sensors

Communication
Towards Real-Time Analysis of Gas-Liquid Pipe Flow: A
Wire-Mesh Sensor for Industrial Applications
Philipp Wiedemann 1, * , Felipe de Assis Dias 1 , Manuel Trepte 2 , Eckhard Schleicher 1, * and Uwe Hampel 1,3

1 Institute of Fluid Dynamics, Helmholtz-Zentrum Dresden-Rossendorf, Bautzner Landstraße 400,

01328 Dresden, Germany
2 Teletronic Rossendorf GmbH, Bautzener Landstraße 45, 01454 Radeberg, Germany
3 Chair of Imaging Techniques in Energy and Process Engineering, Technische Universität Dresden,
01062 Dresden, Germany
* Correspondence: [email protected] (P.W.); [email protected] (E.S.)

Abstract: Real-time monitoring of gas-liquid pipe flow is highly demanded in industrial processes
in the chemical and power engineering sectors. Therefore, the present contribution describes the
novel design of a robust wire-mesh sensor with an integrated data processing unit. The developed
device features a sensor body for industrial conditions of up to 400 ◦ C and 135 bar as well as real-time
processing of measured data, including phase fraction calculation, temperature compensation and
flow pattern identification. Furthermore, user interfaces are included via a display and 4 . . . 20 mA
connectivity for the integration into industrial process control systems. In the second part of the
contribution, we describe the experimental verification of the main functionalities of the developed
system. Firstly, the calculation of cross-sectionally averaged phase fractions along with temperature
compensation was tested. Considering temperature drifts of up to 55 K, an average deviation of 3.9%
across the full range of the phase fraction was found by comparison against image references from
camera recordings. Secondly, the automatic flow pattern identification was tested in an air–water two-
phase flow loop. The results reveal reasonable agreement with well-established flow pattern maps
for both horizontal and vertical pipe orientations. The present results indicate that all prerequisites
for an application in industrial environments in the near future are fulfilled.

Citation: Wiedemann, P.; de Assis

Keywords: wire-mesh sensor; two-phase flow; flow pattern identification; online data processing
Dias, F.; Trepte, M.; Schleicher, E.;
Hampel, U. Towards Real-Time
Analysis of Gas-Liquid Pipe Flow: A
Wire-Mesh Sensor for Industrial 1. Introduction
Applications. Sensors 2023, 23, 4067. In many industrial processes, e.g., oil and gas production, the chemical and phar-
https://doi.org/10.3390/s23084067 maceutical industries, food processing and power generation, multiphase flows of gas
Academic Editor: Heinrich Wörtche and liquids occur in manifold different scenarios and conditions. The knowledge and
understanding of liquid and gas holdup, phase distributions, fluid velocities and flow
Received: 13 March 2023 morphologies are thus often key parameters for precise metering and intelligent process
Revised: 7 April 2023
control. Moreover, they are mandatory for process intensification and sometimes important
Accepted: 11 April 2023
for safe plant operation, e.g., in the case of slug development in pipelines.
Published: 18 April 2023
Nevertheless, there are only a few multiphase flow sensor systems available on the
market for industrial process control. Mainly, such systems have been developed for
monitoring and metering of oil–gas–water mixtures for the oil and gas industry. A recent
Copyright: © 2023 by the authors.
overview can be found in [1]. In the past, the majority of such systems have utilized
Licensee MDPI, Basel, Switzerland. different mixing procedures to homogenize the fluid mixture in order to gather spatially
This article is an open access article and temporally averaged data of the flow composition as well as the mixture flow velocity.
distributed under the terms and More recent developments tend towards fast tomographic sensors for industrial process
conditions of the Creative Commons control, cf. [2]. While fast tomographic X-ray scanners are still very expensive and suffer
Attribution (CC BY) license (https:// from radiation safety issues, electrical-impedance-based systems such as EIT and ECT have
creativecommons.org/licenses/by/ insufficient spatial resolutions. Among different tomographic techniques for multiphase
4.0/). flow measurement, wire-mesh sensors (WMSs), first introduced by Prasser et al. [3], have

Sensors 2023, 23, 4067. https://doi.org/10.3390/s23084067 https://www.mdpi.com/journal/sensors

Sensors 2023, 23, 4067 2 of 13

a high spatial resolution of typically 2 . . . 4 mm and a very high frame rate of up to

