Professional Documents
Culture Documents
Orifice Plate1
Orifice Plate1
2014-01-04
http://www.producao.usp.br/handle/BDPI/46754
Downloaded from: Biblioteca Digital da Produção Intelectual - BDPI, Universidade de São Paulo
Experimental Thermal and Fluid Science 58 (2014) 93–104
a r t i c l e i n f o a b s t r a c t
Article history: The performance of orifice plates in real-time monitoring of oil, gas and water standard flow rates was
Received 4 May 2014 investigated. To this end, a multi-rate test was implemented in two production wells routed individually
Received in revised form 17 June 2014 to a test separator in field operational conditions. The well flow rate was varied in steps by changing the
Accepted 18 June 2014
choke opening.
Available online 28 June 2014
The ranges of fluid properties and flow conditions achieved during the experiment were: wellhead
pressure from 9073 kPa to 13,278 kPa, wellhead temperature from 47.8 °C to 53.5 °C, downstream choke
Keywords:
pressure from 6770 kPa to 7913 kPa, downstream choke temperature from 41.6 °C to 49.1 °C, gas–oil-
Multiphase flow rate measurement
Oil–gas–water flow
ratio from 1144 Sm3 =Sm3 to 2068 Sm3 =Sm3 , water-cut from 4.64% to 58.35%, standard oil specific gravity
Multi-rate test from 0.7988 to 0.8058, standard gas specific gravity from 0.7340 to 0.7550, standard oil flow rate from
Orifice plate 46:86 Sm3 =d to 266:65 Sm3 =d, standard gas flow rate from 62:68 103 Sm3 =d to 296:65 103 Sm3 =d,
Slip model standard water flow rate from 18:06 m3 =d to 159:33 m3 =d.
Black oil model The wells tested showed a different dynamic behavior: while well #2 did not vary significantly the
Petroleum production technology stream composition with flow rate, well #1 produced under gas coning, a near well-reservoir phenome-
non that governs the contribution of the reservoir gas-cap to the total stream composition.
The multi-rate tests generated two data sets with 1424 flow conditions through two flange-tap orifice
plates installed upstream (wellhead) and downstream of a cage choke valve. The ranges of orifice vari-
ables were: orifice diameter from 0.03479 m to 0.0430 m, beta factor from 0.4946 to 0.6507, differential
pressure from 15 kPa to 187 kPa.
The virtual metering system presented in Paz et al. (2010) was used to correlate the experimental data.
The associated model, suitable for differential pressure measuring devices, includes effects such as flow
concentration and slip (through Chisholm’s correlation), generalizing the mass flow rate versus pressure
drop relationship for multiphase flow. The total mass flow rate depends on a set of variables evaluated at
metering conditions: density and viscosity of the liquid and gas phase, mass quality, pressure drop across
the flow meter and geometry (contraction area and beta factor).
The determination of the fluid properties at metering conditions was made by using black-oil correla-
tions. These correlations are based on a set of input variables at standard condition that characterizes the
stream composition such as gas–oil ratio, water–oil ratio and specific gravities of each phase.
A comparison was made between the multiphase flow rates predicted by the model and the ones
simultaneously measured at the test separators. The oil, gas and water standard volumetric flow rate
deviations (coefficients of variation of the root mean square deviations) were below 3:52%.
It was theoretically demonstrated and experimentally verified that a systematic error exists when the
homogeneous model (equal phase velocities) is considered in the formulation, resulting in a flow rate
underestimation. When the homogeneous model was used to correlate the data, this effect increased
the deviation up to 10:5%.
Flow pattern at the wellhead was characterized as intermittent and annular-mist. Lockhart–Martinelli
parameter varied from 0.362 to 0.836; despite of the experimental data being beyond the wet gas region,
the multi-rate tests showed that Chisholm’s over-reading can be successfully extrapolated to these range.
Ó 2014 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.expthermflusci.2014.06.018
0894-1777/Ó 2014 Elsevier Inc. All rights reserved.
