Proceedings of the ENCIT 2012 14th Brazilian Congress of Thermal Sciences and Engineering

Copyright
c 2012 by ABCM November 18-22, 2012, Rio de Janeiro, RJ, Brazil

UNCERTAINTY AND REPRODUCIBILITY ANALYSIS OF A

LABORATORIAL SCALED PIPE PROVER
Marcelo Aiolfi Barone, [email protected]
Rogério Ramos, [email protected]
Márcio Coelho de Mattos, [email protected]
Universidade Federal do Espírito Santo - UFES
Programa de Pós-Graduação em Engenharia Mecânica - PPGEM
Av. Fernando Ferrari, 514, Campus Universitário de Goiabeiras, Vitória, ES, CEP 29075-910

Abstract. The work uses a laboratory scaled prototype of primary pipe prover flowmeter in order to assess conformity
with the requirements established by technical standards and examine the uncertainties of flow measurement performed,
besides to check the reproducibility, the influence of operating parameters such as operating pressure variation and
comparison with gravimetric test. The prototype is plastic made tubing and considering statically defined dimensions.
Inside tubing, an interfering elastomer sphere is driven by flow provided by a centrifugal pump. The detection system uses
infrared light sensors able to detect the spheres passage causing any pipeline interference or blockage. The acquisition
data system is automatic and records the sphere transit time through each pulse signal detected as well as the pulses
generated by a calibrated turbine meter, as comparison of the evaluation of process uncertainties.

Keywords: pipe prover, uncertainty, flow measurement, gravimetric method, reproducibility, primary meter.

1. INTRODUCTION

In petroleum industry, the custody transfer operations are characterized by the ownership transfer of an oil company
to another in terms of contract following the technical requirements for fiscal measurement, which consists in measuring
the produced volume by a accredited measuring point production (ANP, 2000). In such transactions, researches for
improvements in measurement systems are essential because of the high capital values and taxes involved.
The standards API MPMS 4.2 (1988) and ISO 7278-2 (1988) establish that the flow meters used for custody transfer
should be calibrated periodically. In turn, the Brazilian National Agency of Petroleum, Natural Gas and Biofuels - ANP,
by its resolution no . 1 (ANP, 2000) regulates the calibration of measurement systems in line, which includes provers,
master meters and other meters previously approved.
The bidirectional pipe prover is a prover type for primary flow checking. According to the Guide to Uncertainty
of Measurement - GUM (2003), a prover of primary flow means that it is capable to estimate the flow rate through
independent variables, in other words, direct relationship between volume and time, while a secondary flow prover does
not correlate these variables directly, requiring intermediate correlations between physical properties of flow, such as
coriolis, ultrasonic, orifice plate or electromagnetic flow meters.
So, the objective of this work is to analyze the uncertainty of flow measurement performed by a laboratory scaled pipe
prover prototype in order to better understand the behavior of flow uncertainties in relation to specific standards, beyond
the influence of operational parameters such as variation of operating pressure and comparison with gravimetric test as
well as test the infrared sphere detection system.

2. BIBLIOGRAPHICAL REVIEW

One of the first pipe prover record is found in 1950, but using a cleaner pig as displacer. After soon the Shell Devel-
opment Company used the sphere as displacer (Dobesh, 1983). According to Pfrehm (1962), before the development of
such provers, the calibration was done by tank provers or master meters.
Proceedings of the ENCIT 2012 14th Brazilian Congress of Thermal Sciences and Engineering
Copyright
c 2012 by ABCM November 18-22, 2012, Rio de Janeiro, RJ, Brazil

