You are on page 1of 6
FLOMEKO'98 Effects of Swirl and Velocity Profile Asymmetry on Flow Conditioner Performance for Orifice Meters TB. Morrow* and K. A. Behring 1I* *Soothwest Reseach Insitute, Division of Mechanical & Flids Engineering, San Antoni, 1exas, USA Abwiract 4. performance tatt wae developed to asstss the bility of flow conditioners to provide proper flow conditions upstream of an orifice plate that minimize orifice meter ‘alltion error. Two different types of flow meter Intaleion are roncidereds am incallaion downstream of ‘tro 90" elbows in perpendicular planes that produces swirl in the meter tube, and an installation downstream of a gate valve closed 50% against the flow, producing an asymmetric vial velocity profile in the meter tube. Unstream meter tube lengths of 45 and 17 pipe diameters are considered. ‘Results are presented fora bare meter tube, two tube bundle straightening vane designs, and two proprietary flow conditioner designs for a single value of orifice diameter ratio, B=047. Flow conditioner performance is expressed 85 AGy, the percentage change in discharge coefficient froma reference valve obtained with 100 diameters of straight pipe upstream of the orifice plate. Inder terms—Flow conditioner, orifice meter, installation fects, performance tes, swir, axial profile asymmetry. 1. PROBLEM DESCRIPTION Orifice meters are fabricated and used in conformance with industry standards such as the American Gas Association (AGA) Report No. 3 [1,2], ot the ‘International Standards Organization (150) 5167 [3]. The standards specify an equation for calculating orifice discharge coefficient, Cs, and specifications for the ‘minimum unsteam and downstream meter tube lengihs for ero additional uncertainty due to installation eror. It is well known that A.G.A. Report No. 3 and ISO 5167 disagree significantly on the minimum requirements for straight upstream length for several installations typical of field configurations. The reference Cy data for the orifice coefficient ‘equation was acquired with long upstream lengths (70D to 10OD) of suight pipe, ox with 44D of stilt pipe and « flow conditioner upstream of the orifice plate. However, orifice meters in th field are often installed downstream of pipestings (tees, elbows, valves, reducers or ‘expanders) or connected toa header that supplies paralle) meter runs. Research has shown that tee, elbows and valves can produce an asymmetric axial velocity profile in the meter tuhe. while elbows in perpendicular planes and headers can produce @ significant swirl velocity component. When a tube bundle straightening vane is used, AG.A. Rey! Nu. 3 petits shinier siniuiwn upstream meter tube lengths (17D instead of 45D at B = 0.75 for an installation downstream of a partly closed valve), while ISO 5167 recommends a minimum upstream meter tbe length of 45D. Research has shown that while tifiee tte 429 straightening vanes remove swirl from the upstream flow, they may not provide the proper shape 10 the axial velocity profile to minimize installation error. New flow conditioner designs [4-8] promise 10 remove swirl from the meter tube while shaping the Telocity profile to simulate « fully developed, turbulent pipe flow. A flow conditioner performance test was developed {9} to determine the amount of installation fertor associated with good and worst case (switl and/or axial profile asymmetry) installauons. UL BwEsTigaTions A serie of 102mm (tinch) diameter orifice meter installation effects tests was performed flowing natural gas in the Gas Research Institute (GRD) Metering Research Facility (MRF) High Pressure Loop (HPL), A. Test conditions ‘The flow conditioner performance tests were carried out ‘ucing an orifice meter the with an orifice fitine and upstream meter tube Jengths' of 17, 29 or 45 pipe diameters (D). Orifice differential pressure measurements ‘were made using two pairs of flange taps located 180° apart onthe orifice fiing. Orifice meter calibrations were performed at a nominal pressure of 1.38 MPa (200 psia) for four values of B ratio: 0.40, 0.60, 0.67 and 0.75. Cy values were measured fora single value of Reynolds