94 S.R.V. Campos et al. / Experimental Thermal and Fluid Science 58 (2014) 93–104
capacity and the gas pressure at the pipeline between wells and measurements GOR; WOR, water-cut WC and total mass flow rate
the plant. W exp are calculated as:
Q g 0 exp
GOR ¼ ð1Þ
Q o 0 exp
2.2. Multi-rate experiments
Q w 0 exp
An experimental campaign was carried out with two wells WOR ¼ ð2Þ
(respectively #1 and #2). The experimental setup for each well,
Q o 0 exp
shown in Fig. 1, comprises two ISO 5167, flange-taps orifice plates
located respectively upstream and downstream of a choke valve. In WOR
WC ¼ ð3Þ
this way, it was possible to obtain simultaneously two measure- 1 þ WOR
ments of the same total mass flow rate at different pressures;
because of vaporization effects, the flows have different qualities W exp ¼ Q o 0 exp qg 0 GOR þ qo 0 þ qw 0 WOR ð4Þ
and void fractions upstream of the corresponding plates. No flow
conditioners were used before the plates. Measured variables in Main sources of uncertainty in instrumentation are noise and
each plate were temperature, pressure and pressure drop. Pres- sensor drift. Typical uncertainties in the measured variables,
sures and temperatures were measured online, while GOR, according to supplier instrumentation sources and laboratory pro-
water–oil ratio WOR and densities of the individual phases were cedures, are shown in Table 1. Although uncertainties associated to
simultaneously measured in the well production tests. the instrumentation are low, the overall uncertainty associated to
The multi-rate tests were performed in accordance with the fol- well testing is higher due to the requirement of stabilized flow in
lowing phases: commissioning, pre-operation and tests. a large volume of the multiphase mixture in the pipeline and sep-
In the commissioning phase pressure, temperature and differ- arator where, in addition, level must be controlled. According to
ential pressure instrumentation shown in Fig. 1 were checked Falcone et al. [9], well testing with a separator has an accuracy
and calibrated. The test separator single-phase flow meters were between 5% and 10%.
also checked and calibrated. After multiple steps, the choke opening was returned to its ref-
In the pre-operation phase, the fluid mixture coming from the erence condition. The multi-rate test configuration parameters for
well was routed to the gravitational three-phase test separator. each well are shown respectively in Tables 2 and 5. The geometric
The pre-operation phase had two objectives: (a) to clean the test parameters for each well are presented respectively in Tables 3 and
circuit from any resident fluid from other wells and (b) to create 6. Tables 4 and Table 7 show the range of wellhead and production
a stable reference for the multi-rate test initialization. The stabil- parameters. The flow regimes presented in these tables are the
ization period was about 12 h. ones predicted by the flow pattern maps built in Section 5. The
In the test phase, wellhead and test separator variables were Lockart–Martinelli parameter was calculated from the model out-
recorded in the SCADA (Supervisory Control and Data Acquisition) lined in Section 3. Samples of oil, gas and water were taken daily
1
system with a sampling rate of 1 min . at the separator outlets for laboratory tests. Oil and water standard
The averaging procedure was made through the PIMS (Process densities were determined by using thermohydrometers, while gas
Information Management System) available in the field, which standard density were determined by using chromatography.
uses a swinging door compression algorithm fed by the SCADA sys- In each step, pressure, temperature and fluid composition var-
tem. Data compression limits the amount of storage for each vari- ied accordingly to reservoir near wellbore behavior, well comple-
able by discarding values which are not necessary to adequately tion and surface facility response.
reconstruct the history of that parameter. Because of the long dura- As an example of the well dynamic response to the multi-rate
tion and slow reservoir process response to the step disturbance, test, Fig. 2 shows characteristic pressures (upstream pressure P u ,
the data sets were constructed with a time step of 30 min. This downstream pressure Pd and pressure drop across the choke DP c )
time step was chosen for post-processing analysis. It is possible as a function of time for well #2, considering the steps made in
to decrease the time step to 1 min and make the PIMS time step the choke opening Sc . A decrease in the wellhead pressure and in
equal to the field SCADA system. Therefore, each experimental data the choke pressure drop are observed as the choke valve opens,
set is the result of the compression algorithm for high-frequency while the downstream choke pressure increases. Figs. 3 and 4 show
data acquisition from a SCADA system. In this way, data from respectively GOR; WC and differential pressure across the upstream
1424 operating conditions were taken. and downstream orifice plates, respectively DPu and DPd , for the
Simultaneous oil, gas and water individual volumetric flow rate same multi-rate test. It was observed that GOR and WC remained
measurements at standard conditions (respectively Q o0 exp ; Q g0 exp fairly constant for different production rates, while the orifice plate
and Q w0 exp ) were made at the separator outlets by using respec- differential pressure signals remained quite steady for metering
tively turbine, orifice plate and magnetic flow meters; from these purposes.
