You are on page 1of 12

Flow Measurement and Instrumentation 46 (2015) 218229

Contents lists available at ScienceDirect

Flow Measurement and Instrumentation


journal homepage: www.elsevier.com/locate/flowmeasinst

Design of a Conductance and Capacitance Combination Sensor for


water holdup measurement in oilwater two-phase ow
Hao Wu, Chao Tan n, Xiaoxiao Dong, Feng Dong
Tianjin Key Laboratory of Process Measurement and Control, School of Electrical Engineering and Automation, Tianjin University, Tianjin 300072, China

art ic l e i nf o a b s t r a c t

Article history: Oilwater two-phase ow widely exists in the process of petroleum industry. The liquid holdup mea-
Received 30 November 2014 surement in horizontal pipeline is very important and difcult. In this work, a Conductance and Capa-
Received in revised form citance Combination Sensor (CCCS) system with four conductance rings and two concave capacitance
4 May 2015
plates is designed and validated for its measurement performance of in situ water holdup through dy-
Accepted 22 June 2015
Available online 24 June 2015
namic experiments. A set of fast electronic switches controls the conductance rings and the capacitance
plates alternatively set up each own sensing eld in the same sensing volume. This conguration ensures
Keywords: the water holdup estimation in the range from 0% to 100% regardless of ow direction. A set of quick
Conductance sensor closing valves was used to acquire the in situ holdup for the on-line calibration of the CCCS system. The
Capacitance sensor
theoretical correlations of conductance sensor and capacitance sensor were established to make the real-
Water holdup measurement
time measurement convenient. A real-time measurement method by CCCS system is provided based on
Oilwater two-phase ow
Quick closing valve the fusion of the conductance and the capacitance measurement without ow pattern recognition. This
Sensor fusion method delivers an average error of 1.06% for the CCCS system measuring the water holdup of oilwater
two-phase ow, with a standard deviation of 0.038 and a relative error less than 7 5%.
& 2015 Elsevier Ltd. All rights reserved.

1. Introduction instantaneous changes of phase fraction.


The electrical method is widely used in ow parameters mea-
Oilwater two-phase ow widely exists in nature and industry surement due to its simple structure and fast response. The elec-
process, such as production wells and transportation in petroleum trical method can be divided into conductance method, capaci-
industry and chemical industry. For petroleum industry, the real- tance method, impedance method and electromagnetic method
time phase fraction is important to ensure oil production safety based on their sensing principle and different electrical properties
and exploration economy. Therefore, the measurement of phase of each phase. The conductance method and the capacitance
fraction plays an important role in such cases. Because of the method are the most commonly used for the two-phase ow
difference of viscosity and density of the oil and the water, the measurement.
mixture uid exhibits several ow patterns during the ow pro- There are many structures of electrical sensors, wire-mesh
cess in a horizontal pipe. The ow patterns change with many sensors, as a particular structure of electrical sensors, can obtain
factors including the ow rate of each individual phase, the phase phase fraction and phase distribution through capacitance or
fraction and pipe diameter [1,2]. Accurate water holdup mea- conductance measurement [9]. Since the wire-mesh sensor is an
intrusive sensor, small deformation of the mesh induced by the
surement is therefore affected by the complex ow patterns. Many
uid supercial velocity changes will lead to erroneous results.
methods have been developed to measure the liquid holdup, such
There are some other sensor structures for recognizing the ow
as quick closing valves [1], radiation sensors [3], microwave sen-
patterns and measuring the water holdup, such as parallel-wire
sors [4], wire-mesh sensors, ultrasonic sensors [5,6] and electrical
conductivity probes [10], arc-shaped conductance method with
sensors [7].
the guard electrodes [11,12], conductance probes array [13] and so
Quick closing valve (QCV) is a widely-applied technique in
on.
multi-phase ow study [1,8], which provides a reliable online
Many studies have been carried out for the two-phase ow
calibration of in situ phase fraction. However, it is limited by the
volume fraction measurement by the electrical methods. For
non-continuous measurement thus not applicable to measure the conductance method, continuous phase is electrically conductive,
and non-continuous phase has a different conductivity from the
n
Corresponding author. continuous phase. The phase fraction is obtained by measuring the
E-mail address: tanchao@tju.edu.cn (C. Tan). mixture's conductivity between two sensing electrodes. Fossa

http://dx.doi.org/10.1016/j.owmeasinst.2015.06.026
0955-5986/& 2015 Elsevier Ltd. All rights reserved.
H. Wu et al. / Flow Measurement and Instrumentation 46 (2015) 218229 219