10,000 fps. The biggest drawback of WMSs is their intrusive character [4–7], which limits
their application to processes and flow scenarios with no large particles streaming with the
fluid mixture in order to prevent electrode destruction or sensor clogging. Additionally,
WMSs so far have always used high-speed data recording with offline analysis afterwards,
being unsuitable for online process control. Thus, during the last two decades WMSs have
mainly been utilized in experimental test facilities emulating different kinds of industrial
applications, such as steam generation in thermal power plants [8], pipe flows [9,10],
chemical reactors [11], thermal separation units [12,13], etc. Even if the sensor is of intrusive
nature, there is still a large field of real industrial processes being suitable for the application
of WMS technology—in case it can become a robust 24/7 sensor system and it overcomes
offline data processing. As one example, in direct steam generating solar power plants,
there is a need for metering the amount of generated steam along the absorber tube system
as well as for flow regime identification to control the system efficiently, cf. [14].
A first concept of a WMS targeting industrial applications was proposed in [15]. This
sensor was equipped with an additional electronics unit, which controls the data acquisition
and performs internal data reduction. More precisely, the calculation of cross-sectionally
averaged phase fraction and the identification of flow patterns in the vertical upward
gas–liquid flow were implemented. The achievements in [15] act as basis for our present
work, which aims at providing an industrial type of wire-mesh sensor (indWMS) that is
applicable to a much wider range of processes as well as to variable process conditions.
Therefore, the newly developed indWMS provides a robust sensor body and electronics
housing along with a redesigned electronics unit that allows for almost real-time data
processing, including internal temperature compensation and data reduction by calculation
of cross-sectionally averaged phase fractions and automatic flow pattern identification
for vertical and horizontal pipe orientation. Moreover, a contemporary user interface
and an automatic calibration routine were implemented for maximum user comfort. The
developed device is described in detail in Section 2. Verification experiments and their
results are presented in Section 3.

2. Design of an Industrial Wire-Mesh Sensor

2.1. Design of a Robust Sensor Body
While many experimental test facilities for multiphase flows are operated with air
and water under ambient conditions, the industrial type of wire-mesh sensors has been
designed to withstand pressures of up to 135 bar and temperatures of up to 400 ◦ C for
maximum applicability in practical processes, e.g., in solar thermal power plants, feed
pipes of distillation columns, pipe lines, etc. The sensor system consists of a stainless-steel
body flanged into the pipe at the position of interest and a sensor slot-in unit, as proposed
in [16]. This design allows an easy exchange or maintenance of the sensor unit without
dismantling the facility and decouples the purpose of electrode insulation and pressure
barrier from each other. The slot-in sensor consists of a stainless-steel frame, which seals
against the sensor body via a graphite gasket and a ceramic inlay accommodating the
individually spring-mounted wire electrodes, cf. [17,18]. Upstream and downstream the
slot-in unit a pressure and a temperature sensor are integrated into the system, respectively.
The single stainless-steel electrodes run out of the sensor body via special eight-channel
glands for pressure sealing and into a newly designed IP56-housing accommodating the
electronics with its user interface via a 3.6-inch display with 4 . . . 20 mA connectivity for
data transfer to any DAQ or real-time controller and an additional Ethernet interface for
debugging and firmware updates. In order to achieve water tightness (IP56), the system is
passively cooled by means of large aluminum heat spreaders. Visualizations of the sensor
system are depicted in Figure 1.
Sensors 2023, 23, 4067 3 of 13

(a) (b) (c)

Figure 1. Industrial type of wire-mesh sensor: (a) 3D-CAD visualization of the sensor design;
(b) photograph of the prototype; 1(c) sectional view.

2.2. Electronics Design

The electronic unit of the indWMS system consists of a FPGA-based controller unit,
the transmitter and receiver blocks, an output driver section and the user interface. A block
diagram of the new electronic system is shown in Figure 2.

480×480 pixel
color display
up to 64 transmitter
electrode drivers 4 context sensitive keys

driver

driver
...

void fraction
Logic
driver

to process control system

pressure

temperature
fluid pressure
Memory
fluid temperature
flow pattern v

amp. ADC flow pattern h1

Microprocessor
flow pattern h2
amp. ADC

FPGA based controller 6× 4...20 mA interface

...

24 V supply
amp. ADC Power
up to 64 receiver
electrode amplifiers
and ADC’s

Figure 2. Block diagram of the electronic unit.

The sensor grid itself comprises two planes of up to 64 parallel wire electrodes spanned
with an equal distance over the cross-section of the pipe. The two electrode planes, trans-
mitter and receiver plane, respectively, are oriented 90 to each other and have a small axial
gap. The transmitter electrode drivers send a single bipolar voltage pulse of 6 µs length to
each of the transmitter electrodes subsequently while the inactive transmitter electrodes are
kept on ground potential. The receiver electrode amplifiers transform the resulting currents
through the fluid in each of the receiver electrodes into a fully parallel voltage and convert
it into a 16-bit digital value. After the last transmitter electrode is activated, the next frame
Sensors 2023, 23, 4067 4 of 13

starts. The frame rate of the measurement is currently limited to 5 kHz (for 16 transmitters).
The controller unit is realized in an FPGA, which contains the programmable logic for
parallel data processing in a pipeline architecture, the memory and a microprocessor for
higher-level algorithms and communication protocols. This central controller unit realizes
the synchronized timing of the transmitter and receiver unit. It also controls the user interac-
tion via a 3.6-inch color display and a four-button context-sensitive keyboard. The display
unit presents the measured and calculated parameters, instantaneous cross-sectional void
fraction, pressure and temperature, and gives a graphical visualization of the identified
flow pattern. It also shows the current menu functions for user access to set up the system.
The local instantaneous void fractions are calculated as averaged values over a period
of one second. A ring buffer with a length of 64 s is utilized to derive the flow patterns
instantaneously by using a fuzzy identification algorithm, cf. Section 2.3.2. The 4 . . . 20 mA
interface can transfer all measured and calculated parameters to an external process control
system. The identified flow regimes in vertical orientation are coded by means of a single
value in a linear 4 . . . 20 mA scale, while in horizontal scenarios two 4 . . . 20 mA signals
are necessary to transfer the polar coordinates representing the regime.
The system also includes a type K thermocouple as the temperature sensor, which is
primarily utilized in the context of temperature compensation, cf. Section 2.3.1. Along with
an additional pressure transducer, the local state variables at the position of the indWMS are
provided and allow for external calculation of the corresponding thermodynamic properties
of the fluids. Furthermore, alerts are generated on the basis of these measurements as soon
as a predefined temperature or pressure is reached, cf. Section 2.4.
The power unit is based on a wide range 10 . . . 35 VDC DC-DC converter supplying
all the required voltage levels of the system components.