Table 1
Typical uncertainties in the measured variables.
Table 2 Table 7
Well #1 multi-rate test parameters. Well #2 range of wellhead and production parameters.
Table 4
Well #1 range of wellhead and production parameters.
between steps compared to the others. Although a dynamic
Well parameters Min–max value Unit response was found in the stream composition after a choke open-
Choke opening Sc 50.7–78.2 % ing, the sampling periods were sufficiently short to capture this
Upstream pressure P u 9073–12187 kPa effect.
Upstream temperature T u 47.82–51.82 °C
Upstream orifice plate pressure drop DP u 15.46–187 kPa
Downstream pressure P d 6770–7306 kPa 3. Model
Downstream temperature T d 41.62–49.12 °C
Downstream orifice plate pressure drop DP d 17.48–168.16 kPa
Upstream Lockart–Martinelli parameter v 0.386–0.836 –
3.1. Conservation equations
Gas–oil ratio GOR 1229–2068 –
Water-cut WC 40.06–58.35 % The model presented in this Section is a thorough version of the
Oil flow rate Q o 0 exp 46.86–146.17 Sm3/d one presented in Paz et al. [21]. It considers steady, one-dimen-
Gas flow rate Q g 0 exp 62.68–237.33 103 Sm3/d sional subsonic flow of gas, oil and water. Normal viscous stresses
Water flow rate Q w 0 exp 31.63–159.33 Sm3/d
and wall shear stresses are neglected in the converging portion of
Oil specific gravity 0.8017–0.8058 –
Gas specific gravity 0.7340–0.7550 – the flow meter. As oil and water have similar densities, it is consid-
Flow regime Intermittent and – ered that the relative velocity between these phases is neglected,
annular mist resulting a single velocity ul for the liquid (oil plus water) phase.
The relative velocity between the gas and liquid phase is taken into
account by means of a slip ratio S defined as:
ug
Table 5 S¼ ð5Þ
ul
Well #2 multi-rate test parameters.
Test parameter Value Unit where ug is the velocity of the gas phase. With these approxima-
Number of operating conditions 450 –
tions, mass and momentum conservation equations can be written
Number of steps 5 – as:
Number of days 9.5 days
Averaging period 30 min x þ xo þ xw ¼ 1 ð6Þ
( " #)
W 2t @ 1 1x x ð 1 xÞ @P
xþ þ S ¼ þ qm g s ð7Þ
A @s A S qg ql @s
Table 6
Well #2 orifice plates geometric parameters.
where
Fig. 5. Well #1 Gas–oil ratio and water-cut response to the step disturbances.
Fig. 2. Well #2 pressure responses to the step disturbances.
Fig. 6. Well #1 pressure drop responses across the orifice plates to the step
disturbances.
Fig. 3. Well #2 Gas–oil ratio and water-cut responses to the step disturbances.
x ql
a¼ ð10Þ
x ql þ ð1 xÞS qg
xo q l qg
ao ¼ S ð11Þ
x ql þ ð1 xÞS qg qo
xw ql qg
aw ¼ S ð12Þ
x ql þ ð1 xÞS qg qw
DP ¼ P1 P 2 þ qm g s ðs2 s1 Þ ð15Þ where ReDm is the mixture Reynolds number and lg ; lo and lw are
respectively the viscosities for the gas, oil and water. For an ISO
d 5167, flange-taps orifice plate, as used in the experimental valida-
b¼ ð16Þ
D tion (see Section 2), the resulting correlation is [15]:
2
where A2 ¼ p4 d is the cross sectional area of the contraction, DP is !0:75
2:1 8 2:5 106
the pressure drop corrected by the hydrostatic column and b is C D ¼ 0:5959 þ 0:0312 b 0:184 b þ 0; 0029 b
the diameter ratio (beta factor of the flow meter). In order to take ReD m
!