Redilla (1977) detailed the construction of unidirectional and bidirectional provers. Two years later, Su (1979) made
several tests to determine statically what would be the minimum number of runs needed for a meter calibration under a
desired confidence level.
Gyory (1984) compared, using the bidirectional prover, the rules of three entities: Physikalisch-Technische Bunde-
sanstalt - PTB of Germany, National Office Measures Hungary - OMH and American Petroleum Institute - API, even
before an international agreement for calibration and expression of uncertainty.
Comstock (1985) studied the usual methods of provers calibration in lines, the method of water draw and the master
meter. In 1990, the author studied the interpolation pulses by three methods more usual: quadruple timing method, phase
locked loop method and the double timing method (Comstock, 1990). Ten years of technological advances were reported
by Jakubenas (1995). According to the author, these advances enabled the pipe prover greater reliability and accuracy of
data in addition to reducing the cost of the prover.
Garcia and Sherief (1999) studied about the methodology used for calibration using the water draw method.
Silva (2004) studied the concepts and calculation procedures for uncertainty and its propagation, determination and
classification. The author also presented some types of meters and made some considerations about the respective ways
for calculating the uncertainty expressions. In turn, Ribeiro (2010) reports and computes correctly the uncertainty of a
meter and discusses the calibration errors.
Lavezzo (2010) designed and built a laboratory pipe prover prototype in order to assess its experimental uncertainty
and performance, fulfill the requirements established by the rules.

3. DESIGN OF PIPE PROVER

The main pieces of the pipe prover are sketched in Fig. 1. The prover consists of a U-form plastic pipe, 53 mm internal
diameter and 9 m in length, equipped with four pairs of infrared transmitter/receptor sensors spaced 3 m each other. The
system is completed with valves, tank, electrical centrifugal pump, flowmeter turbine type and a special arrangement for a
four-way valve operation. But the equipment was dismantled during a civil works in the lab where it is installed, leading
to the need to test its reproducibility after its reassembly.

Figure 1. Pipe prover basic scheme and main parts.

For the measurements, it is used an interfering elastomer sphere which travels inside the pipe. Its interference should
be about 2% larger than the inner pipe diameter, which is filled with the working fluid. This feature guarantees the
tightness of the sphere even in motion avoiding errors of measurement, as provided by standard ISO 7278-2 (1988).
A centrifugal pump drive the flow which propels the sphere inside the tube at a constant flow velocity. The automatic
data acquisition system is automatic designed and constructed specifically for this application and it is composed of four
pairs of transmitter-receiver sensors of infrared light, installed on the tubing external wall, preventing pipe blockage and
interference in sphere movement, as shown in Fig. 2.
Proceedings of the ENCIT 2012 14th Brazilian Congress of Thermal Sciences and Engineering
Copyright
c 2012 by ABCM November 18-22, 2012, Rio de Janeiro, RJ, Brazil

PVC tubing Ø60mm O’ring gasket

Infrared transmi er Infrared receiver

Fluid flow
metering
cross ‐sec on

Glass window for Centralizer socket

ligh ng target

Figure 2. Detail of pipe prover cross section at infrared sensor system position, showing no interference on fluid flow Pipe
prover basic scheme and main parts (Lavezzo, 2010).

Data acquisition system records the sphere transit time through each pulse signal detected at distances, as well as the
pulses generated by the turbine meter.
Aiming flow reversion, ISO 7278-4 (1999) states the use of a four-way valve or a system of four valves working in a
special arrangement. The latter was installed in the prototype due its lower costs (Lavezzo, 2010).
The valve system used consists of four ball valves connected in pairs, and operated in parallel by a crank, in such way
that provide the flow reversion inside the U-tube, as shown in Fig. 3.

Crank to
Ball valves
(4 units) synchronize and
operate valves Cranck rotated
in 90o

Spheres
rota on
axis

Branch 1 Branch 2 Branch 2 Branch 1

Pressure line Pressure Line Return line Return line Branch 1 Branch 2 Branch 2 Branch 1
Pressure line Pressure Line Return line Return line
(a) (b)
Figure 3. Top view sketch of four-way valve assembly, showing two operational positions (Lavezzo, 2010):
Position (a): Pressure on branch 1 and return flow by branch 2;
Position (b): Pressure on branch 2 and return flow by branch 1.

