الکتروتکنیک-برق صنعتی

برای دیدن متن مرتب شده  به این ادرس بروید

http://www.benthamscience.com/eeng/samples/eeng%201-1/Ahmed.pdf

Recent Patents on Electrical Engineering 2008, 1, 1-13 1
1874-4761/08 $100.00+.00 © 2008 Bentham Science Publishers Ltd.
Innovative Techniques For Two-Phase Flow Measurements
Wael H. Ahmed* and Basel I. Ismail**
*Component Life Technology, Atomic Energy of Canada Ltd., Canada, **Department of Mechanical Engineering,
Lakehead University, Canada
Received: November 5, 2007; Accepted: November 27, 2007; Revised: November 30, 2007
Abstract: Two-phase flow is important in an increasing number of applications and industries. For example, the safety
analysis codes for the nuclear energy industry require closure relations for the vapor-liquid interfacial transfer terms,
while accurate two-phase pressure drop models are necessary to design the piping systems in the oil and gas industry.
Also, two-phase flow occurs in heat exchangers, steam generators, chemical reactors, oil transportation and many other
process equipments. In addition, development of accurate and suitable instrumentations for on-line monitoring and
measurement of the solids concentration and velocity in gas-solid two-phase flows has proven to be a challenging problem
with many scientists and engineers worldwide developing novel techniques for this application. This paper presents a
review on the electrical-based measurement techniques for gas-solid and gas-liquid two-phase flows along with the most
recent patents developed for two-phase flow measurements. Also, development of a novel method for the design of
capacitance sensors for void fraction measurement and flow pattern identification was presented in detail.
Keywords: Two-phase flow, gas-solid flow, gas-liquid flow, novel design method, capacitance sensor, recent patents, void
fraction.
1. INTRODUCTION
Online, continuous, two-phase flow measurement is often
necessary, particularly in the oil and gas, nuclear energy and
chemical processing industries. Reliable measurements of
the void fraction and flow pattern identification are important for accurate modeling of two-phase systems. Void
fraction can be measured using a number of techniques,
including radiation attenuation (X or
-ray or neutron beams)
for line or area averaged values, optical or electrical contact
probes for local void fraction, impedance technique using
capacitance sensors and direct volume measurement using
quick-closing valves. The use of the different techniques
depends on the applications, and whether a volumetric
average or a local void fraction measurement is desired. The
radiation attenuation method can be expensive and from a
safety aspect difficult to implement, while intrusive probes
disturb the flow field. On the other hand, the impedance
measurement technique is practical and cost-effective
method for void fraction measurement. The technique is
non-intrusive and relatively simple to design and implement.
Impedance or capacitance sensors have been used successfully to measure time and volume averaged void fraction,
and its instantaneous output signal has been used to identify
the flow pattern [1]. In this paper, electrical-based techniques
for gas-liquid and gas-solid flows measurements are
reviewed. The paper is divided into two parts: In the first
part, gas-solid flow measurements with their important and
relevant application to pneumatic conveying systems are
discussed. However, the detailed description of different
techniques cannot be covered exhaustively here. Therefore,
detailed consideration is focused on the development of a
novel method for the design of a capacitance sensor used in
*Address correspondence to this author at the Component Life Technology,
Atomic Energy of Canada Ltd., Canada; Tel: +1(613) 584-8811;
Fax: +1(613) 584-8031; E-mail: ahmedw@aecl.ca
gas-liquid two-phase flow measurements in the second part
of the paper.
2. RECENT DEVELOPMENTS IN TWO-PHASE
FLOW MEASUREMENTS
2.1. Gas-Solid Two-Phase Flows
Gas-solid Two-phase flows, particularly those solids in
the forms of powder and grain transported in gases (typically
air) through a pipeline, are important in pneumatic conveying which has widespread applications across many industries including chemical, food processing, cement, mining,
petrochemical, pharmaceutical, semiconductor, and in transporting pulverized coal in fuel lines of thermal power plants.
Moving fluidized-beds may also be regarded as a form of
gas-solid flows. This wide application has led to extensive
research on gas-solid two-phase flow systems.
Gas-solid flows (also known as particulate flows), such
as those in pneumatic conveying systems, are usually
classified as dilute- or dense-phase flow. The dilute-phase
flow can be normally achieved with low concentrations of
solids (typically below 10%) and the simultaneous condition
of high gas velocities in the gas-solid flow mixture [2]. In the
dilute-phase conveying, the solid particles are fully dispersed
or suspended in the flow and no deposition occurs. The
velocity at which solids are traveling in a dilute-phase
conveying is important. If they are traveling too slowly,
solids can drop out from the flowing suspension and pipe
blockage can occur. If they are traveling too quickly this can
lead to unnecessarily high pipeline wear and power
consumption due to system pressure drop [3]. Flow patterns
in a dense pneumatic conveying system, however, show
many interesting features and analogies with gas-liquid
flows, such as slug, plug, and stratified flow regimes [4].
Dense-phase flow in pneumatic conveying systems have the
relative benefits of a low air requirement and hence energy
demand, low pipeline erosion and low product degradation 2 Recent Patents on Electrical Engineering, 2008, Vol. 1, No. 1 Ahmed and Ismail
[5]. However, dense-phase pneumatic conveying may lead to
unstable flows, caused by insufficient air velocity, and
deposition. These unstable flows and deposition often cause
pipe blockage and vibration that can be extremely difficult to
remedy.
The fundamental system components of a conventional
pneumatic conveying system are: an air mover, a feeder for
the solids material to be conveyed, the conveying pipeline,
and a receiver to disengage the conveyed solids and carrier
gas (typically air). Local solids concentration and velocity
are two of the most important parameters characterizing gassolid flows. Development of accurate and suitable instrumentations for on-line monitoring and measurement of the
solids concentration and velocity in gas-solid two-phase
flows has proven to be a challenging task with many scientists and engineers worldwide developing novel techniques
for this application. This is particularly due to spatial and
temporal fluctuations of both the solids concentration and
velocity during the pneumatic transportation [6]. The results
of these measurements can be used to ensure efficient and
economic pneumatic conveying operation, prevent blockage
and hazardous damages and maintain safety standards. This
paper presents a review on a range of existing and recent
techniques used to measure solids concentration and velocity
in gas-solid flows, such as those widely used in pneumatic
conveying systems.
2.2. Measurement Techniques for Gas-Solid Flow
Systems
The measurements and visualization of gas-solid flows,
such as those involved in pneumatic conveying systems,
could be performed using a range of noninvasive and nonintrusive electrical impedance, optical transmission [7],
radiography and radiometric sensing [8], and other measurement techniques. The application of capacitance sensors for
measuring local solids concentrations in a gas-solid flow was
established decades ago [9, 10]. Chang et al. [11] tested
strip- or ring-type capacitance transducers that could be used
in non-electrically conductive (dielectric) pipes to nondestructively detect powder flow or measure powder
thickness using the dielectric constant difference of the gas
and powder. Capacitance impedance rings are noninvasive,
simple and relatively inexpensive for implementation and
provide high temporal resolution. However, those devices
are difficult to calibrate and care must be taken when using
correlation models for different gas-solid flow patterns to
convert electrical impedance to solid concentrations [12]. In
a dense-phase pneumatic conveying where density gradient
exists in the gas-solid flow caused by a non-uniform
distribution of solids in a pipe cross section, electrical
capacitance tomography (ECT) can be effectively used [4; 5;
13]. In an electrical process tomography, the capacitancesensing electrodes are installed at equal intervals around the
periphery of the pipe, in order to obtain the threedimensional distribution of solids along the direction of the
gas-solid flow. The electrodes are usually installed in a
noninvasive way, i.e. outside the pipe made of dielectric
material [14]. The process tomography or tomographic
imaging technique has the potential of producing crosssectional images of the distribution of flow components in a
conveying pipeline from which solids velocity and
concentration profiles of a section of a pipeline can be
