THE DANIEL SENIORSONIC MULTIPATH GAS FLOWMETER THE DANIEL SENIORSONIC METER AND FLOWCONDITIONERS

Danel SenorSonc The Use of Flow Condtoners Page 1 THE DANIEL SENIORSONIC MULTIPATH GAS FLOWMETER THE DANIEL SENIORSONIC METER AND FLOWCONDITIONERS

Danel SenorSonc The Use of Flow Condtoners Page 2 1. Introducton It s becomng ncreasngly common to use flow condtoners wth ultrasonc meters for a majorty of ther applcatons. However, the overall effect on the meter s not been clearly known. Certan brands of flow condtoners may mprove the SenorSonc meter s nstalled performance. However, often they are not needed, and some can actually cause the overall performance to be less than f no condtoner were used. Most users have assumed that usng a flow condtoner mproves meter performance. Ths document dscusses the advantages and dsadvantages of usng flow condtoners wth the Danel SenorSonc meter, and what effect they may have on the calbraton results. The use of flow condtoners n natural gas meterng has been accepted for many years. Ther prmary use n the past has been wth orfce and turbne meters as both are known to be senstve to swrl. The most common flow condtoner n servce s the 19-tube bundle. Ths desgn conssts of 19 small tubes, usually the same dameter, welded together and arranged n a pattern that fts nsde the ppe. The devce s located upstream of the meterng element at a dstance varyng from 5 to 11 dameters, dependng upon the prmary element and user preference. A sgnfcant amount of work has been done n recent years to mprove on the tradtonal 19-tube bundle desgn. These new hgh performance condtoners take the tradtonal flow condtoner to a new level. They attempt to solate the meter from a second upstream ppng effect typcally known as non-symmetrcal velocty profle (also knows as an asymmetrcal profle). Brands of hgh performance condtoners nclude Gallagher, Nova, Vortab and Zanker. Research n recent years has shown the effects of a non-symmetrcal velocty profle on orfce meters to be sgnfcant. A non-symmetrcal profle s one where unequal veloctes exst n the ppe at the same dstance from the ppe wall. These nonsymmetrcal velocty profles are generated by varous ppng confguratons ncludng headers, elbows, tees, flters, valves and other devces. The dfgure shown above shows two dfferent symmetrcal velocty profles n a ppelne. A flow profle wth non-symmetry would have a non-unform shape.

Danel SenorSonc The Use of Flow Condtoners Page 3 The Danel SenorSonc ultrasonc flowmeter was desgned to be nstalled wthout flow condtoners. A sgnfcant amount of research durng the past 15 years has proven the desgn works well wth no flow condtoner. However, many users stll prefer to use flow condtoners as nsurance that the flow profle (hence the meter s accuracy), at the tme of calbraton, wll reman the same once the meter s nstalled. The belef that the flow condtoner totally solates the upstream effects from the meter s not always true. Durng the past few years GRI has funded extensve testng on ultrasonc meters at the Southwest Research Insttute n San Antono. The prmary purpose s to determne how well an ultrasonc meter performs n typcal ppng confguratons, both wth and wthout the varous flow condtoners currently avalable. A.G.A. 9 requres a manufacturer to mantan meterng accuracy wthn ±0.3% from calbraton to nstallaton wth typcal ppng desgns. As orfce meters typcally can be effected by much more than 0.3%, t was assumed by the customer that an ultrasonc meter would also be mpacted sgnfcantly. Followng s a bref summary of test data recently released to the GRI commttee members. It shows the effect on accuracy wth no flow condtoner, wth varous brands of flow condtoners,and wth dfferent upstream ppng confguratons. The numbers reflect the weghted error dfference from the test condton relatve to the orgnal calbraton (typcally wth 100D of straght ppe and the same flow condtoner nstalled). In other words, each result from the varous upstream tests s compared to a baselne test wth the same flow condtoner nstalled n the same locaton (relatve to the meter), but wth 100D upstream n leu of the nstallaton effect. 12 ANSI 600 SENIORSONIC Installaton Effect Sngle Elbow Elbows Out Elbows In Meter Orentaton 0deg 90deg 0deg 90deg 0deg 90deg Bare at 10D -0.07-0.02 0.53-0.04 0.02 0.24 Bare at 20D -0.13-0.11 0.05 0.10 0.11 0.12 19 Tube -0.35-0.37 0.13 0.37-0.15-0.22 Vortab 0.10 0.03-0.02 0.02 0.07 0.14 CPA 50E -0.15-0.13-0.23-0.30 0.03 0.04 GFC 0.04 0.00-0.17 0.23 0.36 0.41 PASSES AGA 9 REQUIREMENT FAILS AGA9 REQUIREMENT Numbers that are n the green background are results that meet or exceed A.G.A. 9 accuracy requrements (less than 0.3% nstallaton effect). Two meter orentatons were tested (meter mounted normal, then rotated 90 degrees) to smulate the dfferent nstallatons that mght be encountered n the feld. Important to notce s the result of No Condtoner, 20D. The results clearly show the SenorSonc meter easly meets the nstallaton effects requrements of A.G.A. 9 n all tests. In fact, the Danel SenorSonc was the only meter tested that meets the A.G.A. 9 nstallaton accuracy requrement wthout flow condtonng.

