Accurate and reliable flare testing methods

Recently the Texas Commission on Environmental Quality (TCEQ), through The University of Texas (UT) initiated a project utilizing John Zink facilities to better understand the operating envelope of varying flare technologies across various process conditions. A large scale flare testing environment was necessary to obtain accurate data, specifically with regard to emissions and combustion performance. The results of this testing offer objective data that may be used to improve and understand flare operation within the entire industry. For many years flare system performance predictions have been based on historical results from a few limited test points, as well as theory. This basis has also been used to define regulatory operation envelopes. As additional industry scrutiny calls for more controlled and defined limits of operation, the need for accurate and proven test results grows. Flare testing has always been difficult to conduct due to the unique challenges full-scale systems pose with their wide range of process conditions at intermittent durations, high flow-rates, and elevated position of operation. This paper will describe the recent UT/TCEQ testing of large-scale flares and the methods used to obtain reliable data for air-assisted, steam-assisted, and high-pressure flares.

Accurate and Reliable Flare Testing Methods By Zachary K odesh1, Scott Fox2, J a m e s Franklin3 Presented a t th e American Flare Research Com m ittee International Combustion Symposium Houston, T exas, S e p te m b e r 18-21, 2011 ABSTRACT: Recently t h e T exas Commission on Environmental Quality (TCEQ), th rough T he University of T exas (UT) initiated a project utilizing John Zink facilities to b e tte r u n d e rsta n d th e operating enve lo p e of varying flare technologies across various process conditions. A large scale flare testing e n vironm ent w as n e c essary to obtain a c c u ra te d a ta , specifically with regard to em issions and com bustion perform ance. T he results of this testing offer objective d a ta t h a t m ay be used to improve and u n d e rsta n d flare operation within th e en tire industry. For m an y y ea rs flare system perform ance predictions have been based on historical results from a few limited t e s t points, a s well a s theory. This basis has also been used to define regulatory operation envelopes. As additional industry scrutiny calls for m o re controlled a n d defined limits of operation, th e need for a c c u ra te and proven t e s t results grows. Flare testing has always been difficult to c onduct d u e to th e unique challenges full-scale sy ste m s pose with th eir wide ra n g e of process conditions a t interm ittent durations, high flow-rates, and elevated position of operation. This p a p e r will describe th e re c e n t UT/TCEQ testing of large-scale flares and th e m e th o d s used to obtain reliable d a ta for air-assisted, ste a m -a s s iste d , and high-pressure flares. 1 T ec h n o lo g y M a n a g e r - Flare Systems Division, John Zink C o m p a n y , LLC 2 Vice Presid ent - Flare Systems Division, John Zink C o m p a n y , LLC 3 D ire c to r A f t e r m a r k e t Sales - Flare Systems Division, John Zink C o m p an y , LLC © 2011 by John Zink Company, LLC Page 1 Introduction mm T h e John Zink C ompany, LLC (JZ) has a long history of flare testing, including t h e testing of flare emissions. In 1982, t h e John Zink t e s t facility w as th e site of th e Chemical M anufacturers Association (CMA) s ponsored flare em issions te s ts , t h e results of which w e re utilized in th e d e v e lo p m e n t of th e c u rren t federal regulations governing flares. In 2006, JZ conducted em issions testing of high p re ssu re flares for tw o s e p a r a te clients. It w as during t h e s e te s ts t h a t JZ first developed and em ployed a conical sam p le collector.4 In S e p te m b e r 2010, JZ h osted th e TCEQ s p o n so red Flare Study to ev alu ate emissions of large scale industrial flares a t turndow n conditions. During th e s e e n d e a v o rs JZ developed an e xpertise in performing flare em issions te sts. Photo 1, 2006 High Pressure Flare Emissions Test In preparation of th e TCEQ te s t, JZ upg rad ed its testing capability by adding special low flow t e s t s ta n d s for both assisted These flares. air and test s te a m s ta n d s w ere specially designed to a c c o m m o d a te a wide ra n g e of flare tips and w e re purposely built s h o rt to facilitate plume sampling. flow m e te rs w e re m e a s u r e m e n t accuracy added at to low Additional improve flow experienced during em issions te sts. Photo 2, Low Flow Flare Emissions Test Stands 4 V a r n e r , Va n c e, Scott Fox, R o b e rt Sc hw artz, and Russell W o z n ia k . "Pressure-Assisted Flare Emissions Testing." A m e r ic a n - Japan ese F la m e Research C o m m itte e s I n t e r n a ti o n a l Sym po sium (2 0 0 7 ) . Print. Page 2 rates Flare Inputs For a n y flare te st, w h e th e r em issions will be m e a s u re d or not, it is im portant to accurately know th e inputs to t h e flare. At a minimum th e m e a s u re d inputs should include: • Fuel Composition • Fuel Flow Rate • Fuel T e m p e ra tu re • Assist Media Flow Rate • Assist Media T e m p e ra tu re T h e p re ssu re of th e fuel and assist m edia a r e also inputs t h a t can be m e a s u re d b u t a re of secondary im portance. P ressure m e a s u r e m e n ts can be used a s a quality check of th e flow rates. Fuel Composition and Flow Rate For te s ts w h e re em issions will not be m e a s u re d , fuel com positions te n d to be p ure fuels. In such c ases, valves a r e closed to isolate fuels not in use and th e potential for contam ination is low. Most flare te s ts n ot involving em issions testing a r e usually of a relatively high flow ra te in o rd e r to d e te rm in e t h e sm okeless, radiation a n d noise perform ance of a tip. To achieve th e s e high flows, JZ uses a large volum e ta n k which is purged th e n pressurized with th e t e s t fuel. depressurizing much th e g r e a te r through a than ta n k Rapidly g e n e r a te s could be flows achieved reasonably sized vaporizer or than could be supplied by th e natural g a s provider. If a mixture of fuels will be te s te d , th e s a m e ta n k is utilized with fuel c o nstituents added sequentially then "stirring" th e ta n k via a c o m p re ss o r.5 Photo 3, Fuel Processing Facility W hen em issions will be te s te d th e composition of th e fuel b e c o m e s critical. Additional s te p s should be tak en to e n s u re no contam ination of th e fuel can occur. For th e TCEQ sp o n so re d testing, all branches to/from t h e fuel h e a d e r which w ere not in use w e re isolated with blinds. Not only d o e s blinding p rev en t contam ination via leaky valves, it also elim inates alternative fuel path s th a t could divert fuel to a location o th e r th a n t h e flare tip being te ste d . While em issions te s ts can be conducted with pure fuels, it is m ore likely an em issions t e s t will utilize a fuel containing tw o o r m o re com p o n e n ts. Emission testing of flares also te n d s to b e a t a lower flow rate th a n o th e r flare te s ts d u e to th e need for t h e t e s t to be conducted over a relatively long period (5-10 m inutes). T he long t e s t tim e is required to allow a re p resentative sam p le of th e plume be e xtracted and analyzed. Given th e low flow rate, long t e s t tim e, and likely fact th e fuel is a mixture, it is b e s t to u se a ded icated flow m e te r for each fuel com p o n e n t. JZ em ploys multi-variable orifice flow m e te rs for this function, devices w h o se operation and accuracy a re well d o c u m e n te d . To e n s u r e th e various fuel 5 For m o re i n f o r m a t i o n on large scale flare tes t facilities see: Baukal, Charles E. In d u stria l C o m b ustion Testing. Boca Ra ton: CRC, 2 0 1 1 . Print. Page 3 co m p o n e n ts a re well blended, a static mixer should b e em ployed in th e fuel h e a d e r d o w n stre a m of all c o m p o n e n t injection points. For t h e TCEQ te s ts , to validate th e composition