4. BASE VOLUME

The pipe diameter is obtained by measurement with calibrated instruments. To measure the length between the sensors
it is used a tape laser and, in order to determining the length statically, many measurements were performed. Thus, the
base volume was calculated between the sensors (Lavezzo, 2010).
To calibrate the pipe prover base volume, it is used the gravimetric weighing method (ISO 4185, 1980), which consists
in moving the sphere between the sensors pair, while the fluid is diverted into a container installed on a balance. At the
operation end, it is recorded transit time, temperature, mass and fluid density accumulated in the container. According
to Young (2009), this method was adopted by the National Institute of Standards and Technology (NIST) since 1997 and
in 2003 the National Institute of Metrology, Quality and Technology (INMETRO) approved the norms NIE-DIMEL-043
(2003) and NIE-DIMEL-045 (2003) concerning to the gravimetric method.
Proceedings of the ENCIT 2012 14th Brazilian Congress of Thermal Sciences and Engineering
Copyright
c 2012 by ABCM November 18-22, 2012, Rio de Janeiro, RJ, Brazil

5. OPERATION

The operation begins by inserting the sphere in the launcher with the four valves system arranged so that the flow
inside the pipe is in opposite direction of launching the sphere. Then turns on the pump in order to initiate the flow until
equalize the temperature of the system.
Once prover temperature is equalized, the four-valve system crank is actuated to change the flow direction. The
sphere will accelerate at the beginning of the metering section in order to accelerate until reaches the uniform motion, in
clockwise direction, passing through the series of four optical sensors, blocking the light detection and triggering the start
of timing pulses. Flow through turbine is recorded as well.
To reverse the flow direction, the crank should be actuated starting a new transit time cycle, now in the Counterclock-
wise direction. After a few runs, the flow rate and its uncertainty are calculated statistically from the data obtained.
One of the advantages of bidirectional pipe prover consist to allowed the evaluation and estimation of hysteresis errors
by the time records in clockwise (outward) and counterclockwise (return) directions.

6. UNCERTAINTIES EVALUATIONS

According to GUM (2003) there are two uncertainty type, that can evaluated with statistic (Type A) and other evaluated
with experience or calibration certificate (Type B). In this work, the uncertainties are considered type A, so the uncer-
tainties are obtained from a frequency distributions where the best available estimate of the expected value of a quantity
that varies randomly and for which n independent observations (xi,k ) have been obtained under the same conditions of
measurement, is the average x of the n observations, demonstrated in Eq. (1):
n
1X
x̄i = xi,k (1)
n
k=1

Then the measurand may be given by Xi = x̄i ± ∆xi , so a good estimative of Xi is represented by x̄i with a variations
xi,k that, due to random effects, its experimental variance has to be defined by Eq. (2):

n
1 X 2
s2 (xi,k ) = (xi,k − xi ) (2)
n−1
k=1

The uncertainty that represent the variations xk , in average, will be the square root the experimental variance x2 (xi,k )
divided by n observations as is the Eq. (3) (GUM, 2003):

s2 (xi,k )
u2 (xi ) = s2 (xi ) = (3)
n

In order to determine the uncertainty of each variable that defines the flow, it is necessary to combine each result ac-
cording to the mathematical relationship that define the measurand. Ribeiro (2010) suggest basic rules for the uncertainty
propagation, which simplifies the calculations and does not require the partial derivatives use:

• In addition and subtraction operations, the uncertainties are propagated by the sum of absolute uncertainty or stan-
dard uncertainty.
qX
2
uc = u(xi ) (4)

• In multiplication and division operations, the uncertainties are propagated by the sum of the relative uncertainties.
s
u X  u (xi ) 2
Pc = (5)
xi xi
Proceedings of the ENCIT 2012 14th Brazilian Congress of Thermal Sciences and Engineering
Copyright
c 2012 by ABCM November 18-22, 2012, Rio de Janeiro, RJ, Brazil

Although the combined standard uncertainty can be used to express the uncertainty of a measurement result in some
commercial, industrial or regulatory sense, it is often necessary to report the uncertainty estimative as an interval around
the measurement result which is expected to cover a large fraction of the distribution of values that could reasonably be
attributed to the measurand. Then, an expanded uncertainty is obtained by multiplying the combined standard uncertainty
by a correct factor t-student.

