Pressure-assisted flare emissions testing

Due to increased concern regarding air emissions of highly-reactive volatile organic compounds (HRVOC) in the Greater Houston area, emissions from various sources in the area are now subject to much more scrutiny by the Texas Commission on Environmental Quality. One potential source of emissions is from flares whose design and operation is addressed by the EPA through the Federal Code of Regulation 40-CFR Section 60.18. In particular, this standard sets forth the relationship between volumetric heating value of the flare gasses and the maximum allowable exit velocity. Depending on the specific flare type and gas heating value, maximum allowable exit velocities are generally limited to less than about 400 feet per second. This code was originally developed around non-assisted, air-assisted, and steam-assisted flare technology. Pressure assisted flares are not specifically addressed by the regulation. Pressure assisted flares often operate with an exit velocity of about Mach 1.0, which is much higher than allowed by 40-CFR-60.18. In partnership, John Zink and Dow set out to demonstrate that the destruction efficiency for pressure assisted flares is equal to or better than other flare types, even though the exit velocity is higher. Several different pressure assisted flare burners were tested using various gas mixtures, and the emissions from the flames were captured and measured.

P r e s s u r e - A s s is t e d F la re E m is s io n s T e s tin g By V an ce V arner1, Scott Fox2, Robert Schw artz3 and Russell W ozniak4 P re s e n te d at the American - J a p a n e s e Flam e R e s e a rc h C om m ittees International Symposium Hawaii, O ctober 2007 ABSTRACT: Due to increased concern regarding air em ission s of highly-reactive volatile organic c o m p o u n d s (HRVOC) in the G re ater Houston area, em issio ns from various s o u rc e s in the a re a are now subject to much m ore scrutiny by the T e x a s Commission on Environmental Quality. O n e potential s o u rc e of em ission s is from flares w h o se design and operation is a d d re s s e d by the EPA through the Federal C o de of Regulation 40-C FR Section 60.18. In particular, this stan d ard s e ts forth the relationship b etw een volumetric heating value of the flare g a s s e s a n d the maximum allowable exit velocity. D epending on the specific flare type a n d g a s heating value, maximum allowable exit velocities are generally limited to less than ab ou t 400 feet per seco nd . This c o d e w a s originally dev elop ed around non-assisted, air-assisted, a nd s te a m -a s siste d flare technology. P re s su re a ss is te d flares are not specifically a d d re s s e d by the regulation. P re s s u re a ss is te d flares often o p e ra te with an exit velocity of abo ut Mach 1.0, which is much higher than allowed by 40-CFR-60.18. In partnership, John Zink a n d Dow s e t out to d e m o n stra te that the destruction efficiency for p re s su re a ss is te d flares is equal to or better than other flare types, ev en though the exit velocity is higher. Several different p re s su re a ss is te d flare burners w e re te ste d using various g a s mixtures, and the em ission s from the flam es w e re captured and m easu red. 1Presenting Author, The Dow 2 Chemical Company; John Zink Company LLC; "Corresponding Author, John Zink Company LLC, Tulsa, OK, USA, [e-mail]; 4The Dow Chemical Company, © 2007 by John Zink Company, LLC Page 1 In tro d u c tio n A n um b er of p ro c e ss industry operations in the T e x a s gulf c o a st a re a are served by flares that provide safe, effective disposal of combustible g a s e s . Recently, the efficacy of a small n um b er of th e s e flares c a m e into question. T h e Dow Chemical C om pany (Dow) is a m o ng the co m p a n ies that u s e flares. Dow a n d a flare designer/supplier, Jo h n Zink C om pany LLC (Zink), partnered in undertaking a te st program that would determ ine the destruction removal efficiency (DRE) of the flares in question. This test program w a s a significant undertaking ex ec u te d at Zink's te st facility located in Tulsa, O klahom a a nd w itn