of th e various fuel supplies, analytical analysis of th e delivered fuels w as required from th e g a s suppliers. Since Tulsa natural g a s (TNG) w as o n e of th e c o m p o n e n ts for th e TCEQ te s ts and th e r e w as concern of possible variation of th e TNG composition from d ay to day, TNG sam p le s w ere draw n daily and analyzed. As a final quality check for any em issions te s t, a sam p le of blended fuel should be analyzed to verify th e actual fuel composition t h a t re a c h e s th e flare. Solely relying on flow m e a s u r e m e n t for determ ination of composition allows to o m any opportunities for error. Fuel Temperature T h e fuel t e m p e ra tu r e should be m ea su re d a s close to th e flare tip a s is possible. T he te m p e r a tu r e of th e fuel will affect th e fuel velocity exiting th e tip a s well a s impacting th e flammability limits of th e fuel. For t h e TCEQ te s ts , t h e fuel te m p e ra tu re w as m e a s u re d ju s t prior to t h e fuel entering th e flare t e s t stand. This location allowed for a relatively high fuel velocity across th e te m p e r a tu re e le m e n t which improves t h e resp o n siveness of th e m e a su re m e n t. Assist Media Flow Rate and Temperature Accurately m easuring t h e flow of a ssist m edia, be it s te a m or air or in s o m e c a s e s natural g as, is crucial in u nderstanding a flare's perform ance. Given th e d e m o n s tra te d ability to adversely affect flare c om bustion by t h e application of excessive a ssist m edia, a ssist m edia flow b e com es nearly a s critical a m e a s u r e m e n t a s th e fuel. For steam -flow m e a s u re m e n t, JZ uses both differential orifice flow m e te r s a s well a s an ultrasonic flow m eter. Both m e th o d s a re p re ssu re an d te m p e r a tu r e c o m p e n sa te d . The orifice m e te rs a re utilized for large ste a m flows, while th e ultrasonic m e te r is used for o u r low flow em issions t e s t rig. Air flow m e a s u r e m e n t for sta n d a rd air-assisted flare stacks is problematic given th e lack of suitable flow conditioning. For th e low flow airassisted t e s t stan d , a long conditioning duct with internal baffling w a s built to a c c u ra te air flow m e a s u re m e n t. e n s u re In o rd e r to a s s u re even air flow up th e sh o rt stack, a special air distributor w as designed in th e Photo 4, Air Assisted Flare w ith Air M easurem ent Duct b a se of th e stack. T h e m e a s u r e m e n t of assist m edia te m p e r a tu r e is im portant in o rd e r to know th e exit velocity of th e m edia. For air-assisted flares th e te m p e ra tu re of th e air delivered to th e flare is simply logged. For s te a m -a ssiste d flares th e s te a m te m p e ra tu re is also im portant for u n derstanding th e quality of s te a m reaching t h e flare tip. For te s ts requiring low s te a m flow, it is possible for th e te m p e r a tu r e of t h e s te a m to d ro p to t h e point t h a t condensation begins to occur. If condensation occurs b e tw e e n th e s te a m flow Page 4 m e te r a n d t h e flare tip, th e n less ste a m is reaching th e flare tip th a n is m e a su re d by th e flow m eter. During t h e TCEQ te s ts , it w a s n ec essary b e tw e e n s o m e te s ts to increase th e s te a m flow th rough a line to increase its piping te m p e ra tu re . T he s te a m flow w as m odulated back to th e low ra te a t th e s ta rt of a te st. T h e residual h e a t of th e piping kept th e s te a m a b o v e 212°F during th e co u rse of th e te st. Flare Outputs For non-em issions flare te s ts , th e typically m e a s u re d o u tp u ts include6: • T herm al Radiation • Noise • Sm okeless Capability • Flame Stability and Flame Quality • Tip Metal T e m p e ra tu re s (W hen testing for flam e burn-back in th e flare tip.) For flare em issions te s ts , th e primary o u tp u t of concern is emissions. Emissions Measurement T h e m e a s u r e m e n t of em issions from a flare plum e is not a simple task. W hen designing a plume sampling sy stem , certain a sp e c ts m u s t