obtained. Brown et al. [15] developed a tomographic
technique for imaging gas-solid flow distributions in
pneumatic conveying pipelines. Their technique utilized
ultrasonic transmission-mode measurements constrained to
the megahertz region. They performed image reconstruction
using a back-projection method implemented with standard
graphics algorithms. Brown et al. [15] applied their technique to simulate reconstructions of dense and dilute distributions and obtained results demonstrating the capabilities
and limitations of their technique. They also addressed
aspects of transducer array design. Recently, Steiner et al.
[16] proposed a dual-mode ultrasound and electrical capacitance process tomography sensor for measurements of gassolid flow. In their method, the ultrasound image is used as a
priori information for the finite element method-based
capacitance reconstruction algorithm. They [16] reported that
a significant improvement in image quality can be achieved
with the proposed dual-mode tomography process sensor.
Jiang and Xiong [17] introduced a novel noninvasive
electrostatic method for measurement of the solids velocity
and mass flow rate in gas-solid flow. In their system, the
measurement of the velocity and mass flow rate were
realized by an electrostatic sensor capable of detecting the
electrostatic signals of solid powder in the gas-solid flow.
The transit time of powder was measured by cross-correlating electrostatic signals, and the velocity was calculated.
By correlating the voltage on electrostatic sensor and the
fixed mass flow rate, the voltage-mass flow rate curve of the
sensing system was established. Jiang and Xiong [17]
concluded that electrostatic method can be used to trace and
measure the velocity of the solid phase, and the method is
capable of performing on-line measurements of the mass
flow rate of the gas-solid flow.
3. CAPACITANCE SENSORS FOR GAS-LIQUID
TWO-PHASE FLOWS
The principle of the capacitance method is based on the
differences in the dielectric constants of the two phases in
the flow, and the capacitance measured across the sensors is
dependent on the volume ratio of the two phases. There are,
however, several disadvantages of the impedance technique,
which are sometimes difficult to resolve. For example, the
capacitance measurement is sensitive to the void fraction
distribution or flow regimes due to the non-uniformity of the
electrical field inside the measuring volume. This, however,
can be compensated by first identifying the flow pattern. The
measurement is also sensitive to the changes in electrical
properties of the two phases due to temperature. The noise
due to the electromagnetic field around the sensor and
connecting wires can significantly affect the signal and needs
to be minimized through proper design of the sensor shield.
The sensitivity of the capacitance meter is also improved by
Andrade [18]. He used a full power supply potential swing
on the shield electrodes to allow for the use of simpler shield
electrode design. Using the full swing improves the
capacitance background and balances such noises as the wire
capacitances. Furthermore, it is difficult to resolve the
change in phase distribution within the sensor measurement
volume. The sensitivity to the flow pattern can be increased
through better design of the sensors. For example, Hung et Innovative Techniques For Two-Phase Flow Measurements Recent Patents on Electrical Engineering, 2008, Vol. 1, No. 1 3
al. [19] used eight electrodes along the circumference of the
tube to obtain a tomographic image of the two-phases. On
the other hand, homogenizing the electric field in the axial
direction minimizes the error due to the void distribution in
the measuring volume. This could be done by using rotating
type sensors as suggested by Merilo et al. [20] and Lucas
and Simonian [21] or by using helical electrodes as discussed by Geraets and Borst [22] and Kenneth and Rezkallah
[23]. The electrical capacitance tomography technique was
also implemented recently by Gamio et al. [24] to image
various two-phase gas-oil horizontal flows in a pressurized
pipeline. They emphasized the potential of this technique for
real-time flow visualization and flow regime identification in
practical industrial application at high pressure operating
conditions. Electrical Impedance Tomography (EIT) is also
introduced in a recent patent by Wang [25] as a new signal
processing method. The method is used on-line to obtain
accurate estimates of the local disperse phase volumetric
flow rate, the mean disperse and continuous phase volume
fractions and the distributions of the local axial, radial and
angular velocity components of the disperse phase.
Several different configurations of capacitance sensors,
including flat plate, concave, ring, helical and multiple
helical wound in contact or isolated from the fluid, have
been widely investigated [23, 26-29]. However, there are
fewer studies on the optimization of the design to obtain
good signal to noise ratio and high sensitivity to the different
flow patterns. Elkow and Rezkallah [23] compared the
performance of concave and helical type sensors and
determined that the problems associated with helical type
sensors, including the nonlinear response, poor sensitivity
and poor shielding, can be eliminated by using the concave
type sensors. The accuracy of the concave parallel sensors
can be improved by having both electrodes of equal length to
decrease the non-uniformity of the electric field between the
two electrodes and eliminate the non-linear response. Based
on several tests, they also recommended that the distance
between the electrodes and the shield should be large relative
to the separation distance between the two electrodes in
order to improve the immunity to stray capacitance.
In this paper, a systematic method for the design of
capacitance sensors for void fraction measurement and flow
pattern identification is presented. Two different configurations of the sensors are considered: concave and ring type
Fig. (1). For the ring types sensor each electrode covers the
entire circumference, except for a small gap to facilitate the
installation of the sensor around the tubes, and are separated
in the axial direction of the tube Fig. (1a), while in the concave sensor, two brass strips are mounted on the tube circumference opposite to each other Fig. (1b). The difference
in the electrode geometry results in different electric fields
within the measurement volume and hence in the sensitivity
and response of the sensors. The two geometries are
analyzed for the signal to noise ratio and the sensitivity to
the void fraction and flow pattern. Experiments were
performed to validate the design theory and to evaluate the
sensor characteristics in actual operating conditions using
air-oil two-phase flow in a horizontal pipe. To more
objectively identify the flow pattern, the probability density
function (PDF) and power spectral density (PSD) of the time
trace of the void fraction signal is analyzed in a similar
manner to that done by Jones and Zuber [30].
4. CAPACITANCE SENSOR DESIGN
The capacitance sensors need to be designed for accurate
measurement of the void fraction, and have sufficient time
response to detect the variations for the different flow
patterns. The value of the capacitance across the sensors is
related to the phase distribution and dielectric properties of
the two-phases. The relationship between the capacitance
and the void fraction is dependent on the dielectric values of
the two phases, cross-sectional area of the sensors, and the
separation distance between the two electrodes. A dielectric
material placed between two conductor plates acts as an
insulator to increase the charge storage capabilities because
the dielectric contains charged molecules that are randomly
oriented. When an external field is applied across the two
plates, the charged molecules align themselves with the
electric field and produces dipoles, where the positive
charges of each molecule are in the direction of the applied
field and the negative charges oppose the field. An internal
Fig. (1). Schematic of electrodes configurations.
Capacitance
meter
Elec (1)
Elec (2)
Elec (1)
Capacitance
meter
Elec (2)
a) Ring type b) Concave type
Capacitance
meter
Elec (1)
Elec (2)
Elec (1)
Capacitance
meter
Elec (2)
a) Ring type b) Concave type 4 Recent Patents on Electrical Engineering, 2008, Vol. 1, No. 1 Ahmed and Ismail
electric field, opposite in direction of the external electric
field, will result with a consequent reduction of the overall
electric field and the overall potential. It is important to
understand the theory of the capacitance sensor technique to
properly optimize the sensor design. For simplicity, this can
be illustrated using a simple parallel plate capacitor to show
the effect of a dielectric material on the capacitance.
Neglecting edge effects, the uniform electric field between
the two plates of the capacitor in vacuum space without the
dielectric medium, with charge density f
and the permittivity of free space,
o = 8.854x10
-12
C
2
N
-1
m
-2
, can be written
as:
E =