U = t (ν) × uc (y) (6)

Where, ν is degrees of freedom obtained by ν = n − 1 with confidence interval, which can be obtained the correction
factor t-student using the appropriate table (see GUM (2003)).
So the combined uncertainty relative to test the prover reproducibility and pressure variation effects will be composed
by the combined uncertainty of thermal expansion - uc (∆VT ) - and prover pressure - uc ∆VP , transit time - uc (t) - and
the measured volume basis - uc (Vb ), as demonstrated in the equation below:
s 2  2  2  2
uc (t) uc (Vb ) uc (∆VT ) uc (∆VP )
uc (y) = + + + (7)
t Vb Vb Vb

And for the gravimetric method, the combined uncertainty relative is compounded by the mass measurement - uc (M )
- and the three-way valve - uc (MV 3V ), as in Eq. (8).
s 2  2
uc (M ) uc (MV 3V )
uc (y) = + (8)
M M

7. RESULTS AND DISCUSSION

One objective of this work is to certify the system reproducibility after the reassembly of the pipe prover. Thus this
work follows the procedures adopted by Lavezzo (2010), which uses twenty cycles (corresponds to a run-way and another
back in the prover) in three levels of flow (maximum, intermediate and minimum).
It adopted the rejection criterion of Chauvenet, according to Mendes and Rosario (2005), which determines the mag-
nitude of the deviation di of a particular measurement xi if the average is greater than a value dch (limit rejection of
Chauvenet) the measurement should be rejected.

7.1 Reproducibility

After applying the criterion of rejection, the results obtained will be compared with data from Lavezzo (2010). These
will be presented in graphs with three levels of flow combined with their absolute uncertainties, as can be seen in Fig. 4.

0,360 0,360
Average Flow (l/s) - Lavezzo (2010)

Average Flow (l/s) - Lavezzo (2010)

0,320 0,320

0,280 0,280

0,240 0,240

0,200 0,200

0,160 0,160

0,120 0,120

0,120 0,160 0,200 0,240 0,280 0,320 0,360 0,120 0,160 0,200 0,240 0,280 0,320 0,360
Average Flow (l/s) - Present Average Flow (l/s) - Present

(a) Clockwise Direction (b) Counter Clockwise Direction

Figure 4. Comparing data for reproducibility test.
Proceedings of the ENCIT 2012 14th Brazilian Congress of Thermal Sciences and Engineering
Copyright
c 2012 by ABCM November 18-22, 2012, Rio de Janeiro, RJ, Brazil

Tab. 1 show the analysis numerically, since it is difficult to observe the uncertainties traces at the graph.

Table 1. Comparison data for reproducibility test.

Present Work Lavezzo (2010)

Level Average Uncertainty Average Uncertainty
Direction Flow Flow Absolute Relative Flow Absolute Relative
(l/s) (l/s) (%) (l/s) (l/s) (%)
Maximum 0,328 0,0003 0,10 0,329 0,0003 0,11
Clockwise Intermediate 0,240 0,0002 0,09 0,229 0,0004 0,16
Minimum 0,150 0,0001 0,10 0,139 0,0007 0,53
Counter Maximum 0,327 0,0003 0,10 0,329 0,0004 0,10
Clockwise Intermediate 0,240 0,0002 0,09 0,228 0,0004 0,18
Minimum 0,150 0,0001 0,10 0,135 0,0015 1,08
Complete Maximum 0,327 0,0002 0,05 0,330 0,0004 0,12
Cycle Intermediate 0,240 0,0001 0,04 0,228 0,0003 0,13
Minimum 0,150 0,0001 0,04 0,142 0,0002 0,17