2.3. Data Processing

2.3.1. Temperature Compensation and Calculation of Cross-Sectionally Averaged
Phase Fraction
The measurement principles of wire-mesh sensors rely on discriminating phases by
differences in their electrical properties, cf. [3,19–21]. In the current version of the ind-
WMS, the conductivity-based measurement principle according to [3] was implemented
so far, focusing initially on gas–liquid flows with at least one conductive phase (water),
which are encountered in many industrial applications, e.g., generation of process steam or
water–steam cycles in power plants. In the conductivity-based approach, local (i, j) instanta-
neous (k) phase fractions αi,j,k are usually quantified by normalizing the measured voltage
meas , which represent the conductivity distribution in the pipe’s cross-section
signals Ui,j,k
liquid
with respect to a single-phase reference measurement of liquid Ui,j . However, since
the electrical conductivity of water is known to exhibit a strong temperature dependence,
the temperatures of reference and two-phase measurement need to coincide. This may
be achieved straightforwardly in most lab applications by recording several reference
measurements if temperature drifts occur. In contrast, this is seldom possible in most
industrial processes, since flexible operations with continuous transitions between partial
load and overload scenarios would lead to a tremendous effort for reference measurements
only. Furthermore—and much more seriously—pure single-phase liquid flow is hardly
obtained at saturation temperature, which is usually the region of interest for two-phase
flow measurements. Therefore, the temperature compensation method proposed by [22]
was implemented in the real-time data processing unit of the indWMS system. The adapted
formulation for the quantification of phase fractions reads
meas ( T )
Ui,j,k
gas
αi,j,k ( T ) = 1 −  
liquid
 . (1)
F T, Tre f ·Ui,j Tre f

Here, F denotes the temperature compensation factor that is calculated on the ba-
sis of the ISO 7888 model [23] using the temperatures of the reference and two-phase
Sensors 2023, 23, 4067 5 of 13

measurement, Tre f and T, respectively, cf. [22]. Hence, the single-phase reference matrix
liquid
Ui,j recorded at an arbitrary reference temperature Tre f can be converted to a reference
matrix at temperature T corresponding to the present conditions of the two-phase mea-
meas . A prerequisite for applying Equation (1) is that the gain settings of the
surement Ui,j,k
transimpedance amplifiers are identical for the reference and the two-phase measurement.
Details on adjusting the amplifiers are given in Section 2.4.
gas
All values of αi,j,k are subsequently limited to the interval [0 . . . 1]. Eventually, the
cross-sectional average is calculated as

hαk i = ∑ ∑ ai,j αi,j,k ,

gas gas
(2)
i j

with ai,j denoting the share of a pixel i, j with the pipe’s cross-section, cf. [15]. For the
output via display and 4 . . . 20 mA signal, a temporal average hα gas i is calculated from the
last k = 1 . . . 5000 frames and updated every second.

2.3.2. Flow Pattern Identification

Algorithms for flow pattern identification were implemented for vertical and horizon-
tal pipe orientations according to [15] and [24], respectively. Both algorithms are based
on the analysis of statistical features of the measured three-dimensional phase fraction
gas
distributions αi,j,k and provide the final identification result in a fuzzy manner. Based on a
so-called degree of membership, the output describes the likelihood of a certain measure-
ment to belong to one or more predefined flow patterns. According to [25], four major flow
patterns were selected for each pipe orientation, namely bubbly, slug, churn and annular
flow for vertical and bubbly, intermittent, stratified and annular flow for horizontal pipe
orientation. The identification algorithms use a time span of 64 s as input from the ring
buffer and the result is updated every second. In this way, transitions of flow patterns
can also be captured almost in real time. The results of the flow pattern identification are
represented graphically on the display (see Section 2.4) and are coded as 4 . . . 20 mA output
signals as well (cf. Section 2.2).

2.4. User Interface

The user interface with display and context-sensitive keys was integrated to provide
instant information about the present flow conditions and the sensor status as well as to
configure the sensor during the commissioning process. After switching on the power
supply, a boot screen is shown temporarily. Subsequently, the sensor system directly
activates the measurement protocol and switches to the main screen, in which current data
of temperature, pressure and cross-sectional gas phase fraction are displayed as numerical
values, cf. Figure 3. The identified flow pattern is visualized in a fuzzy manner, as proposed
by [15] and [24] for vertical and horizontal sensor positions, respectively. Additionally,
alerts are displayed automatically, if certain events occur, cf. Section 2.2 and Figure 3. Using
the keys next to the display, the user can switch to and navigate through the configuration
menu. Here, three configurations are available and need to be set when putting the sensor
into operation:
1. Switching between horizontal and vertical orientations leads to activation of the
respective flow pattern identification algorithm and its visualization scheme on the
main screen as well as to a change in the display orientation.
2. Adjusting the expected maximum temperature of the two-phase flow is required for
the calibration routine, which is explained below.
3. Running the calibration routine.
Sensors 2023, 23, 4067 6 of 13

Figure 3. Display screens and menu structure of the indWMS.

