Sunteți pe pagina 1din 15

Flow Measurement and Instrumentation 16 (2005) 113127

www.elsevier.com/locate/flowmeasinst

Measurements of oilwater separation dynamics in primary separation


systems using distributed capacitance sensors
Artur J. Jaworskia,, Tomasz Dyakowskib
a School of Mechanical, Aerospace and Civil Engineering, University of Manchester, Sackville Street, PO Box 88, Manchester M60 1QD, United Kingdom
b School of Chemical Engineering and Analytical Science, University of Manchester, Sackville Street, PO Box 88, Manchester M60 1QD, United Kingdom

Received 29 July 2004; received in revised form 24 January 2005; accepted 14 February 2005

Abstract
This paper provides an overview of research activities conducted and results obtained within an industrially driven research programme
managed through UMIST (currently The University of Manchester) over the recent years. Its main objective was to design, construct
and demonstrate new instrumentation technologies which could allow reliable measurement of the separation process in the oil and gas
extraction industries with the view of transferring the technology developed in academia into the oilfield environment. The paper describes
the developments starting from the early work, which aimed at the laboratory qualification of the distributed electrical capacitance sensors
embedded in the parallel-plate interceptor of the transparent oilwater separator located at the university Pilot Plant. This is followed by the
description of the design, analysis and industrial evaluation of the pre-prototype capacitance probe, containing 24 distributed sensors, in the
live oilfield conditions of the BP Exploration facility at Wytch Farm, Dorset, UK. Finally, the most recent advances, including industrial tests
of the prototype multi-electrode capacitance probe carried out in the National Engineering Laboratory in Glasgow are reported. The probe
was designed and built to include a mechanical deposit prevention system and to meet the flame proof standards of the oil and gas extraction
industry. The three case studies presented demonstrate the process of evolution of the measurement technology, which originally stemmed
from ECT, and its transfer into the live process environment.
2005 Elsevier Ltd. All rights reserved.

1. Introduction and background


This paper presents an overview of activities focused on
the design, construction, prototyping and industrial testing
of instrumentation for on-line and in situ characterisation
of separation dynamics in the three-phase horizontal
separators, common to many oil and gas extraction
installations, both on-shore and off-shore, to separate the
gas, oil and water components of heterogeneous mixtures.
This work stemmed from the research activities related
to the applications of industrial process tomography at
UMIST. In particular, the measurement techniques described
are based on chargedischarge circuits developed for the
electrical capacitance tomography (ECT). Also in a broad
sense the measurement devices discussed make use of
Corresponding author. Tel.: +44 (0) 161 275 43 52.

E-mail address: artur.jaworski@manchester.ac.uk (A.J. Jaworski).


0955-5986/$ - see front matter 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.flowmeasinst.2005.02.012

arrays of distributed sensors, which, although different from


applying electrodes on the circumference of a pipe or
a vessel as is the case in tomography, enable far more
comprehensive imaging of multiphase flow than pointwise sensors currently dominating industrial applications.
The activities described in this paper began at UMIST
towards the end of 1995, under research funding obtained
from the Engineering and Physical Sciences Research
Council (EPSRC), grant number GR/L10994 and Design
and Instrumentation for Process Separation Systems
(DIPSS): a managed research programme between UMIST,
University of Sheffield, University of Nottingham and others
funded by a consortium of oil and gas extraction and
petrochemical companies. Some background work carried
out at UMIST dates from even earlier as described for
example in Ref. [1]. The majority of the scientific research
and development in the area of electrical capacitance
measurements described here was carried out between

114

A.J. Jaworski, T. Dyakowski / Flow Measurement and Instrumentation 16 (2005) 113127

1996 and 1999. In 2000 the activities were focused


on commercialisation of the products developed earlier
through formation of a UMIST spin-off company: Oilfield
Technology Ltd (OTL), that has obtained official funding
for further development of a capacitance based level sensor,
which incorporates a device for preventing the sensors
fouling. The latter activities culminated in industrial tests
carried out in May 2004.
The paper is organised in several sections which include:
the introduction to the primary separation processes,
followed by the principles of electrical measurements
adopted in the probes and sensors presented in this
paper, and case studies, which illustrate the development
of measurement techniques. There are three case studies
discussed here which include the following: (i) development
of capacitance sensors embedded in a parallel plate
interceptor, capable of on-line and in situ monitoring of the
oilwater separation process in a research separator at the
UMIST Pilot Plant Laboratory; (ii) design of a segmented
capacitance probe for use in gasoilwater separators and
subsequent trials conducted at the BP oil and gas extraction
facility, Wytch Farm in Dorset, UK; (iii) tests of the OTL
segmented capacitance probe, conducted at NEL, which
utilises electrical capacitance measurement, and includes the
deposit prevention mechanism. The section containing the
case studies will be followed by a short section including
summary, conclusions and future work.
2. Overview of measurement challenges in primary
separation processes in oil and gas extraction
Fig. 1 shows a simplified schematic of a three-phase
separator. A mixture of gas, oil and water (very often
accompanied by solid particulates) enters the vessel from the
left. The mixture is separated under the action of buoyancy
forces, thanks to a sufficiently long residence time, and
then divided into individual streams by means of a weir
arrangement. Separation of gas from the system is relatively
easy due to a large difference in density between gas and
liquid phases although it is known in some instances that
foam (a mixture of gas and oil) which has not completely
collapsed, may be carried over resulting in the presence of
micron-size liquid droplets in the gas stream.
A far more difficult and cumbersome issue in the
operation of horizontal separator vessels is the presence of
oilwater mixtures referred to as emulsions. The physics
of the oilwater separation is governed, in the micro-scale,
by two main mechanisms: settling and coalescence [2].
Both depend very strongly on the droplet diameter of the
dispersed phase. In broad terms, settling relies on the density
difference between separating phases, which in the case
of oil and water is relatively low and thus makes the
separation more difficult. In addition to this, the drag force
experienced by the droplets is governed by Stokess law,
which implies that the smaller the droplet diameter the
more difficult it is for the droplet to join the continuous

Fig. 1. Schematic of a three-phase water separator.