designed a conductance probe for measuring the liquid fraction in realize the fast and accurate acquisition of the conductance and
gasliquid two-phase ows and evaluated its performance. The the capacitance data. Theoretical correlations of the conductance
device consisting of two pairs of electrodes (two ring electrodes sensor and the capacitance sensor have been improved to make
and two plate electrodes) was placed on the internal wall of a the real-time measurement convenient. The CCCS system can
cylindrical duct and ushed to the pipe surface [14]. Then, Devia measure the full range of in situ water holdup without ow pat-
and Fossa designed and optimized an impedance probe for void tern recognition. The installation method and the measurement
fraction measurement for different gasliquid ow patterns [15]. results are not inuenced by the ow direction based on its
Lucas and Mishra proposed a local four-sensor probe to measure symmetrical structure. The non-intrusive water holdup measure-
the swirl inuence for bubble ow in vertical pipe, and analyzed ment method can reduce pressure loss and processing difculty.
the impedance and time sequence of electronic measurement Therefore, the CCCS system has application prospect in petroleum
system [16]. Xu et al. measured the water hold-up of stratied oil industry.
water ow by a conductance array consisting of 24 needle-like
electrodes that mounted on 12 supporting arms arising from the
central shaft [17]. Zhai et al. numerically designed and optimized 2. Sensor structure
the geometry of a ring conductance probe for measuring the
conductance of oilwater mixtures in horizontal pipes [18]. Dang The test section of CCCS is shown in Fig. 1, which consists of
et al. used the conductance sensor to measure the ow rate of oil four stainless alloy ring-electrodes (conductance sensor) and a
gaswater three-phase ow [19]. pair of copper plates (capacitance sensor). All the electrodes are
For the capacitance method, the non-continuous phase should placed on a Plexiglas pipe of 50 mm inner diameter. The con-
have a different relative permittivity compared with the con- ductance rings and the capacitance plates of the CCCS set up each
tinuous phase [20]. Sami et al. found that a concave-plate capa- own sensing eld in the same sensing volume alternatively. The
citance sensor has a higher sensitivity for void fraction measure- switch time is short enough to make sure the uid condition re-
ment in two-phase ow [21]. Huang et al. discussed the con- mains almost the same in the sensing volume during the switches
ductivity effects on capacitance measurements, and concluded between conductance and capacitance. This structure is non-in-
that the conductance effects can be reduced by using semi- trusive and can deal with the incoming ow at either direction.
conductor switches [22]. Xie et al. used nite element model to
prove that the pipe wall is an important factor to the sensitivity 2.1. Conductance sensor
distribution of the concave capacitance sensor [23]. Jaworek and
Krupa increased the excitation frequency up to 80 MHz to solve The four probes conductance sensor is designed to measure the
the capacitance measurement problem when the continuous water holdup in water continuous oilwater two-phase ow. The
phase is conductive [24]. Strazza and Demoria obtained the re- detail of conductance sensor is shown in Fig. 2(a). The ring elec-
lationship between the accurate phase fraction and the axis length
trodes are embedded in the inner wall of a Plexiglas pipe, where
of the capacitance plate for the core-annular oilwater two-phase
E1 and E2 are the exciting electrodes; S1 and S2 are the measuring
ow [25]. Ye and Peng suggested an optimal structure of helical
electrodes. In a pipe of 50 mm inner diameter, the distance be-
capacitance sensor by simulation and experiments for gasliquid
tween E1 and E2 is 200 mm, the distance between S1 and S2 is
two-phase ow in small diameter tubes [26,27]. Kerpel used the
60 mm, the distance between E1 and S1, E2 and S2 is 70 mm, the
capacitance sensors to evaluate the gaswater two-phase ow
width of E1 and E2 is 5 mm, and the width of S1 and S2 is 3 mm.
behaviors in up and downstream of a sharp return bend [28]. Dos
This conguration has been numerically optimized and experi-
Reis et al. compared the advantages and disadvantages of different
mentally tested in previous work [30].
capacitance sensors, such as helical electrodes sensor, concave
The measured electrical impedance across an electrode pair
electrodes sensor and double ring sensor [29].
immersed in a conducting liquid is resistive when the frequency of
However, in the oilwater two-phase ow, phase inversion
the alternating current (AC) exciting signal is high enough (for tap
happens when water holdup ranges from 0% to 100%. When water
water, 10100 kHz) [14]. In this system, the conductance electro-
holdup is large, water is continuous phase, and oil disperses in the
des E1 and E2 are connected to a constant electric current signal
water, and vice versa. Capacitance method can deal with full range
source that generates a 20 kHz square wave of 2 mA peak-to-peak
holdup measurement with careful conguration of the sensing
amplitude (Fig. 2(b)). The electric potential drop increases with the
circuit, but the sensitivity is relatively low for the water con-
overall resistance between E1 and E2 under the constant exciting
tinuous ow, especially for high water holdup situation. Con-
current. Therefore, the voltage collected between S1 and S2
ductance method can deal with the high water holdup situation
(Sensor A) is directly related to the water holdup.
with relatively high sensitivity, but it is incapable to deal with the
low water holdup situation. Since capacitance sensor and con-
ductance sensor are sensitive to phase inversion condition, it is 2.2. Concave capacitance sensor
necessary to design an electrical method to realize water holdup
measurement for full range of oilwater two-phase ow with high A pair of capacitance electrode is designed to measure water
sensitivity. Thus a new Conductance and Capacitance Combination
Sensor (CCCS) system is presented based on the different sensi-
tivity ranges of capacitance and conductance method. It takes the
advantages of the conductance sensor and capacitance sensor in
different ow conditions, and measures the water holdup from 0%
to 100% with high sensitivity. The measurement complementary
information deals with phase inversion, and redundant informa-
tion and data fusion improve the accuracy of water holdup
measurement.
The congurations of both the conductance sensor and the
capacitance sensor of the CCCS system have been optimized. The
measurement system and control software were designed to Fig. 1. Structure of Conductance and Capacitance Combination Sensor (CCCS).
220 H. Wu et al. / Flow Measurement and Instrumentation 46 (2015) 218229

[24]. However, the signal integrity of the high frequency analog


circuit is difcult to guarantee. Because as the frequency of the
circuit increases to higher than 50 MHz, the signal integrity will be
inuenced violently by the materials and the parameters of the
electronic components, the layout of the Printed Circuit Board
(PCB) and the way of routing high-speed traces. The signals on the
PCB will suffer from many problems such as delay, reection,
crosstalk, time sequence and oscillations. When the exciting signal
frequency is not very high, the received signal that has relation-
ship with the value of the capacitance still has a response with the
phase fraction change when the water is the continuous phase
[29,32]. Therefore, we used 1 MHz and 20 V peak-to-peak ampli-
tude sine voltage as the exciting signal in capacitance sensor to
measure the water holdup (Fig. 3(b)).