Running the calibration routine basically means that the current flow conditions
 are
liquid
treated as the reference state ( Tre f ← T ) and that the reference matrix Ui,j Tre f of
the completely liquid-filled pipe is recorded. However, since the indWMS accounts for
temperature compensation in order to provide reliable measurements over a wide range
of operating conditions, the amplifier gains in the receiver circuit need to be adjusted
in an anticipatory manner at the calibration stage already. More precisely, the amplifier
gains must be configured to meet an optimal range of signal response, e.g., to avoid
overdrive, at the targeted operating temperature, or rather the corresponding electrical
conductivity of the liquid. As the absolute electrical conductivity and its temperature-
induced change is assumed to be unknown in most applications, an iterative procedure
was implemented in the indWMS system to allow for an autonomous adjustment of
amplifier gains. Here, firstly the relative change
 in the
 electrical conductivity is estimated
by a maximum compensation factor Fmax Tmax , Tre f , which is based on the user-defined
maximum temperature Tmax of the two-phase process and the one of the present reference
state Tre f . As Fmax applies to both conductivity and the measured voltage signal in the
receivers, cf. [22], a maximum permissible voltage Umax can be estimated from the total
ADC range. Subsequently, reference data are acquired and compared against Umax . Based
on the result, the amplifier settings may become modified until agreement is obtained with
a defined tolerance range. The flow chart of the calibration routine is depicted in Figure 4.
Sensors 2023, 23, 4067 7 of 13

start

load 𝑇𝑚𝑎𝑥 and

default amplifier settings

measure 𝑇 and store as 𝑇𝑟𝑒𝑓

calculate 𝐹𝑚𝑎𝑥
from 𝑇𝑚𝑎𝑥 and 𝑇𝑟𝑒𝑓

calculate 𝑈𝑚𝑎𝑥 from 𝐹𝑚𝑎𝑥

measure 𝑈𝑖,𝑗

no increase
0.3 𝑈𝑚𝑎𝑥 < max(𝑈𝑖,𝑗 ) < 0.9 𝑈𝑚𝑎𝑥
amplifier gain

yes

𝑙𝑖𝑞𝑢𝑖𝑑
store 𝑈𝑖,𝑗 as 𝑈𝑖,𝑗

store amplifier settings

stop

Figure 4. Flow chart for autonomous adjustment of amplifier gains and acquisition of the reference
data set.

3. Experimental Verification
3.1. Temperature Compensation
3.1.1. Experimental Procedure
To verify the implemented temperature compensation algorithm, the cross-sectionally
averaged phase fractions, which are measured by the indWMS at different temperatures,
need to be compared against reference data from a temperature-independent measurement
technique. For that purpose, the experimental setup shown in Figure 5a was utilized.
Here, the indWMS was mounted horizontally between flanges with transparent inspec-
tion glasses, allowing for visual observation of the measurement plane. Tap water with a
conductivity of κ25◦ C = 395 µS/cm was circulated through the system at defined tempera-
tures by means of a Lauda ProLine RP870 thermostat. In a first step, the pipe was filled
with liquid of 24 ◦ C completely and the start-up routine of the indWMS, i.e., autonomous
adjustment of amplifier settings and reference measurement (cf. Sections 2.3.1 and 2.4),
was performed. Then, the liquid temperature was increased step-wise with the thermostat.
After a stable temperature was obtained in the sensor at each temperature level, different
filling heights of liquid were adjusted by injecting a gas layer of pressurized air. This proce-
dure artificially emulates a two-phase condition below the actual saturation temperature of
the liquid. At the same time, the liquid flow was paused for ensuring a smooth gas–liquid
interface. For each set of temperature level and filling height, the 4 . . . 20 mA signals of the
Sensors 2023, 23, 4067 8 of 13

indWMS were recorded for at least one minute with a sampling frequency of 2 Hz using a
National Instruments CompactDAQ Chassis and a PC. In parallel, a photo of the sensor’s
measurement plane was recorded against backlight using a 48 mega pixel camera. The
photos had a resolution of approximately 25 px/mm in the area of interest and served as a
basis for evaluating the indWMS data.

indWMS

PC
air

DAQ

light
source camera

water
thermostat

(a) (b)

Figure 5. (a) Experimental setup for verifying phase fraction measurement combined with tempera-
ture compensation; (b) images of 1pipe cross-section recorded by high-resolution camera: (top) raw
image, (bottom) cropped and scaled image with calculation of Aliquid .

