Flow Measurement

and Instrumentation
Flow Measurement and Instrumentation 19 (2008) 181187
www.elsevier.com/locate/flowmeasinst

Calibration tests of pulse-Doppler flow meter at national standard loops

Kenichi Tezuka a, , Michitsugu Mori a , Takeshi Suzuki a , Yasushi Takeda b
a Thermal-Hydraulics & Fluid-Structure Dynamics Research Group, R&D Center, Tokyo Electric Power Company, 4-1, Egasaki-cho, Tsurumi-ku,

Yokohama, 230-8510, Japan

b Laboratory for Flow Control, Hokkaido University, Nishi 8, Kita 13, Kita-ku, Sapporo, 060-8628, Japan

Received 2 November 2006; received in revised form 10 October 2007; accepted 8 November 2007

Abstract
Calibration tests of UdFlow, the ultrasonic pulse-Doppler flowmeter manufactured by the Tokyo Electric Power Company, were conducted at
the national standard loop in Mexico, CENAM (The Centro National de Metrologia), in order to evaluate the accuracy of the flowmeter. Four
ultrasonic transducers were mounted circumferentially on the surface of 100 and 200 mm stainless steel pipes to measure four velocity profiles.
Flow rates can be obtained by integrating each measuring line and averaging them. Air was injected upstream of the measuring point to provide
bubbles as ultrasonic reflectors. Tests were conducted at five different flow rates with Reynolds numbers from 200,000 to 1,200,000. Tests were
repeated six times at each flow rate to evaluate repeatability. In addition, a take-off and put-back test was carried out on the 100 mm pipe at a
flow rate of 3000 L/min to evaluate reproducibility. The values of the CENAM loop are based on the average of weighing time while those of the
ultrasonic-Doppler flow velocity-profile flowmeter are based on the time average of instantaneous values. The calibration tests found a deviation
of less than 0.3% between the two devices in terms of the average of the values recorded in six rounds of measurement. Measurement at a different
Reynolds number showed that the overall average deviation between the two devices was less than 0.3%.
c 2007 Elsevier Ltd. All rights reserved.

Keywords: Ultrasonic-Doppler; Velocity profile; Flowmeter; Industrial application; Calibration

1. Introduction
The feedwater (FW) systems of power plants generally
utilize fluids at high temperatures and/or high pressures within
large pipes. Flow rate measurements are usually made with
meters such as flow nozzles or transit-time ultrasonic flow
meters that employ profile factors (PF), which are correction
factors used to adjust the measured value in the field based
on a factory test. However, determining the PF under the same
flow conditions and configurations as the large pipe diameters
and curves found in the field is impractical, so certain errors
in measurement arise. In fact, it is impossible at the present
time to determine a PF by a high-precision calibration loop
using a weighing method under such high temperature and
pressure conditions as those found in the typical FW system.
Consequently, the PF has to be determined with a Reynolds
number (Re) smaller by an order of magnitude than that in
Corresponding author. Tel.: +81 45 613 6156; fax: +81 45 613 7899.

E-mail address: tezuka.kenichi@tepco.co.jp (K. Tezuka).

c 2007 Elsevier Ltd. All rights reserved.
0955-5986/$ - see front matter
doi:10.1016/j.flowmeasinst.2007.11.001

the actual plant. Conventional ultrasonic flowmeters correct all

indeterminate errors with a PF as described in Fig. 1 [1]. To
remove these errors, efforts are needed to simply eliminate
the PF by determining flow rates based on the true flow
profiles in the piping [2,3]. In order to achieve a highly
accurate flow measurement and eliminate the PF, the flow
profile is required [4]. We have conducted field tests using
UdFlow, a flow-metering system that uses ultrasonic pulseDoppler velocimetry, [5] to obtain instantaneous flow profiles
and flow rates in the water circulation system at a thermal
power plant [6]. Obtaining flow rate measurements in nuclear
FW systems requires even greater accuracy (within 0.5%)
to monitor the thermal power and to reduce measurement
uncertainty for any power uprate. Integration of instantaneous
velocity profiles, obtained by performing continuous linemeasurements along the piping, can provide an accurate flow
rate measurement, acting as an advanced flowmeter, superior
to the conventional flowmeter that uses a PF. Conventional
ultrasonic flowmeters based on the transit-time method depend
largely on the accuracy of the PF, as it is a factor in determining

182

K. Tezuka et al. / Flow Measurement and Instrumentation 19 (2008) 181187

Fig. 1. Algorithms of conventional flowmeters and ultrasonic pulse-Doppler

velocimetry.