f
o
(1)
When a dielectric material is inserted to fill the entire
space between the plates, the dielectric is polarized, and a
polarization charge of density
p appears on the two surfaces.
For a uniform electric field out from a plane, the electric
field can be written as:
E =
( )
f
+
p
o
(2)
The polarized charge density is:

p
= 
o
E
(3)
where  is the electric susceptibility of the material. Finally
we can write the capacitance as:
C =

o
( ) 1
 A
d (4)
and the dielectric constant
is defined as:
 = ( ) 1
 (5)
The permittivity
is sometimes used to characterize the
dielectric behavior of the matter and is related to the
permittivity of the free space as
 = 
o

(6)
In order to mathematically represent the output signal as
a function of the void fraction, the capacitance measured by
a transducer for two-phase flow can be treated as an approximation of a parallel or series combination of capacitors
with different dielectric constants. In this work, the performance of two different types of sensor configurations is
investigated, and a design method to optimize the sensor
performance is developed. The sensor output is sensitive to
the flow pattern and to incorporate the flow pattern in the
theoretical analysis, it is important to clearly classify the
flow patterns. To minimize the number of flow patterns, only
the basic flow patterns are considered here. For horizontal
pipes: bubbly, stratified, slug, and annular flow are used and
for vertical pipes: bubbly, slug, and annular are considered.
Churn flow is difficult to simulate in our analysis.
4.1. Ring Type Sensors
If we consider only stratified, annular and long slug flow
patterns among the basic flow patterns listed before, the
electric field between two ring electrodes in free space can
be schematically presented in Fig. (2a). Assuming the electrical field is shielded outside the tube and neglecting the
radial electric field, the distribution can be approximated to
that shown in Fig. (2b). Therefore the electric field can be
assumed approximately constant in the axial direction and
the two rings equivalent to parallel flat disks. The equivalent
capacitance circuit method can be used to analyze this
problem [11]. In this method, the two phases are modelled as
series or parallel capacitors between the electrodes. This
equivalent circuit is based on the distribution of the two
phases inside the channel and is illustrated in Fig. (3). By
considering the two electrodes as two imaginary disks in
parallel with a separation distance (d), the two phases distributed horizontally will be equivalent to two capacitances in
series Fig. (3a), while the two phases distributed in the
vertical direction will be equivalent to two parallel capacitances Fig. (3b). The equivalent circuits for different twophase distributions are shown Fig. (4a) through d as
discussed by Chang et al. [11]. The theoretical output
Fig. (2). Schematic of electrical field profile generated by two ring electrodes.
2
Electrode 2 Electrode 2
d
Electrode 1 Electrode 1
Imaginary planes
a) Actual b) Assumed Innovative Techniques For Two-Phase Flow Measurements Recent Patents on Electrical Engineering, 2008, Vol. 1, No. 1 5
capacitance as a function of void fraction for annular,
stratified and long slug flow can be obtained by considering
the equivalent circuit method represented in Fig. (4a & b),
and the total equivalent capacitance for both circuits can be
calculated as:
CT
= Cg
+CL
+Cw
(7)
For annular or stratified flow
CT
=
Kg