be considered: • Wind variations and resulting plume m o v e m e n t • Ensuring th e sampling device d o e s not influence th e flare com bustion process • Height a t which th e plume will be captured • Designing th e sampling sy ste m to w ithstand high te m p e ra tu r e s g e n e ra te d by th e flare • Designing a sampling device a n d /o r m ethodology t h a t results in a repre se n ta tiv e a v e ra g e Destruction Removal Efficiency (DRE) or Combustion Efficiency (CE) determ ination of th e flare. For th e 1982 CMA te s ts , a single point probe w as utilized. For th e 2006 high pre ssu re flare te s ts , JZ developed a conical plume conditioning device to g ra b a larger portion of th e plum e and pre ssu re successful hom ogenize flares, this given th at it. For high te c h n iq u e high proved p re ssu re flare plum es have a high velocity and a r e well b e haved in m o d e ra te winds. TCEQ te s ts , JZ For th e 2010 collaborated with th e University of T exas and A erodyne Research Inc. to further refine t h e conical plume conditioning device. By using an air driven eductor, a large volum e of g a s th rough th e device. Photo 5, Sample Collector Photo from TCEQ 2010 Flare Study Draft Report Presentation is draw n T h e g a s e s entering th e device a re well mixed by ta b s located in th e elbow of th e collector. T herm ocouples 6 Sc hw artz, Ro bert, Charles Baukal, M a h m o u d Fleifil, and Jianhui Hong. "Industrial-Scale Flare Testing." C h em ic a l E ngineering Progress M a y (2 0 0 6 ) : 3 5 - 3 9 . Print. Page 5 located a ro u nd th e inlet co n e provide fe e d b a c k of th e g a s te m p e ra tu re entering th e cone. This te m p e r a tu r e fe e d b ac k assists in properly positioning th e collector. Particularly helpful fee d b ac k for positioning of th e sam p le collector c a m e from tw o FLIR-GF series IR c a m e ra s filtered to d e te c t hydrocarbon em issions. right cluster of four a r e t h e FLIR im ages. In th e photo below, th e two bottom displays on th e T he c a m e ra s w e re positioned approxim ately 90° a p a r t and allowed in sta n ta n e o u s fe ed b a ck a s to th e position of th e inlet cone relative to th e plume. A single view w as n ot sufficient to ascertain if th e cone w as actually in th e plume. T he tw o c a m e ra s to g e th e r provided th e relative position of co n e to plum e in all th r e e spatial dimensions. Photo 6, John Zink Flare Test Facility Control Room Display Not only should th e sam p le collector cone be in th e plume, but it is im portant t h a t all combustion p ro cesses be com plete before th e g a s e s interact with th e sam p le collector. To help e n s u re this goal for t h e TCEQ te s ts , it w as required t h a t th e te m p e ra tu re of t h e g a s e s entering t h e collector not ex ceed 250°F. In keeping with th e idea of eliminating a n y ongoing com bustion p ro cesses in th e sam p le collection sy stem , th e sam p le extraction m e th o d utilized a nitrogen dilution probe w h e re a large volum e of nitrogen (17 tim es th e volum e of sam pled g a s ) w as mixed with th e sa m p le a t th e point of extraction. Mixing such a large volum e of nitrogen with th e s a m p le a rre ste d any ongoing com bustion processes. T h e analytical e q u ip m e n t utilized for g a s analysis should be well m aintained with calibrations occurring a t re c o m m e n d ed intervals to e n s u r e t h e highest accuracy possible. For th e TCEQ te s ts , th e re s p o n s e tim e of th e analytical e q u ip m e n t sampling th e plum e w as very quick, on t h e o rd e r of 1Hz plus travel time. Due to t h e fa st re sp o n se tim e of th e analytical e quipm ent, n e a r real-time fe e d b a c k of th e various m e a su re d species w as possible. T he analysis m eth o d em ployed to d e te rm in e DRE a n d CE eliminated th e concern of plum e dilution with am b ie n t air.7 Alternative m e th o d s discussed for this study included t h e 7 TCEQ 2 0 1 0 Flare Stu dy Final Re port, Sec 7 Accuracy and Precision o f Plu m e Sam pling System M e a s u r e m