Fig. 3. Test loop at NIST.

Table 1
Flow rates measured by a reference meter and by UdFlow at the NIST standard
loop
Average
flowrate of
reference
meter
(L/min)

Average
flowrate of
UdFlow
(L/min)

Deviation Combined
(%)
standard
uncertainty
(%)

Expanded
uncertainty
(%)

69.611

69.634

0.032

0.308

0.1541

Fig. 2. Schematic diagram of the test loop at NIST.

the flow rate of the fluid. This is also true for a conventional
one-point ultrasonic-Doppler flowmeter. Accordingly, these
conventional methods have only a limited scope as they are
effective only in measuring fully developed steady-state flow.
In other words, these methods rely on an approximation that
is really applicable only in a narrow flow range, the so-called
reference conditions [6].
Calibration tests were performed at the national standard
loops of four countries in order to validate the accuracy of
UdFlow, which is based on the measurement of line velocity
profiles, thereby eliminating PFs and giving a more accurate
determination of flow rates.
2. Calibration tests at NIST, NMIJ AND NMI
2.1. Tests at NIST
The flow of water per unit of time can be determined by
collecting the fluid flowing down the test section in a weighing
tank for a set period of time and dividing the collected volume
by the time elapsed. The nominal measurement uncertainty
of the test loop at The National Institute of Standards and
Technology (NIST) in the U.S. is 0.12%. Fig. 2 shows a
schematic diagram of the test loop at NIST and Fig. 3 shows
a photograph.
In these tests, the water flow was measured at the point
where it reached the stage of full development. The UdFlow
was found to match the approved values of the standard loop
with sufficient accuracy. Table 1 compares the approved values
of the NIST standard loop and the corresponding data from the
UdFlow at Re = 400,000. The values of the NIST loop were

Table 2
Flow rates measured by a reference meter and by UdFlow at NMIJ standard
loop
Average
flowrate of
reference
meter
(L/min)

Average
flowrate of
UdFlow
(L/min)

Deviation Combined
(%)
standard
uncertainty
(%)

Expanded
uncertainty
(%)

2000.5
1512.7
986.1

2008.9
1508.2
984.6

0.421
0.294
0.147

0.288
0.078
0.178

0.1442
0.0390
0.0890

obtained by the weighing tank method while those of the UDF

were based on the average of instantaneous values. As indicated
in the table, the test found a deviation of 0.03% between the
two devices in the average values recorded by five rounds of
measurement. From measurements with different Re numbers
it was found that the overall average deviation between the two
devices was no more than 0.2% [7].
2.2. Tests at NMIJ
Calibration tests were conducted on the UdFlow by the
liquid flowmeter calibration facility, a verification loop, at the
National Metrology Institute of Japan (NMIJ). Fig. 4 shows a
schematic diagram of the test loop at NIMJ, and Fig. 5 shows a
photograph.
The calibration tests on the UdFlow were carried out for
water with a measuring instrument attached to the 400A piping
section of the loop at NMIJ. The results of the test at NMIJ are
summarized in Table 2.

K. Tezuka et al. / Flow Measurement and Instrumentation 19 (2008) 181187

183

Fig. 4. Schematic diagram of the test loop at NMIJ.

Fig. 5. Test loop at NMIJ.

The test findings at NMIJ rate the uncertainty of the

tested flowmeter by comparing the average of 10 rounds of
measurement with the reference flow rate. The reference at
NMIJ was based on a weighing method. The NMIJ tests gave
the UdFlow a calibration certificate showing uncertainty ranges
within 0.4%.

Fig. 6. Schematic diagram of the test loop for water at NMI.