3. Theoretical correlations

Fig. 2. The conductance sensor conguration and the exciting current wave form. For horizontal oilwater two-phase ow, different ow rates
(a) Conductance probe conguration and (b) the exciting current. cause different ow patterns, which may lead to different theo-
retical correlations for phase fraction estimation. There are six
holdup in oil continuous oilwater two-phase ow. Two copper horizontal ow patterns, i.e. stratied ow (ST), stratied ow
plates were oppositely placed on the outer wall of the Plexiglas with mixing at the interfaces (ST&MI), dispersion of oil in water
pipe to form the sensing volume. and water (O/W&W), dispersion of water in oil and oil in water
When the continuous phase has very low conductivity, the (W/O &O/W), oil in water emulsion (O/W), water in oil emulsion
measured capacitance is linearly correlated with the water holdup. (W/O) [2]. Different ow patterns describe the different ow
However, in the actual oil production or transportation process, states. The observation and classication of the ow pattern are
the conductivity of the water is high, bringing troubles to the es- conducive to analyze the oil continuous state and water con-
timation of the phase fraction through the capacitance measure- tinuous state and the measurement error. Thus, theoretical cor-
ment. An effective way to decrease the inuence of water con- relations are appropriate to calculate measurement results.
ductivity is to increase the axial length L of the electrodes [25].
Considering the relationship between axial length and sensitivity, 3.1. Conductance measurement correlation
the distance between the conductance rings S1 and S2 (where the
capacitance sensor locates), and the shield effect of the capaci- The conductivity of the tap water is about 0.040.05 S/m, while
tance electrodes, L is selected as 30 mm in the system. The shape that of oil is close to zero. When the electric current ows into the
of electrodes is concavely congured with an opening angle sensing volume, the measured voltage between S1 and S2 is pro-
90 so that it is more sensitive than other electrode structures portional to the holdup of water in the volume between these two
[31]. The capacitance electrodes structure is shown in Fig. 3(a). electrodes.
To deal with the conductive water continuous ow situations, In oilwater two-phase ow, the density of oil is smaller than
Jaworek suggested that raising the frequency of exciting signal to water. The oil will ow on the top of the horizontal pipe when the
80 MHz could effectively eliminate water conductivity inuence ow rate is low. For the ST ow pattern, in order to eliminate the
inuence of water conductivity change and normalize the mea-
surement results, a dimensionless voltage V describing the relative
voltage difference is introduced based on the analytical results
[33]
Vw
V=
Vm (1)

where Vw is the measured voltage in the water lled pipe and Vm is


the measured voltage of the oilwater two-phase ow in the test
section.
Another typical phase distribution is the mixture of water
continuous or oil continuous ow with the other phase dispersed
as bubbles entrained in the continuous phase. For the conductance
measurement, oil droplets of different sizes ow in the water lled
pipe. The relation of the water holdup for dispersed ow Ht and
the conductivity can be expressed according to the Maxwell the-
ory [14]
2Ht
m = w
3 Ht (2)

where sm and sw are the conductance of mixture and water


respectively.
Fig. 3. The capacitance electrodes and the exciting voltage wave form. Since V is proportional with sm/sw, Eq. (2) can be transformed
(a) Capacitance electrodes structure and (b) the exciting voltage wave form. as
H. Wu et al. / Flow Measurement and Instrumentation 46 (2015) 218229 221

3V
Ht =
2+V (3)

The dimensionless voltage V has a good match with the water


holdup in stratied ow static calibration. However, in dynamic
calibration, the response characteristic of the conductance probes
is inuenced by ow distribution. And Ht calculated with Maxwell
equation has good linear trend with the water holdup except the
bubble ow. The bubble ow can be tted by a linear equation
Hb = a1 V + b1 (4)

where Hb is the water fraction for bubble ow for the water


holdup larger than 90%, a1 and b1 are the slope and the intercept of
the linear tting parameters which can be obtained by calibration,
respectively.
In the conductance measurement, the relatively strong effect of
phase distribution to the response of the sensor's output has been
considered. The prediction of holdup has deviations, which can be
compensated by
aV + b
Ht* =
2 + cV (5)
*
where Ht is the water fraction for oilwater two-phase ow, a, b
and c are tting parameters. For stratied ow calibration, a 2,
b 0, c0, for dispersed ow, a 3, b0, c 1, and for bubble ow,
a 2a1, bb1, c 0. These parameters will be used to t the Ht*
equation to the theoretical model of each situation mentioned
above.

3.2. Capacitance measurement correlation Fig. 4. Equivalent model and circuit for concave capacitance. (a) Cross section of
the full pipe with water and (b) equivalent circuit.