number foreach B value. Values of Reynolds number were chosen to be approximately equivalent tan orifice pressure Sifferential of 49.8 KPa (200 inches of water or 7.22 ) Baseline Cy measurements were made by placing a ‘SSD long straight spool piece upstream of the inlet tothe 45D long meter tube. The baseline Cy values fell within the 95% confidence intervals for the Reader HarrisGallagher orifice equation specified in AGA. ‘Report No. 3, Part (1. B. Flow conditioners ‘The research objective was to develop a flow conditioner erfoimauce leat. This seguired selecting a set of existing flow conditioner designs and subjecting them to the test procedure, The 19-tube bundle staghtening vane Te nomenclature in AGA, Report No.3 2] s sedi his paper. A fs he opaeam cer take length for bare meter te. Ais the ‘steam meter tbe length wena flow conditioners sled in he ies mie iste datince ftom the downeteam end of the fo ‘Saaidone tote one pate. FLOMEKO’98 escribed in A.G.A. Report No. 3, Part 2 (2) was one obvious choice. Three new flow conditioner designs were Slco selected These art Stuart C-3 tube bundle Straightening vane [4], a NOVA #50E perforated plate flow conditioner [5], and a Gallagher flow conditioner (plate #21 and a 14D segment of 2 19-tube bundle Straightening vane) (6), Unfortunately, budget and time limits prevented us from including more flow conditioners in the test patern, The three new design conditioners that were chosen were prototype, pre-commercial models that had. demonstrated good performance in previous GRT MRF flow conditioner tests in meter tubes with a short upstream length, A'= 17D. A test witha bare meter tube (ith no flow conditioner) was includ in the test pattern to show the level of performance improvement that could be obtained by installing a flow conditioner. The sliding flow conditioner technique was used t0 vary ine loeatlon, ©, of te Now comlicvacy i he wien tube. This involved attaching a small diameter rod to the back of each flow conditioner, and passing the rod though a seal in an upstream elbow or tee. The rod was calibrated to allow placing the flow conditioner at specific lcations upstream of the orifice plate. C. Inatalotion tort nrvangamente Four pes of orifice meter field installation configurations were simulated: 1. Good flow conditions [10]. This test will show ‘whether a flow conditioner installed in a meter tube in which the axial velocity profile is close to fully developed, and the swirl angle is low, introduces @ remirhation that causes a significant deviation from the baseline C, value, 2. Two 90° elbows in perpendicular planes installed irectly upstream of the meter tube. This configuration produces # swirl velocity eomponont in the meter tbe [10-12], Installations downstream of two 90° elbows out-of-plane separated by less than 10 D of straight pipe are specifically addressed in AGA. Report No. 3, Part 2 [2]. 3, A panially closed valve installed upstream of the meter tube. Research has shown that a gate that partially obstructs the flow passage can produce @ swongly asymmetric axial velocity profile in the meter tube [8]. Installations downstream of a partially closed valve are also addressed in A.G.A. Report No. 3, Part? [2 4. Highly disturbed flow conditions representative of @ meter tube installed downstream of a header. Research on header installation effects has shown that swirl angles of up to 230" can occur, ang tat @ header may also produce an asymmetric axial velocity profile (7,13]. Rather than construct and testa specific header configuration. it was decided to simulate the effect of a header by using a swirl ‘eenerator to produce high swirl angles, and to turn the swirling flow through a tee into the inlet of the ster tube (9) TL RESULTS OBTAINED ‘This paper presents the flow conditioner and bare meter tube test results for two performance tests, (1) two 90° 420 clbows in perpendicular planes, and (2) a gate valve closed 50% against the flow. Sliding flow conditioner results are shown fora single value of B = 0.67, and for Upstream meter tube lengths of A'= 17D and 43D. The Reynolds number was approximately 2.7*10° for these tests, Bare meter tube results are shown for A = 17D and 45D, ain for B - 040, 0.60. 