2.3. Tests at NMI

Further calibration tests were conducted on the UdFlow at
the Netherlands Meetinstituut (NMI). The tests were carried out
for water and kerosene with the 150A piping section of the loop.
Schematic diagrams of the test sections for water and kerosene
are shown in Figs. 6 and 7.
Photographs of the test sections for water and kerosene are
shown in Figs. 8 and 9.
Both the calibration facilities have their standard uncertainty
set at 0.02% of the reference flow rate. The results of
the test at NMI for water and kerosene are summarized in
Tables 3 and 4.
The uncertainty of a flowmeter tested at NMI is determined
by comparing the average of three test runs with the reference,
which was a turbine flowmeter. The UdFlow was given an NMI

Fig. 7. Schematic diagram of the test loop for kerosene at NMI.

calibration certificate showing uncertainty ranges within 0.59%

for water and 0.73% for kerosene.
2.4. Calibration tests at CENAM
Following improvements to the UdFlow System, calibration
tests were carried out at CENAM, using ultrasonic transducers

184

K. Tezuka et al. / Flow Measurement and Instrumentation 19 (2008) 181187

Fig. 10. Schematic diagram of the test loop at CENAM.

Fig. 8. Picture of the test loop for water at NMI.

Fig. 11. CENAM test Loop.

Fig. 9. Picture of the test loop for kerosene at NMI.

Table 3
Flow rates measured by a reference meter and by UdFlow in the NMI standard
loop for water
Average
flowrate of
reference
meter
(L/min)

Average
flowrate of
UdFlow
(L/min)

Deviation Combined
(%)
standard
uncertainty
(%)

Expanded
uncertainty
(%)

1275.2
953.8
630.8

1276.7
953.6
632.0

0.120
0.014
0.179

0.556
0.775
0.611

0.2779
0.3873
0.3055

Table 4
Flow rates measured by a reference meter and UdFlow in the NMI standard
loop for kerosene
Average
flowrate of
reference
meter
(L/min)

Average
flowrate of
UdFlow
(L/min)

Deviation Combined
(%)
standard
uncertainty
(%)

1280.8
954.6
642.9

1276.5
956.2
639.4

0.333
0.169
0.534

0.0960
0.3632
0.1919

Expanded
uncertainty
(%)

0.192
0.726
0.384

Fig. 12. Mounting for four beam ultrasonic transducers on the test pipe.

clamped on the surface of 100 mm and 200 mm stainless steel

pipings. Fig. 10 shows the pipe arrangement at CENAM. The
CENAM loop contains a straight pipe long enough to produce
a developed flow, an inlet header and pumps, and weighing
systems. CENAM has two weighing systems: 1500 kg and
10,000 kg. Fig. 11 shows a photograph of the test loop at
CENAM. Fig. 12 shows the mounting device for four beam
ultrasonic transducers installed on the test pipe.
Air was injected upstream of the measuring point with
a volumetric void fraction of approximately 0.04%. We
confirmed that the effect of air injection on the flow rate was

K. Tezuka et al. / Flow Measurement and Instrumentation 19 (2008) 181187

The experimental standard deviation of the mean of the meter

factor at each flow is given by Eq. (5).

Table 5
Ultrasonic parameters in the CENAM test
Ultrasonic parameters

Value

Basic frequency
Pulse repetition frequency
Cycle per pulse
Sampling interval per profile
Number of data points per profile
Number of pulse repetitions

1 MHz
10003000 Hz
4 waves/pulse
250 ms
128
128

negligible as measured by the weighing system of CENAM, by

comparing flow rates with and without air injection. Table 5
shows the major ultrasonic parameters used in the CENAM
test.
Tests were conducted at five different flow rates with
Reynolds numbers between 200,000 and 1,330,000, and were
repeated six times at each flow rate to evaluate repeatability.
In addition, a take-off and put-back test was carried out on
the 100 mm piping at a flow rate of 3000 L/min to evaluate
reproducibility. The values from the CENAM loop are based
on the average of multiple weighing times while those from
the UDF were based on the average of instantaneous values.
References [810] provide the definitions of the individual
values discussed in the following sections:
2.4.1. Relative errors
The relative error e in flow rate qv between the primary
standard and the meter under test is defined by Eq. (1).