phase. Droplet coalescence is the mechanism which allows


the drops to become bigger by joining together and thus
speed up the separation process, however the ability of
the droplets to coalesce is very often impeded by various
chemical compounds typically present in the crude oil [3],
as well as chemical additives introduced into the process
such as defoamants or corrosion inhibitors [4]. Separation of
solids is not discussed in this paper, although it needs to be
appreciated that this also provides a significant engineering
challenge.
In the macro-scale, the dynamics of the droplet
settling and coalescence within an oilwater separator
can be characterised by the position of the interfaces
between oil, oilwater mixture and water. In particular the
measurement of the thickness of the unseparated mixture
layer and its variation along the length of the oilwater
separator can provide crucial information about the separator
performance. From the separator operation point of view, the
composition and the stability of oilwater mixtures, entering
a vessel, may vary widely over relatively short periods.
Therefore, precise control over the levels of individual
phases inside a vessel is required to prevent accidental
cross entrainments of one phase into the production stream
of another, as this can seriously affect the rigs overall
efficiency. Monitoring of these macro-scale properties of the
flow inside the separator in an on-line fashion provides the
motivation for the research and development work presented
in this paper. It should be emphasised, however, that there are
no fundamental difficulties in applying the instrumentation
described in this paper to many other industrial processes.
3. Measurement principles
There are a number of techniques available which could
be applied to monitor the multiphase flows or measure
the phase distribution within the three-phase separators.
These include, for example, electrical capacitance, electrical
conductivity, ultrasonic, gamma ray or microwave methods;
excellent reviews of these methods were given in
Refs. [57]. The selection of a particular measurement
technique has to be based on a thorough analysis of all
contributing factors. These include, inter alia, the ability
of the technique to distinguish between various phases

A.J. Jaworski, T. Dyakowski / Flow Measurement and Instrumentation 16 (2005) 113127

and provide quantitative information about mixture content;


practicality of the sensors for a given application (e.g.
their size, robustness etc.); safety issues (e.g. radiological
or explosion hazards involved). In the work presented,
the authors selected the electrical techniques primarily due
to the low cost of sensor manufacture, their robustness
and reliability, and ease of sensor implementation within
the restricted spaces of the separator internals. The high
contrast between the dielectric properties of gas, oil and
water played an equally important role.

Electrical measurements of the volume fraction of one


liquid in another in the heterogeneous mixtures were the
subject of numerous studies [8,9]. The theoretical work
attempting to link the electrical properties of a two-phase
mixture with the volume fraction of one material dispersed
within another was first presented by Maxwell [10]. In his
calculations Maxwell assumed that small spheres of one
material were uniformly distributed in the continuous phase
of another material, and that an otherwise homogeneous
electrical field was disturbed by their presence. The spheres
were assumed to be of equal diameter, and small compared
to the distance between them. The Maxwell model resulted
in the following relation
eff = 1

21 + 2 2cv (1 2 )
.
21 + 2 + cv (1 2 )

dielectric dispersion to indicate that electrical properties


differ in low and high frequency regions. An example of such
an analysis can be found in the work of Hanai [9].
For the purpose of such analysis the dielectric
permittivity and electrical conductivity are discussed
simultaneously, in terms of complex dielectric permittivity
. This concept is fully explained by Hanai [9]. Generally,
in a system of two electrodes the electrical charge is induced
by polarisation and conducted through the medium. So the
total charge is
Q = U C + U G I ( j ) = U [C + G I ( j )]

3.1. Electrical properties of heterogeneous mixtures

(1)

Here 1 and 2 denote the dielectric permittivity of the


components, eff the effective dielectric permittivity of
the mixture, and cv the volume fraction of material 2
in material 1. Similar formulae were also obtained for
calculating the effective conductivity of a mixture. Many
others, [1113], developed various models for calculation
of the electrical properties of mixtures of two different
materials. In many of these approaches heterogeneous
systems are treated as either ideal dielectrics or as purely
conducting materials. This is equivalent to solving the
electrostatic problem for two materials with dielectric
properties 1 and 2 , without considering the electrical
conduction, or conversely, to solving a DC conduction
problem for materials with conductivities 1 and 2 ,
while neglecting dielectric properties. Moreover, as the
electrostatic and DC problems have similar mathematical
form the formulae derived are mathematically similar. When
looking at Eq. (1) or similar formulae, an obvious property is
the lack of frequency dependence of the electrical properties.
In reality none of the heterogeneous systems is an
ideal dielectric or ideal conductor and the full analysis,
taking into account all material properties i.e. 1 , 2 ,
1 and 2 should be conducted. When such analysis
is performed it is apparent that the effective electrical
properties are functions of frequency due to the phenomenon
called interfacial polarisation (induction of charges on fluid
internal interfaces). Mathematically this effect is called

115

(2)

where U , I , C, G and are voltage, current, capacitance,


conductance and frequency, respectively. The same relationship can be presented as
Q = U C

(3)

where, by definition, C is the complex capacitance


understood as
C = C j (G I ) = C j C  .

(4)

In this form, the conductive phenomena are seen as the


imaginary part of the complex capacitance.1 Considering the
arrangement of two flat plate electrodes of area S, spaced
at distance d from one another, the complex capacitance
can be linked to the dielectric permittivity, , and electrical
conductivity, , by the following relationship



(5)
C = 0 +
j 0 d
where 0 is the dielectric permittivity of the vacuum. Thus
complex permittivity is defined as



= j
(6)
= j  .
=+
j 0
0
Following the analysis originally conducted by Wagner [14], it can be shown that the complex permittivity of
the mixture is linked to the volume fraction of dispersed
spheres of one fluid (with permittivity 2 ) in another (with
permittivity 1 ) by the following equation
21 + 2 2cv (1 2 )
21 + 2 + cv (1 2 )
l
l h
+
= h +
1 + j
j 0

= 1

(7)

or in terms of complex conductivity


21 + 2 2cv (1 2 )
21 + 2 + cv (1 2 )
j (h l )
+ j 0h
= l +
1 + j

= 1

(8)

1 It is also possible to introduce the concept of complex conductance


and treat capacitance as the imaginary part of it, i.e. G = jC .
Similarly, complex permittivity can be replaced by complex conductivity
= j0 .

