Measurement 124 (2018) 81–90

Contents lists available at ScienceDirect

Measurement
journal homepage: www.elsevier.com/locate/measurement

Void fraction measurement using concave capacitor based sensor – T

Analytical and experimental evaluation
⁎
Amit Pal , B. Vasuki
Department of Instrumentation and Control Engineering, National Institute of Technology, Tiruchirappalli, India

A R T I C LE I N FO A B S T R A C T

Keywords: Two-phase flow in pipelines is very common in the oil and gas industries. Void fraction is one of the important
Concave capacitive sensor parameters in two-phase flow. Measurement of void fraction plays an important role in various areas of in-
Two-phase flow dustries. Two types of measurement techniques are commonly used in electrical methods for macroscale range-
Void fraction variable capacitance type and variable conductance type. A comparatively low cost and non-invasive design of
Electrical model
the concave capacitive sensor is analytically evaluated and experimentally validated to measure the void fraction
Non-invasive technique
Uncertainty analysis
inside the non-metallic pipe. The design is effective to reduce the fringing effect. The electrical model of the
concave capacitance sensor is analyzed, which involves all the parasitic capacitances along with the cable ca-
pacitances. The capacitance of the sensor varies for the whole range of void fraction (0–100%). If the two phases
are distilled water and air, the capacitance value varies from 94.0134 pF to 3.4377 pF experimentally for void
fraction variation of 0–100%. Uncertainty analysis for the proposed sensor design along with the uncertainty
budget is presented, for the whole range of measurement. The standard uncertainty is less than 0.97 pF for the
whole range of measurement. The procedure for analysis is also discussed in this paper.

1. Introduction
Nowadays void fraction measurement in two-phase flow become
necessary in various fields such as petrochemical industries, oil in-
dustries, etc. Sometimes it is very necessary for micro range applica-
tions, for example in drugs industries [1]. Different techniques are used
there to measure the void fraction of two immiscible fluid phases inside
the pipe like invasive methods (wire mesh method, optical probe
methods [2], etc.) and non-invasive methods (Gamma-ray attenuation
method, electrical methods). Void fraction measurement is also im-
portant at some crucial processes like heat transfer/exchange in nuclear
power plant [3–5]. Two-phase flow referred as the flowing of two dif-
ferent and immiscible fluids, which may be a liquid-liquid mixture or
gas-liquid mixture. The amount of each fluid in the mixture can be
determined by measuring void fraction in cross-section of a pipe. By
analyzing the value of void fraction in any pipe, one can directly per-
form the qualitative and quantitative analysis of the mixture of the
fluids flowing inside the pipe. Currently, researchers are working in the
area of void fraction measurement for improvement in design, to reduce
the output error and to find the relation between the designed and the
actual output using mathematical models.
To measure the void fraction, there are two types of methods, in-
vasive method and non-invasive method. The selection of measurement
method depends on the type of fluids and type of electrodes used for
measurement. In the invasive type of method, the electrodes are di-
rectly in contact with the fluid inside the pipe, and the impedance is
measured [6] like wire mesh sensor [7–9]. Earlier researchers have
analyzed that the wire mesh sensor can be used for multiphase flow also
[8]. X. Chen et al. have discussed about designing of coaxial type ca-
pacitive sensors to measure the void fraction [10]. Researchers have
done Finite Element Method (FEM) analysis for optimization of the
geometry of capacitive sensor. Water hold up is analyzed and the result
is compared with the set values from Quick Closing Valve (QCV). It has
higher sensitivity because the inner electrode is in direct contact with
the fluids inside. For flow characterization, optical method can also be
used. In this method, the mesh-like structure of optical fibers is placed
perpendicular to the flow direction [11]. The simpler method to mea-
sure the void fraction is to just put two electrodes inside the flow, and
by measuring the conductance or impedance between these probes, the
value of void fraction can be determined. As in this method, the probes
are in direct contact with the fluid, it is a very accurate method for both
the non-conductive and conductive type of fluids. Sometimes in case of
conductive liquid, the combination of conductance probe and concave
electrode capacitance sensors are also used [12].
Currently, researchers are focussed on the non-invasive type mea-
surement of void fraction because of its various advantages. As it does

⁎
Corresponding author.
E-mail addresses: amitpal.pdh@gmail.com (A. Pal), bvas@nitt.edu (B. Vasuki).