In order to provide numerical values from the image references, further processing
was accomplished using the graphics software Inkscape v0.92. The following steps were
conducted manually for each individual measurement point:
1. Image cropping;
2. Image scaling according to the pipe’s inner diameter (green line in Figure 5b);
3. Approximating the gas–liquid interface in the stratified state within a concentric circle
of the pipe’s inner diameter (taking menisci at the pipe wall into consideration);
4. Calculating the area A gas or Aliquid of the remaining geometry (blue area in Figure 5b).
Finally, the cross-sectional phase fraction was calculated as follows:

A gas Aliquid
hα gas i = = 1− . (3)
A pipe A pipe

The absolute uncertainty of hα gas i from image processing is estimated to be lower

than 0.01.
For the cross-sectional phase fractions measured by the indWMS, temporal averages
of the time series are reported in the following section. The averaging periods were chosen
in a way to exclude noticeable temperature drops that particularly occurred at higher
temperature levels. In this way, a maximum uncertainty of 1.8 K was achieved for the
temperature measurements. The relative uncertainty of phase fraction measurements
with wire-mesh sensors was assessed with 10.5% by Tompkins et al. [6] by comparison
with other measurement techniques and accounting for a large variety of complex flow
patterns. However, much lower uncertainty can be expected here, since all measurements
are conducted under an ideal stratified condition and steady state.
Sensors 2023, 23, 4067 9 of 13

3.1.2. Results and Discussion

The cross-sectional gas phase fractions that were obtained at various temperature
levels from the indWMS and the temperature-independent image reference are compared in
Figure 6. Since the majority of the measurements exhibit an absolute deviation of less than
±5% while the electrical conductivity of water and thus the measured voltage signals are
expected to increase by a factor of up to 2.5 in the investigated temperature range, cf. [22],
it can be concluded that the implemented temperature compensation works excellently.

1.0 80

0.8
65
⟨αgas ⟩ indWMS

0.6

T [°C]
5%

50
+

5%

0.4
−

35
0.2

0.0 20
0.0 0.2 0.4 0.6 0.8 1.0
⟨αgas ⟩ image reference

Figure 6. Comparison of cross-sectional gas phase fractions hα gas i obtained by the industrial wire-
mesh sensor and the camera image.
1
All minor discrepancies that are observed for measurements of up to 65 ◦ C, i.e., a tem-
perature drift of 41 K, are assumed independent of the temperature compensation method,
since they show the identical trend when compared to the results of the measurements
at the reference temperature of 24 ◦ C. For high liquid levels, i.e., low hα gas i, some pixels
on the top of the pipe were observed to exhibit a signal indicating liquid, despite being
located in the stratified gas zone. Consequently, slightly lower hα gas i are obtained from the
indWMS here. As no water was observed visually in that region, we suppose an influence
of the wire fixations in the sensor slot-in unit. On the other side, positive deviations of the
indWMS at very low liquid levels, i.e., high hα gas i, might be related to the intrusive nature
of the measurement technique. Here, liquid lamellas were observed visually and in the
images between the pipe wall and the outer wires above the horizontal gas–liquid interface.
Since they are considered analogously to menisci in the image reference (cf. Figure 5b),
along with the uncertainty of the 2D projection regarding whether they actually stick to
both wire planes and are consequently measured as liquid, overestimation of hα gas i can
be explained.
In contrast to the above descriptions, obviously larger and exclusively positive devi-
ations of up to +10% are observed at a temperature level of 80 ◦ C. As the implemented
temperature compensation algorithm has proven to work satisfactorily for temperature
drifts of up to 60 K in single-phase flow (according to [22]), the increased deviations in
Figure 6 must be explained by the gradual onset of degassing or boiling. Due to locally
higher temperatures in the heating section of the thermostat, micro-bubbles of gas or
vapor are formed and transported to the test section inside the continuous liquid phase.
This phenomenon was also observed visually, but is not captured in the photos nor the
image-processing procedure. However, the indWMS data reflect the visual observation by
higher gas fractions. The effect is more dominant at higher liquid levels, i.e., low hα gas i,
due to a higher share of the affected cross-sectional area. It can be stated that the increased
deviations at the temperature level of 80 ◦ C originate from the experimental setup and
procedure, while the temperature compensation is assumed to work properly. The above
finding also shows that a sufficient distance needs to be kept between Tre f and T = Tsat
when running the calibration routine in practical applications.
Sensors 2023, 23, 4067 10 of 13

In summary, it can be concluded that the implemented algorithms for temperature

compensation and phase fraction calculation work satisfactorily. Temperature drifts of up
to 55 K can be compensated with an average deviation of only 3.9% across the full range of
phase fractions. However, improvement of the wire fixations is necessary to recover the
desired accuracy. Moreover, transient analysis is needed in future studies to determine
the signal response characteristics related to thermal inertia of the system, particularly the
thermocouple used for the temperature measurement.

3.2. Real-Time Flow Pattern Identification

3.2.1. Experimental Procedure
Investigations for the verification of the flow pattern identification algorithms were
performed using the experimental facility described in [24]. While the indWMS was
installed at 75 L/D downstream the gas injection for horizontal flows, only 36 L/D were
feasible for vertical pipe orientation due to height limitations. The experiments and data
acquisition were conducted according to [24]. Several measurements were performed at
different combinations of gas and liquid flow rates to cover all possible flow patterns and
phase fractions. Measurement uncertainty with respect to the resulting superficial velocities
is presented in [24]. At each measurement point data were sampled over 180 s with the
indWMS to allow for three cycles of data buffering. Since the flow pattern remained
constant at each measurement point and the signals of the indWMS showed negligible
fluctuations, temporal averages are reported in the following evaluation.