116

A.J. Jaworski, T. Dyakowski / Flow Measurement and Instrumentation 16 (2005) 113127

Fig. 2. Permittivity and conductivity of oilwater mixtures as a function of


water cut [17].

where coefficients h , l , h , l and are as defined


in [14]. From Eqs. (7) and (8) it can be seen that both the
permittivity and the conductivity of a mixture are functions
of frequency of the applied electrical field. Moreover it can
be shown that Eq. (1) represents a limit of Eq. (7) for high
frequencies. Similar considerations of the electrical models
for heterogeneous mixtures were developed by Rao and
Ramu [15] and Ramu and Rao [16].
Of course, for practical applications, formulae such as
those given by (7) and (8) can be simplified after considering
the orders of magnitude of all parameters. For example
Hammer et al. [17] suggested simplified expressions for
permittivity and conductivity of mixtures of crude oil and
process water in a frequency independent form. For the oil
continuous phase the permittivity and conductivity are given
as:
1 + 2
1 + 2
o
; mo = oil
.
(9)
= oil
m
1
1
For the water continuous phase the permittivity and
conductivity are given as
w
m
= water

2
;
3

mw = water

2
.
3

(10)

Here is the water fraction in the mixture. Eqs. (9) and


(10) are illustrated in Fig. 2 taken from Ref. [17].
3.2. Summary of current methods for sensing interface
levels
In the simplest case of a single-phase detection a
capacitance based device for phase detection may consist
of a single electrode made in the form of a dipstick.
Here the varying interface level is determined from the
varying capacitance to earth. This type of probe is widely
marketed by a number of manufacturers, such as described in
Refs. [1820]. Of course, the introduction of multiple phases
(such as oil, water and their heterogeneous mixtures) into
the vessel, equipped with these devices, leads to ambiguous
readings.

Fig. 3. Examples of probes for interface level detection: (a) single-electrode


probe, (b) arrangement of separate cells, (c) multiple source electrodes
with a common detector.

Several possible solutions to this problem have been


suggested. Refs. [21] and [22] describe two designs
of multiple capacitance sensors forming vertical sets of
identical cells. In this approach, the detection of individual
phases is quite straightforward. The individual sensors (or
cells) are evenly distributed across all phases and the
resulting set of output signals allows capacitance profiles
to be obtained. These are directly related to the configuration
of layers inside a vessel. Yang et al. [1] developed a slightly
different design, which utilises an array of excitation (or
source) electrodes facing a common detection electrode. The
above mentioned probe designs are shown schematically
in Fig. 3. Schller et al. [2325] proposed a probe in the
form of a cylinder, with a number of ring-like segmented
electrodes along the length, which gives a profile of
capacitance between each electrode and infinity. The spatial
resolution of the devices based on the concept of segmented
electrodes is of the order of half of the electrode spacing.
In general, the capacitance methods work very well for
detection of multiphase flows of dielectric materials such as
gas, foam and oil. The conducting fluids, however, such as
produced water, may short the sensors so that effectively
the measured capacitance becomes 50% of the value for
the electrode insulation. The appearance of this problem is
certainly related to the actual electronic circuitry used for
measurements. For example, Schller et al. [2325], seem
to be able to measure the oilwater mixture composition, for
both oil-continuous and water-continuous mixtures. Others
have used electrical admittance measurements in order to
obtain both components of electrical impedance electrical
capacitance and conductance and to be able to monitor
mixtures of arbitrary continuity [26].
Somewhat outside the scope of this paper, it is worth
mentioning that there are alternative devices for measuring
the distribution of phases in separators, based on other
techniques. Tracerco have constructed a vessel profiler
based on an array of gamma-ray sourcedetector pairs [27],
while ABB [28] have reported a probe with distributed

A.J. Jaworski, T. Dyakowski / Flow Measurement and Instrumentation 16 (2005) 113127

117

measurement cells based on inductive circuits to measure


electrical conductivity of oilwater mixtures.
4. Case studies
4.1. Application of distributed capacitance sensors in the
parallel-plate interceptor
In the case study presented in this section, the capacitance
sensors have been incorporated in the packed bed of parallel
plates of the Pilot Plant separator at UMIST. The packed
bed, also known as a parallel-plate interceptor (PPI), is a
common feature of many commercial vessels [29]. It is
usually located in the vessels central part, downstream of
the flow distributors and consists typically of a series of
plates, parallel to the vessel axis, but inclined at an angle
of 45 to the horizontal, thus dividing the separator into
a number of individual channels. It is believed that the
top and bottom walls of an individual channel provide a
means for the coalescence process to take place on their
surfaces and effectively shorten the path of particles of the
dispersed phase [2]. Locating the sensors within the packed
bed seemed a natural choice because of its important role in
an effective separation process.
The work presented had two objectives. Firstly, it served
as a proof of concept in early stages of development of
the capacitance probes discussed later, by addressing the
need for monitoring oilwater separation processes in situ
and on-line. Secondly, it was believed that the instrument
developed could provide an insight into the physics of
the separation processes. Modelling of these processes is
exceedingly difficult due to very complex boundaries of
the separator vessels and the resulting flow patterns, as
well as the still imperfect models of droplet settling and
coalescence, which are available at the moment. Zhang and
Davies [30] showed, for example, that, in a single channel,
there is strong re-circulation due to buoyancy forces, in the
cross section perpendicular to the main flow, which reduces
the drop settling velocity in part of the cross section and
results in a reduction of the separation rate. The capacitance
instrument described could in future enable the experimental
validation of such numerical calculations.
4.1.1. Experimental facilities
The case study on monitoring oilwater separation was
conducted at the Pilot Plant facility at UMIST, equipped
with a cylindrical horizontal oilwater separator (Fig. 4)
2.55 m long and 0.75 m in diameter. The separator is
made of transparent PVC, and works under atmospheric
pressure. The oil (Shell Catenex 11) and tap water used
in the experiments had relative permittivity 2.1 and 79
respectively; conductivity 1013  cm1 and 3.2
106  cm1 , respectively.
They were stored in two separate tanks (2 m3 each).
The media were introduced into the flow loop by means
of two pumps with fully controlled flow rates. An in-line

Fig. 4. Transparent oilwater separator (seen with capacitance meter inside


a metal box).

Fig. 5. Details of the Pilot Plant separator layoutdimensions in


millimetres.

mixer located just before the inlet to the separator provided


a high shear required to produce fine oilwater dispersions.
The typical drop size distribution was between 100 and
200 m. The oilwater mixture entered the separator through
a 38 mm inlet, at the centre of the circular endplate (Fig. 5).
The flow was distributed by a baffle plate, after which it
entered a PPI. At the downstream end, oil left by a pocketlike weir arrangement, while water left by the outlet located
at the bottom of the separator. Separated oil and water
were recycled back to the storage tanks. The flow rig could
usually operate for a few hours without signs of phase crossentrainment.
The parallel-plate interceptor (PPI) inside the UMIST
separator was made of 18 PVC plates, which were 0.5 m
wide, 1.0 m long and 3 mm thick. Eight stainless steel rods
supported the plates, which were placed one above another
with 25 mm spacing in between. The whole arrangement
was placed at an angle of 45 to the horizontalthe plane

118

A.J. Jaworski, T. Dyakowski / Flow Measurement and Instrumentation 16 (2005) 113127

Fig. 7. Capacitance readings as a function of water contents and phase


continuity.