https://doi.org/10.1016/j.measurement.2018.04.010
Received 7 July 2017; Received in revised form 15 February 2018; Accepted 4 April 2018
Available online 06 April 2018
0263-2241/ © 2018 Elsevier Ltd. All rights reserved.
A. Pal, B. Vasuki
not have any direct contact with the fluid inside, it is very safe, and it
can be used for a variety of liquids (conductive or non-conductive).
Also, the design of the outer electrodes can be changed or modified,
based on the flow pattern inside the pipe. In non-invasive type mea-
surement researchers are focussing onto two types of methods, one is to
use X-rays [13] or Gamma rays [14] to measure the void fraction, and
other is to put some metallic electrodes, outside the pipe and make a
concave capacitive sensor. Capacitance-based sensors are more prac-
tical in most types of fluids because of its simple design and accurate
measurement. Mostly three types of electrodes are used for measure-
ment of the void fraction; helical electrodes, double ring type electrode
and concave type electrode. Most of designs and techniques, considered
for void fraction measurement is based on concave electrode type ca-
pacitance design due to its simple construction and wide applications
[15–18]. Research has also been carried out for optimizing the geo-
metry of the concave electrode based capacitive sensor using FEM
technique [19]. A comparison has been reported between double ring
type and concave electrode type capacitive sensors for different con-
ditions and shows that the double ring type sensor has better sensitivity
and the concave capacitive sensor has better linearity [20,21]. It has
also been reported that the performance of the ring type sensor doesn’t
change much according to the air/oil distribution and concave elec-
trode-type sensor performs better for the detection of flow regime [21].
Comparison between helix type and concave electrode based sensor is
also reported in [22] and it is shown that the concave electrode capa-
citance sensor has better linearity. Double helix capacitance-based
sensor along with the guard electrode is designed and analyzed for
horizontal two-phase flow, and it is calibrated based on parallel wire
capacitance probe [23]. Different techniques are used to measure this
capacitance such as phase shift detection [24], trans-impedance am-
plifier [25], etc. Multiple concave electrodes are used to form more
than two capacitances to increase the sensitivity and to get the 3D
image of the void inside the pipe to achieve better results, and the
method is known as Electrical Capacitance Tomography [26].
The designing and development of the concave electrode based
capacitive sensor have been discussed very briefly for air-water void
fraction measurements, reported in [25,27]. In the first part, hardware
has been designed along with the guard electrodes, and the cable ca-
pacitances have also being considered in the equivalent circuit. ARM
microcontroller and excitation circuits have been used for signal con-
ditioning. The capacitance value is nearly linear with respect to the
void fraction. In second part FEM simulation based calibration method
is reported for the same sensor.
In this work, a more sensitive and simple design, a concave elec-
trode based capacitive sensor for measuring void fraction is discussed.
Instead of considering only two concave electrodes and two guard
electrodes across the receiver electrode, a different design is presented
here with four concave electrodes and four guard electrodes. These
concave electrodes are used to form two capacitors, and between every
two electrodes, a grounded guard electrode is present. All setup is en-
closed with a grounded outer metallic sheet. In this proposed design a
highly precise non-invasive sensor is described, where the exciter
electrode and receiver electrode can be interchanged. It is a low-cost
sensor, and its implementation is also very easy. In the equivalent
model of the capacitors, the cable capacitances with respect to ground
are also considered. The capacitance changes in the full range of the
void fraction, from 0% to 100%. The output of the sensor (capacitances
values) is evaluated theoretically using mathematical calculations for
different void fractions and same is experimentally validated for dis-
tilled water and air as the two phases. As two individual capacitors are
measuring the same void fraction in the acrylic pipe, its sensitivity is
very high and the output is very less affected by the environmental
effects due to guard electrodes and outer shield. Orientation of the
sensor plays a major role in introducing error while measuring void
fraction as the output i.e. change in the capacitors depends on the
combination of the individual capacitances forming inside the pipe.
82
Measurement 124 (2018) 81–90
2. Design and mathematical evaluation of the concave capacitive
sensor
2.1. Geometry of sensor
The sensor is designed to measure the void fraction of the two-phase
liquid mixture flowing through a pipe. As it is a non-invasive method,
the sensor will be fixed outside the pipe, and it is not in direct contact
with the fluid flowing inside. An acrylic pipe of outer diameter ‘Dp’ and
inner diameter ‘dp’ and length ‘Lp’ is taken such that Lp ≫ he, here ‘he’ is
the length of the concave electrodes. The thickness of the pipe wall is
‘tp’ (2tp = Dp − dp). Four concave electrodes (A, B, C, and D) are placed
on the opposite sides of the pipe such that they will form two parallel
plate capacitors. The length of the all four electrodes are same, i.e. ‘he’
and with a thickness of ‘te’. The width of the electrodes (le) was decided
in such a way that the capacitors will fully cover the whole circular area
inside the pipe. So if electrodes are mounted on the outer surface of the
pipe, then the angle which the two ends of the electrode are making
with the center of the pipe (θ) is slightly less than 90°. The width of the
π
electrode can be given by equation (le) = 4 dp . As these concave elec-
trodes are placed symmetrically on the outer surface of the pipe making
two parallel plate capacitors (CAC and CBD), there is still some distance
left between every two plates. This gap between electrodes makes an
acute angle θ′ with the center of the pipe. This angle can be given by
( ). In this gap, metallic (copper) guard elec-
expression θ′ = 90 1− D
dp
p
trodes [27] are placed. The two ends of the guard electrode make an
angle θ″ with the center of the pipe. This angle θ″ chosen in such a way
that θ″ < θ′ (approx. 1° difference from both the sides). In the end,
these guard electrodes are connected to the ground potential. So, four
concave copper electrodes and four guard electrodes are placed on the
outer surface of the acrylic pipe symmetrically. This whole setup is then
covered with an outer metallic (copper) grounded shield (O) of dia-
meter ‘o’, such that o > Dp. To get the output of the capacitances CAC
and CBD, four 75 Ω coaxial cables [25] with BNC connectors are used.
One cable for each electrode. The woven copper shields of these cables
are also grounded. The complete geometry of the concave capacitance
sensor is shown in Fig. 1
2.2. Design of the sensor: Electrical model
In the above geometry of sensor (Fig. 1), four concave electrodes A,
B, C, and D are used. Among these electrodes, two electrodes A and B