The relative permittivity of oil is about 2 and the relative per-


mittivity of water is about 80, so the capacitance of oilwater two- relation between the measurement values of conductive water and
phase ow shows different responses when holdup changes. For non-conductive water is obtained. By dening a parameter
the measurement of oilwater two-phase ow with puried wa- |1/Zw1|/|1/Zw2|, then taking into Eq. (6), the revised RCD can be
ter, water holdup should be calculated with relative capacitance expressed as
difference (RCD) to eliminate the stray capacitance
VcmVco
CmCo VcmVco RCD* =
RCD = = Vcw/ Vco (9)
CwCo VcwVco (6)
The trend of RCD* is different between the oil continuous ow
where the Co and Cw are the values of capacitance when the pipe is and the water continuous ow. However, RCD* is linearly depen-
full of oil and water respectively. Vco and Vcw are the measured dent on the water holdup. When the water holdup increased, RCD*
values that are proportional to Co and Cw. will increase in oil continuous ow, but will decrease in water
However, for the oilwater two-phase ow with tap water, the continuous ow. Thus, the measurement results of the concave
conductivity of the water will affect the holdup estimation. The capacitance for the water holdup of the oilwater two-phase ow
RCD no longer correctly calculates the water holdup in low fre- can be calculated by
quency (less than 80 MHz [24]). The ideal Cw is difcult to acquire
[34], because when the pipe is full of conductive water, the Cw Ho = k RCD* + d (10)
measurement is not reliable. The measurement results when the
where k and d are slope and intercept of the linear tting para-
pipe is full of non-conductive water (puried water) can be used to
meters which should be determined separately in the water con-
solve this problem. The impedance analysis for the concave ca-
tinuous ow and oil continuous ow respectively.
pacitance and the equivalent circuit for the full pipe with con-
ductive water and non-conductive water are shown in Fig. 4.
where Cp is the capacitance of the pipe wall. Rw and Cw1 are the
4. Experimental system design
resistance and capacitance of conductive water respectively, and
Cw2 is the capacitance of non-conductive water. The impedance of
In order to obtain the water holdup of the same volume of oil
the full pipe with conductive water and non-conductive water can
water two-phase ow by alternatively using capacitance sensor
be expressed by
and the conductance sensor, the data collection system includes
2 Rw the conductance sensor exciting and signal conditioning module,
Z w1 = +
jC p 1 + jCw1R w (7) the capacitance sensor exciting and signal conditioning circuit, the
switch circuit, and the data acquisition system. The system
2 1 structure is shown in Fig. 5.
Z w2 = + For the conductance sensor, the exciting circuits use a square
jC p jCw2 (8)
wave signal source and a voltage controlled current source (VCCS)
Because Vcm is related to capacitance between electrodes, the which consists of two operational ampliers to realize the 2 mA
222 H. Wu et al. / Flow Measurement and Instrumentation 46 (2015) 218229

Fig. 5. The CCCS system structure.

Fig. 6. The conductance circuit structure.

peak-to-peak current signal of 20 kHz frequency. The corrosion instrumentation ampliers to produce the 20 V peak-to-peak
and deposition caused by electrolytic effects of the conductance voltage of 1 MHz frequency. The measured capacitance Cm is an
electrodes can be minimized by the connection with a high input extremely small value which consists of the capacitance of pipe
impedance electrical circuit with the alternating current. The wall and oilwater two-phase uid. To obtain accurate Cm, the
conductance circuit structure is shown in Fig. 6. exciting signal source Vi is connected with one of the capacitance
The impedance matching is realized in a circuit in the signal- electrodes, and the other capacitance electrode is connected to the
conditioning unit as shown in Fig. 7. A capacitance of 220 nF and a C/V (capacitance to electric voltage) transfer circuit and the signal
resistance of 2.2 M are connected to the core cable to receive the processing circuit. The output voltage by the C/V transfer circuit Vo
effective AC signal. A 30 resistance is connected to the screen can be calculated by the following equation:
cable to avoid the inuence of the stray capacitance between the
jVi R f
cable and the screen. Vo = Cm
For the capacitance measurement, Huang designed a new ca- 1 + jC f R f (11)
pacitance transducer for industrial applications providing a pro- where Rf and Cf are the feedback resistance and feedback capaci-
grammable measurement frequency ranging between 100 kHz tance respectively. The amplitude of output voltage Vo is propor-
and 5 MHz [35]. Dean and Rane designed a digital frequency- tional to Cm because the value of Vi, , Cf and Rf are constant in the
locked loop system for capacitance measurement [36]. In this circuit.
work, an AC based capacitance measurement method, which is When jR f C f > > 1
commonly used in Electrical Capacitance Tomography (ECT) sys-
Cm
tem, is adopted [37,38]. The capacitance measurement circuit Vo = Vi
structure is shown in Fig. 8. Cf (12)
The exciting signal source circuit uses an FPGA to control two
Then, the signal conditioning circuit output voltage Vs, will pass
Direct Digital Synthesizer (DDS) chips, band-pass lters and the
through a phase sensitive demodulation circuit. Another DDS cir-
cuit sources a 20 V Peak-to-peak voltage and 1 MHz frequency sine
wave for demodulation. The output of the demodulation Vd can be
calculated by

Vd = Vs *VDDS = A sin (t + )*B sin (t + )


= AB [cos ( ) cos (2t + + )] (13)

where A, , B and are the amplitude and phase of the Vs and VDDS
respectively. When the Vs and VDDS has the same phase, = :
Vd = AB [1 cos (2t + + )] (14)
Fig. 7. Impedance matching circuit in signal conditioning unit.
H. Wu et al. / Flow Measurement and Instrumentation 46 (2015) 218229 223

Fig. 8. The capacitance measurement circuit structure.

When Vd passes through a low-pass lter, the output signal Vdc interference to the capacitance measurement.
is proportional to the measured capacitance Cm In order to test the repeatability and the stability of CCCS sys-
tem for the conductance and capacitance measurement, the static
Vdc = AB (15)
calibration of laminar ow has been done. For the conductance
In the switch circuit, an FPGA controls the switch sequence sensor, the gaswater two-phase laminar ow calibration is shown
alternatively to stimulate the conductance sensor and the capaci- in Fig. 11(a). For the capacitance sensor, the gasoil two-phase
tance sensor. The switch frequency is 500 Hz so that the acquisi- laminar ow calibration is shown in Fig. 11(b).
tion data of each period consists of sufcient length of both the From the calibration results, the CCCS system shows good
sensitivity in phase fraction response under different two uids
conductance data and the capacitance data for ow conditions
combinations in both the conductance and the capacitance data.
analysis [39]. In order to make the capacitance measurement more
The signal-noise ratio (SNR) of the system is higher than 70 dB
accurate, the T-type switch is used, as shown in Fig. 9. The FPGA
exhibiting a good repeatability and stability.
controls each switch in the sequence that when switches 1 and
2 are turned on, switches 3 and 4 are turned off; when switches
3 and 4 are turned on, switches 1 and 2 are turned off. This pro-
5. Experimental result and analysis
cedure avoids the inuence of the stray capacitance when the
switches are in between on and off status. A series of dynamic calibration experiments have been con-
The data acquisition system is designed with a PXI (Peripheral ducted to test the performance of CCCS in phase fraction mea-
Component Interconnect (PCI) extensions for Instrumentation) surement of oilwater two-phase ow. QCV method was used for
data acquisition card and a graphical interface. The sampling fre- the in situ water holdup to evaluate the CCCS system.
quency of conductance and capacitance is 200 kHz and the sam-
pling time is set 20 s for each experimental condition. The preci- 5.1. Experimental facility
sion of A/D conversion is 16-bit and the sampling rate is 200 kHz
per channel. The software preprocesses the collected signals of The experiments were conducted on the oilgaswater multi-
conductance and capacitance and shows the real-time acquired phase ow experimental facility of Tianjin University, as illustrated
waveform under each experimental condition. in Fig. 12. The total length of the horizontal pipeline, which is
The photos of CCCS sensing system and the sensors are shown manufactured by stainless steel, is approximately 16.6 m with in-
in Fig. 10. The copper guard shield should be set very close to the ternal diameter of 50 mm. It consists of three horizontal parts: two
concave capacitance electrodes. And a copper screen shield covers legs with the length of 7.2 m and 7.3 m respectively, and a hor-
guard shield and electrodes to avoid the electromagnetic izontal U-bend with the length of 2.1 m. A water tower with height