essed by represen tativ es of the US EPA and the T e x a s Commission on Environmental Quality (TCEQ). B ackground C oncern ab ou t em issio ns in the Houston - G alveston a re a prom pted the TCEQ a n d a large nu m ber of re s e a rc h e rs to collect ground-level and airborne em ission s d a ta along the T e x a s gulf coast. T h e first study w a s co nd ucted in the late s u m m e r of 2000 (T exas Air Quality Study I). A follow-up study w a s co nducted in the late s u m m e r of 2 0 06 (T exas Air Quality II). As a result of T e x a s Air Quality Study I, several volatile organic compounds (VOC's) w ere d e te cte d in concentrations that e x c e e d e d the levels that w ere predicted b a s e d on plant annual em ission inventory reports. S o m e of th e s e V OC's, known to b e major contributors to ozone, h av e b e e n d e sig n a te d a s Highly Reactive Volatile Organic C o m p o u n d s (HRVOC's) by the TCEQ. Ethylene, propylene, butenes, and 1,3bu tadien e a re a m on g the hydrocarbon c o m p o u n d s d e sig n a te d a s HRVOC's. Dow o p e ra te s several flares that handle c o m p o u n d s d e sig n a te d a s being HRVOC's. T he a p p a re n t discrepancy betw een the plant reports and the aerial m e a s u re m e n ts c a u s e d the TCEQ to se a rc h for potential s o u rc e s of HRVOC Page 2 em issions. The TCEQ identified certain ty pes of p ro c e ss equipm ent a s being s u s p e c t for und er reporting of em issio ns - flares, vents, cooling tow ers and fugitives. In the c a s e of flares, any u n d e rstatem e n t or ov e rstatem e n t of flare DRE w a s probably d u e to a lack of specific know ledge of flare DRE's. TCEQ R e g u la to r y R ev iew T he TCEQ developed regulations to a d d re s s HRVOC em issio ns in the a re a from flares, cooling towers, vents, and fugitive so u rces. For flares, the regulations m a n d a te that all flares m e e t the requirem ents outlined in the C o de of Federal Regulations, Title 40, Part 60.18 (Part 60.18) w hen com busting HRVOC materials. Part 60.18 contains both requirem ents and gu id an ce for the operation of certain typ es of flares: stea m -assist, air-assist a nd n o n -a ssiste d flares. However, there a re no requirem ents or g uid ance provided for p re s su re -a s sis te d flares su ch a s the Dow flares in question. A key regulatory issue a d d re s s e d in Part 60.18 is the maximum allowable flare g a s velocity at the burner exit. Previous testing h a s show n that th ere is a relationship betw een g a s exit velocity a n d flame stability. T h e s e te sts also d e m o n stra te that flame stability is a key p a ra m e te r that determ in es flare em issions. In turn, there is a relationship betw een g a s heating value a nd exit velocity. Actually, relating heating value to g a s exit velocity is a convenient simplification b e c a u s e several other factors also influence flame stability. Fuel characteristics a re d is c u s s e d in m ore detail later in this paper. It is sufficient here to note that, in general, the m ore stable the flame the higher the DRE. Conversely, a flare flame that is n e a r the point of blowing out will g e n e ra te higher em issions. T he impact of Part 60.18 on flare g a s exit velocity can be d e m o n stra ted a s follows: A ss u m e a g a s with a heating value, a s determ ined by the m e th o d s in the regulation, g re a ter than 1000 Btu/scf (37.23 MJ/scm). For this heating value the maximum g a s exit velocity for a s te a m -a ssiste d , air-assisted or n o n -a ssiste d flare is 400 feet p er s e c o n d (122 m/s). Even if th e s e flares a re firing p ro p a n e with a Page 3 heating value of 23 1 5 Btu/scf (86.20 MJ/scm), the maximum allowable g a s exit velocity is still 400 feet per second; (122 m/s) this co rre sp o n d s to a Mach n u m ber of approximately 0.5. In contrast, p re s su re -a s sis te d flare burners generally o p e ra te at sonic exit velocities at the tip. (Mach. 1.0 is 820 feet p er se c o n d [250 m/s] for propane.) Dow h a s re q u e ste d approval from TCEQ for the u s e of p re s su re -a s sis te d flares for com busting HRVOC c o m p o u n d s in