acts as exciter electrodes and other two electrodes C and D act as
measuring electrodes. As these electrodes are placed very near to each
other, to reduce the fringe field effect at the edges, some metallic
plates/electrodes (copper plates) are placed in between every two
electrodes. These guard electrodes are also grounded, to minimize the
stray capacitance by minimising the stray electric field. Also, the whole
setup is shielded by a sheet of copper, and the sheet is also connected to
ground potential. The cross-sectional view of the concave capacitive
sensor is explained in Fig. 2.
In this configuration of capacitors, we get many capacitors between
electrodes A and C, similarly between electrodes B and D and also we
got some stray capacitances with respect to ground. It is shown as an
electrical model in Fig. 3(a) and its simplified electrical model is shown
in Fig. 3(b).
Here CAO, CBO, CCO, CDO are the capacitance between concave
electrodes (copper) A, B, C, D and the grounded outer cylinder (O).
These capacitances can be given by the following expression:
πdp
4
he
CAO = CBO = CCO = CDO = ε 0 εg ∗
o−(Dp + t e) (1)
where ɛ0 is absolute permittivity (in F/m), ɛg is relative permittivity of
gas, dp is inner diameter of the pipe (acrylic pipe) (in m), he is length of
A. Pal, B. Vasuki
Fig. 1. Geometry of the concave capacitance sensor.
Fig. 2. Cross-sectional view of the concave capacitive sensor.
the concave electrodes (in m), o is the diameter of the outer shielding
metallic cylinder (in m), Dp is the outer diameter of the pipe (acrylic
pipe) (in m) and te is thickness of the concave electrode (in m). The
above capacitance can be measured experimentally.
Capacitances CAGr1 and CBGr1 are the capacitances between concave
electrodes A and B with the guard electrode 1. Similarly, capacitances
CAGr2 and CDGr2, capacitances CCGr3 and CDGr3 and capacitances CCGr4
and CBGr4 are the capacitances between concave electrodes A and D, C
and D and C and B with the guard electrode 2, guard electrode 3 and
guard electrode 4 respectively. All these capacitances (CAGr1, CBGr1,
CAGr2, CDGr2, CCGr3, CDGr3, CCGr4 and CBGr4) are stray capacitances, and
its values depend on the geometry and various other parameters. CAP,
CBP, CCP, and CDP are capacitances between the acrylic pipe inner wall
and A, B, C and D concave electrodes respectively. The value of these
capacitances are equal due to the symmetric geometry
(CAP = CBP = CCP = CDP).
CAC is the capacitor between the inner wall of acrylic pipe from A
and C concave electrode sides. Similarly capacitor between the inner
Fig. 3. (a) Electrical model of concave capacitance sensor without cable capacitance, (b) Simplified electrical model of concave capacitance sensor without cable
capacitance.
83
Measurement 124 (2018) 81–90
wall of acrylic pipe from B and D concave electrode sides denoted as
CBD. From Fig. 3(a) it is clear that capacitances CAGr2, CAGr1, and CAO
are connected in parallel. So let capacitance CAG equal to the parallel
combination of these three capacitances. In a similar way, the parallel
combinations of three capacitances CBGr4, CBGr1, CBO, capacitances
CCGr4, CCGr3, CCO and capacitances CDGr2, CDGr3, CDO are represented as
CBG, CCG and CDG respectively. Capacitance CAP is in series with capa-
citance CCP, and capacitance CBP is in series with capacitance CDP. It is
clear that the capacitance values of CAP, CBP, CCP and CDP are equal
(CAP = CBP= CCP = CDP). Let consider the series combination of capa-
citances CAP − CCP and capacitances CBP - CDP are CP1 and CP2 respec-
tively. Fig. 3(b) shows the simplified and complete electrical model of
the concave capacitance sensor without considering the capacitance of
connecting cables.
For measuring capacitance, connection between the electrodes and
external measurement systems has to be done by some wires. These
wires can also contribute in some stray capacitances. So here coaxial
cables are used, and the outer woven copper shield is grounded. So we
get a stray capacitance between the wire and the grounded outer woven
copper shield only. Fig. 4(a) demonstrated the cable capacitance along
with concave capacitance sensor [27]. Here CCA, CCB, CCC, CCD are the
capacitance of the cables which are connecting electrodes A, B, C and D
respectively with respect to ground.
The electrical model of the concave capacitances along with the
cable capacitances is shown in Fig. 4(b). It is clear from Fig. 4(b) that
the capacitance CAG is in parallel with capacitance CCA. Let capacitance
CA equals to the parallel combination of capacitances CAG and CCA.
Similarly, the parallel combination of the capacitances CBG and CCB,
capacitances CCG and CCC and capacitances CDG and CCD are CB, CC and
CD respectively. The overall simplified electrical model of the concave
capacitance sensor is shown in Fig. 5.
In this simplified electrical model, except the value of CAC and CBD,
all the other parasitic capacitance values are constant and not vary with
time, in a constant temperature environment. The change in
A. Pal, B. Vasuki
Fig. 4. (a) Schematic diagram of concave capacitor sensor with cable capacitances, (b) Electrical model of concave capacitance with cable capacitance.
Fig. 5. Simplified electrical model of concave capacitance sensor with cable
capacitance.
temperature will affect the sensor design and measurement due to two
major parameters. First, the dielectric constants of the two fluids will
change with a change in temperature, and the coefficient of thermal
expansions of electrode material (copper), pipe material (acrylic) and
fluids (distilled water and air) has to consider, to analyse the tem-
perature effect on void fraction measurement. However, in present
work, constant temperature environment maintained so only capaci-
tance value CAC and CBD will vary according to variation in the void
fraction.
2.3. Analytical evaluation of concave capacitive sensor (for void fraction
0% and 100%)
As per above electrical model as discussed in Section 2.2, it is clear
that only capacitance CAC and CBD is variable and all the other capa-
citances are constant. Now capacitances CAC and CBD are evaluated for
the two extreme conditions (i.e. void fraction = 0% and void frac-
tion = 100%). Due to the same geometrical design of the concave ca-
pacitance sensor (as shown in Fig. 2), the capacitances CAC and CBD are
equal in these extreme conditions of void fractions.
The capacitance value of CAC and CBD is calculated using the ana-
lytical method when the void fraction is 100% (i.e. pipe is empty). As at
100% void fraction, both the capacitances are equal, so first CAC is
calculated alone. Fig. 6 shows the geometry of the concave capacitance
sensor when only CAC is considered.
Distance (z) between two points on the opposite inner wall of the
pipe in front of two concave electrodes A and C with the same value of x
is given by the following expression:
z= (y2−y1)2
(2)
As the center of the circle is considered at the origin (0, 0) and both
the coordinates of the ends of the straight vertical line having the same
value of ‘x-coordinates’, their ‘y-coordinates’ are equal in magnitude but
different in sign, (y2 = −y1). By putting this value of y2 in Eq. (2), the
distance between two points (z) equal to twice of the value of y1 i.e.
z= 2y1. At given value of ‘x-coordinate’, the value of ‘y-coordinate’ can
84
Measurement 124 (2018) 81–90