into account other effects not considered in the analysis, a discharge L1 b4 L2 3
coefficient C D is defined as: þ 0:09 0:0337 b ð24Þ
D 1 b4 D
W
CD ¼ ð17Þ where L1 ¼ L2 ¼ 0:0254 m. The correlation is valid for the following
Wt
range of variables: 0:05 m 6 D 6 1 m;d P 0:0125 m;0:2 6 b 6 0:75
where W is the real total mass flow rate. The final expression for the and ReD m P 1:26 106 b2 D ðmÞ.
mass flow rate results:
2 312 3.3. Slip ratio
2 DP
W ¼ C D A2 4 x 1x 5 ð18Þ
x þ 1x 4 There is an extensive bibliography on flow of two-phase mix-
S q þ q S 1b
g l
tures through orifices. Chisholm [5] proposed the following corre-
Eq. (18) can be regarded as a generalization of the mass flow rate lation for the slip ratio for incompressible two-phase mixtures
versus pressure drop relationship corresponding to a differential through sharp-edged orifices:
pressure flow meter for multiphase flow. It can be easily verified 8 1
>
>
< qql
4
that Eq. (18) reduces to the one corresponding to single-phase flow g
for v<1
for the limit values of mass quality (x ¼ 0 or x ¼ 1) and to homoge- S¼ h i1 ð25Þ
>
>
neous flow model for S ¼ 1. For the homogeneous model, the pre- : 1 þ x ql 1 2 for v>1
q g
dicted mass flow rate W h results:
where v is the Lockart–Martinelli parameter, defined as:
1
2 DP qmh 2 !12
W h ¼ C D A2 ð19Þ 1 x ql
1 b4 v¼ ð26Þ
x qg
1 x 1x
¼ þ ð20Þ According to Eq. (25), the slip ratio only depends on the liquid
qmh qg ql
to gas density ratio for high mass qualities (v < 1), while it is also
where qmh is the mixture density for the homogeneous model. a function of the mass quality for low mass qualities (v > 1). It can
be verified that Chisholm’s correlation gives a slip ratio indepen-
3.2. Discharge coefficient dent of the geometry of converging portion of the flow meter, in
agreement with the model assumptions. Moreover, it can be easily
In single-phase flow, the discharge coefficient is mainly depen- verified that the slip ratio is continuous for v ¼ 1. Fig. 7 shows the
dent on the Reynolds number and beta factor. The dependence on slip ratio as a function of the gas mass quality for different density
Reynolds number is very weak (except for low Reynolds numbers), ratios qql . It can be observed that the slip factor increases for low
g
so the performance of differential flow meters is rather insensitive qualities, reaching a constant value that increases as the density
to fluid viscosity. Discharge coefficients are very close to unity for ratio increases.
venturi and nozzle flow meters, showing that dissipation effects
are almost negligible. In orifice plates dissipation effects are not 3.4. Comparison between homogeneous and slip models
important either, but discharge coefficients are quite less than
unity due to vena contracta effects (diameter of the vena contracta It is interesting to make a comparison between the theoretical
is less than diameter of the obstruction). mass flow rate calculated with the homogeneous model W th ,
Although there is no assurance that a multiphase flow follows obtained by making S ¼ 1 in Eq. (14), and the one calculated with
the single-phase pattern, the assumption of negligible viscous dis- the slip model W ts . For the same pressure drop and flow meter
sipation was assumed in other studies [16]. Besides, in the pioneer geometry, the ratio of these quantities can be expressed as:
paper of Chisholm [4] studying incompressible two-phase mix- 8 h i912
ql
W t h < x þ S S þ qg S x =
1x
tures through orifices, the assumption of equal contraction coeffi-
cients (ratio of cross sections at the vena contracta and at the ¼ ð27Þ
Wt s : 1 þ ql 1 x ;
throat) was made for the liquid and gas phases. qg
Based on the previous considerations, dissipation and vena con-
Using Chisholm’s correlation for the slip factor, Eq. (27) is a
tracta effects are estimated in the model by a discharge coefficient
function of the mass quality and the ratio of the liquid and gas den-
C D calculated from correlations corresponding to single phase flow,
sities. The resulting mass flow rate ratio is shown in Fig. 8, where it
using a homogeneous model to evaluate the viscosity of the mix-
can be observed that the homogeneous model gives total mass
ture lm , this is:
flow rates systematically lower than the slip model. The ratio of
theoretical flow rates decreases as the density ratio increases, this
C D ¼ C D ðReDm ; bÞ ð21Þ
is, as pressure decreases.