Danel SenorSonc The Use of Flow Condtoners Page 4 The reference paper by GRI s attached to ths document for nformaton. Addtonally, only one flow condtoner even comes close to the same results. Ths summary s the frst real commercal result showng the SenorSonc meets or exceeds A.G.A. 9 nstalled performance requrements wth no flow condtoner. Interestngly, no other USM manufacturer s devce passed ths AGA9 requrement. Flow condtoners often have another effect on an ultrasonc meter. In the process of changng the velocty profle, they often create a flow profle that can bas the meter and cause t to regster dfferent. Ths dfference can be seen when comparng calbraton results of a meter wth and wthout the flow condtoner. In other words, f a meter needed to be adjusted by 0.2% relatve to the calbraton faclty, the amount of adjustment may be on the order of 0.5% wth the flow condtoner nstalled. The ntent of a flow condtoner s to return a swrlng, non-symmetrcal flow profle to a condton many consder equal to 100D of straght ppe (symmetrcal wth no swrl). Ths equvalent length of straght ppe s thought to (and data has been obtaned to verfy t) elmnate any upstream flow dsturbances and create a very consstent, reproducble profle. However, often these hgh performance flow condtoners generate a profle that s somewhat dfferent than the 100D baselne profle. The prmary purpose of a meter calbraton s to determne the amount of error from the reference (laboratory). Ths dfference can also be thought of as the meter s accuracy. However, other ssues should be consdered when revewng calbraton results. Besdes absolute accuracy, lnearty and repeatablty are mportant to the customer. Thus, not only s the amount of adjustment mportant, but the queston s often asked What mpact does the flow condtoner have on the meter s lnearty? Only a lmted amount of nformaton s avalable to answer the queston regardng lnearty and repeatablty wth all types of nstallaton condtons. However, the results from several calbratons at a varety of laboratores, all usng a sgnfcant numbers of straght ppe upstream, suggest there s no detrmental effect on the SenorSonc s lnearty and repeatablty. The only documented effect s to the meter s adjust factor. That s, f a flow condtoner s used, regardless of brand, the Meter Factor wll be dfferent than f t were calbrated wthout the flow condtoner, but the lnearty and repeatablty wll reman vrtually unchanged. At frst glance ths shft n Meter Factor doesn t appear to be a problem. Once the meter s calbrated, overall accuracy remans the same. However, some users nterpret a meter that s outsde the basc as found A.G.A 9 accuracy requrement of ±0.7% for 12-nch and larger, and ±1.0% for 10-nch and smaller, to not be n complance. Ths poses a problem for the manufacturer snce some flow condtoners can mpact some meter desgns by a sgnfcant amount. The desgn used n the SenorSonc has been shown to be somewhat nsenstve to most flow condtoner desgns. However, there s an effect wth all desgns of flow condtoners. Typcally a bas to the Meter Factor can be approxmately 0.2-0.3%, but ths depends upon meter sze and brand of flow condtoner. That s, the meter error wll be hgher (Meter Factor lower) than f t were calbrated wthout the condtoner.

Danel SenorSonc The Use of Flow Condtoners Page 5 2. Ax Symmetrc Changes To Flow Profle The Danel SenorSonc meter can tolerate ax-symmetrc changes n profle wth no mpact on meter accuracy. AGauss-Jacob numercal ntegraton gves the average flow velocty V avg over the area of the ppe from the weghted sum of the lne averages along the chords V where the weghtng factors, W, are determned for the specfc chord locatons. Note that t s sgnfcant that ntegraton of a unform velocty profle, usng ths method would gve the correct answer. Another nterpretaton of the weghtng factors W s the proporton of ppe area assocated wth each chord to obtan the flow, the sum of whch must be unty. For an ndvdual chord, 2 L t V ( r ) =. 2x t du ud t. t du ud The average gas velocty s calculated usng equ(1) V = 4 = 1 V ( r ) W equ(2) where the wghtng factors are Chord A = 0.1382 Chord B = 0.3618 Chord C = 0.3618 Chord D = 0.1382 Note that the sum of the factors equals unty. Fgure 1 shows the locatons for the four chords.