Fig. 6. Geometry of the concave capacitance sensor considering only capaci-
tance CAC.
be calculated by using the equation of the circle, such as:
2
dp
y1 = ± ⎛ ⎞ −x2
⎜ ⎟
⎝2⎠ (3)
By using the above value of y1 (as it is the function of ‘x’), the dis-
tance between two coordinates (z) which are on opposite inner wall of
the pipe in front of concave electrodes with the same value of x-co-
−d d
ordinate can be calculated. The limits of ‘x’ in this function is p to p .
2 2 2 2
The z is expressed as a function of ‘x’ in following Eq. (4):
2
dp
z= 2 ⎛ ⎞ −x2
⎜ ⎟
⎝2⎠ (4)
To calculate the capacitance between concave electrodes A and C,
the capacitor CAC is divided into infinite parallel plate capacitors. All of
these capacitors are connected in parallel combination. These all par-
allel plate capacitors are identical to each other except the distance
between them which is changing continuously. The summation of all
these capacitances gives the approximate values of CAC. It is evaluated
by dividing the above-said capacitor into a known number of parts and
approximating these parts as parallel plate capacitors with equal area.
The individual capacitance is calculated and at the end, all these small
capacitances are added together (for this calculation MATLAB R2016b
is used). In this manner, the value of capacitance CAC is calculated when
the void fraction is 100%. It is already explained that at extreme con-
ditions of the void fraction, the value of capacitance CAC is equal to CBD.
To find capacitances CAC and CBD for the other extreme condition of
void fraction i.e. 0% void fraction, the values can be multiplied by the
corresponding relative permittivity of the second phase.
The capacitances CP1 and CP2 are the series combination of the two
capacitances formed by the inner surface of the acrylic wall and each
concave electrodes of the capacitances CAC and CBD respectively. The
A. Pal, B. Vasuki
πd
width of both the electrodes are same, i.e. l e = 4P and the distance
between both the electrodes is also same at all the points, i.e. ‘dp’. So
these capacitors CP1 and CP2 also act as a parallel plate capacitor and it
is expressed by Eq. (5):
1 πdp h e
CP 1 = CP 2 = ε 0 εp
2 4 tp
where ɛp is relative permittivity of the pipe material (for acrylic ɛp = 3)
and ‘tp’ is the thickness of the pipe wall (in m) and ‘he’ is the length (in
m) of the concave electrodes.
2.4. Estimation of capacitances between the concave electrodes for different
void fractions
Capacitances for void fractions 0% and 100% is already known, now
further calculations are done to determine the values of the capaci-
tances CAC and CBD at different void fractions (like 10%, 25%, 50%,
75%, 90%).
For 10% void fraction:
Further, the values of capacitances CAC and CBD are individually
calculated for 10% void fraction. First to calculate capacitance CAC, the
area of the gas (AG) is calculated in the cross-section of the pipe as
shown in Fig. 7(a) and (b) using the void fraction formula [28]
(i.e. void fraction =
area of gas (AG)
total cross sectional area (A G + AL) ). Then using this area of
gas (AG), the remaining area of liquid (AL) is calculated.
The relation between the liquid area (AL) and height of the liquid
(hL) is given by following equation Eq. (6) [29]:
2 clear from Table 2, that the value of CTOTAL increases with increase in
dp dp/2−hL ⎞ dP d
Area of liquid (AL) = ⎛ ⎞ cos−1 ⎜⎛
⎜ ⎟ ⎟−⎛ −hL ⎞ 2 P hL−h2L
2
⎝ ⎠ ⎝ dp /2 ⎠ ⎝ 2 ⎠ 2
By putting liquid area (AL) in the above Eq. (6), and using Newton's
Method (for nonlinear systems of equations), the height of the liquid
(hL) from the bottom of the pipe is calculated. The capacitances C1 and
C2 are calculated by using the same method which is used to calculate
the CAC (in Section 2.3). In the previous section, the distance (z) is the
distance between two points on the curve, but here distance (z) is the
distance between one point on the curve and other on the straight line
formed by the liquid level. From the Fig. 7(a) it is well illustrated that
the CAC is the series combination of capacitances C1 (liquid as a di-
electric medium) and C2 (gas as a dielectric medium).
For evaluation of CBD, just the limits of ‘y’ has to changed according
to the two different ranges of ‘y’ (one range for liquid as a dielectric
Fig. 7. (a) Cross-sectional view of concave capacitance sensor for calculation of CAC (b) Cross-sectional view of concave capacitance sensor for calculation of CBD.
Measurement 124 (2018) 81–90
medium and another has gas as a dielectric medium), and the same
method is used (as explained in Section 2.3) for evaluation. The cross-
sectional view of capacitance CBD is given in Fig. 7(b). Similarly the
capacitances CAC and CBD are calculated for void fractions 25%, 50%,
75% and 90%.
(5)
2.5. Complete electrical model and evaluation of the capacitance values
The electrical model for the capacitances formed due to different
void fractions is illustrated in following figure Fig. 8.
From Fig. 8, it is clear that the capacitance CAC is the equivalent
capacitance of the series combination of C1 and C2. Also, the capaci-
tance CBD is the equivalent capacitance of the parallel combination of
C3 and C4.
Considering the medium inside the acrylic pipe is distilled water as
liquid and air as the gaseous medium, the values of the capacitances C1,
C2, C3, and C4 are given in following Table 1.
All the parasitic capacitance with respect to ground (CA, CB, CC, and
CD) can’t be calculated theoretically, as it depends on various arbitrary
conditions. So for theoretical calculation, they are considered as neg-
ligible. Using values of capacitance CP1, CP2, CAC and CBD, the net ca-
pacitance between the terminals A-C and terminals B-D are calculated.
CP1 and CP2 are constant for the whole range of void fraction (CP1 = CP2
are 98.05 pF). CNET AC is the net capacitance between concave copper
electrodes A and C, and it is given by the series combination of CAC and
CP1. CNET BD is the net capacitance between concave copper electrodes B
and D and given by the series combination of CBD and CP2. The parallel
connection of two individual net capacitances (CTOTAL)
(CTOTAL = parallel combination of CNET AC and CNET BD) gives the best
varying result with respect to the variation in the void fraction. It is
the level of the liquid inside the pipe (CTOTAL increases as void fraction
decreases). The solved values of CNET AC and CNET BD are given in fol-
(6) lowing Table 2.
For distilled water-air two-phase flow, the results of the capacitive
sensor results are shown in Table 2, and it is clear that the value of
capacitance is inversely proportional to the void fraction. The same
theoretical analysis is done for a number of different two-phase flow
fluids (Fuels/industrial/agricultural fluids (various oils) vs distilled
water or air) pairs. For analysis liquid-gas and liquid-liquid both types
of pairs are considered. The results obtained from this analysis is shown
in following figures Fig. 9(a) and (b). The variation of capacitance
CTOTAL with respect to void fraction is shown for fluids such as Diesel
(ɛr = 2.1), Gasoline (ɛr = 2), Cotton seed oil or Olive oil (ɛr = 3.1),
Coconut oil (refined) (ɛr = 2.9), Sesame oil or Peanut oil (ɛr = 3) and
Air (ɛr = 1) in Fig. 9(a). Similarly the variation of capacitance CTOTAL
85
A. Pal, B. Vasuki Measurement 124 (2018) 81–90