As Eq. (27) and Fig. 8 are not related to any experimental data,
4W
ReDm ¼ ð22Þ this trend is valid for different pressures, fluids and flow meters
p D lm (orifice plate, nozzle or venturi). Moreover, assuming that the mass
flow rate predicted by the slip model is more accurate than the
lm ¼ lg a þ lo ao þ lw aw ð23Þ one predicted by the homogeneous model and neglecting the
S.R.V. Campos et al. / Experimental Thermal and Fluid Science 58 (2014) 93–104 99
total mass flow rate; this trend was verified in the experiments, Substituting Eq. (29) and (30) in Eq. (28) and neglecting varia-
where systematic deviations appeared when the homogeneous tions in the discharge coefficient (C 0D ’ C D ), it can be found that
flow model was used to correlate the data (see Section 5). the over-reading results:
1
3.5. Wet gas flow and over-reading OR ¼ 1 þ C v þ v2 2 ð31Þ
where:
The definition of wet gas (high mass quality) flow is not unique.
!12 !12
Mehdizadeh et al. [18] presented three different perspectives ql 1 ql
reproduced below: C¼S þ ð32Þ
qg S qg
Reservoir perspective: GOR exceeds 15; 000 SCF=STB with stock The over-reading obtained in Eq. (32) is coincident with the one
tank liquid gravity up to 70 API. deduced by Chisholm [4]. Substituting the slip ratio corresponding
Volumetric perspective: liquid content limited to liquid-to-gas- to v < 1 from Eq. (25) it is obtained the well known Chisholm’s
ratio LGR < 0:2% for stratified and LGR < 0:5% for annular mist over-reading C CH :
flow.
Volume-density perspective: Lockart–Martinelli parameter !14 !14
value range 0:02 < v < 0:3.
ql ql
C CH ¼ þ ð33Þ
qg qg
4. Fluid characterization
From Eq. (18) and (21), it can be seen that the total mass flow
rate depends on a set of variables evaluated at metering condi-
tions: density and viscosity of the liquid and gas phase, mass qual-
ity, pressure drop across the flow meter and geometry (contraction
area and beta factor).
Neglecting mass transfer (vaporization or gas solubilization) in
the water phase and assuming steady state flow and thermody-
namic equilibrium, a material balance yields the mass qualities
for the gas, oil and water phases as:
qg Bg ðGOR Rs Þ
x¼ ð34Þ
qg 0 GOR þ qo 0 þ qw 0 WOR
qo Bo
xo ¼ ð35Þ
qg 0 GOR þ qo 0 þ qw 0 WOR
Fig. 8. Ratio of theoretical mass flow calculated with the homogeneous model to qw 0 WOR
theoretical mass flow calculated with Chisholm’s
correlation, as a function of mass xw ¼ ð36Þ
quality, for different density ratios r ¼ qql .
qg 0 GOR þ qo 0 þ qw 0 WOR
g
100 S.R.V. Campos et al. / Experimental Thermal and Fluid Science 58 (2014) 93–104
qg 0
qg ¼ ð40Þ
Bg
The formation volume factors, gas solubility, gas compressibil-
ity factor and other variables such as viscosities are dependent
on pressure, temperature and properties at standard condition
and can be obtained from suitable correlations; Table 8 shows
the correlations used to determine these variables. Uncertainties
coming from the determination of fluid properties can be reduced
by using a compositional model and/or correlations adjusted to fit
the variables for specific fluids coming from the wells.
Once the total mass flow rate is determined from Eq. (18), the
volumetric flow rates at standard conditions Q o 0 ; Q g 0 and Q w 0 ,
respectively for oil, gas and water phases, can be readily calculated
as:
W
Qo0 ¼ ð41Þ
qg 0 GOR þ qo 0 þ qw 0 WOR
Fig. 10. Oil flow rates for well #1, downstream orifice plate, slip model.
Q g 0 ¼ GOR Q o 0 ð42Þ
ones predicted by the model, given by Eq. (18), (41), (42) and
Q w 0 ¼ WOR Q o 0 ð43Þ (43). The input to the model is a set of variables needed to charac-
terize the fluid (standard gas, oil and water gravity, GOR and WOR),
5. Results used with the black oil model to determine the fluid properties at
metering conditions, as well as geometry variables (orifice and
In this Section a comparison is made between the standard flow pipeline diameters) and flow variables (temperature, pressure
rates measured at the separator (named experimental) and the and pressure drop).