Fig. 6. Position of sensors distributed within the PPI.

of rotation being the transverse cross plane of the separator,


as shown by arrow A-A in Fig. 5.
4.1.2. Fabrication of sensor arrays and measurement
technique
To demonstrate the concept of non-invasive embedded
sensors, it was decided to replace two of the original plates
from the middle of the PPI with a pair of plates with
embedded capacitance electrodes. One of the new plates
(Fig. 6) was equipped with 54 detection electrodes, arranged
in three arrays. These were located close to the inlet (array A,
electrodes 118), in the middle (array B, electrodes 1936)
and close to the outlet from the PPI (array C, electrodes
3754).
The electrode arrays were manufactured as flat inserts
using multi-layer PCB (Printed Circuit Board) technology.
The electrode layer was coated with a 0.5 mm thick epoxy
laminate during PCB manufacture to separate conductors
from the fluids. The inserts were attached inside grooves
machined in the plates, in such a way that the PCB surface
was flush with the plate surface. The second plate contained
three large excitation electrodes, also fabricated in PCB
technology, located opposite the detecting electrode arrays.
The principles of designing the capacitance electrodes
such as those described above are given in detail in Ref. [31].
The working principles of the capacitance measurements
applied are described in detail in Refs. [32,33]. In brief,
the electrical field, imposed by the voltage applied to the
excitation electrodes, scans the space between the two
plates. The electrical charge induced on individual detection
electrodes provides a measure of electrical capacitance (and
thus average dielectric permittivity) of a small cuboidal
region between each detection electrode and the common

Fig. 8. Orientation of phases in a single PPI channel.

Fig. 9. PPI viewed from the separator downstream end. Two plates in the
centre are embedded with sensors.

excitation. As the relation between electrical capacitance


measured and mixture composition is known from ex situ
static calibration (Fig. 7), it is possible to deduce the phase
distribution between plates as schematically shown in Fig. 8.
The photograph in Fig. 9 shows a fragment of the PPI,
equipped with two plates with capacitance sensors, viewed
from the downstream end of the separator.

A.J. Jaworski, T. Dyakowski / Flow Measurement and Instrumentation 16 (2005) 113127

119

Fig. 10. Measurement output on the control PC. Oilwater interface was kept at 15 cm below the weir. (a) Static test; (b) dynamic testtotal flow rate of
120 l/min; oil contents 25%.

The measurements were made using a commercial capacitance meterPTL 120, capable of collecting measurements
from 64 individual channels with an accuracy of 1 fF. The
measurement frequency of the meter was 1 MHz (square
wave). The sampling time per channel was 150 ms. Fig. 4
shows the capacitance meter (contained in a metal box) connected to the plates inside the separator by a series of miniature coaxial cables. These were attached to the edges of the
PCBs and fed through three nozzles in the separator wall.
The output from the measuring unit was displayed in numerical and graphical form on the control PC, located in the
safe area of the Pilot Plant, about 30 m from the separator.
4.1.3. Sample results
The measurement system was subject to extensive static
tests reported fully elsewhere [32,34], and not repeated here.
In these tests, the positions of the interface levels inside
the pack of plates were established simply by looking for
the sharpest local increase in the capacitance between two
neighbouring electrodes (Cn Cn1 ). The accuracy of
interface detection in this way was approximately half of
the electrode height in the vertical direction (in this case
better than 10 mm). The levels determined agreed very
well with the visual observations through the transparent
separator walls, which could be made with an accuracy of
approximately 34 mm.
During the tests in dynamic flow conditions, the mixture
of oil and water was pumped into the separator vessel. The
progress of the separation process was monitored on-line
using the capacitance sensors and recorded by the control
PC. Typically, the oil fraction within the mixture was fixed
at 25%, 50% or 75% by volume. The total flow rate through
the vessel was varied between 60 and 160 l/min. During
each experiment, the outlet conditions from the vessel were
controlled such that the oilwater interface level was kept
at a steady and fixed position at the downstream end of the
separator vessel. The purpose of this procedure was to obtain
steady-state flow conditions through the vessel.
Fig. 10 gives an example of the measurement output
as seen on the screen of the control PC. This comprises
three charts, one for each sensor array. Each chart contains
the electrode number, value of measured capacitance and

a capacitance plot vs. electrode number. The vertical bar


between the electrode number and capacitance plot is multicolour. White corresponds to air (measured capacitance
about 0.3 pF), yellow corresponds to oil (measured values
between 0.4 and 0.5 pF), blue, at the bottom, corresponds
to water (capacitance values above 7 pF). The area between
is shown as shades of green and corresponds to oilwater
mixture presence. The measured capacitance values are
between those for oil and water. Fig. 10(a) corresponds
to a static test with no flow through the separator, while
Fig. 10(b) corresponds to a dynamic situation.
In Fig. 10(b) array A at the inlet to the PPI indicates
a thick layer of oilwater mixture (approximately between
electrodes 6 and 12). As the oilwater mixture flows along
the PPI, the separation process takes place. This is indicated
by a much thinner layer of mixture on array B (between
electrodes 26 and 29). Readings from array C are essentially
identical to those which would be obtained for two static
layers of oil and water. It can be concluded therefore that the
separation process is complete by the time the flow reaches
the end of the PPI.
A range of dynamic flow tests were carried out and the
results presented here must be only fragmentary ones, to
illustrate the capabilities of the measurement system. Results
such as those in Fig. 10(b) can be presented as a time series
to observe temporal changes of the flow pattern within the
separator. Fig. 11 shows the experimental results for the total
flow rate through the vessel 120 l/min and oil content of
25%. Here the parameter varied was the position of the
oilwater interface, at the downstream end of the separator,
in relation to the weir, which was set as 7.5, 15 and 22.5 cm
below the lip of the weir.
It is worth noting that when the interface level is 7.5 cm
below the weir, the water rich mixture is introduced into
the water phase and most of the separation is probably
taking place upstream of the packed bed. When the interface
level is 15 cm below the weir the concentration distribution
resembles the one in Fig. 10(b), with the steady separation
progress along the PPI. In the last case when the position
of the interface is 22.5 cm below the weir the water-rich
mixture is introduced into the oil phase. Here arrays B and C
are practically free of the mixture layer. Surprisingly, array B