Fig. 9. The switch circuit structure.


224 H. Wu et al. / Flow Measurement and Instrumentation 46 (2015) 218229

Fig. 10. Photos of CCCS system and sensors. (a) CCCS sensing system circuit and
(b) photo of CCCS.
Fig. 11. Calibration of conductance probes and concave capacitance. (a) Calibration
of conductance probes and (b) calibration of concave capacitance.

of 32 m provides stable water pressure for the facility, about


0.32 MPa. The frequency converter has also been built to make the diameter). The uid in the QCV shut down section is about
oil pressure stable, which can be adjusted from 0.1 MPa to 0.4 MPa. 1820 mL, including the uid left in the valves' body. Two mea-
For oilwater two-phase ow experiments, oil and water in the suring cylinders with 1000 mL measuring range and 10 mL scales
oil tank and water tank were pumped into the entrance separately were used to separate and measure the shut down uid, which
through standard ow meters (less than 1% precision on average) gives a measuring precision less than 0.55%.
of each phase installed on oil and water inlet pipes respectively. The experimental conditions are shown in Fig. 14, which is
Oil and water are well mixed at the beginning of the experimental based on Trallero's Flow pattern transition boundaries, where Jso
pipeline by an ejector. At the outlet point of the testing pipeline, and Jsw are the supercial velocity of oil and water respectively.
the mixed uid will ow into the oilwater separation tank to wait A computational method of relative error is introduced as the
for the oil and water separate and then pumped back to oil tank standard of accuracy, which dened as
and water tank. The CCCS and the QCV were installed in the
n Holdupmeas Holdupqcv
downstream of the experimental pipe, leaving sufcient length of 1
pipe for ow pattern development. The water ow rate was be-
Er = 100%
n 1
Holdupqcv (16)
tween 0.04 m/s and 2 m/s, the oil ow rate was between 0.07 m/s
and 2 m/s. The inlet pressure was about 0.2 MPa, and the inlet where n is the sample size, Holdupmeas and Holdupqcv is the water
temperature was about 20 C. holdup value measured by CCCS system and QCV respectively.
The structure of the QCV is shown in Fig. 13. The manual ball
valves are open throughout the experiments. When oilwater 5.2. QCV results
mixed uid ow to the experimental pipeline at a stable status,
the electromagnetic switch valve 1 opens, and electromagnetic During the experiments, the QCV section captures the two-
switch valves 2 and 3 close at the same time. The time of valve phase mixture for three times under each experimental condition,
turning on or off is less than 0.8 s. The check valve prevents the and the average of the three water volume fractions was used for
backward pressure from the downstream. The pressurized gas, calibration. The water holdup results measured by QCV and the
provided by a gas pump, pushed the captured two-phase mixture inlet reference water holdup measured by standard meters are
out of the shut down section (800 m length and 50 mm inner shown in Fig. 15.
H. Wu et al. / Flow Measurement and Instrumentation 46 (2015) 218229 225

Fig. 12. Oilwater two-phase ow experimental facility structure.

Compared with the inlet reference results, the average devia-


tion and the standard deviation of QCV results are 1.43% and 3.23%.
According to Fig. 15(b), most of the deviations are less than 4%.
When the reference water holdup increased from 40% to 80% with
the pattern of ST and ST&MI, the relative deviation is large. Ac-
cording to the QCV measurement results, the slip ratio between
water and oil is not very large. The QCV results can be used to
calibrate the CCCS system.

5.3. CCCS results and analysis

The CCCS system is installed on the upstream pipeline of the


QCV. The measurement results of conductance sensor are shown
in Fig. 16 versus the QCV measured in situ water holdup, which
only includes the water continuous ow. The water holdup is
calculated by Eq. (5).
According to the results, when water holdup is less than 30%,
oil is the continuous phase. The measured electric voltage is very
small which is only correlated with the system impedance. The
Fig. 14. Experimental conditions based on Trallero's ow pattern transition
voltage is not used to calculate the holdup (calculation results are boundaries.
larger than 100%). When water holdup is higher than 30%, water is
the continuous phase, the measurement is proportional to the voltage. RCD* shown in Fig. 17 does not include the ow pattern of
water holdup after the impendence matching in the circuit. The ST and ST&MI at slow ow rate (the mixture ow velocity is less
conductance measurement correlation can t the real water
than 0.3 m/s). According to the results of other ow patterns, it is
holdup value.
shown that when water holdup is less than 30%, oil is the con-
The measurement results of the capacitance sensor are shown
tinuous phase with water dispersed in the oil. The more water
in Fig. 17.
According to Eq. (10), the values of k and d are 0.91 and  0.01 owing inside oil phase, the larger capacitance of the mixture will
in oil continuous ow and  1.85 and 0.86 in water continuous be. Thus, the measurement voltage has an increasing trend when
ow. The value of for RCD* calculation is 0.38, which depends on water holdup is under 30%. When water holdup is higher than
thickness and materials of the pipe wall. The red circles represent 30%, water is the continuous phase with oil dispersed in the water.
the values of RCD* which directly respond to the measurement Since water is conductive in the experiments, the mixed uid

Fig. 13. Structure of QCV.