0
Ag
d
+
KL


0
AL
d
+
Kw


0
Aw
d (8)
CT
=
Kg

0
Ag
d
+
KL

0 ( ) 1 Ag
d
+
Kw

0



4
( ) D + t
2
 D
2
( )
d (9)
The void fraction is defined as

=
VG
VT
=
AG
 d
AT
 d
=
AG
AT
(10)
where VG and VT are the volume of gas and total volume
respectively.
By dividing equation (9) by the cross-sectional area of
the pipe results in
CT
=

D
2

o
4d
Kg
 + KL ( ) 1 + Kw
1+
t
D






2
1








(11)
where, t is the pipe wall thickness and
w is the dielectric
constant of the pipe material.
4.2. Concave Type Sensors
A similar analogy is used for the concave type sensors,
however, an equivalent geometrical shape is introduced (Fig.
5a through c), where the change of the electrical field due to
the curvature of the electrode is neglected [26]. Then, the
total capacitance for stratified or annular flow pattern can be
written as:
Fig. (3). Capacitance circuit equivalent to two-phase flow distribution.
Fig. (4). Equivalent capacitance circuits for typical flow regimes (Adapted from Chang et al. [11]).
Cg CL
CL
Cg Liquid
Gas
d
x
d
Ag AL
a) Series circuit b) Parallel circuit
Cg CL
CL
Cg Liquid
Gas
d
x
d
Ag AL
a) Series circuit b) Parallel circuit
3
c) Bubbly flow d) Slug Flow
a) Annular flow b) Stratified flow
CG6 Recent Patents on Electrical Engineering, 2008, Vol. 1, No. 1 Ahmed and Ismail
CT
=
1
CW
+
1
CL2
+
1
Cg
+CL1
+CL1
+
1
CW
+
1
CL2





(12)
The equivalent geometrical parameters can be defined as
d =  ( ) D
2  t e = d =  ( ) D
2  t
f =
1
2
D
 ( ) D
2  t a =
1
2
( )
D  2 z
(13)
Based on the two phases distribution shown in Fig. (5),
and using the capacitance circuit analogy, the capacitances in
equation (12) can be written as:
Cw
= w

o
a
L
t
Cg
=  g

o
h
L
e
CL1
=  L

o
a
L
e
CL2
=  L

o
a
L
f
(14)
The above analysis may be used to estimate the
capacitance of such a sensor by substituting the permittivity
of the free space and the dielectric constant of the liquid as
well as the geometrical parameters listed in equation (13) as
a function of the void fraction. The set of equations (12) to
(14) relate the capacitance to the void fraction and can be
used to investigate the sensitivity and/or to compare different
designs.
4.3. Design Method Applied to Different Flow Patterns
Although the theoretical model discussed above considered only stratified and annular flow, the analysis could be
extended to other flow patterns in order to design and
optimize the capacitance sensor. As an example, plug flow is
considered, and in particular the case where an elongated
bubble with a length lG in a horizontal pipe is passing
through a pair of ring type sensors as shown in Fig. (6a). The
bubble can be assumed as a horizontal cylinder with cross
sectional area AG and occupying a length l between the two
electrodes. Assuming a constant electric field between the
electrodes, the equivalent capacitance circuit using the same
analogy as before is shown in Fig. (6). In this case
CT
=
Cg
+CL1
Cg
CL1
+CL2
+Cw
(15)
and the volumetric void fraction is different from the cross
sectional void fraction and can be calculated as

=
VG
VT
=
AG
 lG
AT
 d
(16)
The capacitances in equation (15) can be written as
Cw
=
w
o


4
( ) D + t
2
 D
2
( )
d
CL1
=
 L

o
Ag
d
l
CL2
=
 L

o ( ) At

Ag
d
Cg
=

g
o
Ag
l
(17)
The corresponding relation between the void fraction and the
output signal of the capacitance sensor in the case of
elongated bubble can be written as
Fig. (5). (a),(b) Geometrical simplification of the concave type capacitance sensor, (c) Equivalent capacitance circuits for annular and core
flow regimes.
Fig. (6). a) Geometrical simulation of elongated bubble in ring type sensor.
b) Equivalent capacitance circuit.
z
D
Elect(2)
Elect(1)
63.5 mm
h
e
D
f
a
t
Cw
Cg CL1
CL2
CL2
Cw
CL,2
(a) (b) (c)
z
D
Elect(2)
Elect(1)
63.5 mm
h
e
D
f
a
t
Cw
Cg CL1
CL2
CL2
Cw
CL,2
(a) (b) (c)
CW
CG CL1
CL2
D
l
t
AG
AL
Liquid
Gas
Elect2 Elect1
d
(a) Elongated bubble simulation b) Equivalent circuit
CW
CG CL1
CL2
D
l
t
AG
AL
Liquid
Gas
Elect2 Elect1
d
(a) Elongated bubble simulation b) Equivalent circuitInnovative Techniques For Two-Phase Flow Measurements Recent Patents on Electrical Engineering, 2008, Vol. 1, No. 1 7
CT
=
o
d