qvmeter
1 100(%).
(1)
e=
qvprimary standard
2.4.2. Meter factor
The Meter Factor (MF), MFi (q j ), for a single measurement
i at flow q j is defined by Eq. (2).
MFi (q j ) =

q jprimary standard
,
q j meter

(2)

where, :
qvprimarystandard volumetric flow rate determined by the
primary standard under line conditions
qvmeter time-averaged volumetric flow rate measured by
the meter under line conditions, over the calibration interval.
The arithmetic mean value of MF for a series of
measurements at flow rate q j is defined by Eq. (3).
MF j (q j ) = MF j =

n
1X
MFi (q j ).
n i=1

n

2
1 X
MFi (q j ) MF j (q j ) .
n 1 i=1

1
sdm(MF j ) = s(MF j ),
n

(5)

where n is the number of repeated tests at flow j. The meters

repeatability, i.e., the short-term stability, can be quantified as
the experimental standard deviation of the mean at each test
flow; the largest repeatability is considered a limit for the meter
for all of the flows tested.
2.4.4. Combined standard uncertainty
The combined uncertainty of the meter factor is the rootsum-squared (RSS) of the type A and type B uncertainties
that contribute to the meter factor, and the type A contributing
uncertainties (i.e., those uncertainties obtained using statistical
methods) are considered to be uncorrelated [11].
u A (MF j ) = sdm(MF j ).

(6)

The type B uncertainties are those obtained by methods other

than statistics, i.e., estimates, etc. and for the meter under test,
these are taken to be the combined uncertainty attributed to the
respective facility, at the 68% confidence level, i.e.,
u B (MF j ) = UC (Facility),

(7)

where UC is the combined standard uncertainty attributed to the

respective CENAM Facility. That is 0.04(%).
The combined standard uncertainty for the meter is written:
q
UC (MF j ) = u 2A (MF j ) + u 2B (MF j )
q
= sdm2 (MF j ) + UC2 (Facility).
(8)

2.4.5. Expanded uncertainty

The expanded uncertainty of the mean value of meter factor
U is given by Eq. (9):
U (MF j ) = kUC (MF j ).

(9)

The expanded uncertainty U can also be expressed using a

coverage factor k based on the t-distribution for eff degrees
of freedom with a level of confidence of approximately 95%.
In the CENAM test, the coverage factor k is 2.57 for five
degrees of freedom. This uncertainty was evaluated according
to Reference [9].

(3)

2.4.3. Meter factor uncertainties

The experimental variance of observations from repeated
measurements at flow rate q j is defined by Eq. (4).
s 2 (MF j ) =

185

(4)

2.4.6. Reproducibility
Reproducibility is defined in terms of the standard deviation
of the mean of multiple sets of runs performed under essentially
the same flow conditions after specific, typical changes in test
conditions given by Eq. (10).
v
u
m 
X
u
2
1
sdn[MF j ] = t
MFi (q j ) MF j ,
(10)
m(m 1) i=1

186

K. Tezuka et al. / Flow Measurement and Instrumentation 19 (2008) 181187

Table 6
Summary of test results with 100 mm pipe for five different flow rates with six
runs at each flow rate
Average
flowrate of
reference
meter
(L/min)

Average
flowrate of
UdFlow
(L/min)

Deviation Combined
(%)
standard
uncertainty
(%)

998.5
1502.2
1998.4
2497.2
3001.4

998.1
1504.1
1994.9
2499.8
3005.6

0.035
0.124
0.174
0.105
0.139

0.0993
0.0716
0.0702
0.0427
0.0412

Expanded
uncertainty
(%)

0.199
0.143
0.140
0.085
0.082

Table 7
Summary of test results with 200 mm pipe for five different flow rates with six
runs at each flow rate
Average
flowrate of
reference
meter
(L/min)

Average
flowrate of
UdFlow
(L/min)

Deviation Combined
(%)
standard
uncertainty
(%)

Expanded
uncertainty
(%)