certain applications. T h e T C E Q 's review of Dow's re q u e st resulted in a nu m ber of qu estio ns and re q u e sts for information. Lacking regulatory guidance, and with a limited am o un t of publicly available perform ance d a ta on p re s su re -a s sis te d flare burners, Dow and TCEQ a g re e d to conduct a series of flare efficiency te sts in support of Dow's claimed DRE's. T h e s e te sts w e re to d e m o n stra te that the DRE's of Dow's p re s su re -a s sis t flare burners are equal to or better than flares that m e e t the exit velocity restrictions contained in Part 60.18. S m o k e l e s s Flaring S m o k e from a flare re p re se n ts incomplete com bustion c a u s e d by insufficient air. For th o s e g a s e s that will sm o k e w hen flared, s o m e m e a n s of assisting the mixing of com bustion air and the g a s is required to ach ieve effective disposal (sm o k e le ss burning). S te a m -a s sis t a nd forced air-assist are commonly u sed to provide the mixing energy. However, there is a n o th e r a ss is t m ean s, p re ssu re assist, which is very effective. A p re s su re -a s sis te d flare burner ta k e s a d v a n ta g e of the w a s te g a s p re s su re (energy) to c re ate a condition w h ereb y air is drawn into contact with the gas, and mixed to g ether with the g a s in such a m a n n e r a s to achieve s m o k e le s s burning. Exit velocities up to a n d including sonic a re utilized. In general, a higher the exit velocity results in m ore air mixed with the w a s te g a s stream prior to burning. However, the a d v a n ta g e s of high exit velocities can be lost if the flare d e sig n e r fails to provide a d e q u a te flame stability. P re s su re a ss is te d flare burners m ust be carefully de sig n e d for stability. Designing a flare for stable burning requires a c o m preh ensive know ledge of the com bustion characteristics of the g a s to be burned a n d a general understanding Page 4 of fluid dynam ic principles. F lam es can be stabilized mechanically, aerodynamically, or by using a combination of both techniques. The Dow p re s su re -a s sis t flare burners are aerodynamically stabilized and o p e ra te at g a s exit velocities of Mach 1.0. At th e s e high exit velocities the burners remain stable ev en in high wind conditions. By their nature, p re s su re -a s sis te d flare burners h ave a sm aller outlet a re a a s co m p ared to a corresponding conventional ste a m or air-assist flare. This a re a can be utilized to design a burner with a single exit port or divided into a nu m ber of ports strategically arra n g e d to provide stability and s m o k e le s s flaring. In practice, p re s su re -a s sis te d flare burners a re u s e d a s a single flare burner or in groups of m any identical burners. Multiple burner s y s te m s commonly employ a burner staging technique in order to greatly e x p an d the capacity ran ge of effective burning. Dow flares, employing this staging technique, h av e b e e n in service for over 30 years. D e v e lo p m e n t o f a T e s t P ro to c o l Zink h a s b e e n an active participant in nu m ero u s flare em ission s test program s a n d o p e ra te s a flare test facility that is well suited for flare em issio n s studies. Dow recognized the a d v a n ta g e s in working with Zink and turned to Zink for support in preparation of a te st protocol a n d te st execution. Dow, Zink, TCEQ, US EPA a nd a third-party analytical service worked together to develop a te st protocol that would guide the testing of flare burners representative of th o se in service at various Dow locations. Preparation of the protocol required careful consideration of e a c h application to a s s u r e that the te sts would h av e a meaningful relationship to actual operating conditions. A review of Dow's p re s su re -a s sis t flare burner u s a g e determ ined that two flare burner d e sig n s with significantly different exit a re a s, to geth er with the proper fuel selection would re p re se n t all of the Dow of p re s su re -a s sis t flare burner applications in question. Both flare burners a re multi-armed with o n e or more ports