53

2. IS ID

2nd International S mposiurn on Ultrasonic Doppler Methods

for Fluid Nlechanics and Fluid Err ineerin•_ September 20-22, 1999

Paul Scherrer Insitut, 5252 Villipen PSI. S itzerland

Development of a new flow metering system using UVP

- (2) Comparison with weight measurement at NIST -

Y. Takeda', N. Furuichi', M. Mori'

M. Aritomi’’ and H. Kikura’

1. Paul ScherrerInstitute, CH 5232 Villip•en, Switzerland
2. Tokyo Electric Power Company. Tsurumi ku. Yokohama 223. Japan
3. Tokyo Institute ot Twhnology, 2-12- I OhoMyarna, Me uro-ku. Tokyo 152, Japan

I.INTRODUCTION

A new flow metering system ultrasonic Doppler method (UVP) has been develOped by Takeda et at "1' '. In this system, a flow rate is obtained by an integration of instantaneous velocity profile measured by C VP or er a pipe diameter. ThiS ,system has a many advantages. One is that pressure !Oss IS ROr Caused becuuse the transducer can be set outside pf the wall and other one is that can be applied to opaque liquid. arid so on. Especially, it is not necessary the process of interpolation or averaging which using other ultrusonic flow meter because a flow rate be estimated directly by using instantaneous velocity profi ie.

Kikura et at. "’ clarified a characteristic of ultrasonic propagation through the metallic wall and indicated that a flow rate can be measured from the outside of the stainless steel when the basic frequency of the ultrasonic pulse is carefully selected. Mori et al. " reported the result of a flow rate measurement in the stainless steel pipe (250A and 400A) which is more realistic configuration. The error rate that obtained by compare a flow rate measured by this method with that by other flow meter (orifice flow meter and electnc flow mater) was )ess than 0.*7r at steady flow condition. Taishi et at. ‘ was indicated that this method has a good sensitivity for a transient flow rate. Thus. it is suggested that this method can be applied to how metering system with a high accuracy.

For the realization this method, it is necessary the more information of the accuracy. Especially, it is necessary that the accuracy that compared ab.solutely flow rate is obtGned. At the .LIST (National Institute of Standard and Technology) in USA, there is the system that can be measured weight flow rate. In this paper we report the result that compared absolutel y flow rate measured by this new method u'ith that measured by weight flow rate by the NUT standard calibradon system.

1. EXPERIMENTAL APPARATUSThe u ater facility of the NIST standard calibration system that consi sts of a reservoir, pump, meter runs and neight tank is shown in Fig. 1. The system is usually operated as a constant tlo« tacility over the test section. The junction that switches the flow channel to the weight tank or the reser volr is set at the downstream a( the test section. An operation that the junction is switched does not affect to the fJof COFldiiion. F)ows up to 3Srn'/min ( I0.000gal/min) can be provided und maximum Reynolds number is about 4S) w hen the pipe 254mm in diameter is used. The weight tank capacity is about 20m' and w'eight of water inside it can be measured. The method of flow rate measurement is below. Water thorough the test secoon is stored in this tank in some pcriod54Test Secti00ResepoirFip•, 1. Schematic of the experimental apparatus - NIST calibration standard system-Table 1. CFP pnanieters

   Reynolds Number	400K	2.6M
   Transducer	4MHz	lM HZ
   Startin* Depth	10 l mm	340nini
   Channel Distance	2.22mm	1.48mm
   Maxirnurn Depth	326mm	340mm
   RF gain	4,4	1, l

