Performance Test of Steam-Assisted and Pressure Assisted Ground Flare Burner Arrays with Passive FTIR

Paper from 2013 AFRC conference titled Performance Test of Steam--Assisted and Pressure Assisted Ground Flare Burner Arrays with Passive FTIR by D. Pearson.

Performance Test of Steam-Assisted and Pressure Assisted Ground Flare Burner Arrays with Passive FTIR Dan Pearson - Clean Air Engineering Scott Evans - Clean Air Engineering Ruth Cade - Marathon Petroleum Company Jim Manning - Marathon Petroleum Company Brandon Monistere - Marathon Petroleum Company Page 2 of 11 Abstract Marathon Petroleum Company (MPC) and Clean Air Engineering (CleanAir) conducted performance testing on the Garyville ground flare burners at the John Zink test facility in Tulsa, Oklahoma from August 3rd to August 11th, 2012. The performance testing was conducted with Passive FTIR technology. Two test series were conducted on SKEC (steam-assisted) burners and two test series were conducted on LRGO (pressure-assisted) burners. The purposes of the test were to determine the NHVcz required to maintain high combustion efficiency and to demonstrate that both types of burners can achieve compliance with necessary combustion efficiency requirements at velocities greater than 400 ft/s. The test results show that when Tulsa natural gas (TNG) is used as fuel, the SKEC burners maintain high combustion efficiency until NHVcz falls below 300 BTU/scf. This is consistent with past PFTIR studies on steam-assisted flare tips. When TNG is used as fuel, the LRGO burners maintain high combustion efficiency until the NHVcz falls to a point where the flame is extinguished. Under all conditions tested, these burners maintained an average CE equal to or greater than 99%. Under all conditions tested, the flame extinguishes at an NHVcz of approximately 550 BTU/scf. This test program demonstrated that as long as the LRGO burners are lit, they maintain high combustion efficiency. Both the LRGO and SKEC burners maintain high combustion efficiency at exit velocities much higher than those allowed by 40 CFR 60.18 and 63.11. The flares operated efficiently at velocities up to sonic which would represent the maximum velocity achievable. Page 3 of 11 Introduction Marathon Petroleum Company (MPC) and Clean Air Engineering (CleanAir) conducted performance testing on the Garyville ground flare burners at the John Zink test facility in Tulsa, Oklahoma from August 3rd to August 11th, 2012. This test was the third performance test on refinery flares by MPC, following the tests in Texas City, TX, in 2009 and Detroit, MI, in 2010. The main objective of the test was to develop an operating envelope for both types of the John Zink burners used on the Garyville ground flares. The performance tests were conducted using Passive Fourier Transform Infrared (PFTIR) instruments developed by Industrial Monitor and Control Corporation (IMACC). The specific analytical method used for these tests is the same method used and validated during testing conducted by the Texas Commission on Environmental Quality (TCEQ) in 2010. This memo summarizes the test results and compares them to results from previous flare tests. The purpose and major benefit of an assisted flare is to significantly reduce the amount of smoke (visible emissions) that would otherwise be created by combustion. The burners tested in this program include both steam-assist and pressure-assist. In a typical system, an assist medium such as steam is injected into the flare combustion zone to deliver educted air as well as mixing energy. Pressure assisted flares rely on increased exit velocity of the flare gas to educt more air into the gas. Over-assist is a generic description of an undesirable operating condition possible in assisted flare systems. For example, in an over-steaming scenario, the amount of steam and educted air introduced into the combustion reaction zone diminishes, rather than promotes, the efficiency of the combustion process if introduced in large enough quantities relative to the flare gas flow rate. The operating envelope of a flare is bounded by excess smoke (too little assist) on one side and decreased combustion efficiency resulting in excess emissions of volatile organic compounds (VOCs) (too much assist) on the other. Smoke suppression is easily monitored by visually observing whether smoke is present. However, the ability to measure or even identify excess emissions caused by over-assisting is a more difficult task. Standard emission estimation techniques have generally assumed a 98% combustion efficiency or higher when calculating VOC emissions from flares regardless of assist rate. Regulatory requirements for flares are contained in 40 CFR §60.18 and §63.11. These requirements were developed from a series of flare emissions tests led by the United States Environmental Protection Agency (US EPA) from 1983 - 1986. The requirements include maintaining a flare pilot, operating with a minimum net heating value of 300 BTU/scf in the vent gas, operating at exit velocities of less than 60 ft/s (or up to 400 ft/s depending upon the vent gas net heating value), and operating with a limited amount of visible emissions. However, a flare can be operated in compliance with these requirements and still have low combustion efficiency due to over-assisting. Prior to the recent refinery tests of flare performance, including the US EPA tests in the mid-1980s, flare tests were conducted on pilot-scale test flares or on flares operating at moderate to high vent gas loads. However, a flare typically operates at low vent gas loads (i.e. high turndown) under normal conditions until a process upset or other operating Page 4 of 11 condition requires the operator to flare waste gas. Thus, the flare normally operates at high turndown for the majority of the operating year, a condition for which there was little to no available performance data. In the past, measuring the combustion products from a flare was difficult and dangerous. However, recent