per arm. Page 5 The te st protocol covered a nu m ber of s u bjects including the following: • Fuel composition • T est equipm ent p ro c e ss design • S a m p le acquisition and analysis • QAQC activities • Meteorological data T h e s e subjects a re d is c u s s e d below. Fuel composition: The flare burners s ele c te d for testing a re in applications w here they burn ethylene, propylene, or a mixture of both. At the time of the tests, Zink's te st facility could not handle ethylene. Therefore, Zink pro po sed conducting the flare efficiency te sts using propylene or mixtures of propylene and nitrogen. Zink's pro po sed fuel selection w a s supported by the following technical analysis: As noted earlier in this paper, burner flame stability is important if a flare burner is to achieve good em ission s perform ance. T e sts conducted by the EPA, CMA and Zink, EPA a nd EER a n d the EPA with Dupont clearly d e m o n stra te that the burning efficiency of a flare is directly related to flame stability. The te sts co nd ucted by the EPA, CMA a n d Zink in 1982, sta te the following in the project report: "Flaring low BTU content gases at high exit velocities, causing flare flame lift off, may result in low combustion efficiencies..." The report on the te sts con du cted by the EPA and EER in 1985 s ta te s the following: "Destruction efficiencies greater than 98 percent were attained when the gas heating value w a s at least 1.2 times the minimum gas heating value required for stability." The report on the te sts con du cted by the EPA a nd Dupont in 1996 s ta te s the following: "All the measurements of destruction efficiencies at conditions more Page 6 stable than lift off were above 99 percent. Further, control efficiencies greater than 98 percent were found at hydrogen contents below the lift off curve." In sum m ary, all of the cited re fe re n c es d e m o n stra te a positive link betw een flare flame stability and high flare efficiency. Therefore, a g a s that h a s a higher propensity to c re ate a stable flame is m ore likely to achieve higher burning efficiencies. G a s properties that play a key role in determining burner flame stability include the following: flammability limits, flame s p ee d , a n d ignition tem perature. A p re s su re -a s sis te d flare burning a g a s that h a s wide flammability limits is e a s ie r to stabilize than o n e with narrow flammability limits. For exam ple, hydrogen is much e a s ie r to stabilize than m e th a n e; o n e re a so n is hydrogen's much wider flammability limits (U pper Flammability Limit [UFL] = 4.0 % and Lower Flammability Limit [LFL] = 74.2 %) vs. m e th a n e (UFL = 5.0 % a n d LFL = 15 %). S tated a n o th e r way, it is less difficult to achieve a flam m able mixture of hydrogen a n d air, than a flammable mixture of m e th a n e a n d air. A nother important property that contributes to hydrogen being m ore stable than m e th a n e is flame s p ee d . A higher burning velocity m e a n s that the com bustion reactions occur at a much fa ster rate, thus promoting a m ore stable flame. Hydrogen h a s a much higher laminar burning velocity than m e th a n e over the entire flammability range. For exam ple, hydrogen h a s a maximum laminar burning velocity in air of 9.3 feet p er se c o n d (2.83 m/s) vs. m e th a n e at 1.48 feet p er s e c o n d (1.45 m/s). Ignition tem p e ra tu re is a n o th e r important property that gives hydrogen superior stability. Hydrogen h a s an ignition tem p e ra tu re of 1062°F vs. 1170°F (572 °C / 632 °C) for m ethan e. In a continuum of com bustion the lower ignition tem p e ra tu re s u g g e s ts m ore rapid ignition of the g a s air mixture and a more stable flame. Page 7 The analysis d isc u s s e d ab o v e d e m o n stra te s that a com parison of certain fuel properties can b e useful in estimating the relative stability perform ance of different fuels. Applying the s a m e analysis to ethylene a n d propylene sho w s that ethylene h a s wider flammability limits, a lower ignition tem p e ra tu re and a faster flame s p ee d . Therefore, it w a s a c c e p te d that propylene, a s a fuel, will not