Fig. 8. Electrical model of concave capacitance at different void fractions (like 10%, 25%, 50%, 75% and 90%).

Table 1 decreases as void fraction increases because the relative permittivity of

Numerical values of C1, C2, C3, and C4 for distilled water as a liquid and air as the fluid of measurement (such as Diesel, Gasoline and all the other
the gaseous medium. oils) is more than that of the other fluid present, i.e. air (ɛr = 1). In
Void Height of liquid hL C1 (pF) C2 (pF) C3 (pF) C4 (pF) Fig. 9(b) the final capacitance value CTOTAL increases as void fraction
Fraction (m) increases because the relative permittivity of the fluid of measurement
(such as Kerosene, transformer oil, and all the other oils) is less than
0% 0.094 166.9 – – 166.9
that of the other fluid present, i.e. water (ɛr = 80).
10% 0.0793 188.8 24.01 3.682 × 10−2 164.0
25% 0.066 229.9 7.701 0.4904 127.7
50% 0.047 333.8 4.172 1.043 83.45
75% 0.028 616.0 2.873 1.596 39.23 3. Hardware implementation of the proposed concave capacitive
90% 0.0147 1920.0 2.360 2.049 2.946 sensor
100% 0 – 2.086 2.086 –

The hardware for the proposed concave capacitive sensor is de-

signed and built to perform the experiments in static conditions. The
Table 2
Numerical values of CNET AC, CNET BD and CTOTAL for distilled water as a liquid experimental setup is made of transparent acrylic pipes/tubes and a
and air as the gaseous medium (Theoretical). copper sheet of 0.3 mm thickness. From the design point of view, the
hardware can be divided into three parts. First is the capacitive sensor
Void Fraction (%) CNET AC (pF) CNET BD (pF) CTOTAL (pF)
portion (middle), second is the arrangement for inlet and outlet of the
0% 61.77 61.77 123.5 distilled water and the arrangement for the cable connections up to LCR
10% 17.50 61.36 78.86 meter (HIOKI IM3536) and third is the acrylic cuboid portion connected
25% 6.925 55.55 62.48 with the pipe for ultrasonic sensor (HC-SR04) to measure the height of
50% 3.955 45.38 49.34
the liquid in acrylic pipe. The whole setup is mounted on the plywood
75% 2.779 28.83 31.60
90% 2.302 4.753 7.055 structure and it is supported by four nut-bolts in the corners as leveling
100% 2.042 2.042 4.084 screw. The perfect horizontal level is achieved by adjusting these nut-
bolts and by using sprit-level. To achieve accurate level inside the pipe,
two marked rulers are fixed on both sides of the pipe. The hardware
with respect to void fraction is shown for fluids such as Transformer oil implementation of the proposed concave capacitive sensor is shown in
(ɛr = 2.2), Kerosene (ɛr = 1.8), Almond Oil (ɛr = 2.8), Castor oil following Fig. 10(a).
(ɛr = 2.6) and Water (ɛr = 80) in Fig. 9(b) [30].
It is clear from Fig. 9(a) that the final capacitance value CTOTAL

140
11.5 120
Capacitance (pF)
Capacitance (pF)

100
9.5
80
7.5 60
40
5.5 20
0
3.5
0 10 25 50 75 90 100
0 10 25 50 75 90 100
Void Fraction (%)
Void Fraction (%)
Ctotal(Diesel & Air) Ctotal (Transformer oil & Water)
Ctotal(Gasolin & Air) Ctotal (Keroscene & Water)
Ctotal(Cotton seed/olive oil & Air) Ctotal (Almound oil & water)
Ctotal(Coconut oil& Air)
(a) (b) Ctotal (castor oil & water)
Ctotal(Sesame/Peanut oil & Air)
Fig. 9. (a) Capacitance Vs Void Fraction (Different fluids and air). (b) Capacitance Vs Void Fraction (Different fluids and distilled water.