120

A.J. Jaworski, T. Dyakowski / Flow Measurement and Instrumentation 16 (2005) 113127

Fig. 11. Illustration of the separation patterns for varying oilwater interface levels. Total flow rate 120 l/min; oil content 25%.

indicates that there are some regions of dispersed water close


to electrode 26, whereas array C indicates a similar region on
a slightly lower level of electrode 48. This could perhaps be
explained by an odd flow pattern where the dispersed water
is raised within the bed to some height above the continuous
phase.
4.2. Design and industrial testing of UMIST segmented
capacitance probe for detection of oilwater interface in
oilwater separators
The ultimate objective of the development of capacitance
measurement techniques in the Pilot Plant separator,
described in Section 4.1, was the transfer of the technology
to more realistic oilfield conditions. Of course retro-fitting
capacitance sensors into a PPI device would be practically
impossible for live separators, however the technology
developed could easily be adapted for a probe configuration,
similar to the one described in [1,2125]. According to the
requirements of the industrial sponsors, the probe should fit
small vessels (1.02.0 m in diameter), working at moderate
pressures and temperatures (up to 20 bars and 80 C). The
diameter of the probe was limited by the dimension of the
vessel entry nozzle, 50.8 mm (or 2 in. in the units accepted
within the industry), while the required spatial resolution
of identifying interface level positions was prescribed to be
at least 10 mm. Due to the moderate operating conditions
required, it was decided to build the probe using dielectrics
as structural materials. This has its rationale in the preferred
distribution of the sensing electrical field, as described

below, although it should be stated that for high pressure


and high temperature applications one should rather consider
metal as a more appropriate structural support. The design
procedures included numerical simulations of the electrical
field to maximise the sensor performance, and selection of
reliable technology and materials for their manufacture [31].
4.2.1. Numerical simulations
Fig. 12(a) shows the cross section of a probe, similar to
that presented by Yang et al. [1], which served as a starting
point for the numerical analysis. The probe consists of a
stainless steel duct with two slots machined along its length,
to allow process media to enter the sensing space. The
capacitance sensors are potted inside two opposite halves
of the probe with an epoxy resin compound. The stainless
steel duct serves as both the mechanical protection and the
electrical shield. Fig. 12(b) and (c) show the results of the
simulations performed using Ansoft EM Field Modeller.
Fig. 12(b) corresponds to the probe immersed in a medium
of low relative dielectric permittivity ( = 1). Fig. 12(c)
corresponds to the probe immersed in a medium of high
dielectric permittivity ( = 80). The graphs show the lines
of equal potential. Here the source electrode was set at
a potential of 1 V, whereas the stainless steel body and
the detector were both set at 0 V. Clearly, the change of
the medium leads to a dramatic change in the character of
electrical field between electrodes, from almost uniform for
a low permittivity medium, to highly distorted one for a
high permittivity medium. A number of simulations were
performed to identify the causes of this behaviour and to
develop a better probe design.

A.J. Jaworski, T. Dyakowski / Flow Measurement and Instrumentation 16 (2005) 113127

121

Fig. 12. Cross-section of the reference probe [1] (a); equi-potential linesprobe in medium of = 1 (b); equi-potential linesprobe in medium of = 80 (c).

Fig. 13. Cross-section of the probe designed in the current study (a); equi-potential linesprobe in medium of = 1 (b); equi-potential linesprobe in
medium of = 80 (c).

The following conclusions were drawn. Firstly, the lines


of electrical field (that are orthogonal to the equi-potential
lines) are pulled away from the detector towards the metal
casing of the probe in the high dielectric permittivity media
(say > 15). This can clearly be seen in Fig. 12(c).
Secondly, the above effect is unavoidable, but can be reduced
by covering the shield with a layer of dielectric material,
and placing it behind the source electrodes. Thirdly, the
depth to which both the source and detection electrodes are
embedded is critical: the thinner the electrode coating the
less curvature of electric field is observed. In addition, an
increased width of the source electrode in relation to the
detector allows the shifting of the electrical field curling
away from the detectors centre.
Fig. 13(a) shows the design of the probe based on the
conclusions presented. The probes body is made of a
dielectric material. The PCB (printed circuit board) inserts
described later are glued into the prepared grooves. The
source electrode is shielded from behind by an electrode
placed on the back of the same PCB. The electrodes are
embedded 0.5 mm below the surface. Fig. 13(b) and (c)
show the results of numerical simulations for the corrected
probe design. Clearly, the change of the process media no
longer leads to distortions of the electrical field.
It is worth mentioning that because, in the current design,
there is a common source electrode and segmented detectors,
the electrical field can be considered two dimensional.
This is not the case for an arrangement with segmented
source and common detector [1], where the electrical field
is highly three dimensional. Of course, the validity of such
simulations is limited in the presence of conductive media.

4.2.2. Probe fabrication


The capacitance sensors were built using the PCB
technology. This allowed a compact design of electrical
connections within a multi-layer structure and a reliable
pressure seal. In addition, the PCBs are known to have some
outstanding temperature, pressure and media compatibility
specifications. The layout of electrodes is shown in
Fig. 14(a). The detection PCB has 24 electrodes, distributed
along the length of 406 mm and surrounded by a guard
electrode. The source PCB consists of two layers of plain
copper conductor (excitation and external shield). The
inserts were coated with a thin layer of epoxy laminate
(0.25 mm), to insulate the electrodes from the process media.
Fig. 14(b) shows the general view of the UMIST prototype
probe. Its diameter is 48 mm and the total length is 974 mm.
The PCBs are encapsulated in a dielectric material (Tufnol),
as seen in Fig. 14(a). The active length of the probe is
406 mm. A number of spacers keep the dielectric casings
at a constant distance of 10 mm. The sensitive part of the
probe screws into a stainless steel extension, which is fixed
to a flange. The signal wires which are soldered to the PCBs
on the low-pressure side run inside the extension and are
connected to the capacitance meter. The probe can be fitted
with extensions of various lengths to allow insertion into
different vessels.
4.2.3. Experimental set-up
The field trials of the probe took place at BP Exploration
Wytch Farm, in Dorset, UK. The probe was inserted into
their Bridport Separator V1104, through a 3 in. (76.2 mm)
nozzle at the bottom of the vessel, 300 mm upstream from

122

A.J. Jaworski, T. Dyakowski / Flow Measurement and Instrumentation 16 (2005) 113127

Fig. 15. Bridport separator (top); probe inserted into the separator and the
box of electronics (bottom).
Fig. 14. PCB boards used as electrode arrays in the probe (a); assembled
UMIST probe (b).