D
2
4
w
1+
d
t





 1





+ L
1
d
l












+
l
2
 g
+
l
2
d
l
1






 L




,
where
> 1
l
d
.
The above analysis can be extended to different flow
regimes, where the level of complexity of the analysis
depends on the complexity of the flow regime and the
required accuracy of the modelling.
5. DESIGN AND CHARACTERIZATION OF THE
SENSORS
Based on the theoretical analysis, ring and concave type
sensors were designed and constructed for a tube having
inside and outside diameter of 12.7 and 15.8 mm, respectively. The schematic diagrams with the basic dimensions of
the sensors are shown in Figs. (7 & 8). The ring type sensor
consists of two pairs of active electrodes made of 5 mm
brass strips spaced 2 mm apart to increase the signal to noise
ratio by amplifying the absolute value of the capacitance
circuit. This geometry provides sufficient volumetric
resolution to resolve the characteristics of the slug flow
regimes. Two separate half hollow cylinders made of acrylic
are used for housing the electrodes as show in Fig. (7a & b),
and the unit is shielded using a 0.5 mm thick grounded brass
electrode to eliminate stray capacitance between any of the
electrodes, circuit and the wires. The housing is sandwiched
together over the test section pipe and fastened using acrylic
screws to make sure the electrodes are firmly in contact with
the tube surface. The noise on the signal due to the
surroundings and orientation of the connection wires was
eliminated in the present design by increasing the separation
distance between the electrodes and the shield relative to the
separation distance between electrodes as suggested by
Elkow and Rezkallah [23]. The length of the electrode for
the concave type sensor is taken as 10 mm, and both
electrodes are equal in length to decrease the nonlinearity in
the sensor response as recommended by Elkow and
Rezkallah [23]. Two pairs of electrodes were used with
spacing of 6 mm to have the same spatial resolu-tion as the
ring type for comparison. In the experimental work reported
here, the distance between the electrodes and the shielding
was experimentally optimized to reduce the noise due to
ground current and electric field leakage. The output
capacitance from the sensors is measured using a Boonton
72B capacitance meter operated at an excitation frequency of
1 MHz. The accuracy of the meter is + 4% in the 10 pF
range with meter resolution of at least 0.001 of the range.
The analog output voltage signal from the capacitance meter
is sampled using a 16-bit A/D converter at a sampling rate
up to 2 KHz.
The sensors were calibrated over the full range of void
fractions to verify the theoretical design approach. A static
calibration was performed using simulated stratified and
annular flow patterns, and a dynamic in situ calibration was
performed to investigate the effect of the noise from the
surroundings such as from other operating devices on the
output signal of the sensors. It should be noted that the effect
a) Schematic of the ring type capacitance sensor.
b) Solid model.
Fig. (7). Ring type capacitance sensor.
Fig. (8). Geometrical simplification of the concave type capacitance
sensor.
of the operating temperature was taken into consideration in
the calibration tests. The calibration tests were used to
compare the performance of the two types of sensors based
on better sensitivity for the same spatial resolution as
described in details by Ahmed [31].
5.1. Sensor Calibration
Canvas based Phenolic rods with the same dielectric
constant as oil (
=5.45) are used to represent the liquid in the
static calibration at ambient temperature. The solid rods with
Elct(2)
A
Elct(1)
A
Sec A-A
d= 5 mm
50.8 mm
63.5 mm
D= 15.8 mm
z
D
Elect(2)
Elect(1)
63.5 mm8 Recent Patents on Electrical Engineering, 2008, Vol. 1, No. 1 Ahmed and Ismail
a diameter equal to the inside pipe diameter of the test
section (12.7 mm) are machined to mimic different flow
regimes. The lengths of these pieces are the same and equal
to 13 cm. Calibration was carried out for two different flow
patterns, namely annular and stratified flow. To represent
annular flow, holes of different diameter were drilled
through the center of each piece. Nine pieces were generated
to simulate different void fractions. For stratified flow,
eleven separate sections were milled off the rod. The value
of void fraction for each piece can be calculated using the
geometry.
A comparison between the results from the theoretical
models and the experimental data for the off-line calibration
is presented in Figs. (9a & b) for both ring and concave
sensors. The experimental results show that the relation
between the capacitance and void fraction for both types of
sensors is approximately linear. The deviation between the
model and experimental results is due to the assumption of
a) Static calibration for ring type capacitance sensor.
(b) Static calibration for concave type capacitance sensor.
Fig. (9). Static calibration for capacitance sensors.
perfect shielding and also due to a non-uniform electrical
field between the electrodes, with the discrepancy being
higher for the concave type sensor. This is because the
assumption of a constant electric field between the electrodes
is more applicable for the ring type sensor than for the
concave type. The non-linearity of the model results for the
concave type sensor is due to the electrical circuit analogy
which gives rise to a series capacitance circuit, where the
total capacitance is a non-linear function of the individual
capacitances. For the ring type sensors, the total capacitance
between the electrodes is a summation of capacitances in
parallel. The calibration data showed a correlation coefficient of 0.928 and a precision error of 0.04 with 95% confidence level for the ring type sensor, while the agreement
between the experiments and the theory for the concave
sensor was within 18%.
Measurements were performed with the sensors installed
on the air-oil two phase flow loop which discussed in details
by Ahmed [32]. The density and viscosity of the oil used are
973.05 kg/m
3
and 0.026 Pa.s, respectively. The air and oil
are then mixed through an annular mixer. The oil flows on
the outside of an inner perforated pipe, while the air flows in
the inner pipe and enters the oil stream through 380 0.79-mm
diameter perforations. The air-oil mixture passes through the
horizontal test section, which has a total length of 3.5 m.
Tests were performed with single phase oil (a=0) and
single phase air (a=1) to check the range of the output signal
of the sensors. A number of tests were performed for
stratified flow by adjusting the depth of the oil in the pipe to
check the linearity of the capacitance-void fraction relation.
Results show that the sensitivity is the same for both the in
situ and static calibrations, however, the absolute values of
the capacitances at zero and 100% void for the two cases are
off by 5 to 10%. Therefore, the calibration curves were
checked against the dynamic results before the experimental
data were taken. Repeatability tests were also performed and
the results indicated that the variation in void fraction is
between 2% to 3.5%.
The capacitance measurements in the air-oil flow tests
are in the range 0.1 to 15 pF, and proper shielding against
stray capacitance is required to obtain a good signal to noise
ratio (SNR). The range of the noise frequencies due to the
equipment and electrical devices near the test rig were
determined using a frequency analyzer and was found to be
lower than 100Hz in this case. The electronic circuit was
modified to eliminate the environment noise using a high
pass filter.
The void fraction can be related to the normalized
capacitance (Cn) defined as:
Cn
=
Call
oil

Cmeasured
all
oil

all
air
(18)
The correlation between the void fraction and normalized
capacitance is found by applying a linear regression to the
calibration data for the ring type sensor and is given by,