perform a s well a s ethylene and that the DRE's obtained from testing propylene would be lower than they would b e if ethylene had b e e n u s e d a s fuel. In order to d e m o n stra te the ran ge of perform ance a n d flexibility of their p re ssu re a ss is te d flare burners, Dow decided that e a c h of the sele c te d te st flare burners should be te sted on a fuel with a heating value lower than the heating value of g a s e s flared at their facilities. The reduction in heating w a s achieved by blending propylene with nitrogen. Combining the s ele c te d te st flare burners and the test fuels results in the te st matrix show n in Table 1. Test A B C D E Flare Burner Large Large Large Small Small Target Flow Lbs/hr 5,000 8,000 5,000 1,200 1,200 Fuel HC V V Kg/hr 2,270 3,640 2,270 550 550 HC+N2 V V V Table 1 T est Matrix T est equipm ent p ro c e ss design: After establishing the te st matrix, it w a s n e c e s s a ry to select/design the equipm ent a n d instrumentation n e c e s s a ry to carry out the te st plan. Zink's existing flare test a n d develo pm ent c e n te r provided the infrastructure required to support the test program. The a rra n g e m en t of equipm ent is show n in Figure 1. T h e propylene c o m p on en t of the fuel arrived on site via tanker truck a n d w a s off loaded into a liquid sto ra g e tank. Nitrogen arrived in g a s e o u s form in cylinder truck loads. Nitrogen w a s u s e d a s a purge g a s a s part of the system safety plan a nd a s a c om p on en t in the te st fuel. During testing the flow of propylene or nitrogen w a s m e a su re d using an orifice metering system . A static mixer in the fuel line a s s u re d a h o m o g e n o u s mixture. T he rate of Page 8 propylene and nitrogen consum ption w a s very high and timely replenishm ent required careful planning. T est flow rates w e re regulated and recorded by the control room b a s e d co m p uter network. S a m p le acquisition and analysis: Acquisition of a representative flue g a s sam p le w as an important e lem ent in the test program. Previous te st program s exp erienced difficulty with this element. Zink e n g in e ers devised a sam p le collector that con cen trated the flue g a s and protected the sa m p le suction point from the wind. T he sam p le collector w a s supported by a 180 foot (55 m) tall c ra n e a n d maintained in position by guy line tenders. T he position of the collector, relative to the flare flame, w a s constantly monitored using flue g a s tem perature, flue g a s oxygen level and visual sighting. From time to time, the observations would indicate the n e e d to reposition the collector to e n h a n c e the quality of the sam ple. The handling crew would then work in coordination with the c ra n e op erator to achieve a better position for the collector. A continuous flue g a s sam p le w a s drawn through the s am p le probe located within the collector. T he flue g a s sam p le p a s s e d through a h e a t traced and insulated s am p le line a n d into the analytical service instrument trailer. The flue g a s sam p le w a s divided with a portion going to the total hydrocarbons analyzer. T he remaining portion of the sam p le w a s dried a n d directed to an aly zers for m easuring CO, O2, C O 2 a n d NOx. The analyzer outputs w ere routed to a d ata logger. T he m eth o d s a n d p ro c e d u re s u s e d for the flue g a s analysis w e re in a c c o rd a n c e with US EPA guidelines and methodology. For exam ple, the analysis u s e d for total hydrocarbons followed the requirem ents of US EPA Method 25A. QA/QC activities: Quality a s s u r a n c e a nd control pro c e d u re s followed US EPA requirem ents a n d included calibration checks, zero and s p a n ch ec k s and background m e a su re m e n ts . P rete st a nd post te st calibrations took place with e a c h te st run. Certain p a ra m e te rs su ch a s fuel composition and flue g a s composition w ere ch eck ed using two different methods. Page 9 Meteorological data: Wind sp ee d , am bient w et and dry bulb tem perature, local atm osph eric p re s su re and humidity w e re o bserved a n d d a ta logged throughout