4000.7
6006.8
8005.2
9998.6
12002.1

4010.1
6012.0
7998.6
9980.5
12011.4

0.233
0.087
0.083
0.182
0.078

0.150
0.089
0.097
0.139
0.106

0.0748
0.0446
0.0486
0.0695
0.0528

where j is the flow for which the changed conditions tests

were done again and m is the total number of repetitions
performed at essentially the same test flow.
Specific sets of tests for the UdFlow were made to assess the
meters performance under these conditions. Typical changes
needed by most meter users are turning the flow off and then
turning it back on, and then repeating the tests; this effect
can be quantified as TOTO (turn-offturn-on) reproducibility.
For clamp-on type meters, another typical change is quantified
by TOPB (take-offput-back) reproducibility. The conditions
changed in these tests at CENAM include, for a single flow,
TOTO and TOPB reproducibility. Longer term meter stability is
quantified by the experimental standard deviation of the mean
for these tests.
Tables 6 and 7 summarize the test data on the 100 mm and
200 mm pipe diameters under ideal flow conditions. Fig. 13
shows an example of a velocity profile at CENAM test for
100 mm pipe with the flowrate of 3000 L/min. Velocity profiles
adjacent to the pipe wall are corrected based on the theoretical
model. The average flow velocities varied from 2 m/s to
6 m/s for both the pipes. The Reynolds numbers were run
up to 1,330,000 for the 200 mm pipe, and 641,000 for the
100 mm. From the measurements conducted with different
Reynolds numbers, it was found that the overall average
deviation between the two devices was less than 0.3%. The
table lists the mean values of meter factor and errors calculated
from 6 runs at each flow rate. An expanded uncertainty of the
average of meter factors is also calculated for the worst case.
Reproducibility was evaluated at a flow rate of 3000 L/min with
100 mm piping, giving a value of 0.66 105 .

Fig. 13. Example of a velocity profile at CENAM test for 100 mm pipe with
the flowrate of 3000 L/min.

3. Concluding remarks
Calibration tests of UDF were conducted at four national
standard loops: NIST in the United States, NMIJ in Japan,
NMI in the Netherlands, and CENAM in Mexico, to evaluate
the accuracy of this new type of flow-metering system. The
velocity profiles at NIST, NMIJ, and NMI showed deviations
within 0.5%. Following improvements to the UDF system,
the maximum spreads in individual MF test results in the mean
values from the UDF are from 0.17% to +0.14% for 100 mm
diameter pipe and from 0.18% to +0.23% for 200 mm
diameter pipe over the range of Reynolds numbers tested at
CENAM. At CENAM the short-term stability (repeatability)
and longer-term stability (reproducibility) are both considered
good, i.e., better than 0.3% under these test conditions.
Further testing will be needed to better quantify reproducibility
characteristics.
The expanded uncertainty for the UDF meter factor in these
tests at CENAM is bounded by 0.26% as a worst case for
100 mm pipe and by 0.19% as a worst case for 200 mm pipe.
These values are computed for 95% confidence levels with a
coverage factor k of 2.57. The air bubble injection used for these
tests had a negligible effect on the CENAM primary standard
measurements.
References
[1] Mori M, et al. Industrial application experiences of new type flowmetering system based on ultrasonic-Doppler flow velocity-profile
measurement. In: Third international symposium on ultrasonic doppler
methods for fluid mechanics and fluid engineering. EPFL; 2002.
[2] Takeda Y, et al. Development of flow rate measurement using ultrasonic
Doppler method (1) Theoretical background. Fall Meeting of AESJ 1998;
F16:343.
[3] Mori M, et al. Development of ultrasonic-Doppler velocity profile method
for flow rate measurements of power plant. ICONE-7 1999;FP7429.
[4] Mori M, et al. Effects of inner surface roughness and asymmetric pipe
flow on accuracy of profile factor for ultrasonic flow meter. ICONE-14
89729. In: Proceedings of ICONE 14. 2006.
[5] Mori M, et al. Development of a novel flow metering system using
ultrasonic velocity profile measurement. Experiments in Fluids 2002;32:
15360.
[6] Takeda Y. Measurement of velocity profile of mercury flow by ultrasound
Doppler shift method. Nuclear Technology 1987;79:1204.

K. Tezuka et al. / Flow Measurement and Instrumentation 19 (2008) 181187

[7] Takeda Y, et al. Development of a new flow metering system using
UVP, Preliminary performance assessments using NIST flow standards.
In: Proceedings of ASME FEDSM00. ASME 2000. Fluids Engineering
Division Summer Meeting.
[8] ASME/ANSI MFC-9M-1988. Measurement of liquid flow in closed
conduits by the weighing method. New York (NY): Amer Soc of Mech
Engrs.

187

[9] ISO/IEC/BIPM/OIML. Guide to the expression of uncertainty in

measurement. 1995.
[10] International vocabulary of basic and general terms in metrology. ISO;
1999.
[11] Mattingly GE, Yeh TT. Effects of pipe elbows and tube bundles on
selected types of flowmeters. Flow Measurement and Instrumentation
1991;413.