0.67 and 0.75. The Reynolds ‘numaber varied from 910° for 8 = 0.40 to 3.6*10° for B = (0.75. The GRIMRF test report [9] contains the results of the other performance tests forall combinations of B and ‘upstream meter tube length, A. Installation error for zero additional uncertainty “To determine te installation error caused by a particular installation with a specific flow conditioner, it is necessary to have a baseline or reference Cy value for the same B ratio and Reynolds number. The percentage shift jn Cs from the baseline value measured in the reference. testis caloulated as AC, =(C, a) Basline Cy values may be caleulated fom the Reader-Hari/Gallager (RG) equation for orifice cicient in A.G.A. Report No.3. Part 111. However. fan. experimental investigation of installation er referenced to the RG equation would be influenced by laboratory bias eras inthe calibrations of reference flow eters and transducors that would affect only the measured C, value. A second flow laboratory, with different experimental biases, might be unable to replicate the experimental result ‘The procedure used i ths study requires that both the installation tests and the baseline calibration tests be performed inthe same laboratory using the same orifice plates, over the same flow range, with the same reference flow meters. Then bias eror caused by a reference flow reer affects both the installation test and the baseline Calibration test by the same amount. The effects ofthese auclated Vasc” aro minimized when the inctllaion error is presented as ACy ‘The obvious, desired value of AC, is 0% for zero additional uncertainty. However, experience in performing baseline caibrauon tests the URI MIRE nas Shown that baseline Cy values can be repeated only to within about £0.19. An orifice meter installation test will give values of Cy that are repeatable to within about 40.1%. Since a baseline C, value could be atthe lower tend ofits range of variation, and an installation test could be atthe upper range of its variation, its sensible to set the upper and lower ius for AC, wo Ue wpyosinatcly 20.2%. Later, when reviewing the test resus, the ACe limits were increased to half of the 95% confidence interval 2a) in the RG equation at infinite Reynolds umber, specified in AGA. Report No. 3, Par (1 ‘This slighty larger acceptance range reflects the greater variability in Cz values found at high and low values of B [Ayple! value for Cy i 40.29% fr P = 0.67 Cosa)! Cates B, General Results ‘The flow conditioner performance test results for any of the four conditioners tested showed that orifice meter installation error, AC,, was a fonction of Braised to an exponent of approximately 3.5 (that is AC ~ B°*) as suggested by Reader-Haris ct al. [7] The sliding ow Conditioner results for b= 0.40 show litle ettect ot the ‘upstzeam disturbance. ‘The magnitude of the installation error reaches a maximum for B = 0.75. The magnitude of the installation eror alsa varins with the flow conditioner type, the upseam disturbance, the upstream meter tube length, A’, and the flow conditioner location in the meter tube, C. For some combinations of flow conditioner, and upsteam alsturpance, ine miaynlade of di lust terror also depends upon which pair of pressure taps is used to measure differential pressure. For bare meter tubes, when swirl was present upstream of the orifice plate, the installation error did not follow the AC, ~ B°* model. The installation tests with two 90° out-of-plane elbows, and high swil with a tee produced the ype 1" eis” described by Mattingly and ‘Yeh {11}. In the bare meter tube tests with type 1” swirl, AC; values were either within or close to the acceptable range for B = 075, but outside the acceptable range for B ‘040, 0.60 and 0.67 for all three values of upstresm meter tube length © Twn 90° elbows in perpendicular planes Figure 1 shows the test arrangement for a flow conditioner performance test in a meter tube installed downstream of two 90° elbows in perpendicular planes. ‘Since the 90" elbows are