the water outlet. The typical throughput of the separator


was 700800 barrels per day of the oilwater mixture.
The temperature of the media was 42 C and the working
pressure was 5.9 bar (gauge).
The top of Fig. 15 shows a general view of the separator.
It is 1.9 m in diameter and has a length of 4.5 m. The bottom
of Fig. 15 shows the nozzle, with the probe already inserted,
and the local box of electronics. The measurements were
controlled by a remote PC located in the Equipment Room.
The communication, over a distance of 200 m, was realised
by two separate lines: one controlled the multiplexer which
switched the measurements between consecutive electrodes;
the other carried the analogue signal (measured value of
capacitance) back to the PC. The measurements from the
probe were compared with the existing float device LIC 1101
(also referred to as a displacer).
4.2.4. Results
Examples of the probe output as displayed on the control
PC are shown in Fig. 16. This is very similar to the graphical
interface shown earlier in Section 4.1. Each graph contains
a table with the electrode number and its reading in pF, a
vertical spectrum bar for easier interpretation of readings
and two profiles showing the measured capacitance versus
electrode number (lower horizontal scale 010 pF) and the
difference in readings between each pair of neighbouring
electrodes (upper horizontal scale 5 to +5 pF). In the

Fig. 16. Examples of UMIST probe readings.

experiments presented, the set of 24 electrodes was scanned


approximately every 30 s.
The interface level was obtained from the sharpest
increase in the capacitance profile. This was made with
an accuracy corresponding to half of the electrode spacing
(i.e. 8.5 mm), which was adequate for the application
discussed. The objective of the experiments, such as those
depicted in Fig. 16, was to check the probes response
while the oilwater interface was traversed along the sensing
range. Probe readings were compared with the reference
devicethe float LIC 1101. While these can be compared
in the time domain as explained in [31], here a cumulative
result of five days of testing is shown in the form of a
correlation line, where the interface level from the UMIST

A.J. Jaworski, T. Dyakowski / Flow Measurement and Instrumentation 16 (2005) 113127

123

Fig. 17. Correlation between UMIST probe readings and displacer readings.
Fig. 18. Photograph of the probe removed from the separator.

probe is plotted against that from the float device (Fig. 17).
Most of the experimental data lie close to the regression line
marked in the figure, whose equation was determined as:
HUMIST = 0.981UFLOAT + 115.3

(11)

where HUMIST is the interface position according to


the UMIST probe, and HFLOAT is the interface position
according to the float (both expressed in millimetres above
the bottom of the separator). The number 115.3 represents
the offset between the two (also expressed in millimetres).
There are also two sets of points in Fig. 17: one along
a vertical line (LIC1101 = 372 mm) and another along
a horizontal line (UMIST Probe = 735 mm); these
correspond to the oilwater interface position beyond the
ends of the float range. The data below the regression
line was obtained for a high rate of water draining. The
UMIST probe was only 300 mm from the water outlet,
and it is likely that a strong vortex close to the probe was
responsible for a local interface displacement. Of course,
such a displacement would not affect the displacer located
externally to the vessel. The 115 mm offset between the
probe and the float simply reflects the fact that the float was
not calibrated prior to the tests. The measurements from the
UMIST probe rely on the physical contact of the media with
the individual electrodes and therefore are more reliable in
terms of interface measurements.
At the end of the five day period the probes readings
became suspicious, and eventually were not able to indicate
the position of the oilwater interfaces. It is thought that
the rapid draining of the vessel and the subsequent filling
it up with the media led to the situation where the sludge,
normally present at the vessels bottom, was agitated and
clung to the probes control surfaces. This is a common
pitfall of most of the electrical based instrumentation, as the

sludge forms a conductive film disabling the measurements.


However it was most likely accelerated by the frequent and
rapid interface movements. The cause of the probes failure
was confirmed when it was removed from the separator;
Fig. 18 shows the photograph of the probe surface covered
in the deposit.
In general terms, the tests showed that the principle of
interface monitoring using segmented detection electrodes
is correct. The results obtained from the UMIST probe and
the displacer agreed very well, apart from an offset due to
inaccurate calibration of the float. However, the tests also
revealed that significant design effort is needed in order to
devise a method of keeping the surface of the probe clean
over prolonged periods of time to avoid electrode fouling.
4.3. OTL probe for vessel profilingtests at NEL
The experience gained during the construction of the
sensors embedded in the PPI of the Pilot Plant separator and
development and testing of the UMIST interface level probe
has been used in the construction of the OTL probe (model
CIP2003). However a number of modifications have been
made to reflect the needs of the equipment for oilfield use.
These are listed below:
The probe was manufactured with a stainless-steel body
for future use at elevated temperatures and pressures.
The printed circuit boards, which are conceptually similar
to those in UMIST probe, had to be modified to meet
the criteria of electrical safety certification, the main
difference being 2 mm thick lamination of the electrodes
to meet the appropriate flame proof norms.

124

A.J. Jaworski, T. Dyakowski / Flow Measurement and Instrumentation 16 (2005) 113127

Fig. 20. Schematic of the NEL multiphase flow facility.

Fig. 19. Sensing head of the OTL probe, CIP2003.

The probe was equipped with a mechanism for online and in situ deposit removal, inducing a mechanical
excitation of sensor surfaces to prevent deposition on the
surface [35].
The probe was equipped with a purpose built capacitance
measurement circuit (to replace the PTL120 capacitance
meter) working on a similar principle but with enhanced
robustness and suitable for housing in a flame proof box.
Overall the probe described has been designed to meet
flame proof equipment specification as follows: EEx m II
T6 Tamb 0 C to 80 C.
4.3.1. Probe description
The photograph of the sensing part of the probe is shown
in Fig. 19. It contains 16 detector electrodes embedded
within a 4 mm thick PCB (in the left half of the probe)
and a single excitation electrode (in the right half).
The detector electrodes are distributed along the sensing
length of 400 mm, which provides the spatial resolution
for interface measurements of about 12 mm. The PCBs are
positioned within the steel body of the probe in such a
way that there is a 12 mm wide gap, which can be entered
by the fluids for the purpose of interface measurements.
A more detailed description is not provided because the
overall measurement technique is similar to that described in
Sections 4.1 and 4.2. The flange which can be seen at the top
of the sensing part of the probe mates with a similar flange
on an extension a circular pipe with appropriate flange

Fig. 21. NEL separator and insertion of the probe.

fittings that is used to secure the probe in the separator.