Danel SenorSonc The Use of Flow Condtoners Page 6 Fgure 1 Chord Locatons For SenorSonc Usng a power law, shown n equaton 3, dfferent flow profles can be generated as shown n Fgure 2. v V max y = R 1 n equ(3) These profles are calculated usng n = 7 and n = 12. Both of the profles represents an average velocty of unty. Usng the SenorSonc algorthm,.e. takng the pont velocty at each chord locaton and ntegratng usng the weghtng factors gven above, t s proven that the meter s completely mmune to ax-symmetrc changes n profle. Ths s demonstrated n the tabulated calculaton n Fgure 3.

Danel SenorSonc The Use of Flow Condtoners Page 7 Fgure 3 SenorSonc Immunty to Changes n Ax-Symmetrc Profle Another manufacturer uses a fve path confguraton where three of the velocty measurements are made through the centre secton of the flow usng a sngle bounce traverse. Ths type of measurement requres a correcton based upon Reynolds number. However, Reynolds number correctons rely on a fully developed flow profle and accurate knowledge of the ppe wall roughness. Wth these chord confguratons, f the flow s not full developed,.e. a flow condtoner s ntroduced, then the Reynolds number correcton wll be erroneous. In addton, f the ppewall roughness changes, then there s no way of knowng ths and agan the Reynolds number correcton wll be errouneous. Ths theoretcal approach hghlghts the fundamental advantage the Danel SenorSonc four path meter has over other meters such as a fve path cnetre lne measurng devce. The Danel meter samples the flow profle and does not make any assumptons about the nature of the profle. 4. Swrlng Flow Profle The followng s an explanaton on how the Danel SenorSonc copes wth swrlng gas flow. To ad the reader, a quck remnder on the chord arrangement and ntegraton technques for the SenorSonc s provded.

Danel SenorSonc The Use of Flow Condtoners Page 8 The meter s symmetrcal about the centre lne wth chords A and D postoned 0.809R from the centre lne and chords B and C beng postoned 0.309R from the centrelne. These locatons are fxed such that each meter sze s geometrcally smlar. For these locatons the chord weghtng factors are; W A = 0.1382 W B = 0.3618 W C = 0.3618 W D = 0.1382 Chords A and B are crss-crossed as are C and D The average gas velocty s calculated by ntegratng the product of measured velocty and weghtng factor for each chord..e. V = 4 = 1 V ( r ) W BULK SWIRL If there s a bulk swrl wth angular velocty, ω, around the ppe axs then at a radus, r, ths wll gve a velocty of ωr perpendcular to r. The component along the chord wll be ωr cosθ cos45 as shown n the fgure below,.e. the swrl component along

Danel SenorSonc The Use of Flow Condtoners Page 9 the chord s constant at ωz cos45. So on chords A and B we have swrl components S A and S B gven by S A S B = (0.707 ω).0.809r = (0.707 ω).0.309r Applyng the weghtng factors for the chords as per the ntegraton, we fnd that V SA V SB = 0.1382. (0.707 ω).0.809r = 0.1118. (0.707 ωr) = 0.3618 (0.707 ω).0.309r = 0.1118. (0.707 ωr) Thus the contrbuton of swrl on these two chords s equal, but s opposte drectons snce the chords are crossed. Ths means that ther contrbutons exactly cancel each other. Smlarly, the swrl components on the C and D chord exactly cancel each other. Ths feature s unque to the SenorSonc desgn where weghtng factors are nversely proportonal to the chord radal locaton from the centre.

Danel SenorSonc The Use of Flow Condtoners Page 10 Concluson The SenorSonc meter was desgned to be nstalled wth no flow condtoner. Some of today s hgh performance flow condtoners may mprove overall nstalled accuracy. However, not all flow condtoners mprove the SenorSonc s performance. Usng a flow condtoner, regardless of the brand, wll mpact the nstrument s Meter Factor. The bas typcally causes the meter to over-regster by 0.2-0.3%, but s somewhat dependent upon meter sze and brand of flow condtoner. The mpact of ths bas may cause the amount of adjustment to exceed A.G.A. 9 requrements. It s Danel s opnon that any mpact that causes the meter s adjustment (Meter Factor) to be outsde of A.G.A. Report No. 9 does not cause our meter to be n non-complance.