CO2 Concentration ppm Derived NOX Emission Factor lb/MM Btu Derived CO Emission Factor lb/MM Btu CE% 1 2183 3.09 3.8 7,052 0.0680 0.0836 99.96 2 2183 2.16 8.5 4,719 0.0710 0.2793 99.82 3 2183 3.46 13.8 8,159 0.0658 0.2623 99.82 4 2183 1.96 75.3 6,616 0.0459 1.7651 98.80 8 2183 1.45 61.1 5,400 0.0416 1.7548 98.81 7 2183 1.62 7.9 5,224 0.0481 0.2345 99.84 5 2183 2.09 4.1 6,115 0.0530 0.1040 99.94 67 2183 3.77 N/A 3,758 0.1556 N/A N/A 17 2183 1.00 6.1 3,493 0.0444 0.2708 99.84 50 2183 0.50 16.7 4,220 0.0184 0.6137 99.45 56 2183 0.58 7.8 3,120 0.0288 0.3877 99.70 61 2183 1.32 398.4 6,273 0.0326 9.8495 82.18 55 2183 0.38 171.0 2,012 0.0293 13.1807 68.95 57 294 2.68 5.0 6,945 0.0598 0.1117 99.90 11 -­‐ 3.69 7.1 5,269 0.1086 0.2090 99.83 11a 305 3.31 4.7 6,677 0.0769 0.1092 99.93 11b 342 4.17 8.6 8,158 0.0793 0.1635 99.85 11c 364 4.00 11.6 8,210 0.0756 0.2191 99.82 59 -­‐ 1.41 49.9 5,413 0.0404 1.4297 98.49 59a 192 1.30 62.1 5,575 0.0362 1.7275 98.11 59b 232 1.62 25.4 5,090 0.0494 0.7739 99.32 60 298 0.99 28.3 3,685 0.0417 1.1910 98.92 51 309 0.57 34.1 3,347 0.0264 1.5800 98.66 16 -­‐ 1.87 7.7 4,059 0.0714 0.2942 99.75 16a 339 1.39 6.1 3,236 0.0666 0.2923 99.74 16b 408 2.42 9.6 5,291 0.0709 0.2814 99.75 16c 519 1.57 7.3 3,419 0.0712 0.3311 99.74 16d 634 2.28 7.9 4,458 0.0793 0.2748 99.78 54 209 5.00 6.8 7,115 0.1090 0.1482 99.90 23 267 5.90 4.5 8,465 0.1081 0.0824 100.01 52 268 0.68 16.1 2,622 0.0402 0.9523 98.82 53 209 2.83 23.9 5,741 0.0764 0.6456 99.40 26 481.6 5.34 5.5 6,270 0.1321 0.1360 99.97 65 159.0 2.40 20.3 4,878 0.0763 0.6454 99.57 28 157.0 8.16 3.2 6,078 0.2082 0.0817 99.94 31 22.7 4.02 27.9 4,568 0.1365 0.9472 99.17 66 158 0.97 129.4 2,432 0.0619 8.2517 61.94 29 -­‐ 1.06 180.9 2,179 0.0754 12.8752 61.60 29a 168 1.09 146.6 1,529 0.1106 14.8696 55.14 29b 146 1.04 213.9 2,808 0.0574 11.8137 65.60 64 282 1.24 8.6 3,282 0.0586 0.4064 99.74 62 153 0.68 90.2 3,076 0.0343 4.5477 94.18 63 289 1.57 19.9 4184 0.0582 0.7376 99.37 33 83 0.74 15.8 1857 0.0618 1.3195 98.24 32 -­‐ 1.75 22.7 3702 0.0733 0.9510 98.87 32a 294 0.63 12.2 1761 0.0555 1.0744 98.91 32b 228 2.39 28.8 4811 0.0770 0.9284 98.86 Air & Nonassisted Low Btu Table A-­‐1. Derived NOx and CO Emission Factor Using CMA Test Data Steam Assisted Low Btu Steam Assisted High Btu Air & Nonassisted High Btu