In order to quantify the performance of the model in predicting
the experimental values, the coefficient of variation CV of the root
mean square deviation RMSD is used for a set of N experimental
Table 8
Correlations used for fluid characterization.
data corresponding to the variable X, defined as:
Table 10
Well #1 fluid property average values.
Table 9
CV percentage values for the wells, slip model.
ð 1 xÞ W
jl ¼ ð47Þ
ql A
Well CV %, upstream CV %, downstream
Oil Gas Water Oil Gas Water
#1 1.48 1.30 1.40 2.60 2.49 2.54 xW
jg ¼ ð48Þ
#2 3.12 3.12 3.52 2.22 2.25 2.47 qg A
Although it was not possible to observe the flow pattern in
comparison with the experimental data, the data predicted with operational conditions during the experiment, the boundaries
the model showed an excellent agreement, considering that the between the different flow regimes were drawn using the software
model does not have any fitting constant coming from the experi- given in Shoham [23]. Tables 10 and 11 show the average fluid
mental data set. Besides, the correlations used for the fluid proper- properties used to draw the flow pattern maps, respectively shown
ties were standard ones (not fitted for the specific fluids at Urucu in Figs. 13 and 14 for the multi-rate test data in each well. For all
field), so uncertainties in the calculations coming from the fluid the maps, a horizontal tube of diameter D ¼ 0:0584 m and surface
characterization were not minimized. The CV values shown in roughness ¼ 3: 104 m were considered.
Table 9 for the orifice plates located upstream and downstream
of the choke valve are below 3:52% for any phase.
Table 11
Eq. (44) accounts only for the uncertainty corresponding to the Well #2 fluid property average values.
correlation of the model-calculated and experimental mass flow
Fluid property Value Unit
rates; overall or total uncertainty was not addressed. Total uncer-
tainty includes propagation of the uncertainties corresponding to Liquid density 732.05 kg/m3
the input variables through all the equations, as well as the uncer- Liquid viscosity 4:74 104 kg/m/s
Surface tension 0.070 N/m
tainties corresponding to the correlations. Uncertainties corre-
Gas density 120.21 kg/m3
sponding to the input variables and to the experimental mass Gas viscosity kg/m/s
1:76 105
flow rates are not high. The major contribution to the overall
102 S.R.V. Campos et al. / Experimental Thermal and Fluid Science 58 (2014) 93–104
1 Intermittent
jl (m/s)
0.1
Annular
Wavy
0.01
Stratified
0.001
0.1 1 10 100
jg (m/s)
Fig. 16. Oil flow rates for well #1, upstream orifice plate, homogeneous model.
1 Intermittent
jl (m/s)
0.1
Annular
Wavy
0.01
Stratified
0.001
0.1 1 10 100
jg (m/s)
Fig. 17. Oil flow rates for well #1, downstream orifice plate, homogeneous model.
these cases only after the flow at the reservoir after a choke open-
ing variation stabilizes its composition.
The total mass flow rate and the volumetric flow rates for each
individual stream at standard condition calculated with the VMS
showed an excellent agreement with the experimental data (from
test separator), with a CV ðRMSDÞ below 3:52%; these deviations
are less that the errors encountered in well testing with separators
(between 5% and 10%, according to Falcone et al. [9]). It is impor-
tant to notice that the model does not have any fitting constant
coming from the experimental data set. The correlation of the
experimental data neglecting slip between the gas and liquid
phases showed systematic deviations, increasing the deviation up
to 10:5%; this verifies the theoretical prediction that the experi-
mental data is underestimated by the homogeneous model.
Despite of the experimental data being beyond the wet gas
region, the multi-rate tests showed that Chisholm’s over-reading
can be successfully extrapolated to higher values of the Lock-
Fig. 19. Oil flow rates for well #2, downstream orifice plate, homogeneous model. hart–Martinelli parameter (v < 0:836).