As can be observed in the Fig. 4 and Tab. 1, the flow rates approached those used by Lavezzo (2010), and these
differences were due to different operators, while the combined uncertainty absolute behaves similarly and should only
be attempted for the minimum flows than in previous work have an absolute uncertainty larger than the current one. The
differences observed may be related to the fact that this job has a greater concern with impurities that can pass to the
prover, thinking about it, a filter was installed between the tank and the turbine, thereby retaining the impurities that can
be carried by the pump and taken to the prover. To ensure the efficiency of the filter is adopted the procedure to check and
clean it at the beginning of each experiment.
The ISO standard 7278-2 (1988) states that the bidirectional pipe prover uncertainty of three runs should be 0,02%.
Observing the Tab. 1, the relative uncertainties the current work are shown higher than established by norm and stable
when compared to the uncertainties obtained by Lavezzo (2010).

7.2 Gravimetric Method

According to Lavezzo (2010), as described in section 4. , the prover base volume has 19,85 liters with combined
absolute uncertainty the 0,02 liters. The applying results the gravimetric method can be seen in Fig. 5 it is noted that the
volumes does not present the same values, considering clockwise direction (19.86 liters) and counter clockwise direction
volume varies significantly with the level of flow set.
21,20

20,80
Volume (l/s)

20,40 Clockwise Volume

Counter Clockwise Volume
Volume - Lavezzo (2010)

20,00

19,60

Minimum Intermediate Maximum

Flow Level

Figure 5. Data obtained from the gravimetric method.

Proceedings of the ENCIT 2012 14th Brazilian Congress of Thermal Sciences and Engineering
Copyright
c 2012 by ABCM November 18-22, 2012, Rio de Janeiro, RJ, Brazil

Table 2. Gravimetric Method Results.

Clockwise Direction Counterclockwise Direction

Flow Average Uncertainty Average Uncertainty
Level Volume Absolute Relative Volume Absolute Relative
(l) (l) (%) (l) (l) (%)
Maximum 19,87 0,0360 0,18 21,08 0,0142 0,07
Intermediate 19,86 0,0233 0,12 20,78 0,0137 0,07
Minimum 19,86 0,0212 0,11 20,43 0,0161 0,08

7.3 Pressure Test

For the pressure test it was used three pressure levels (0,50 , 0,75 and 1,00 kg/cm2 ) and for each pressure is performed
for three levels of flow (minimum, intermediate and maximum). As can be seen in Fig. 6, the average flow decrease as
pressure increase, as expected, and the uncertainty increase.

Table 3. Experimental average data for test pressure.

Clockwise Direction Counter Clockwise Direction

Pressure Flow Average Uncertainty Average Uncertainty
Level Flow Absolute Relative Flow Absolute Relative
(kg/cm2 ) (l/s) (l/s) (%) (l/s) (l/s) (%)
Maximum 0,278 0,0003 0,12 0,276 0,0004 0,13
0,50 Intermediate 0,194 0,0003 0,16 0,193 0,0003 0,16
Minimum 0,125 0,0005 0,39 0,124 0,0007 0,58
Maximum 0,242 0,0003 0,13 0,241 0,0005 0,19
0,75 Intermediate 0,165 0,0008 0,47 0,163 0,0011 0,65
Minimum 0,107 0,0008 0,76 0,106 0,0006 0,54
Maximum 0,189 0,0005 0,27 0,190 0,0007 0,35
1,00 Intermediate 0,130 0,0004 0,32 0,130 0,0011 0,84
Minimum 0,084 0,0006 0,70 0,083 0,0013 1,60

0,280 0,280
Pressure 0,50 kg/cm² Pressure 0,50 kg/cm²
Pressure 0,75 kg/cm² Pressure 0,75 kg/cm²
Pressure 1,00 kg/cm² Pressure 1,00 kg/cm²
0,240 0,240
Average Flow (l/s)