86
A. Pal, B. Vasuki
Fig. 10. (a) Experimental setup – Concave capacitance based void fraction measurement system and (b) Experimental setup – Concave capacitance based void
fraction measurement system (with LCR meter and BNC connections).
3.1. Capacitive sensor part
A copper sheet of thickness 0.3 mm is used to design the main ca-
pacitive sensing part. The rectangular pieces (four electrodes and four
guard rings) are cut as per the dimensions described before in Section
(2.1) and placed on the surface of the acrylic pipe (Internal dia-
meter = 94 mm, outer diameter 100 mm) and secured by very thin
plastic tapes. A PVC pipe (Internal diameter = 104 mm) is cut into half,
and inner curve surface of both the pieces are covered with copper
sheet to make an openable shield to cover the whole capacitive sensor.
By using wires, the guard electrodes and outer copper shield is shorted
together, which is the measurement ground for the whole system.
3.2. Inlet/outlet and arrangement for cable connections
For inlet and outlet of the setup, acrylic tubes (Internal dia-
meter = 15 mm and outer diameter = 20 mm) is mounted on two dia-
metrically opposite sides such that top opening is used as inlet and
bottom opening is used as outlet for the fluids. The bottom opening is
87
Measurement 124 (2018) 81–90
secured by a ball valve to drain and adjust the level inside the pipe. For
connecting the coaxial cables to the capacitor’s electrodes, four female
BNC straight jack (75 Ω) are mounted on the acrylic pipe. For con-
necting this setup to LCR meter, BNC plug-plug RG59 cables (75 Ω) are
used. On the other end also female BNC connector straight jack (75 Ω)
are mounted on a small piece of perfboard where the crocodile clips of
LCR meter are connected as shown in Fig. 10(b).
3.3. Acrylic cuboid section for ultrasonic sensor
To compare the obtained results with another sensor, an ultrasonic
sensor (HC-SR04) is mounted on another side of the setup such that the
level inside the pipe and the cuboid portion are same, which can be
measured through the ultrasonic sensor. By measuring level the void
fraction can be found using Eq. (6) and void fraction formula (Section
2.4). Arduino Mega 2560 is used with the ultrasonic sensor to measure
the level.
A. Pal, B. Vasuki Measurement 124 (2018) 81–90

Fig. 11. Photograph – liquid level in pipe for various void fractions.

Table 3 40 5
Values of CNET AC, CNET BD, CTOTAL and confidence interval for distilled water as
a liquid and air as the gaseous medium (experimental). 4
30

Void Fraction (%)

Capacitance (pF)

Void Void fraction CNET AC CNET BD CTOTAL (pF) Confidence

fraction from (pF) (pF) interval (pF)
3
(%) ultrasonic (Mean) (Mean) (95%) 20
sensor (%) 2
(Mean)
10
1
0 0 45.710 47.651 94.013 ± 0.923
10 8.594 12.894 41.793 57.843 ± 2.022
25 21.208 5.031 38.651 45.472 ± 0.611 0 0
50 46.682 3.557 30.165 31.992 ± 1.219 0 20 40 60 80 100
75 71.818 2.773 11.684 14.760 ± 0.763
Void Fraction (%)
90 85.599 2.059 2.565 4.261 ± 0.225
100 99.534 1.706 1.654 3.437 ± 0.133 Error in CNETAC (pF)
Error in CNETBD (pF)
Error in CTotal (pF)
Table 4
Error in Void Fraction measured by Ultrasonic sensor (%)
Uncertainty of measurement budget with combined uncertainty and expanded
uncertainty. Fig. 12. Error analysis of the void fraction measurement system (with respect to
theoretical observations).
Sources Units Standard Uncertainty

Proposed concave capacitance based sensor (max) % 0.967

measurement, this particular frequency is chosen for all measurements
Measuring ruler mm 0.289
Mitutoyo Digimatic Micrometer mm 2.89 × 10−4 of capacitance. The level of the distilled water is shown in Fig. 11 for
Hioki LCR Meter (IM3536) pF 2.89 × 10−5 different void fractions.
Combined standard uncertainty uc (CTotal) 1.0093 The mean capacitance values CNET AC, CNET BD and CTOTAL for dis-
Expanded uncertainty (K = 2, 95% confidence) U 2.0186 tilled water as a liquid and air as the gaseous medium are presented in
Table 3. Confidence interval is calculated for twenty sets of readings for
95% confidence interval and is also presented in Table 3.
3.4. Results and discussions

Experiments are performed at room temperature (28 °C) in the span 3.5. Uncertainty analysis of proposed measurement system
of five days and twenty sets of readings are acquired for each void
fraction. Distilled water is filled from inlet to the appropriate level (HL) Standard Deviation of individual measurements S(CTOTAL) is calcu-
(measured by the marked ruler in the ends), and capacitances CAC and lated by the following expression Eq. (7) [32]:
CBD are measured for each level. Reading of ultrasonic sensor is also
n
noted for these levels using Arduino Mega. As LCR meter (Hioki ∑i= 1 (Ci−Cm)2
S(CTOTAL) =
IM3536) is highly sensitive at 9 kHz frequency for capacitance n−1 (7)