The semi-transparent block on the top of the sensing head is
a potting compound that secures the cabling coming out of
the sensing head.
4.3.2. Experimental set-up
The preliminary experiments to test the performance of
the OTL probe were conducted in a three-phase gravity
separator at the TUV NEL Ltd multiphase flow facility [36].
The separator is part of the multiphase flow facility to study
flows in pipelines and test the multiphase flow meters. It
is about 2.8 m in diameter and 12 m long and can contain
approximately 35 m3 of water and 25 m3 of oil. A schematic
of the NEL flow facility is shown in Fig. 20. The probe was
inserted through a 6 in. (152.4 mm) nozzle from the top of
the vessel (as shown in Fig. 21), which is about 600 mm
downstream from the end of the PPI used in this vessel.
The length of the extension attached to the probe was such
that the sensing length of the probe would provide interface
measurements in the region from 300 to 700 mm below the
lip of the weir plate, which itself is 1900 mm above the
bottom of the vessel. The electronics to perform capacitance

A.J. Jaworski, T. Dyakowski / Flow Measurement and Instrumentation 16 (2005) 113127

measurements was placed on top of the flange attached to


the probe extension and connected to the control PC located
on the side of the separator.
The oilwater interface level in the separator could be
measured independently by using a pressure transducer
placed at the bottom of the vessel. This type of measurement
is of course limited to the situation when the vessel is filled
up to the weir lip level and in addition there is a clear
interface present such as in the static situation. The fluids
used in the flow facility are as follows: the oil is a mixture of
Forties and Beryl crude oil, topped to remove light ends and
increase flashpoint to 75 C, with kerosene added to restore
original viscosity; the water is a solution of magnesium
sulphate MgSO4 7H2 O with concentration of 50 kg/m3.
The temperature of the fluids is controlled by means of heat
exchangers and can be set to between 1040 C with the
accuracy of 1 C. For safety reasons, nitrogen is used as
the gas phase. The total flow rate through the vessel can be
in excess of 100 m3 /h although most of the flow tests were
carried out for flow rates in the range of 5070 m3 /h. The
concentration of water (water cut) can be adjusted between
0 and 100% with accuracy of 1%. Typically, in the current
tests, the water cut was varied in steps of 10% between pure
water and pure oil.
4.3.3. Sample results
Fig. 22 shows an example of the capacitance measurements as displayed on the control PC for static conditions.
Here a clear interface between oil and water is located approximately between electrodes 6 and 7 (with some water
most likely partly covering electrode 6 as can be interpreted
from its increased reading). During preliminary testing, this
type of profile was obtained in order to cross-reference the
interface readings with pressure transducer measurements.
Fig. 23 shows the correlation between the probe readings
and the readings from the pressure transducer (which is conceptually somewhat similar to those presented in Fig. 17).
It is clear that both devices indicate similar interface level
positions. It should be noted that the scatter of data is most
likely due to the unknown level of fluid as it is flowing over
the weir, which clearly affects the pressure readings.
Fig. 24 shows an example of a capacitance profile
obtained for the mixture flow rate of 60 m3 /h and the water
concentration of 30%. It can be seen that the capacitance
readings change gradually between oil rich mixture at the
top and water rich mixture at the bottom of the sensing
range of the probe. In principle, it is possible to establish the
relationship between the measured capacitance and water cut
such as described with reference to Fig. 7 earlier.
Finally, some experiments were conducted to establish
whether mechanical excitation technique is able to dislodge
the deposits from the surface of the probe. The process
fluids used at NEL are relatively clean and no problems
were encountered that would indicate that sludge or scaling
affected the readings. Therefore some ex situ tests were
conducted. These included coating of the probe with a

125

Fig. 22. Sample capacitance profile from the OTL probe.

Fig. 23. Correlation between OTL probe and pressure transducer readings.

Fig. 24. Capacitance profile for water cut of 30% and the flow rate 60 m3 /h.

thick layer (around 5 mm) of bitumen and immersing


the probe in a small cylinder containing crude oil taken
out of the separator. Fig. 25 shows the close-up of the
electrode surface with bitumen and the same area after
application of mechanical excitation. As can be seen the
mechanical excitation provided within the probe proved to

126

A.J. Jaworski, T. Dyakowski / Flow Measurement and Instrumentation 16 (2005) 113127

invaluable during our stay at NEL. The sponsors of the


research activities described including EPSRC, the DIPSS
consortium, Oilfield Technology Ltd and the Small Business
Service of the DTI are gratefully acknowledged.
References

Fig. 25. Removal of bitumen from the probe surface.

be successful in removing the deposit within a few minutes


of the experiment duration.
5. Conclusion
This paper presents an overview of the recent developments in the measurement of oilwater separation dynamics and composition of the heterogeneous mixtures. Three
case studies have been presented which dealt with various
aspects of the measurement. Firstly, a viability study of using
embedded capacitance sensors distributed in the internals of
the Pilot Plant separator has been presented. This was followed by the description of industrial tests of the segmented
capacitance probe carried out in the BP oil and gas exploration facility. Pitfalls of the method have been discussed,
and particular attention has been drawn to the problem of
conductive deposit that may occur on the control surfaces
of the measurement devices. The third and final case study
concerned an improved design of the segmented capacitance
probe, where mechanical excitation is used to prevent sensor fouling. The results of this investigation are somewhat
inconclusive as in the NEL flow facility where the probe has
been tested the problem of deposition does not occur. Nevertheless the probes ability to remove the deposit was tested
by applying bitumen coating in ex situ conditions. Further
work is still required, especially in the area of developing robust sensors, capable of withstanding high temperatures and
high pressures (perhaps of the order of 150 C and 150 bar,
respectively, or more), as well as incorporating dual or multiple measurement modalities in order to obtain a more comprehensive picture of the mixed state in the multiphase flow.
Some preliminary work to address these issues has been recently reported by the authors in Ref. [37].
Acknowledgments
The authors would like to thank the personnel of NEL,
Glasgow, for their help in preparing the test and during the
period of testing. Our thanks are directed in particular to
Dr Ming Yang and Mr Richard Harvey, whose help was