It is worth mentioning the modular formulation of the model
leading to Eq. (18). In order to calculate the set of variables at
Table 12 metering conditions, a different slip correlation suitable for other
CV percentage values for the wells, homogeneous model.
flow pattern configurations and/or a different model for the fluid
Well CV %, Upstream CV %, Downstream characterization (for instance, a compositional model) could be
Oil Gas Water Oil Gas Water used. To overcome the limitation of constant fluid composition
#1 8.71 8.86 8.80 9.19 9.33 9.28
during the measuring period, the mass quality in Eq. (18) could
#2 8.21 8.14 8.69 10.0 9.93 10.5 be determined on real time from a concentration (void fraction)
meter located upstream of the orifice plate.
A reduction in the uncertainties is expected by utilizing PVT
correlations fitted for the specific fluids at the field, as well as by
information coming from the analysis above should be taken as a
using conditioners located upstream of the measuring device, pro-
rough estimate.
vided that the additional pressure drop introduced does not sub-
Regarding the over-reading considerations made in Section 3.5,
stantially reduce the well production.
Fig. 15 shows that the Lockhart–Martinelli parameter correspond-
It is worth mentioning that the VMS based on the orifice plates
ing to the experimental data ranged from 0.362 to 0.836, beyond
was implemented in all the wells at Urucu field for real-time mon-
the wet gas region (v < 0:3); as there is an excellent agreement
itoring of well production; for stabilized wells, the metering sys-
in the flow rate prediction, the multi-rate tests showed that Chis-
tem is an alternative to lower the frequency of well tests needed
holm’s over-reading (Eq. (33)) can be successfully extrapolated to
to monitor well production (usually once a month), as the set of
these range.
variables at the reference conditions (GOR; BSW and densities)
It is worth mentioning that there are other wells operating in
may not vary significantly with production rate and during the per-
the wet gas region; for these wells, the VMS is working fine in esti-
iod between well tests. For these cases, the orifice plate is used as a
mating the production.
stand alone instrument, fed with the concentration and reference
In order to investigate the influence of the slip ratio on the flow
variables obtained from batch well tests.
rate, Figs. 16–19 show the comparison of the experimental data
with the homogeneous model prediction. Larger values of the CV
are observed (up to 10:5%, see Table 12), as well as systematic Acknowledgements
deviations in which the homogeneous model underestimates the
experimental data. This is consistent with the theoretical analysis This work was granted by Petróleo Brasileiro S. A. (Petrobras).
made in Section 3.4. The authors wish to thank to Hermano Barreto, Francisco
Albuquerque, Daniel Wagner and João Batista (Petrobras Amazonas
Operational Unit) and the field operation personnel involved in the
multi-rate experiments. The authors also wish to thank to Conselho
6. Conclusions Nacional de Desenvolvimento Científico e Tecnológico (CNPq, Brazil)
and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior
The performance of orifice plates in real-time monitoring of oil, (CAPES, Brazil).
gas and water standard flow rates was investigated using field data
corresponding to two wells located at Urucu field (Solimões basin, References
Amazonas, Brazil), where the flow rate was varied by step changes
in the choke opening. [1] H.D. Beggs, J.R. Robinson, Estimating the viscosity of crude oil systems, JPT
(1975) 1140–1141.
The determination of the fluid properties at metering conditions [2] S.R.V. Campos, A.F. Teixeira, L.F. Vieira, S. Sunjerga, Urucu field integrated
is made by using black-oil correlations. These correlations use a set production modeling, in: Proceedings of SPE Intelligent Energy Conference and
of input variables at standard condition that characterizes the fluid Exhibition, Paper SPE 128742-MS. SPE, Utrech, Netherlands, 2010, p. 21.
[3] N. Carr, R. Kobayashi, D. Burrows, Viscosity of hydrocarbon gases under
composition such as the gas–oil ratio, the water–oil ratio and the
pressure, Trans. AIME 201 (1954) 270–275.
specific gravities of each phase. Although the flow meter can be [4] D. Chisholm, Flow of incompressible two-phase mixtures through sharpe-
used in transient measurements, the main limitation with this for- edged orifices, J. Mech. Eng. Sci. 9 (1) (1967) 72–78.
mulation is that the fluid composition must remain constant dur- [5] D. Chisholm, Research note: two-phase flow through sharpe-edged orifices, J.
Mech. Eng. Sci. 19 (3) (1977) 128–130.
ing the measuring period. As many wells at Urucu field [6] A.G. Collins, Petroleum Engineering Handbook, SPE, Dallas, Ch. Properties of
experience gas and water coning, the flow meter can be used in produced waters, 1987.