88
A. Pal, B. Vasuki
Uncertainty in Capacitance (pF) 1.2
1 0.5
0.8
0.6
0.4
0.2
0 0
0 20 40 60
Void Fraction (%)
Standard Uncertainty in CNETAC
Standard Uncertainty in CNETBD
Standard Uncertainty in Ctotal
Standard Uncertainty in level measured by Ultrasonic Sensor
Fig. 13. Comparison of standard uncertainty among CNET AC, CNET BD, CTOTAL
(Experimental values) and level measured by Ultrasonic sensor values.
where Ci is the individual measurement of capacitances, Cm is the mean
value of the measurement capacitances and ‘n’ is the number of sets of
the measurement. The standard deviation of the mean is calculated by
dividing the standard deviation of individual measurements by the
square root of the number of measurement performed. This standard
deviation is considered as ‘Standard Uncertainty due to Repeatability’,
and it is given by the following equation i.e. Eq. (8).
S(CTOTAL)
u(CTOTAL ,Rep) = S(C TOTAL(m) ) =
n (8)
where ‘u(CTOTAL ,Rep) ’ is standard uncertainty due to repeatability,
‘S(CTOTAL(m) ) ’ is the standard deviation of the mean, ‘S(CTOTAL) ’ is the
standard deviation of individual readings and ‘n’ is total number of
readings.[31] Type – A standard uncertainty is calculated for the pro-
posed sensors for different void fractions. To find the combined un-
certainty, uncertainties of the all the measurement devices are also
calculated. The rectangular distribution function is assumed for calcu-
lating standard uncertainty for the measurement devices. The calibra-
tion range of each device is divided by ‘ 3 ’ to convert every rectangular
distribution in standard uncertainty. By combining all the standard
uncertainty components combined standard uncertainty (uc(CTotal)) is
calculated. The individual standard uncertainties of measurement de-
vices and combined standard uncertainty is presented as uncertainty of
measurement budget and is given in Table 4.
From the theoretical and experimental observations, errors are calcu-
lated with respect to the theoretical values for both the systems, i.e. ca-
pacitive sensor and ultrasonic sensor (Error = |CTheoretical−CExperimental|) ,
and it is shown in following Fig. 12. Comparison of standard uncertainty
for, the level measurement from ultrasonic sensor, CNETAC, CNET BD and
CTOTAL values are presented in Fig. 13.
4. Conclusion
For measuring void fraction, a simple concave capacitor-based
sensor design is proposed. The electrical model includes all the capa-
citances along with the stray capacitances due to connecting cables. For
reducing the fringing effect, four guard electrodes are also placed. The
proposed design has a symmetrical geometry, so the concave electrodes
of this sensor have interchangeable exciter electrode and receiver
electrode. It is theoretically evaluated and experimentally confirmed
Measurement 124 (2018) 81–90
0.6 that capacitance values changes according to the change in the void
Uncertainty in Height (mm)
fraction. The theoretical analysis is repeated for several numbers for
different immiscible, nonconductive fluid pairs (liquid-gas or liquid-li-
0.4 quid two-phase flow). From Tables 2 and 3, it is clear that the net ca-
pacitance between the two concave copper electrodes A–C and B–D also
0.3
give individually significant capacitance change in their respective
0.2 range. As change in capacitance is continuous, the resolution of the
system depends on the resolution of the capacitance measuring system.
0.1 Results obtained from this analysis clearly represents that the change in
capacitance value is sufficiently high and it is measurable (in the range
of pF) and the system is highly precise as the standard uncertainty is
80 100
less than 0.97 pF for the whole range of measurement.
Conflict of interest
None of the authors have any conflict to declare.
Acknowledgements
The authors would like to acknowledge the support from MHRD,
Government of India.
References
[1] Pieter Gijsenbergh, Maarten Driesen, Philippe Jourand, Robert Puers, Integrated
void fraction sensors for two-phase, microfluidic systems, Procedia Eng. 47 (2012)
643–646.
[2] F. Murzyn, D. Mouaze, J.R. Chaplin, Optical fibre probe measurements of bubbly
flow in hydraulic jumps, Int. J. Multiph. Flow 31 (2005) 141–154.
[3] Marcelo S. Rocha, Eduardo L.L. Cabral, José R. Simões-Moreira, Capacitace sensor
for void fraction measurement in a natural circulation refrigeration circuit, in:
International Nuclear Atlantic Conference – INAC 2009, 2009.
[4] Volker Grabenstein, Anja-Elsa Polzin, Stephan Kabelac, Experimental investigation
of the flow pattern, pressure drop and void fraction of two-phase flow in the cor-
rugated gap of a plate heat exchanger, Int. J. Multiph. Flow 91 (2017) 155–169.
[5] H. Asano, N. Takenaka, T. Fujii, H. Hayashi, T. Wakabayashi, Visualization and void
fraction measurement of gas-liquid two-phase flow in a commercial plate heat ex-
changer by thermal neutron radiography, IEEE Trans. Nucl. Sci. 52 (2005).
[6] H.C. Yang, D.K. Kim, M.H. Kim, Void fraction measurement using impedance
method, Flow Meas. Instrum. 14 (2003) 151–160.
[7] Ragna Kipping, Rosmer Brito, Eckhard Scheicher, Uwe Hampel, Developments for
the application of the Wire-Mesh sensor in industries, Int. J. Multiph. Flow 85
(2016) 86–95.
[8] H.F. Velasco Peña, O.M.H. Rodriguez, Applications of wire-mesh sensors in multi-
phase flows, Flow Meas. Instrum. 45 (2015) 255–273.
[9] E. Roitberg, L. Shemer, D. Barnea, Measurements of cross-sectional instantaneous
phase distribution in gas–liquid pipe flow, Exp. Therm Fluid Sci. 31 (2007)
867–875.
[10] X. Chen, Y.F. Han, Y.Y. Ren, H.X. Zhang, H. Zhang, N.D. Jin, Water holdup mea-