= 0.756 Cn
(19)
with a correlation coefficient of 0.95.
0 0.2 0.4 0.6 0.8 1
1.5
2
2.5
3
3.5
Void fraction (%)
Capacitance (pF)
Model
Annular simulation
Stratified simulation
0 0.2 0.4 0.6 0.8 1
0
0.5
1
1.5
2
2.5
Void fraction (%)
Capacitance (pF)
Model
Annular simulation
Stratified simulationInnovative Techniques For Two-Phase Flow Measurements Recent Patents on Electrical Engineering, 2008, Vol. 1, No. 1 9
5.2. Time Constant
The definition of the time constant of the capacitance
sensor here is referred to the time interval required for the
sensor-meter to change 70% from one state or condition to
another. The time constant for both types of capacitance
sensors was obtained experimentally by applying a unit step
signal and recording the sensor response Fig. (10). The time
constant for both types of capacitance sensors using equation
(20) is found to be approximately 40 μs, corresponding to a
dynamic response of 25 kHz. Fast Fourier transform analysis
of the current flow fluctuations in the present study showed
that the major fluctuations in the void fraction were below 1
kHz, thus the response of the sensors is more than adequate
Yt

Yintial
Ymax

Yintial
= 1
e
T

(20)
5.3. Sensitivity
The sensitivity of the sensor can be defined as:
Senistivity =
Call
liquid