104 S.R.V. Campos et al. / Experimental Thermal and Fluid Science 58 (2014) 93–104
[7] J.R. Dempsey, Computer routine treats gas viscosity as a variable, Oil Gas J. [18] P. Mehdizadeh, J. Marrelli, V.C. Ting, Wet gas metering: trends in applications
(1965) 141–143. and technical developments, in: SPE Annual Technical Conference and
[8] P.M. Dranchuk, J.H. Abu-Kassem, Calculation of z-factors for natural gases Exhibition, No. SPE 77351, SPE, San Antonio, Texas, 2002, pp. 91–105.
using equations of state, JCPT (1975) 34–36. [19] J.T.H.Ng, E.O. Egboah, An improved temperature-viscosity correlation for crude
[9] G. Falcone, G.F. Hewitt, C. Alimonti, Multiphase flow metering. principles and oil systems, in: Proceeding of the 34th Annual Technical Meeting of the
applications, Developments in Petroleum Science, vol. 54, Elsevier, Oxford, UK, Petroleum Society of CIM, CIM, Banff, 1994.
2010. [20] J.L.G. Oliveira, J.C. Passos, R. Verschaeren, C.V. der Geld, Mass flow rate
[10] G. Falcone, G.F. Hewitt, C. Alimonti, B. Harrison, Multiphase flow metering: measurements in gas–liquid flows by means of a venturi or orifice plate
current trends and future developments, SPE 74689 (2002) 77–84. coupled to a void fraction sensor, Exp. Thermal Fluid Sci. 33 (2009) 253–260.
[11] V. Faluomi, E. Dellarole, M. Bonuccelli, L. Antico, Virtual metering system for [21] E.F. Paz, J.L. Baliño, I. Slobodcicov, D.F. Filho, Virtual metering system for oil
oil and gas field monitoring: design, implementation and applications, in: 5th and gas field monitoring based on a differential pressure flowmeter, in:
North American Conference on Multiphase Technology. BHR Group, Banff, Proceeding of the SPE Annual Technical Conference and Exhibition, No. Paper
Canada, 2006, pp. 91–105. SPE-133895-PP, SPE, Florence, Italy, 2010, p. 10.
[12] C. Fischer, Development of a metering system for total mass flow and [22] A. Rasmussen, Optimizing field production using multiphase and wet gas
compositional measurements of multiphase/multicomponent flows such as metering, World Oil 74 (2007) 99–103.
oil/water/air mixtures, Flow Measur. Instrum. 5 (1994) 31–42. [23] O. Shoham, Mechanistic Modeling of Gas–Liquid Two-Phase Flow in Pipes. SPE,
[13] ICT, 1927. International Critical Tables II. Dallas, 2006.
[14] I. Ismail, J.C. Gamio, S.F.A. Bukhari, W.Q. Yang, Tomography for multi-phase [24] M.B. Standing, Volumetric and Phase Behavior of Oil Field Hydrocarbon
flow measurement in the oil industry, Flow Measur. Instrum. 16 (2005) 145– Systems, SPE, nineth ed., Dallas, 1981.
155. [25] R. Steven, A.R.W. Hall, Orifice plate meter wet gas performance, J. Flow Measur.
[15] ISO, ISO standard 5167-1, measurement of fluid flow by means of pressure Instrum. 20 (2009) 141–151.
differential devices, in: International Standards Organization, Geneva, [26] N.T. Thang, M.R. Davis, Pressure distribution in bubbly flow through venturis,
Switzerland, 1991, pp. 91–105. Int. J. Multiphase Flow 7 (1981) 191–210.
[16] G.F. Kojasov, F. Landis, P. Kwame-Mensah, C.T. Chang, Two-phase pressure [27] M. Vasquez, H.D. Beggs, Correlations for fluid physical properties prediction,
drop in multiple thick- and thin-orifice plates, Exp. Thermal Fluid Sci. 15 JPT 32 (1980) 968–970.
(1997) 347–358. [28] C.H. Whitson, M.R. Brule, Phase behavior, SPE Monograph, Henry L. Doherty
[17] W.D. McCain, The Properties of Petroleum Fluids, PennWell Books, Tulsa, 1990. Series. (2000).