surement of oil-water two-phase flow with low velocity using a coaxial capacitance
sensor, Exp. Therm Fluid Sci. 81 (2017) 244–255.
[11] Carlos R. Zamarreno, Cicero Martelli, H.V. Virginia Baroncini, N. Eduardo dos
Santos, Marco J. da Silva, RigobertoE.M. Morales, Pablo Zubiate, Francisco
J. Arregui, Ignacio R. Matias, Single and multiphase flow characterization by means
of an optical fiber Bragg grating grid, J. Lightwave Technol. 33 (2015).
[12] Wu. Hao, Chao Tan, Xiaoxiao Dong, Feng Dong, Design of a conductance and ca-
pacitance combination sensor for water holdup measurement in oil–water two-
phase flow, Flow Meas. Instrum. 46 (2015) 218–229.
[13] Abdullah Abbas Kendoush, Zareh Azat Sarkis, Void fraction measurement by X-ray
absorption, Exp. Therm Fluid Sci. 25 (2002) 615–621.
[14] E. Nazemi, S.A.H. Feghhi, G.H. Roshani, R. Gholipour Peyvandi, S. Setayeshi,
Precise void fraction measurement in two-phase flows independent of the flow re-
gime using gamma-ray attenuation, Nucl. Eng. Technol. (2016) 64–71.
[15] H. Canière, C. T’Joen, A. Willockx, M. De Paepe, Capacitance signal analysis of
horizontal two-phase flow in a small diameter tube, Exp. Therm Fluid Sci. 32 (2008)
892–904.
[16] A. Jaworek, A. Krupa, M. Trela, Capacitance sensor for void fraction measurement
in water/steam flows, Flow Meas. Instrum. 15 (2004) 317–324.
[17] Domenico Strazza, Marco Demorib, Vittorio Ferrari, Pietro Poesio, Capacitance
sensor for hold-up measurement in high-viscous-oil/conductive-water core-annular
flows, Flow Meas. Instrum. 22 (2011) 360–369.
[18] A. Jaworek, A. Krupa, Gas/liquid ratio measurements by rf resonance capacitance
89
A. Pal, B. Vasuki
sensor, Sens. Actuators A 113 (2004) 133–139.
[19] Zhao An, Jin Ningde, Zhai Lusheng, Gao Zhongke, Liquid holdup measurement in
horizontal oil–water two-phase flow by using concave capacitance sensor,
Measurement 49 (2014) 153–163.
[20] Seyed Milad Salehi, Hajir Karimi, Ali Akbar Dastranj, A capacitance sensor for gas/
oil two-phase flow measurement: exciting frequency analysis and static experiment,
IEEE Sens. J. 17 (3) (2017).
[21] Wael H. Ahmed, Capacitance sensors for void-fraction measurements and flow-
pattern identification in air–oil two-phase flow, IEEE Sens. J. 6 (5) (2006).
[22] M. Sami, A. Abouelwafa, E. John, M. Kendall, The use of capacitance sensors for
phase percentage determination in multiphase pipelines, IEEE Trans. Instrum.
Meas. IM-29 (1) (1980).
[23] Lusheng Zhai, Ningde Jin, Zhongke Gao, Zhenya Wang, Liquid holdup measure-
ment with double helix capacitance sensor in horizontal oil–water two-phase flow
pipes, Chin. J. Chem. Eng. 23 (2015) 268–275.
[24] A. Jaworek, A. Krupa, Phase-shift detection for capacitance sensor measuring void
fraction in two-phase flow, Sens. Actuators, A 160 (2010) 78–86.
[25] Nikolas Libert, Rigoberto E.M. Morales, Marco Jose, da Silva, Capacitive measuring
system for two-phase flow monitoring. Part1: Hardware design and evaluation,
Flow Meas. Instrum. Vol. 47 (2016) 90–99.
[26] E.J. Mohamad, R.A. Rahim, M.H.F. Rahiman, H.L.M. Ameran, S.Z.M. Muji,
O.M.F. Marwah, Measurement and analysis of water/oil multiphase flow using
electrical capacitance tomography sensor, Flow Meas. Instrum. 47 (2016) 62–70.
[27] Nikolas Libert, Rigoberto E.M. Morales, Marco Jose, da Silva, Capacitive measuring
system for two-phase flow monitoring. Part 2: Simulation-based calibration, Flow
Meas. Instrum. 50 (2016) 102–111.
[28] Wolverine Tube, INC., Engineering Data Book III, Void Fraction in Two-Phase Flows
(Chapter 17).
[29] W.H. Beyer (Ed.), CRC Standard Mathematical Tables, 28th ed. CRC Press, Boca
Raton, FL, 1987. p. 125.
[30] Honeywell dielectric constant table.
[31] JCGM 100:2008 (GUM 1995 with minor corrections), Evaluation of measurement
data – guide to the expression of uncertainty in measurement (accessed 18 Jan
2018).
[32] Ivo Leito, Lauri Jalukse, Irja Helm, Estimation of Measurement Uncertainty in
90
Measurement 124 (2018) 81–90
Chemical Analysis, University of Tartu, Online Course (accessed 18 Jan 2018).
Amit Pal was born in Bhopal, Madhya Pradesh, India in
1991. He received his Bachelor’s Degree in Electronics and
Instrumentation Engineering from L.N.C.T., R.G.P.V.
University Bhopal, Madhya Pradesh in 2013, and the
Master’s degree in Instrumentation Engineering from
National Institute of Technology, Agartala, Tripura in 2016.
He is currently working towards the doctoral degree in the
Instrumentation and Control Engineering Department,
National Institute of Technology, Tiruchirappalli, India. He
has been awarded the HTRA fellowship from MHRD,
Government of India for his Ph.D. study. His current re-
search interests includes Instrumentation systems for mea-
suring Void fraction and flow patterns.
B. Vasuki was born in Thanjavur, Tamil Nadu, India in
1967. She received her Bachelor’s Degree in
Instrumentation and Control Engineering from the
Government College of Technology, Coimbatore, India in
1989, the Master’s Degree in Power Systems, Regional
Engineering College, Tiruchirappalli, in 1994 and the PhD
Degree in Instrumentation and Control from the National
Institute of Technology, Tiruchirappalli, India in 2009. She
worked as a lecturer in the Department of Electrical and
Electronics Engineering, Shanmuga Polytechnic College,
Thanjavur from 1989 to 1995. She joined in the
Department of Instrumentation and Control Engineering as
a faculty member, National Institute of Technology,
Tiruchirappalli, India, in 1997, where she is currently
working as Professor. Her research interests include Instrumentation systems and Interval
Analysis.