Average Flow (l/s)

0,200 0,200

0,160 0,160

0,120 0,120

0,080 0,080

Minimum Intermediate Maximum Minimum Intermediate Maximum

Flow Level Flow Level

(a) Clockwise Direction (b) Counter Clockwise Direction

Figure 6. Experimental average data for test pressure.
Proceedings of the ENCIT 2012 14th Brazilian Congress of Thermal Sciences and Engineering
Copyright
c 2012 by ABCM November 18-22, 2012, Rio de Janeiro, RJ, Brazil

8. CONCLUSION

Lavezzo (2010) through experiments proved that the bidirectional pipe prover had repeatability, and this work contin-
uing to study and confirms the prover reproducibility. Despite the uncertainties are above the established by the standard
they remained stable independently of the flow level.
The gravimetric method, showed that the prover base volume at the counter clockwise direction varies with the flow
level and also differ the base volume in clockwise direction. The values are closes to Lavezzo (2010). The standard API
MPMS 4.9.4 (2010) recommends that the repeatability of the gravimetric method is 0.02%, so the uncertainties obtained
in the experiment were higher than recommended by the standard, this difference may be related to the fact that the
diverter valve be operated manually, leading to increase the uncertainty.
For pressure test, the pressure incresing causes a linear increment in flow as well as the associated relative uncertainties
overcoming the values recommended by ISO 7278-2 (1988).

9. REFERENCES

ANP, 2000. “Portaria conjunta no 01”. Agência Nacional do Petróleo.

Comstock, D.M., 1985. “Calibration of liquid provers”. International School of Hydrocarbon Measurement (ISHM).
Comstock, D.M., 1990. “Theory and application of pulse interpolation to prover systems”. International School of
Hydrocarbon Measurement (ISHM).
Dobesh, E.L., 1983. “Liquid flow provers”. International School of Hydrocarbon Measurement (ISHM).
Garcia, E.N. and Sherief, N.M.A., 1999. “Waterdraw calibration of a meter prover”. Saudi Aramco Journal of Technology,
pp. 25–36.
Gyory, J., 1984. “The determination and the uncertainty of a prover’s base volume”. Measurement, Vol. 2, No. 4, pp. 191
– 196. ISSN 0263-2241. doi:10.1016/0263-2241(84)90052-6.
Jakubenas, P.P., 1995. “Advances in pipe prover technology”. 3rd International Symposium on Fluid Flow Measurement.
Lavezzo, L.F.Q., 2010. “Análise experimental e avaliação das incertezas em medição de líquidos com referência tipo ball
prover”. Universidade Federal do Espírito Santo (UFES).
Mendes, A. and Rosario, P.P., 2005. “Metrologia e incerteza de medição”. Editora Epse.
Pfrehm, R.H., 1962. “The calibrated pipe prover - an improved method for calibrating meters in lact units”. Journal of
Petroleum Technology, Vol. 14, No. 10, pp. 1087–1090. doi:10.2118/343-PA.
Redilla, R.L., 1977. “Design of displacement provers”. FMC Smith Meter Inc.
Ribeiro, M.A., 2010. “Incerteza na medição”. Revista InTech, , No. 118.
Silva, C., 2004. “Incerteza na medição de vazão: Uma técnica simples, mas poderosa, para aumentar a produtividade,
minimizar desperdícios e reduzir vazamentos para o meio ambiente”. Boletim Técnico da Petrobras, , No. 47, pp.
202–232.
Su, S.Y., 1979. “A statistical meter - proving method”. Journal of Petroleum Technology, Vol. 31, No. 9, pp. 1191–1193.
doi:10.2118/7113-PA.
Young, W.R., 2009. “Calibration of liquid provers”. International School of Hydrocarbon Measurement (ISHM).

10. RESPONSIBILITY NOTICE

The following text, properly adapted to the number of authors, must be included in the last section of the paper: The
authors are the only responsible for the printed material included in this paper.