e n t s , p 100-113 Page 6 u se of a tra c e r com pound in th e fuel and a total plum e c a p tu re hood.8 For particulate analysis, a sam p le line s e p a r a t e from th e g a s sam p le line w as used. T he particulate sam p le line w as m a d e of a conductive material t h a t w as ch arg ed to pre v e n t th e particulate m a tte r from sticking to t h e walls of th e sam p le line. During a n y em issions testing, so m e plum e profiling is useful in determ ining t h e sensitivity of th e results to t h e positioning of th e sampling system . Profiling of th e plume w as d o n e during th e TCEQ te sts. Horizontal profiling occurred naturally d u e to wind variations. In s o m e te s ts , th e sam p le collector w as raised an d lowered to d e term in e if emission variations could be d e te c te d b e tw e e n th e upp e r and lower portions of t h e plume. Such vertical profiling did not d e te c t significant variation in em issions w hen th e DRE w as 8 5 % or g r e a te r.9 This lack of emission variation helps verify t h a t th e sam p le collector perform ed a s intended by drawing in a large volum e of g a s and homogenizing it. Quality Assurance Prior to any testing, all instrum entation should be verified a c c u ra te within its operating sp a n . tr u e not only for field instrum ents such transm itte rs, te m p e ra tu re e le m e n ts, etc., analytical e quipm ent. This is as but also During such calibration checks, t h e d a ta logging system should also be checked to verify recorded variables a re labeled. properly scaled and Prior to th e s ta r t of th e TCEQ te s ts , all JZ instrum entation w as verified a c c u ra te with calibration e q u ip m e n t t h a t had recently been factory certified. W hen performing te s ts , it is im portant to replicate points for d a ta quality validation. Replicated d a ta helps identify th e random error associated with th e experim ent. In o rd e r to minimize th e potential for bias appearing in th e replicated d a ta , it is b e st if so m e tim e lag occurs be tw e e n replicated points. sc hedule allows, replicated test If th e t e s t points should be e x ecuted on different days to allow variations in wind, a m b ie n t te m p e ra tu r e , and possibly o th e r variables to Photo 7, Test of Steam Assisted Flare in Progress During TCEQ 2010 Flare Study occur. While it m ay be intuitive th a t all d a ta acquisition s y ste m s should be synchronized, unless a conscience effort is m a d e , this is unlikely to occur. For m ost te s ts th e r e will be m o re th a n o n e acquisition system , an acquisition system for tra n sm itte r d a ta and a s e p a r a te system for video cap tu re . For th e TCEQ te s ts , th e r e w e re no less th a n 10 acquisition sy ste m s w hen all th e video c a m e ra s and re m o te monitoring co m p an ies a re included. A m a s te r clock w as utilized and all acquisition s y ste m s w e re synchronized to it every m orning before testing began. 8 T h e to ta l p lu m e c a p tu r e idea w a s d e e m e d impra ctical d u e t o its likely e ff e c t on t h e c o m b u s tio n process. 9 TCEQ 2 0 1 0 Flare Stu dy Final Re port, Sec 7.2.c.i H e igh t o f sa m p le pro be , p. 109 Page 7 Concluding Remarks Flare testing, like any o th e r experim ent, requires proper planning, quality control, and c o m p e te n t execution in o rd er to obtain useful d a ta . T he challenges of full-scale flare testing a re not insurm ountable. John Zink's investm ent in its large-scale flare t e s t facility, which allows fuel flows from 100 to 150,000 lbs/hr an d associated a ssist m edia, regularly sh e d s new light on flare com bustion processes. The challenge of em issions testing has also been refined over th e yea rs with th e 2010 TCEQ te s ts being th e late st d e v e lo p m en t in m e a s u r e m e n t techniques. u n d erstan d ing of flare com bustion. Ultimately, th e goal of any flare testing is b e tte r Such understanding leads to improved products with validated perform ance. To this end t h e John Zink Com pany is fully com m itted. Page 8