226 H. Wu et al. / Flow Measurement and Instrumentation 46 (2015) 218229

Fig. 17. Water holdup estimate with the capacitance sensor.

becomes an equipotential body by increasing water holdup, and


measurement circuit loses efciency for the capacitance mea-
surement accordingly. When the pipe is full of water, the mea-
surement circuit cannot measure the capacitance at all, so that the
measurement voltage has a downward trend showing this in-
validation process, which can also report the water holdup change.
According to the measurement results of the CCCS, the con-
ductance sensor can measure the holdup in the ST or ST&MI at
slow ow rate and all other conditions when water holdup is
higher than 30%. For the capacitance sensor, all ow patterns ex-
cept the ST or ST&MI with slow ow rate can be measured. All the
effective measurement results with different ow patterns are
shown in Fig. 18(a) and the measurement error of conductance and
capacitance is shown in Fig. 18(b).
Most measurement error of the conductance sensor and the
capacitance sensor is less than 5% and the average error is less
than 1.8%.

5.4. CCCS system real-time measurement method


Fig. 15. Water holdup results and deviation measured by QCV. (a) Water holdup
measured by QCV and (b) deviation of the QCV results. Considering the conductance sensor and the capacitance sensor
have their own high sensitivity measurement range, these two
measurement methods have redundant information in water
continuous ow and complementary information in oil continuous
ow. To integrate the output of the capacitance sensor and the
conductance sensor, a real-time measurement method for in situ
water holdup in horizontal oilwater two-phase ow by using
CCCS is illustrated in Fig. 19.
The CCCS system needs dynamic calibration before the mea-
surement. The theoretical correlations of each method (Eqs. (5)
and (10)) should be used to calculate the water holdup. For the
conductance sensor, the water holdup calculation results will
higher than 100% when the ow condition is oil continuous ow
(water holdup is less than 30% and ow patterns is not ST ow and
ST&MI ow at low ow rate) and it is a wrong result. So in this
situation, the water holdup calculation of the capacitance sensor
should be adopted. If the water holdup calculation of the capaci-
tance are very close to the value that pipe is full of water, the ow
condition is the ST ow and ST&MI ow at low ow rate. In this
situation, the water holdup calculated with the conductance sen-
sor should be adopted. For the other situation, Eqs. (5) and (10) can
Fig. 16. Water holdup estimate of water continuous ow with the conductance be used to calculate the water holdup measured by conductance
sensor. and capacitance. The data fusion method can be used to calculate
H. Wu et al. / Flow Measurement and Instrumentation 46 (2015) 218229 227

Fig. 19. CCCS system real-time measurement method.

two-phase ow. The two sets of sensors measure the phase frac-
tion of two-phase ow within the same sensing volume. This
conguration could avoid the estimate error caused by the mis-
Fig. 18. Water holdup results measured by conductance and capacitance.
alignment of different sensors.
(a) Measurement results with different ow patterns and (b) measurement error.
The CCCS system was experimentally tested on a multiphase
ow loop. The two sets of sensors respond distinctively in the
water continuous ow and the oil continuous ow. The ow pat-
the water holdup for the different situations [40]. In this experi-
tern also affects the selection of the sensors: for the conductance
ment, the measured object of CCCS system is the same one at the
sensor, the sensitive range includes the ow patterns of ST and
same time, and different weights can be set for the conductance
ST&MI at low ow rate, and all the water continuous ow with
sensor and the capacitance sensor based on the insensitive range.
water holdup larger than 30%; for the capacitance sensor, the
In conductance sensitive range, such as ST&MI ow pattern, the
sensitive range includes all the ow patterns without ST and
weight of conductance is higher than capacitance, and in other
situation, the average weight can be used. The number of weights ST&MI at low ow rate. The average error of either the con-
should be chosen by experiments for the CCCS system. The CCCS ductance sensor or the capacitance sensor measuring water
system data fusion results for water holdup and the error in ex- holdup is less than 1.5% and the relative error is less than 5% in
periments is shown in Fig. 20. most conditions compared with QCV.
The relative error by using the CCCS measurement is lower A real-time measurement method of CCCS is provided by in-
than the conductance sensor and the capacitance sensor measured tegrating the outputs of the two sensors, which obtains an overall
separately compared with Fig. 18. Most measurement error is less estimate of phase fraction within the range from 0% to 100% in
than 5%, average error is 1.06% and standard deviation is 0.038, horizontal oilwater two-phase ow. The data fusion based
which verify the feasibility of CCCS measuring the in situ phase method delivered an average error of 1.06% for the CCCS system
fraction of oilwater two-phase ow in real time. measuring water holdup, with a standard deviation of 0.038 and a
relative error less than 5%.
The proposed structure and method of CCCS is veried to be an
6. Conclusions efcient solution to measure the water holdup in horizontal oil
water two-phase ow, as well as a demonstration of integrating
A conductance and capacitance combined sensor (CCCS) is different sensors for a more convenient installation and compre-
designed to measure the full range of water holdup of oilwater hensive measurement in similar applications in industries.
228 H. Wu et al. / Flow Measurement and Instrumentation 46 (2015) 218229