flange connected, there is swaight spacer length of 1.6D between the first and second elbow. The piping upstream of the first elbow consisted of @ 35D straight spool piece and a flow ‘conditioner to minimize swirl and asymanety. Figure 2 shows the installation error for bare meter tube lengths of A= 17D and 45D and B values of 0.40, 0.60, 0.67 and 0.75, For f ~ 0.60, 0.67 an 0°75, thorn ‘was lite difference between the results for the long and short meter tube lengths. For B = 0.40, the results appear to be influenced somewhat more by “type 1” swirl decay, ‘whieh is Reynolds number dependent. Values ot AUy are smaller in absolute magnitude for A= 45D than for A 17D, ACz values are outside the acceptance interval for B = 0.4n, 060 and 087A “erossower point” where ACs (0% appears at approximately B = 0.72. Figure 3 shows the effect of placing 2 19 tube bundle straightening vane in the meter tube for = 0.67. For A 19D, AU, values are acceptable for values of 9D SC S 15D, and for 32D $C 41D. For A’= 17D, ACgis inthe acceptable range only for C = 13D. This result shows that while a suaishtening vane can remove switl, it may not beable to properly shape the axial velocity profile Figure 4 shows the effect of placing a Stuart C-3 tube bundle straightening vane in the meter tube for B = 0.67 For A'= 45D, AC; values ave avceptable for values of 7D SCS 11D, and for 32D $C < 41D. For A’= 17D, ACs is in the acceptable range for 3D 2. Figure 6 shows the effect of placing a Gallagher flow conditioner inthe meter tube for § = 0,67. For A’ = 17D and 45D, AC, values ae acceptable for values of C 2 SD. D. 50% Closed Valve Figure 7 shows the test arrangement for a flow conditioner performance test in a meter tube installed downstream of a gate valve closed 50% against the flow. ‘A 55D long spool piece and a tee was located on the upsteam side of the gate valve. The flow conditioners were traversed along the meter tube length between the Yalve and the orifice plats Figure 8 shows the installation error for bare meter tube lengths of A = 17D and 45D and 6 values of 0.40, 0.60, 0.67 and 0.75. For the long meter tube length, A 45D, the disturbance produced by the valve is dissipated, ‘andthe installation erro is negligible. For the short meter tube, A= 17D, the installation error is significant ang lies ‘outside the acceptable range for B > 0.40, is interesting to compare Figure 2 and Figure 6 ‘The effect of “type 1” swisl produced by two 90" elbows in perpendicular planes persists through upstream lengths, jreatct than 45D. However, the major Gow disturbance produced by a gate valve closed 50% is dissipated, and a eveloped pipe flow is produced within an upstream length of 45D. igure 9 shows the erect of placing 2 19 ube bundle straightening vane in the meter tube for 6 = 0.67 Surprisingly, forthe long meter tube, A'= 45D, the effect, ff tube bundle location is sienificant for 1D < C $ 41D. ‘The optimum locations forthe 19 tube bundle are C =13D and C= 38D. At other locations, the installation error is ‘obviously not reduced in comparison to the bare meter table tesuh fon) 0.67. In the short meter tubes A! = 17D, an optimum value of C does not exist where ACs is within the acceptable limits for both pressure taps. Figure 10 shows the effect of placing a Stuart C-3 tube bundle straightening vane in the meter tube for p= U.0 For A'= 45D, AC, values are acceptable for values of C 26D, For A'= 17D, AC, s in the acceptable range for 3DsC2sD. Figure 11 shows the effect of placing a NOVA #50E ‘low conditioner in the meter tube for b= 0.67. For A’ 45D, AC, values are acceptable for values of C > 7D. For ‘A= 17D, AC, values are acceptable tor C 2 JD, except for C = 10D and 12D, where the AC, value for one pressure tap is 0.014% outside the acceptable range. FFigare 12 shows the effec of placing 2 Gallagher flow ‘conditioner in the meter tube for = 0.67. For A’= 45D, ‘AC values are acceptable for values of C > 3D. For A 17D, AC, values are acceptable for 4D < Cs 9D, TV. ConcLusions Al four flow conditioners removed “type 1” switl from the upstreaus flow, The test report [9} shows that Ce

You might also like