   Fig. 2. Test section detail and coordinate systemand a weight is measured. The weight flow rate is obtained as a result that the volume of water stored in the tank divides by that time. The relative expanded uncertainly for these facilities is 0.12a . The test section has 10.l5m (400in.) length and the pipe diameter is d—253.75mrn (10in.). The measurinp• rep•ion was set a downstream Ud=33 from the nozzle exit.The test section with transducer setting is shown in Fig.2. Two type transducers which frequency of ultrasonic are 1MHz and 4MHz were used. The 1MHz transducer was put outside of the stainless wall and the 4MHz one was put on spatial mount made by Plexiglas which thickness is 2mm. inclination angle of lMHz is 5 degree and 4MHz 0-20 degree with flexibility. A particle was not used as a retâector in this experiment because there is enou•ph small cavltation bubble that generated around the pump in water.
2. EXPERIMENTAL METHODExperimental procedure was according to the one of the NIST. A simultaneous measurement of flow rate by Lfi"P and NIST system was examined five times in one running. The average flow’ rate was determined zs the one that is averaged in each examination. A sampling offlow’ rate by UVP was started at the time when the junction is switched to the tank and continued over storin • to the tank. The method of estimating hOw rate is r same one as Takeda et al"' has reported. Reynolds < numbers are 400K and 2.6M and the transducer of the UVP was used 4MHz and 1MHz, respectively. TypicalRFP parameterx are shown in Table 1. A mean velocityprofile of U component was measured to obtain aninformation of flow condition. This component ofvelmiiy can not be measured directly by UVP so that itwas mCasu red according to the method of flow tTlappipg. {g. 3. Mean x'eloctty profile (be —400K)
3. RESULTS AND DISCUSSIOP
   1. S Jean ›elocity profileThe mean velwity pfofi le in the pipe is sho vr in Fig.3. The frequency of ultffl.son ic is 4MHz and the inclination angle is 12 deÿree. Reynolds number is 400K As shown in this figurc, the velocity measured by CAP is disturbed by the reflcction from the wall of the test section in this experirnen i so that it is difficult estivale the tlow rate directlv. The velocity that i.s measured with the ultrasonic freqoency of l hlHz thorouph the stainless steel is distur bed ut near side of ttansduccr by refiection such as Kïñura et u1.' ' reported. However. if a flou condition is symmetry, a flow rate can be calculated by usine half side velocity profite. To obiain an in formation of the fow condition. ne measured the U-component ’elocity oi verrous po.sition.s. Mean ve)ocity proFile of U-component s shown in Fig.4. As mentioned above, the reflection is too xtrong to measure over a pipe diameter so that a velocity was measured two position which is 0 and 90 degree and the Results are shown only half side from the wall of transducer side to the rZd=0.5. As shoun in fip•ure, velOCit y profite aï Lt- canipanent is agreement with that of the l/7 power laiv in v'arious position.s so that it is supp•ested that the velociiy distribution in the pipe is alniosi sJmmetr
   2. Flow rate measurement

As menfiooed above, the velocity profile at tar side of” the pape wall can not be obtaiii so that a flow rate is estiniated by using one ef near side re*ion at the case ot min* 4MHz frequeiicy. On the other hand. the velocity profils can not be obtained at near sidc at the

.5

:Vertical Ptate, a=12

.Horizontal Rate. a=12

:HorizontaI Rate, a—10

:Horizontal Rate. a=12

: t/7 law

0

case of usin a 1MHz trequeocy becaosC OU the

rim*inc of the stainless steel so that one Of f’.ir

.side is used to estimate flow rate. Typical transient flow rate is shown in Fio 5. Sarnpling interval of flow rate is 72msec. Mean flow rate is 68.181/s and standard der iaiion is 7.3Yc. Various frequency fluctuation can be observed and it has been ctari ted that these fluctuation of flow rate was agreement with that measured by orifice flow meter’6!.

The result of fie=400K that is compared flow rate measured by UVP with that measured by NIST system is shown in Table 1. First column means data name. The value of second colunui is the flow rate that is measured by l_lVP and the value of thud column is wei•qht flow rate by NIST system. In this table. .leven data set.s of measurement result that is examined five timer in one experiment are shows. Especially. the number C, the axerap•e flow rate is

excellent agreement with one measured by c

usin_ «ei ght flow rate and the result is 0.00°%. About other data set as shown the table. the flow rare measured by thix new method is g•ood a•greement with the wei¿ht flow rate by the WIST system. The error fatc in all eX OflfFlC t is only 0. l8Vr. The result about more h1zh Reynolds u umber (=2.6M) tS shour in Tuble 2. The error rate is a httle

0 O.1 0.2 0.3 0.4 0.5

rJd

Fig. 4. V component mean ’elocity

@n is bulk velocity that measured by NIST system and

“a” is inclînatiOn anp•1e of the trunsducer

ZO

8

4C0

Tirne(sui

Fig. 5. Trarisient flow rate

56

thís method is very high as shown in the table.

, A

Differencc y+ error

In this experiment. the inclination ample of tranUucer for I MHz frequency is fixed 5 deyee

A 70.60 3.25

2 70.24 3.22

3 70.76 3.01

8

4 70.6\

\ 109. } 4 6g g7

1 T10.27 70.04

\ 1O.30 70.Q4

T1s0.23 70.04

63 0 90 /

-0.20 4.29X

6.72 - t .03*/‹

-0.57 -0.82T•

        S  70.23 3.\2

TJJ0.78 7007

-0.16 -0.23.

because ue aimed the measurement Or hi•h Reynolds number. As a result the reflection was too strcn• to

measure a flow rate. however as shou'n in a result. it

B 6 0 33

7 70, Ei 3.41

8 70.20 3.39

9 69.8s 3.56

\\t0.87 70.08

1111.42 70.1\

t1f0.32 70.04

1 t0g.gJ 7o.01

6.\2 -0.17T

-0.25 -0.3S%

.g3 .Ü2@o

0.z S 0.21”ó

is possible that the flow rate measures using half side

10

C t 1

69.90 3.38

70.21 3.17

0 7 7 ,      0.17         0.Z4*e   

1113.62 70.25 0.04 0.05?

velocity profile in this configuration. lf we examine with flexfbility of the incTination anule oÉ transducer, it is su ested that we can measure with gore hi*h

’accuracy.