Call
air
all
liquid

all
air
= Call
liquid

Call
air
(21)
and needs to be maximized. The sensitivity depends on the
geometrical shape and gap between the electrodes, which
also affects the spatial resolution of the sensor. The effect of
the design parameters such as the electrode width and
spacing were studied using equations (11) and (12) for both
sensors. For the ring type sensor, the main dimension that
affects the sensitivity is the spacing between the electrodes
(d). The results show that the sensitivity increases as the
spacing decreases Fig. (11), which also results in a better
spatial resolution, with the only limitation being in the
fabrication of the sensor. For the concave sensor, the
sensitivity increases as the electrode separation (z) decreases
Fig. (12a), and the electrode length increases. However,
increasing the electrode length leads to a poor spatial
resolution as shown in Fig. (12b). In general, the sensitivity
of the ring type sensor is found to be higher than the concave
type for the same spatial resolution. The sensitivity of the
ring sensor was found to be approximately 0.75 pF, while for
the concave sensor it was 0.6 pF.
Fig. (10). Capacitance Sensor response to an input step function.
Fig. (11). Effect of electrode spacing on the sensitivity of ring type
sensor.
(a)Effect of electrode separation.
(b) Effect of electrode length.
Fig. (12). The effect of sensor dimensions on the sensitivity for
concave type sensor.
-2
-1
0
1
2
0.4 0.45 0.5 0.55 0.6
Time (µs)
Volt (v)
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
0
5
10
15
d/D
Sensit vi ity (pF)
Elct(2)
d
Elct(1
D
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.2
0.4
0.6
0.8
1
z/D
Sensit vi ity (pF)
z
D
Elect(2
Elect(1
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.2
0.4
0.6
0.8
1
z/D
Sensit vi ity (pF)
z
D
Elect(2
Elect(110 Recent Patents on Electrical Engineering, 2008, Vol. 1, No. 1 Ahmed and Ismail
The effect of the flow regime on the sensitivity of the
capacitance sensor can b estimated. For example, in the case
of elongated bubbles, the sensitivity can be calculated using
equation (18) along with equation (21). The sensitivity as a
function of void fraction for different bubble length to
electrode spacing length ratios for an electrode spacing of 10
mm is shown in Fig. (13). It should be noted that the
sensitivity is also affected by the electrode spacing and for
the ring type sensor an electrode spacing of less than 2 mm is
required for a high sensitivity. For this electrode spacing, the
effect of (l X ) is found to be neglected, which means that
the sensor gives an approximate volumetric void fraction
equal to the cross sectional value for bubbly flow.
Fig. (13). The sensitivity of ring type sensor in case of elongated
bubble flow.
5.4. Measurement Uncertainty
The total flow rate through the test section was maintained within +2% of the average value while the void fraction measurements were taken. The main source of uncertainty in the void fraction measurement was the noise in the
sensor signal from the surrounding equipment such as the oil
pump and the biases of the voltage signal, sensor spacing
and position. This uncertainty was calculated to be in the
range of +6% over the entire range of void fraction.
6. FLOW PATTERN IDENTIFICATION
Flow pattern identification can be performed either by
visual inspection of the flow in a transparent pipe or by
measuring and quantifying the fluctuations of the flow
parameters such as void fraction (Jones and Zuber [30]) or
dynamic pressure (Keska and Williams [33]), which reflect
the flow structure. The flow pattern recognition from the
signal fluctuations can be done by analyzing the probability
density function (PDF) or power spectral density function
(PSD) of the time trace signal [30]. In the present study, the
applicability of the capacitance sensors for flow pattern
identification was investigated. Only the ring type sensor is
considered here since it has a better sensitivity than the
concave type for the same spatial resolution. The time
averaged void fraction and the probability density function
(PDF) of the void fraction signal is used in this instance to
identify the flow pattern. The flow regimes are also obtained
using a high speed camera for validation. Three main flow
regimes and their corresponding PDF distributions are
discussed here for two-phase air-oil through a horizontal
pipe. These three basic flow regimes are: elongated bubble,
slug and annular flow. All the data presented was collected
at 2 kHz over a period of 50 sec.
Figures (14a,b & c) illustrate the time trace, PDF and
void fraction signal respectively for a typical slug flow while
an image of the corresponding flow pattern using the high
speed video system is presented in Fig. (14d). The time trace
is characterized by an intermittent void fraction signal that
fluctuates between a high and low value. The high void
fraction (a=0.29) correspond to the passage of a long gas
bubble, while the low void fraction indicates the passage of
the liquid slug. Small bubbles in the tail of the long bubbles
gives rise to the low void fraction (a=0.04). Although the
capacitance sensor measures the volumetric void fraction
over a tube length of 1.65 D, the sensor seems to be sensitive
to bubbles that are entrained in the liquid slugs as can be
seen in the time trace signal. In order to distinguish between
two signals having double peaks on the PDF diagram, the
power spectral density (PSD) was used. In this case,
additional information about the flow structure was obtained
such as the bubble and liquid slug frequencies. The power
spectral density (PSD) for the slug flow signal shown in Fig.
(14c) is characterized by two dominant frequencies, corresponding to the liquid slug and the gas bubble frequencies. In
Fig. (14c), the slug frequency is approximately 118 Hz,
while the frequency of the small bubbles is 228 Hz. For
elongated bubble and slug flows, a double peak PDF is
obtained, one at low void fraction corresponding to the liquid
slugs and the other at high void fraction due to the gas slugs.
The time trace and PSD in this case are used to distinguish
between the two cases. For elongated bubbly flow, the void
peaks on the time trace signal occur at a higher frequency
than for slug flow (Fig. (15a) through d). This is indicated in
the power spectral density (PSD) of the signal where a
higher frequency of 394 Hz is obtained for the case of
elongated bubble flow Fig. (15c), while a frequency of 228
Hz is obtained for the case of slug flow Fig. (14c).
The PDF for annular flow is characterized with a single
peak at high void fraction as shown in Fig. (16b), while the
time trace shows small oscillations around a high mean value
of void fraction (a=0.54) indicating the unsteady surface
waves of the liquid film. It should be noted that, as the peak
in the PDF becomes narrower, the liquid film thickness
becomes more nearly constant.
7. CURRENT & FUTURE DEVELOPMENTS
Electrical capacitance is an innovative technique for
measurement of two-phase flows. In this paper, a review
including most recent patents developed for two-phase flow
measurements using capacitance-based sensor was presented. Also, in this work, a novel systematic method for the
design of capacitance sensors for void fraction measurement
and flow pattern identification was developed and discussed
in detail. Two types of capacitance sensors, ring and concave
type, have been analyzed. The phase or void distribution is
represented by an equivalent capacitance circuit to estimate
= 16.0
d
l
= 4.0
d
l
= 7.0
d
l
Sensitivity (pF)
0 0.2 0.4 0.6 0.8 1
0.1
0.2
0.3
0.4
Void fraction
= 16.0
d
l
= 4.0
d
l
= 7.0
d
l
Sensitivity (pF)
0 0.2 0.4 0.6 0.8 1
0.1
0.2
0.3
0.4
Void fractionInnovative Techniques For Two-Phase Flow Measurements Recent Patents on Electrical Engineering, 2008, Vol. 1, No. 1 11
Fig. (14). (a) Time trace signal, (b) PDF. (c) PSD and (d) Flow image for intermittent (slug) flow regime (GL1 = 605 kg/m
2
.s and x=0.0077).
Fig. (15). (a) Time trace signal, (b) PDF, (c) PSD and (d) Flow image for elongated bubble flow regime (GL1 = 790 kg/m
2
.s and x=0.001).
0 2 4 6 8 10
0
0.2
0.4
0.6
0.8
1
Time (Sec)
Void Fraction
0 0.2 0.4 0.6 0.8 1
0
5
10
15
20
25
30
Void Fraction
PDF (%)
10
1
10
2
10
3
10
4
10
-3
10
-2
10
-1
10
0
10
1
Freqency (Hz)
PSD
1 in
(a) (b)
(c)
(d)
0 1 2 3 4 5
0
0.2
0.4
0.6
0.8
1
Time (Sec)
Void Fraction
0.2 0.4 0.6 0.8 1
0
5
10
15
20
25
30
Void Fraction
PDF (%)
10
1
10
2
10
3
10
4
10
-3
10
-2
10
-1
10
0
10
1
F (Hz)
PSD
1 in
(a) (b)
(c) (d) 12 Recent Patents on Electrical Engineering, 2008, Vol. 1, No. 1 Ahmed and Ismail
Fig. (16). (a) Time trace signal, (b) PDF, (c) PSD and (d) Flow image for annular flow regime (GL1 = 395 kg/m
2
.s and x=0.15).
the sensitivity of both concave and ring type sensors. The
sensitivity of the ring type sensor increases as the separation
distance between the electrodes decreases, which also
increases the spatial resolution of the sensor. The sensitivity
of the ring type sensor is found to be higher than the concave
type for the same spatial resolution. Both types of sensors
were fabricated and tested in an air-oil flow loop. The void
fraction predictions from the theoretical models for the
sensor design are within 15% of the experimental data.
Calibration of the sensors shows that the output capacitance
is linearly related to the void fraction for both types of
sensors. The volumetric average void fraction is independent
of the flow pattern for both ring type and concave type
sensors, which is useful to obtain accurate measurements of
average void fraction. The capacitance sensors are very
sensitive to flow regime change and can be used to identify
the flow pattern. The mean value and the probability density
function of the instantaneous void fraction signal is used to
identify the flow regime. Both the slug and elongated bubble
flow patterns are characterized by a double peak PDF with
one peak at a high void fraction and one at a low void
fraction. However, the time trace signal and the power
spectral density (PSD) can be used to distinguish between
the two flow regimes. The time trace signal shows that, for
elongated bubble the fluctuations occur at a higher frequency
than for slug flow, which is reflected in the power spectral
density (PSD). A single sharp peak in the PDF at high void
fraction characterizes the annular flow pattern.
NOTATION
A = Area
D = Diameter (mm)
E = Electric field