3.2.2. Results of Cross-Sectional Phase Fraction and Flow Pattern Identification in

Vertical Pipes
The measurement results of the indWMS are plotted against the flow pattern map
of Barnea [26] in Figure 7. Note that the bracketed point in the annular regime is not
related to the readable superficial velocities, but was produced artificially by stagnant
gas and a falling liquid film. Figure 7a shows that the cross-sectional gas fractions have
both plausible values and comprehensible trends when compared to the orientation of
the axes and the expectations derived from the flow pattern map. Moreover, reasonable
quantitative agreement can be attested for the comparison of the measured gas fractions
with experimental data from the literature (refer to Jones and Zuber [27] and Rosa et al. [28]).

10 1.00 10 annular
dispersed dispersed
bubbly bubbly

0.75
1 1 churn
vl,s [m/s]

vl,s [m/s]
⟨αgas ⟩

0.50
annular

annular
bubbly

bubbly
churn

churn
slug

slug

0.1 0.1 slug

0.25

( ) ( )
0.01 0.00 0.01 bubbly
0.01 0.1 1 10 100 0.01 0.1 1 10 100
vg,s [m/s] vg,s [m/s]

(a) (b)

Figure 7. Comparison of measurement results from indWMS against Barnea’s [26] flow pattern map
for vertical pipe orientation: (a)1cross-sectionally averaged gas fraction; (b) flow pattern.

With regard to the flow patterns, Figure 7b shows that the identification results from
the indWMS agree fairly well with the predictions of Barnea’s map [26]. However, prema-
ture transitions of the measured flow patterns are observed when increasing the superficial
gas velocity, particularly at low liquid flow rates. This effect might be related to the
Sensors 2023, 23, 4067 11 of 13

relatively short distance between gas injection and measurement position impeding full
development of the flow.

3.2.3. Results of Cross-Sectional Phase Fraction and Flow Pattern Identification in

Horizontal Pipes
The results of the indWMS measurements in the horizontal pipe are depicted in
Figure 8 along with the flow pattern map of Mandhane [29]. In analogy to the vertical
pipe orientation, the cross-sectional gas fractions in Figure 8a show plausible values and
comprehensible trends when compared to the orientation of the axes and the expectations
derived from the flow pattern map again. A quantitative comparison against the experi-
mental data in [24] reveals excellent agreement of the phase fraction distribution among
the map, which is, however, expected due to the high reproducibility of the test conditions
with the utilized experimental setup.

annular
100 1.00 23 24
annular
21 22 23 24 21 22
10 wavy 0.75
16 17 18 slug
19 20
stratified
vg,s [m/s]

intermittent

intermittent
⟨αgas ⟩

1 11 12 13 14 15 0.50 16
18
bubbly

smooth 20
17
6 7 8 9 10 14
13 9
19
0.1 plug
0.25 6 15
8
1 2 3 4 5 1 11 7 12 5
str

ly
10
ati

bb
0.01 0.00 2
fied

bu
4
0.01 0.1 1 10
vl,s [m/s]

(a) (b)

Figure 8. Comparison of measurement results from indWMS against Mandhane’s [29] flow pattern
map for horizontal pipe orientation: 1(a) cross-sectionally averaged gas fraction; (b) flow pattern in
terms of polar coordinates according to [24] (color indicates most dominant membership).

The flow patterns identified by the indWMS are depicted in Figure 8b using the polar
representation proposed by [24]. The numbers refer to the measurement points in Figure 8a
to allow for comparison against the flow pattern map of Mandhane [29]. It can be seen that
excellent agreement is obtained for the present measurements. In particular, transitional
regimes are captured well by the fuzzy methodology, e.g., point 16 (stratified/annular),
point 17 (stratified/intermittent/annular) and point 20 (intermittent/bubbly). With regard
to points 3 and 4, which appear to be identified incorrectly by the indWMS, we need to
point out that plug flow at low gas velocities features very small gas pockets only (see
hα gas i in Figure 8a also) and is thus sometimes called elongated bubble flow [30] or bubbly-
transitional flow [31] in the literature. In contrast to this subjective interpretation (see
also discussions in [24,32]), the objective differentiation by structural analysis of the phase
distribution of the indWMS is assumed to be more reliable here.
Finally, it can be concluded that the new indWMS system is able to identify flow
patterns in vertical as well as horizontal pipes satisfactorily. However, further quantitative
validation against references from other measurement techniques is necessary.

4. Conclusions
An industrial type of wire-mesh sensor with novel real-time data processing and a
user interface was developed and tested successfully. It was designed for high-temperature
and high-pressure applications and has three main functionalities: (1) temperature com-
pensation, (2) calculation of cross-sectional phase fraction and (3) real-time flow pattern
identification. All these functionalities were demonstrated experimentally for air–water
two-phase flows. In the first demonstration, cross-sectionally averaged phase fractions
Sensors 2023, 23, 4067 12 of 13

along with temperature compensation were evaluated. The experimental results show
that temperature drifts of up to 55 K can be compensated, leading to average deviations
of only 3.9% across the full range of phase fractions. Eliminating shortcomings with the
constructive design of the wire fixation will lead to higher accuracy in the future. The
automatic flow pattern identification was verified in an air–water two-phase flow loop,
showing reasonable agreement with flow pattern maps from the literature for horizontal
and vertical pipe orientation. Further work with the developed indWMS system will focus
on three aspects: (1) quantitative validation of flow pattern measurement by comparison
with other tomographic techniques, (2) study of dynamic flow scenarios, i.e., temporal
evolution of temperature and/or flow pattern, as well as the response behavior of the de-
veloped system, and finally, (3) evaluation of the indWMS in real industrial applications. In
addition, we aim at the implementation of the capacitive and dual modality measurement
principles, higher frame rates and advanced data processing using GPUs in the future.