1 2

13

4

\ S

0 \ 6

17

t8

\9

70.34 3.17

70.38 3.39

70.30 3.40

70 ›6 3.41

69•@ \ 3.22

70.12 3.12

69.67 3.25

69.88 3.21

T T 13.76 70. 6

1\ 13.6\ *0.2's

1t 15.04 70.34

1 t4.l0 70.2B

T1 11.90 70. 14

II 13.85 70.Z7

70.21

t 1t2.82 70.2o

-0.08 -0, 1%

6.13 -0.199

8

*

0.04

0. t2 O. 7”@

0./3 0.d•@@e

0.15 0.2t*s

0.54 0.77 z

0.32 0. S+,

   20  

07 2

13 73 0 26    o. \ 9    0.274  

S.CONCLUSION

A new type flow’ meterin•p system usine ultrasonic Doppler method has ten developed. In this presentation. we reported the result of comparison absolutel y fi ow rate measured by this ne«’ method with weight flow rate by the MST standed calibration system.

B

In this experiment, the pipe flow of NIST

E 21

22

23

24

25

26

27

28

29

:x›

G 31

32

33

34

70.20 3.29

69.97 3.NO

70.13 3.3s

70.j t 3.31

70.@ 3.38

69.76 2.96

69.72 3.23

69.44 3, \ o

sg.sz 3 Z2

Gg.g6 3.07

69.70 3,21

69.36 3.24

69.S4 3.1U

1f0J.7? 69. 0

1102.62 69.56

\ \ 02.g0 69.57

1 \02.85 69.57

1 03.20 69.59

110130 69.60

j 103.51 69.61

11D3. 49 69.6J

103.65 69.62

1f 03.M 69.6z

1 tOt.77 69.50

1102. 62 69.56

1102.90 69.57

T \D2.gS 69.57 |

-0.70 -1.0O”/

&.41 -0.591

4.S6 -0.BO*;

6.54 6.78

-0.77 -1. \ 0?

-0.16 0.23*

-0..06 -0.08*

-0.t \ -0.16

0. \ 8 0.26’»

0.04 0.C6•

-0.46 6.66°A

-0.J 4 •0.2O9

0.22 o.31%

0.03 0.05%

system is symmetry that is fully developed so that a

t)0\\ filte «'as calculated usin* by hal I side vet ocii y

Average

0.13

profile because velocity prOfi)e of another half side was disturbed by the ieRectiOR. The result of fie=400K is very good agreement with the ROw rate measured by the weight flow rate. The error rate in all experiment is only 0.187a. The result of more high Reynolds number (—2.6itt) is a little larger than the result of fie=400K because it was measured from the outside of the stainless pipe. However, when the

Tat›le 3. Flor rafe measurement

	UVP Average	(the) Deviation	Weight GPU	Lis	Oitference	% error
A	1	475.35	25.66	7485.94	472.z4	-3.1	6.66%
	Z	47846	2S.45		472.22	-5.85	-›.24°/
	3	475.10	2S.60	7494.20	472.76	-2.34	6.5O9•
	4 5	474.44 472.0‹	26.95 26.38	7478.94 7483.13	47\.@ 472.cms	-Z.6S a.as	-0.569 o.atop

COridition oL the seeding is better. the accuracy of this method is very high about 0.6°1.

As mentioned above, it is indicated that this new type ROw metefin• system usin z by UVP has ery high accurac}: And this method is multipurpOSe system because flow rate can be measured under high Reynolds

ACKNOWLEDGEMENT

We gratefully acknowledgement the supportinp« work made by Dr.iYIattin*1¿ and NIST crew and Mr.Barba_«allo in PSI.

REFERENCES

( 1 ) Takeda. Y., et at. AESJ fall annual meeting. F 16. 1998 (in Japanese)

(21 Kikura. H., et at. AESJ fall annual meetinp•. F l7, 1998 (in Japanese)

(3) S Lori. N., et aI. AESJ fall annual meeting. F t9, 1998 (in Japanese) [4) Taishi. T., et al. AESJ fall annual meeting. F IS. 1998 (in Japanese)

{5) Tskeda, Y., et al. Proc. of the 3 rd ASME/ASME Joint Fluid Enp•ineering Conference. FEDMS99-7 140. 1999

(6) Kikora, H.. et a!. Proc. of the 3 fd ASMC/JSIV(E Joint Fluid En gineerinp• Conference. FEDMS99-7 14 1, 1995

(7) ñfori. M., et al. ICONE-?, Dó-5, 1990