technological advances have produced remote sensing instruments capable of measuring combustion products such as carbon dioxide, carbon monoxide, and select hydrocarbons without the safety hazards introduced by physically sampling a flare plume. One such instrument is the PFTIR, which characterizes a plume's chemical make-up (carbon dioxide, carbon monoxide, and total hydrocarbons) in units of concentration × pathlength. Using this technology, the absolute concentration cannot be determined from a flare plume, but the product of concentration × pathlength (e.g., ppmv × meters), can be used in combustion efficiency calculations. The PFTIR is a relatively new tool and was recently blind-validated against extractive sampling results for flare plume testing by TCEQ and the University of Texas in 2010. The PFTIR was first used for refinery flare testing at MPC Texas City in 2009. Several accuracy, precision, and bias checks were performed during the recent flare tests to better characterize the PFTIR measurement technique. Because of the practical and logistical difficulties of conducting a test of a ground flare at an actual refinery, this test program was carried out on a test cell at the John Zink test facility in Tulsa, Oklahoma. Three burners of each type were tested each identical to the burners installed at the Garyville facility. Figure 1: John Zink Test Stand (showing SKEC burners installed). Page 5 of 11 Test Conditions Combustion efficiency was measured under several combinations of vent gas composition and flare operating parameters. Each of these combinations is referred to as a "test condition" in this report. The following test conditions were used during this project. Test SN1 SKEC Steam-assist burners. 100% Tulsa natural gas (TNG) Objective: To determine the performance curve of the burner at the minimum flow rate. Test VS1 SKEC Steam-assist burners. Velocity Screening Test Variable TNG/nitrogen mix Objective: To identify the minimum NHVcz needed to maintain a flame over a range of exit velocities. Test PA1 LRGO Pressure-assist burners. Variable TNG / nitrogen mix. Objective: To determine the minimum NHVcz that supports good combustion at sonic velocity. Test PA2 LRGO Pressure-assist burners. Variable TNG / nitrogen mix. Objective: To determine the minimum NHVcz that supports good combustion at minimum velocity. The PFTIR performance test conducted on John Zink's SKEC burners, LRGO burners, and elevated steam flare produced valuable insights into the efficiency performance of ground flare burners under a variety of conditions. Tests were conducted while flaring gases containing a mixture of Tulsa natural gas (TNG), propylene, and nitrogen. For the results presented below, relationships between combustion efficiency and other parameters were analyzed: • VG Composition - Composition of the gas being sent to the flare tip • NHVcz - Net heating value of the combustion zone gas (BTU/scf) • S/VG - Actual steam to vent gas ratio (lb steam/lb vent gas) or (scf steam/scf vent gas) • Exit Velocity - Velocity of gases exiting the flare tip Page 6 of 11 Results A summary of results from the VS1 and SN1 tests is shown in Figures 1-2 through 1-8 below. Figure 2: NHVcz vs. CE for SN1 & VS1 (SKEC Burners) Figure 3: NHVcz vs. CE for SKEC Burners & Other Base Load Test Series Page 7 of 11 Figure 4: S/VG (lb/lb) vs. CE for SN1 & VS1 (SKEC Burners) Figure 5: S/VG (lb/lb) vs. CE for SKEC Burners & Other Base Load Test Series Page 8 of 11 Figure 6: S/VG (scf/scf) vs. CE for SN1 & VS1 (SKEC Burners) Figure 7: S/VG (scf/scf) vs. SKEC Burners & Other Base Load Test Series Page 9 of 11 Figure 8: Tip Velocity vs. Minimum NHVcz (SKEC Burners) A summary of results from the PA1 and PA2 tests is shown in Figures 1-12 and 1-13 below. Figure 1-12: NHVcz vs. CE for PA1 & PA2 (LRGO Burners) Page 10 of 11 Figure 1-13: Tip Velocity vs. NHVcz for LRGO Burners Conclusions The PFTIR test of the SKEC burners and LRGO burners provided data to support the following conclusions. SKEC Burners • When TNG is used as fuel, the SKEC burners maintain high combustion efficiency until NHVcz falls to less than 300 BTU/scf. This is consistent with past PFTIR studies on steam-assisted flare tips. • The SKEC burners maintain high combustion efficiency at exit velocity limits higher than those established by 40 CFR 60.18 and 63.11. At velocities greater than 60 ft/sec and up to sonic (maximum) velocity, the burner operates efficiently until the NHVcz falls to a point where the flame is extinguished. Therefore, the SKEC burners should be exempt from the velocity limits established in 40 CFR 60.18 and 63.11. • The S/VG does not appear to be a factor for the SKEC tip until the exit velocity falls below 60 ft/s. Therefore, when operating at velocities greater than 60 ft/s, an S/VG limit should not apply to the SKEC burners. At velocities less than 60 ft/s, a S/VG limit of 3 lb/lb or 3 scf/scf could be used by the SKEC burners to assure high combustion efficiency. Page 11 of 11 LRGO Burners • When TNG is used as fuel, the LRGO burners maintain high combustion efficiency until the NHVcz falls to a point where the flame is extinguished. Under all conditions tested, these burners maintained an average CE equal to or greater than 99%. Under all conditions tested, the flame extinguishes at an NHVcz of approximately 550 BTU/scf. This test program demonstrated that as long as the LRGO burners are lit, they maintain high combustion efficiency. To ensure the LRGO burners are operating at high combustion efficiency, the NHVcz of the LRGO burners should be kept above 550 BTU/scf. This eliminates any need for Combustion Efficiency Multipliers. • The LRGO burners maintain high combustion efficiency at exit velocities much higher than those allowed by 40 CFR 60.18 and 63.11. The flares operated efficiently at velocities up to sonic which would represent the maximum velocity achievable. Therefore, the LRGO burners should be exempt from the velocity limits established in 40 CFR 60.18 and 63.11.