[6] S. Al-lababidi, Multiphase ow measurement in the slug regime using ultra-


sonic measurements techniques and slug closure model, in: ProQuest, UMI
Dissertations Publishing, Craneld, 2006, p. 169.
[7] T. Yan, D. Sabic, An electrode polarization impedance based ow sensor for
low water ow measurement, Meas. Sci. Technol. 24 (6) (2013) 067002, http:
//dx.doi.org/10.1088/0957-0233/24/6/067002.
[8] M.S. Arney, R. Bai, E. Guevara, et al., Friction factor and holdup studies for
lubricated pipelining-I. Experiments and correlations, Int. J. Multiph. Flow 19
(6) (1993) 10611076, http://dx.doi.org/10.1016/0301-9322(93)90078-9.
[9] M.J. Da Silva, E. Schleicher, U. Hampel, Capacitance wire-mesh sensor for fast
measurement of phase fraction distributions, Meas. Sci. Technol. 18 (7) (2007)
2245, http://dx.doi.org/10.1088/0957-0233/18/7/059.
[10] J.E. Koskie, I. Mudawar, W.G. Tiederman, Parallel-wire probes for measure-
ment of thick liquid lms, Int. J. Multiph. Flow 15 (4) (1989) 521530, http:
//dx.doi.org/10.1016/0301-9322(89)90051-7.
[11] G. Costigan, P.B. Whalley, Slug ow regime identication from dynamic void
fraction measurements in vertical airwater ows, Int. J. Multiph. Flow 23 (2)
(1997) 263282, http://dx.doi.org/10.1016/S0301-9322(96)00050-X.
[12] C.H. Song, M.K. Chung, H.C. No, Measurements of void fraction by an improved
multi-channel conductance void meter, Nucl. Eng. Des. 184 (23) (1998)
269285, http://dx.doi.org/10.1016/S0029-5493(98)00202-7.
[13] J.G. Flores, X.T. Chen, C. Sarica, et al., Characterization of oilwater ow pat-
terns in vertical and deviated wells, SPE Prod. Facil. 14 (2) (1999) 102109,
http://dx.doi.org/10.2118/56108-PA.
[14] M. Fossa, Design and performance of a conductance probe for measuring the
liquid fraction in two-phase gasliquid ows, Flow Meas. Instrum. 9 (1998)
103109, http://dx.doi.org/10.1016/S0955-5986(98)00011-9.
[15] F. Devia, M. Fossa, Design and optimisation of impedance probes for void
fraction measurements, Flow Meas. Instrum. 14 (45) (2003) 139149, http:
//dx.doi.org/10.1016/S0955-5986(03)00019-0.
[16] G.P. Lucas, R. Mishra, Measurement of bubble velocity components in a swir-
ling gasliquid pipe ow using a local four-sensor conductance probe, Meas.
Sci. Technol. 16 (3) (2005) 749, http://dx.doi.org/10.1088/0957-0233/16/3/018.
[17] W.F. Xu, L.J. Xu, Z. Cao, et al., Normalized least-square method for water hold-
up measurement in stratied oilwater ow, Flow Meas. Instrum. 27 (10)
(2012) 7180, http://dx.doi.org/10.1016/j.owmeasinst.2012.05.002, Xu W., Xu
L., Cao Z., et al..
[18] L.S. Zhai, N.D. Jin, Y.B. Zong, et al., The development of a conductance method
for measuring liquid holdup in horizontal oilwater two-phase ows, Meas.
Sci. Technol. 23 (2) (2012) 025304, http://dx.doi.org/10.1088/0957-0233/23/2/
025304.
[19] R.R. Dang, D.S. Zhao, L.P. Li, et al., The ow measurement of multiphase ow
based on the conductance sensor, Mod. Instrum. 01 (04) (2012) 12, http://dx.
doi.org/10.4236/mi.2012.14005.
[20] C.N. Strizzolo, J. Converti, Capacitance sensors for measurement of phase vo-
lume fraction in two-phase pipelines, IEEE Trans. Instrum. Meas. 42 (3) (1993)
726729, http://dx.doi.org/10.1109/19.231597.
[21] M.S.A. Abouelwafa, E.J.M. Kendall, The use of capacitance sensors for phase
percentage determination in multiphase pipelines, IEEE Trans. Instrum. Meas.
29 (1) (1980) 2427, http://dx.doi.org/10.1109/TIM.1980.4314856.
[22] S.M. Huang, R.G. Green, A.B. Plaskowski, et al., Conductivity effects on capa-
citance measurements of two-component uids using the charge transfer
method, J. Phys. E: Sci. Instrum. 21 (6) (1988) 539, http://dx.doi.org/10.1088/
Fig. 20. Water holdup results and error measured by CCCS. (a) Water holdup 0022-3735/21/6/003.
measured by CCCS and (b) CCCS results error. [23] C.G. Xie, A.L. Stott, A. Plaskowski, et al., Design of capacitance electrodes for
concentration measurement of two-phase ow, Meas. Sci. Technol. 1 (1)
(1990) 6578, http://dx.doi.org/10.1088/0957-0233/1/1/012.
Acknowledgment [24] A. Jaworek, A. Krupa, Gas/liquid ratio measurements by rf resonance capaci-
tance sensor, Sens. Actuators A: Phys. 113 (2) (2004) 133139, http://dx.doi.
org/10.1016/j.sna.2004.02.006.
The authors appreciate the support from the National Natural [25] D. Strazza, M. Demori, V. Ferrari, et al., Capacitance sensor for hold-up mea-
Science Foundation of China (Nos. 61227006 and 61473206) and surement in high-viscous-oil/conductive-water core-annular ows, Flow
Meas. Instrum. 22 (5) (2011) 360369, http://dx.doi.org/10.1016/j.
the Science and Technology Innovation Plan of Tianjin
owmeasinst.2011.04.008.
(13TXSYJC40200). [26] J. Ye, L. Peng, W. Wang, et al., Helical capacitance sensor-based gas fraction
measurement of gasliquid two-phase ow in vertical tube with small dia-
meter, IEEE Sens. J. 11 (8) (2011) 17041710, http://dx.doi.org/10.1109/
JSEN.2010.2095004.
References [27] J. Ye, L. Peng, W. Wang, et al., Optimization of helical capacitance sensor for
void fraction measurement of gasliquid two-phase ow in a small diameter
[1] G. Oddie, H. Shi, L.J. Durlofsky, et al., Experimental study of two and three tube, IEEE Sens. J. 11 (10) (2011) 21892196, http://dx.doi.org/10.1109/
phase ows in large diameter inclined pipes, Int. J. Multiph. Flow 29 (4) (2003) JSEN.2011.2116115.
527558, http://dx.doi.org/10.1016/S0301-9322(03)00015-6. [28] K. De Kerpel, T. De Keulenaer, S. De Schampheleire, et al., Capacitance sensor
[2] J.L. Trallero, C. Sarica, J.P. Brill, A study of oil/water ow patterns in horizontal measurements of upward and downward two-phase ow in vertical return
pipes, SPE Prod. Facil. 12 (3) (1997) 165172, http://dx.doi.org/10.2118/36609- bends, Int. J. Multiph. Flow 64 (2014) 110, http://dx.doi.org/10.1016/j.
PA. ijmultiphaseow.2014.04.003.
[3] Z.B. Li, Y.X. Wu, D.H. Li, Gamma-ray attenuation technique for measuring void [29] E. Dos Reis, D. da Silva Cunha, Experimental study on different congurations
fraction in horizontal gasliquid two-phase ow, Nucl. Sci. Tech. 18 (2) (2007) of capacitive sensors for measuring the volumetric concentration in two-
7376, http://dx.doi.org/10.1016/S1001-8042(07)60022-0. phase ows, Flow Meas. Instrum. 37 (2014) 127134, http://dx.doi.org/
[4] E. Nyfors, Industrial microwave sensorsa review, Subsurf. Sens. Technol. 10.1016/j.owmeasinst.2014.04.001.
Appl. 1 (1) (2000) 2343, http://dx.doi.org/10.1023/A:1010118609079. [30] Y.Y. Shi, F. Dong, C. Tan, Conductance probe for the measurement of liquid
[5] M. Meribout, N.Z. Al-Rawahi, A.M. Al-Naamany, et al., A multisensor intelligent volume fraction and axial velocity in gasliquid two phase ow, in: Pro-
device for real-time multiphase ow metering in oil elds, IEEE Trans. In- ceedings of the 9th International Conference on Electronic Measurement &
strum. Meas. 59 (6) (2010) 15071519, http://dx.doi.org/10.1109/ Instruments, ICEMI '09, Beijing, China, 2009, doi: 10.1109/
TIM.2009.2028210. ICEMI.2009.5274438.
H. Wu et al. / Flow Measurement and Instrumentation 46 (2015) 218229 229