[1] W.Q. Yang, M.R. Brant, M.S. Beck, A multi-interface level


measurement system using a segmented capacitance sensor for oilseparators, Meas. Sci. Technol. 5 (1994) 11771180.
[2] M.E. Rowley, G.A. Davies, The design of plate separators for the
separation of oilwater dispersions, Chem. Eng. Res. Des. 66 (1988)
313322.
[3] G.A. Davies, F.P. Nielsen, P.E. Gramme, The formation of stable
dispersions of crude oil and produced water: the influence of oil type,
wax and asphaltene content, in: Proc. SPE Conf., vol. , 69 October,
Denver, CO, USA, 1996.
[4] Y. Ming, A.C. Stewart, G.A. Davies, Interactions between chemical
additives and their effects on emulsion separation, in: Proc. SPE Conf.,
vol. , 69 October, Denver, CO, USA, 1996.
[5] G.F. Hewitt, Developments in multiphase metering, in: Conference on
Developments in Production Separation Systems, 45 March, London,
1993.
[6] R. Thorn, G.A. Johansen, E.A. Hammer, Recent developments in
three-phase flow measurement, Meas. Sci. Technol. 8 (1997) 691701.
[7] E.A. Hammer, G.A. Johansen, R. Thorn, Measurement principles in multiphase meteringtheir benefits and applications, in:
ConferenceThe Future of Multiphase Metering (Conference),
2627 March, London, 1998.
[8] L.K.H. van Beek, Dielectric behaviour of heterogeneous systems, in:
Progress in Dielectrics, vol. 7, 1967, pp. 69114.
[9] T. Hanai, Electrical properties of emulsions, in: P. Sherman (Ed.),
Emulsion Science, Academic Press, London, New York, 1968
(Chapter 5).
[10] J.C. Maxwell, A Treatise on Electricity and Magnetism, vol. I,
Clarendon Press, Oxford, 1873.
[11] J.W.S.B. Rayleigh, On the influence of obstacles in rectangular order
upon the properties of a medium, Phil. Mag. 34 (1892) 481.
[12] O. Wiener, Die theorie des mschkorpers fr dir feld der stationaren
stromung, Abh. d. Sach. Ges. d. Wiss., Math.-Phys. K1 (32) (1912)
509.
[13] C.J.F. Bttcher, Theory of Electric Polarisation, Elsevier, NY, 1952.
[14] K.W. Wagner, Arch. Elektrotech. 2 (1914) 371.
[15] N.Y. Rao, T.S. Ramu, Determination of the permittivity and loss
factor of mixtures of liquid dielectrics, IEEE Trans. Electr. Insul. E1-7
(1972) 195199.
[16] T.S. Ramu, N.Y. Rao, On the evaluation of conductivity of mixtures
of liquid dielectrics, IEEE Trans. Electr. Insul. E1-8 (1973) 5560.
[17] E.A. Hammer, J. Tollefsen, E. Cimpan, The importance in calculating
the permittivity and conductivity in two component mixtures for
image reconstruction, in: Proc. Engineering Foundation Conference:
Frontiers in Industrial Process Tomography II, 912 April, Delft
University of Technology, Netherlands, 1997, pp. 235240.
[18] VEGA Controls Limited, Capacitance level measurement, in: Seminar
series by The Institute of Measurement and Control, 23 February
1998.
[19] J. McIntyre, Sorting out liquid level sensors, I&CS (1992).
[20] Levels of Intelligence, Process Engineering, July 1990.
[21] European Patent Application No 84201889.7, Level GaugeShell
International Research, 1984.
[22] S. Hutzler et al., Measurement of foam density profiles using AC
capacitance, Europhys. Lett. 31 (1995) 497502.
[23] R.B. Schller, K. Olsen, M. Halleraker, B. Engebretsen, Monitoring
of high pressure separation of real hydrocarbon/water mixtures by
single electrode capacitance probes, in: Proc. 3rd World Congress on
Industrial Process Tomography, 25 September, Banff, Canada, 2003,
pp. 750755.

A.J. Jaworski, T. Dyakowski / Flow Measurement and Instrumentation 16 (2005) 113127


[24] R.B. Schller, T. Solbakken, B. Engebretsen, M. Halleraker, Water
concentrations and interface positions by single electrode capacitance
probes (SeCaP), in: Proc. 2nd World Congress on Industrial
Process Tomography, 2931 August, Hannover, Germany, 2001,
pp. 644651.
[25] R.B. Schller, M. Halleraker, B. Engebretsen, Advanced profile gauge
for multiphase systems, in: Proc. 1st World Congress on Industrial
Process Tomography, 1417 April, Buxton, Greater Manchester, 1999,
pp. 126132.
[26] F. Garcia-Golding, M. Giallorenzo, N. Moreno, V. Chang, Sensor
for determining the water content of oil-in-water emulsion by
specific admittance measurement, Sensors Actuators 4647 (1995)
337341.
[27] M. Darwood, K. James, P. Jackson, P. Hewitt, Density profiling in
multiphase systems using gamma ray absorption, in: Proc. 3rd World
Congress on Industrial Process Tomography, 25 September, Banff,
Canada, 2003.
[28] K. Asskildt, P. Hansson, New measuring sensor for level detection in
subsea separators, ABB Rev. (Issue Number 1999/04) (1999) 1117.
[29] J.G. Miranda, Designing parallel-plates separators, Chem. Eng. 84 (3)
(1977) 105107.

127

[30] Q. Zhang, G.A. Davies, Simulation of two-phase flow in cross flow


separator, PHOENICS J. CFD Appl. 9 (1) (1996) 139156.
[31] A.J. Jaworski, T. Dyakowski, G.A. Davies, A capacitance probe for
interface detection in oil and gas extraction plant, Meas. Sci. Technol.
10 (1999) L15L20.
[32] A.J. Jaworski, T. Dyakowski, G.A. Davies, Application of capacitance
sensors for monitoring oilwater separation processes, in: Proceedings
of ASME, HTD, vol. 3615, 1998, pp. 429437.
[33] W.Q. Yang, Hardware design of electrical capacitance tomography
systems, Meas. Sci. Technol. 7 (1996) 225232.
[34] A.J. Jaworski, T. Dyakowski, Application of distributed sensors for
monitoring multiphase processes, Trans. ASME, J. Energy Resour.
Technol. 125 (4) (2003) 253328.
[35] Oilfield Technology Ltd, 3 Peel Moat Road, Heaton Moor, Stockport
SK4 4PL, Tel.: 0161 374 9867 (private communication).
[36] National Engineering Laboratory, A guide to the NEL multiphase flow
facility, TUV NEL Ltd, 2004.
[37] G. Meng, A.J. Jaworski, T. Dyakowski, J.M. Hale, N.M. White,
Design and testing of a thick-film dual-modality sensor for
composition measurements in heterogeneous mixutres, Meas. Sci.
Technol. 16 (2005) 942954.

S-ar putea să vă placă și