Author Contributions: Conceptualization, E.S.; methodology, E.S., P.W. and M.T.; software, P.W. and
M.T.; validation, P.W. and M.T.; formal analysis, P.W. and F.d.A.D.; investigation, P.W. and F.d.A.D.;
resources, E.S. and U.H.; data curation, P.W.; writing—original draft preparation, P.W., F.d.A.D.,
E.S. and M.T.; writing—review and editing, P.W., F.d.A.D., E.S., M.T. and U.H.; visualization, P.W.;
supervision, U.H. and E.S; project administration, E.S.; funding acquisition, E.S. and M.T. All authors
have read and agreed to the published version of the manuscript.
Funding: This research was funded by the German Federal Ministry for Economic Affairs and Climate
Action (BMWK) under grant number ZF4292502WM6. The APC was funded by Helmholtz-Zentrum
Dresden-Rossendorf.
Data Availability Statement: The experimental data presented in this study are available at doi:10.14278/
rodare.2182 upon reasonable request.
Acknowledgments: The authors appreciate technical support by Martin Tschofen, Bernd Franken-
stein, Andreas Waldert and Johannes Beckert.
Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design
of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or
in the decision to publish the results.

References
1. Hansen, L.S.; Pedersen, S.; Durdevic, P. Multi-Phase Flow Metering in Offshore Oil and Gas Transportation Pipelines: Trends and
Perspectives. Sensors 2019, 19, 2184. [CrossRef]
2. Hampel, U.; Babout, L.; Banasiak, R.; Schleicher, E.; Soleimani, M.; Wondrak, T.; Vauhkonen, M.; Lähivaara, T.; Tan, C.; Hoyle, B.;
et al. A Review on Fast Tomographic Imaging Techniques and Their Potential Application in Industrial Process Control. Sensors
2022, 22, 2309. [CrossRef] [PubMed]
3. Prasser, H.-M.; Böttger, A.; Zschau, J. A new electrode-mesh tomograph for gas-liquid flows. Flow Meas. Instrum. 1998, 9, 111–119.
[CrossRef]
4. Wangjiraniran, W.; Motegi, Y.; Richter, S.; Kikura, M.; Aritomi, M.; Yamamoto, K. Intrusive Effect of Wire Mesh Tomography on
Gas-liquid Flow Measurement. J. Nucl. Sci. Technol. 2003, 40, 932–940. [CrossRef]
5. Ito, D.; Prasser, H.-M.; Kikura, H.; Aritomi, M. Uncertainty and intrusiveness of three-layer wire-mesh sensor. Flow Meas. Instrum.
2011, 22, 249–256. [CrossRef]
6. Tompkins, C.; Prasser, H.-M.; Corradini, M. Wire-mesh sensors: A review of methods and uncertainty in multiphase flows relative
to other measurement techniques. Nucl. Eng. Des. 2018, 337, 205–220. [CrossRef]
7. Dias, F.D.A.; Ripa, D.R.; Morales, R.E.M.; da Silva, M.J. Wire-Mesh Sensor Super-Resolution Based on Statistical Reconstruction.
IEEE Trans. Instrum. Meas. 2021, 70, 4503212. [CrossRef]
8. Prasser, H.-M.; Grunwald, G.; Höhne, T.; Kliem, S.; Rohde, U.; Weiss, F.-P. Coolant Mixing in a Pressurized Water Reactor:
Deboration Transients, Steam-Line Breaks, and Emergency Core Cooling Injection. Nucl. Technol. 2017, 143, 37–56. [CrossRef]
9. Parsi, M.; Vieira, R.E.; Torres, C.F.; Kesana, N.R.; McLaury, B.S.; Shirazi, S.A.; Schleicher, E.; Hampel, U. On the effect of liquid
viscosity on interfacial structures within churn flow: Experimental study using wire mesh sensor. Chem. Eng. Sci. 2015, 130,
221–238. [CrossRef]
10. Kesana, N.R.; Parsi, M.; Vieira, R.E.; Azzopardi, B.; Schleicher, E.; McLaury, B.S.; Shirazi, S.A.; Hampel, U. Visualization of
gas-liquid multiphase pseudo-slug flow using Wire-Mesh Sensor. J. Nat. Gas Sci. Eng. 2017, 46, 477–490. [CrossRef]
Sensors 2023, 23, 4067 13 of 13