[31] T. Chen, N. Bowler, J.R. Bowler, Analysis of arc-electrode capacitive sensors for [37] W.Q. Yang, Hardware design of electrical capacitance tomography systems,
characterization of dielectric cylindrical rods, IEEE Trans. Instrum. Meas. 61 (1) Meas. Sci. Technol. 7 (3) (1996) 225, http://dx.doi.org/10.1088/0957-0233/7/3/
(2012) 233240, http://dx.doi.org/10.1109/TIM.2011.2157573. 003.
[32] A. Zhao, N.D. Jin, L.S. Zhai, et al., Liquid holdup measurement in horizontal oil [38] W.Q. Yang, T.A. York, New AC-based capacitance tomography system, IEE Proc.
water two-phase ow by using concave capacitance sensor, Measurement 49 Sci., Meas. Technol. 146 (1) (1999) 4753, http://dx.doi.org/10.1049/ip-
(2014) 153163, http://dx.doi.org/10.1016/j.measurement.2013.11.036. smt:19990008.
[33] P. Andreussi, A. Di Donfrancesco, M. Messia, An impedance method for the [39] C. Tan, P.F. Li, W. Dai, et al., Characterization of oilwater two-phase pipe ow
measurement of liquid hold-up in two-phase ow, Int. J. Multiph. Flow 14 (6) with a combined conductance/capacitance sensor and wavelet analysis, Chem.
(1988) 777785, http://dx.doi.org/10.1016/0301-9322(88)90074-2. Eng. Sci. 134 (2015) 153168, http://dx.doi.org/10.1016/j.ces.2015.04.046.
[34] M. Demori, V. Ferrari, D. Strazza, et al., A capacitive sensor system for the [40] C. Tan, H. Wu, F. Dong, Horizontal oilwater two-phase ow measurement
analysis of two-phase ows of oil and conductive water, Sens. Actuators A: with information fusion of conductance ring sensor and cone meter, Flow
Phys. 163 (1) (2010) 172179, http://dx.doi.org/10.1016/j.sna.2010.08.018. Meas. Instrum. 34 (2013) 8390, http://dx.doi.org/10.1016/j.
[35] S.M. Huang, J. Fielden, R.G. Green, et al., A new capacitance transducer for owmeasinst.2013.08.006.
industrial applications, J. Phys. E: Sci. Instrum. 21 (1988) 251, http://dx.doi.org/
10.1088/0022-3735/21/3/002.
[36] R.N. Dean, A.K. Rane, A digital frequency-locked loop system for capacitance
measurement, IEEE Trans. Instrum. Meas. 62 (4) (2013) 777784, http://dx.doi.
org/10.1109/TIM.2013.2240092.

You might also like