the testing activities. T e s t e x e c u tio n : Preparation, review, modification and approval of the te st protocol required approximately ten m onths to complete. T h e effort invested in the protocol w a s rew arded w hen the test execution portion of the program moved rapidly and without any major problem. Logistics played a major role in the test execution. From maintaining an a d e q u a te te st fuel inventory, to organizing rotating shifts of operating personnel, to coordination of third party services providers such a s the crane, the team m e m b e rs responsible for logistics w e re very busy. Prior to any te st facilities activities, an Environmental Health a nd Safety review of the test s e tu p operation plan w a s conducted. A reas of responsibility w ere defined, a nd work a ss ig n m e n ts m ade. For exam ple, the chief operato r w a s a ss ig n e d the responsibility for interfacing with the local fire d ep artm en t and airport air traffic control. Execution of the test program w a s very exp en siv e and m an p o w er intensive, and lasted several d a y s o n ce it w a s started. Fuel a n d m an p o w er w e re the major c o sts incurred. Approximately 20 operators, technicians, engineers, co m pany re p resentativ es and regulatory a g e n c y o b serv ers w ere involved during the testing. From time to time additional workers w e re required to a ss is t in setu p or breakdow n of equipment. A total of five different com binations of burner a n d fuel w ere tested. Each combination w a s te ste d for three, 15 to 20 minute "on the record" runs. A pre-run period took place before the "on the record" portion of e a c h run. During the pre­ run the fuel flow would be started a nd the flow rate ad justed to approximately the target value. As the te st flow rate w a s established, the sam p le collector w a s lifted a n d guided into position a b o v e the flame. Instruments and d a ta loggers w ere Page 10 c h eck ed for proper function. W hen operation re a c h e d a n e a r s te a d y state condition the run would go "on the record" for 15 to 20 minutes. At the e n d of the "on the record" period the run would continue until the flue g a s analysis team leader confirmed that sampling w a s completed. At this point Fuel flow would be stopped. As show n in Figure 2, testing took place in an o p e n area. Wind velocity varied throughout the testing. The a v e ra g e wind s p e e d for all te st runs w a s about 6 MPH (10 KPH) a n d the maximum about 16 MPH (26 KPH). Wind w a s not a problem during the tests. T h e position of the sam p le collector relative to the flame w a s constantly monitored a nd the collector repositioned a s ne ce ssa ry . R esults a n d C onclusions A ded icated industrial-scale flare test facility offers m any a d v a n ta g e s over "inthe-field" testing. This is ev en m ore important w hen the te st focus is flare em issions. Increased safety and the ability to control and m e a s u re key p a ra m e te rs are a few of the benefits gained. Large sca le flare testing, although very exp ensive a nd difficult to execute, can yield valuable information regarding flare perform ance. The test program reported on herein d e m o n stra ted that the DRE's of the flare burner d e sig n s te sted a re at least a s good a s the p erform ance level's reported for previous te sts of s te a m -a s sist a n d air-assist flares. C om bustion stability is a major factor in flare burner perform ance. A well d e sig n e d and properly o p erated p re s su re -a s sis te d flare burner with a stable flame will achieve 99+% DRE, which is the s a m e or better than the efficiency of th o se flares that m e e t the requirem ents of C od e of Federal Regulations, Title 40, Part 60.18. Wind velocities up to 16 MPH (26 kph) had no identifiable impact on DRE results. Page 11 Water bath vaporizer Surge tank ') Propylene Propylene flow meter Nitrogen flow meter Nitrogen Static mixer Zink sample collector A Data trailer Figure 1 Flare Emissions Testing Equipment Arrangement 0 2007 by John Zink Company, LLC Figure 2 Overall View of Flare Emissions Test © 2007 byJohn ZinkCompany, LLC Page 12