11. Timaeus, R.; Berger, R.; Schleicher, E.; Hampel, U.; Schubert, M. Liquid Flow Visualization in Packed-Bed Multiphase Reactors:
Wire-Mesh Sensor Design and Data Analysis for Rotating Fixed Beds. Chem. Ing. Tech. 2019, 91, 1812–1821. [CrossRef]
12. Brinkmann, J.H.; Plate, D.; Merkel, A.; Seithümmer, V.B.; Biessey, P.; Grünewald, M. Application of a wire-mesh sensor for the
experimental characterization of liquid distribution in packed columns. Chem. Eng. Sci. 2021, 242, 116721. [CrossRef]
13. Vishwakarma, V.; Schleicher, E.; Bieberle, A.; Schubert, M.; Hampel, U. Advanced flow profiler for two-phase flow imaging on
distillation trays. Chem. Eng. Sci. 2021, 231, 116280. [CrossRef]
14. Hoffmann, A.; Schleicher, E.; Keller, L.; León Alonso, J.; Pitz-Paal, R. Application of a single wire-mesh sensor in a parabolic
trough facility with direct steam generation. Sol. Energy 2018, 159, 1016–1030. [CrossRef]
15. Kipping, R.; Brito, R.; Schleicher, E.; Hampel, U. Developments for the application of the Wire-Mesh Sensor in industries. Int. J.
Multiph. Flow 2016, 85, 86–95. [CrossRef]
16. Schleicher, E.; Tschofen, M.; Pietruske, H. Gittersensor-System Zum Charakterisieren Einer Fluidströmung. German Patent
DE102015117084B3, 19 May 2016.
17. Pietruske, H.; Prasser, H.-M. Wire-mesh sensors for high-resolving two-phase flow studies at high pressures and temperatures.
Flow Meas. Instrum. 2007, 18, 87–94. [CrossRef]
18. Pietruske, H.; Sühnel, T.; Prasser, H.-M. Gittersensor. German Patent DE102005019739B3, 26 October 2006.
19. da Silva, M.J.; Schleicher, E.; Hampel, U. Capacitance wire-mesh sensor for fast measurement of phase fraction distributions.
Meas. Sci. Technol. 2007, 18, 2245–2251. [CrossRef]
20. dos Santos, E.N.; Vendruscolo, T.P.; Morales, R.E.M.; Schleicher, E.; Hampel, U.; Da Silva, M.J. Dual-modality wire-mesh sensor
for the visualization of three-phase flows. Meas. Sci. Technol. 2015, 26, 105302. [CrossRef]
21. Dias, F.D.A.; dos Santos, E.N.; da Silva, M.J.; Schleicher, E.; Morales, R.E.; Hewakandamby, B.; Hampel, U. New Algorithm
to Discriminate Phase Distribution of Gas-Oil-Water Pipe Flow with Dual-Modality Wire-Mesh Sensor. IEEE Access 2020, 8,
125163–125178. [CrossRef]
22. Wiedemann, P.; de Assis Dias, F.; Schleicher, E.; Hampel, U. Temperature Compensation for Conductivity-Based Phase Fraction
Measurements with Wire-Mesh Sensors in Gas-Liquid Flows of Dilute Aqueous Solutions. Sensors 2020, 20, 7114. [CrossRef]
23. Deutsches Institut für Normung. ISO 7888: 1985; Water Quality—Determination of Electrical Conductivity, German Version EN
27888: 1993. Beuth Verlag: Berlin, Germany, 1993.
24. Wiedemann, P.; Döß, A.; Schleicher, E.; Hampel, U. Fuzzy flow pattern identification in horizontal air-water two-phase flow
based on wire-mesh sensor data. Int. J. Multiph. Flow 2019, 117, 153–162. [CrossRef]
25. Drahoš, J.; Čermák, J. Diagnostics of gas-liquid flow patterns in chemical engineeing systems. Chem. Eng. Process. 1989, 26,
147–164. [CrossRef]
26. Barnea, D. A unified model for predicting flow-pattern transitions for the whole range of pipe inclinations. Int. J. Multiph. Flow
1987, 13, 1–12. [CrossRef]
27. Jones, O.C., Jr.; Zuber, N. The interrelation between void fraction fluctuations and flow patterns in two-phase flow. Int. J. Multiph.
Flow 1975, 2, 273–306. [CrossRef]
28. Rosa, E.S.; Salgado, R.M.; Ohishi, T.; Mastelari, N. Performance comparison of artificial neural networks and expert systems
applied to flow pattern identification in vertical ascendant gas-liquid flows. Int. J. Multiph. Flow 2010, 36, 738–754. [CrossRef]
29. Mandhane, J.M.; Gregory, G.A.; Aziz, K. A flow pattern map for gas-liquid flow in horizontal pipes. Int. J. Multiph. Flow 1974, 1,
537–553. [CrossRef]
30. Abbagoni, B.M.; Yeung, H. Non-invasive classification of gas-liquid two-phase horizontal flow regimes using an ultrasonic
doppler sensor and a neural network. Meas. Sci. Technol. 2016, 27, 084002. [CrossRef]
31. Roman, A.J.; Kreitzer, P.J.; Ervin, J.S.; Hanchak, M.S.; Byrd, L.W. Flow pattern identification of horizontal two-phase refrigerant
flow using neural networks. Int. Commun. Heat Mass Transf. 2016, 71, 254–264. [CrossRef]
32. Sun, Z.; Chen, Y.; Gong, H. Classification of gas-liquid flow patterns by the norm entropy of wavelet decomposed pressure
fluctuations across a bluff body. Meas. Sci. Technol. 2012, 23, 125301. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.