= Void fraction
l = Electrode length (mm)
t = Pipe wall thickness (mm)
T = Time (sec)
V = Potential difference (volt)
 = Time constant (sec)
 = Dielectric constant
SUBSCRIPT
o = Refer to free space
c = Charge
P = Polarization charge
L = Liquid
G = Gas
T = Total
w = wall
REFERENCES
[1] Lowe DC, Rezkallah KS. A capacitance sensor for the
characterization of microgravity two-phase liquid-gas flows. Meas
Sci Technol 1999; 10: 965-975.
[2] Boothroyed RG. Flowing gas-solids suspensions. Chapman & Hall,
London, 1971.
[3] Green RG, Thorn R. Sensor systems for lightly loaded pneumatic
conveyors. Powder Technol 1998; 95: 79-92.
[4] Jaworski AJ, Dyakowski T. Application of electrical capacitance
tomography for measurement of gas-solids flow characteristics in a
pneumatic conveying system. Measurem Sci Technol 2001; 12:
1109-1119.
[5] Ostrowski K, Luke SP, Bennett MA, Williams RA. Real-time
visualization and analysis of dense phase powder conveying.
Powder Technol 1999; 102: 1-13.
[6] Sen S, Das PK, Dutta PK, Maity B, Chaudhuri S, Mandal C, Roy
SK. Pc-based gas-solids two-phase mass flow meter for
0 5 10
0
0.2
0.4
0.6
0.8
1
Time (Sec)
Void Fraction
0 0.5 1
0
5
10
15
20
25
30
Void Fraction
PDF (%)
10
1
10
2
10
3
10
4
10
-3
10
-2
10
-1
10
0
10
1
Frequency (Hz)
PSD
1 in
(a) (b)
(c) (d) Innovative Techniques For Two-Phase Flow Measurements Recent Patents on Electrical Engineering, 2008, Vol. 1, No. 1 13
pneumatically conveying systems. Flow Measurem Instrument
2000; 11: 205-212.
[7] Harvel GD, JS. Chang. Electrostatic multiphase flow measurement
techniques. Handbook of Electrostatic Processes ch.13 J. S. Chang,
A. J. Kelly, and J. M. Crowley, eds. Marcel Dekker, Inc., New
York, 1995.
[8] Barratt IR, Yan Y, Byrne B, Bradley MSA. Mass flow
measurement of pneumatically conveyed solids using radiometric
sensors. Flow Measurem Instrument 2000; 11: 223-235.
[9] Morse TD, Bellou CD. The uniformity of fluidization, its
measurements and use. Chem Engng Prog 1951; 47: 199-211.
[10] Lanneau KP. Gas-solid contacting in fluidized beds. Trans Instn
Chem Engrs 1960; 38: 125-143.
[11] Chang JS, Girard R, Raman R, Tran EBP. Measurement of void
fraction in vertical gas-liquid two-phase flow by ring type
capacitance transducer. Mass flow Measurements. ASME Press,
New York, 1984.
[12] Tortora PR, Ceccio SL, Trujillo SM, O’Hern TJ, Shollenberger
KA. Capacitance measurements of solid concentration in gas-solid
flows. Powder Technol 2004; 148: 92-101.
[13] Dyakowski T, Edwards RB, Xie CG, Williams RA. Application of
capacitance tomography to gas-solid flows. Chem Eng Sci 1997;
52(13): 2099-2110.
[14] Dyakowski T, Jeanmeure LFC, Jaworski AJ. Applications of
electrical tomography for gas-solids and liquid-solids - a review.
Powder Technol 2000; 112: 174-192.
[15] Brown GJ, Reilly D, Mills D. Development of an ultrasonic
tomography system for application in pneumatic conveying. Meas
Sci Technol 1996; 7: 396-405.
[16] Steiner G, Wegleiter H, Watzenig DA. A dual-mode ultrasound and
electrical capacitance process tomography sensor. IEEE Sensors,
Irvine, CA, USA, page Cat. No. 05CH37665C, 2005.
[17] Jiang T, Xiong Y. Measurement of the velocity and mass flow rate
in gas-solid flow with electrostatic method. J Huazhong Univ Sci
Tech 2005; 33 (1): 93-95.
[18] Andrade, T.L.: US2007134042 (2007).
[19] Huang SM, Plaskowski AB, Xie CG, Beck MS. Tomographic
imaging of two-component flow using capacitance sensors. J Phys
E Sci Instrum 1989; 22: 173-177.
[20] Merilo M, Dechene RL, Cicowlas WM. Void fraction measurement
with a rotating electric field conductance gauge. J Heat Transfer
Trans ASME 1977; 99: 330-331.
[21] Lucas GP, Simonian S. Towards a phase-distribution-independent
impedance volume-fraction meter. Flow Measurem Instrument
1991; 2: 105-114.
[22] Geraets JJM, Borst JC. A capacitance sensor for two-phase void
fraction and flow pattern identification. Int J Multiphase Flow
1988; 14: 305.
[23] Elkow KJ, Rezkallah KS. Void fraction measurment in gas-liquid
using capacitance sensors. Meas Sci Technol 1996; 7: 1153.
[24] Gamio JC, Castro J, Rivera L, Alamilla J, Garcia-Nocetti F,
Aguilar L. Visualisation of gas-oil two-phase flows in pressurized
pipes using electrical capacitance tomography. Flow Measurem
Instrument 2005; 16: 129-134.
[25] Wang, M.C.: US2007570054 (2007).
[26] Kendoush AA, Sarkis ZA. Improving the accuracy of the
capacitance method for void fraction measurement. Exp Thermal
and Fluid Sci 1995; 11: 321.
[27] Hammer EA, Tollefsen J, Olsvik K. Capacitance transducers for
non-intrusive measurement of water in crude oil. Flow Measurem
Instrument 1989; 1: 51-58.
[28] Heerens WC. Application of capacitance techniques in sensor
design (review article). J Phys E Sci Instrum 1986; 19: 897-906.
[29] Atsushi, K.C.: US2007/0103171A1 (2007).
[30] Jones OC, Zuber N. The interrelation between void fraction
fluctuations and flow pattern in two-phase flow. Int J Multiphase
Flow 1975; 2: 273.
[31] Ahmed WH. Capacitance sensors for void fraction measurements
and flow pattern identification in air-oil two-phase flow. IEEE
Sensors J 2006; 6(5): 1153-1163.
[32] Ahmed WH, Ching CY, Shoukri M. Characteristics of air-oil twophase flow across a sudden expansion. In Proc. 4
th
ASME/JSME
Joint Fluid Eng Conf volume 2B, 1765: 1772, Honolulu, HI, USA,
July 6-10, 2003.
[33] Keska JK, Williams BE. Experimental comparison of flow pattern
detection techniques for air-water mixture flow. Exp Thermal &
Fluid Sci 1999; 19: 1-12.