Accepted Manuscript

Analysis of Flow Pattern Transition from Segregated to Slug flow in

Upward Inclined Pipes

Yilin Fan , Eduardo Pereyra , Cem Sarica , Eckhard Schleicher ,

Uwe Hampel

PII: S0301-9322(18)30643-8
DOI: https://doi.org/10.1016/j.ijmultiphaseflow.2019.03.021
Reference: IJMF 2997

To appear in: International Journal of Multiphase Flow

Received date: 3 September 2018

Revised date: 27 January 2019
Accepted date: 25 March 2019

Please cite this article as: Yilin Fan , Eduardo Pereyra , Cem Sarica , Eckhard Schleicher ,
Uwe Hampel , Analysis of Flow Pattern Transition from Segregated to Slug flow
in Upward Inclined Pipes, International Journal of Multiphase Flow (2019), doi:
https://doi.org/10.1016/j.ijmultiphaseflow.2019.03.021

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service
to our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and
all legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

Highlights:
 Experimental study was conducted to investigate the flow pattern transition from segregated to
slug flows step by step.
 The difference between pseudo-slug and conventional slug flows was discussed by analyzing
high-speed videos, 2-D and 3-D wire-mesh sensor data illustration, and structure characteristics.
 Experimental data shows that the structure velocity can be used as an indicator for the flow
pattern transition from conventional slug to pseudo-slug flows.
 New approaches were proposed to model pseudo-slug structure velocity.

T
IP
CR
US
AN
M
ED
PT
CE
AC

1
 ACCEPTED MANUSCRIPT

Analysis of Flow Pattern Transition from Segregated to Slug flow in Upward Inclined
Pipes
Yilin Fan1, Eduardo Pereyra2, Cem Sarica2, Eckhard Schleicher3 and Uwe Hampel3
1
Petroleum Engineering Department, Colorado School of Mines, CO 80401, USA
2
McDougall School of Petroleum Engineering, The University of Tulsa, Tulsa, OK 74104, USA
3
Institute of Fluid Dynamics, Helmholtz-Zentrum Dresden-Rossendorf (HZDR), Germany

Abstract

T
Segregated and slug flows are two of the most common flow patterns encountered in two-phase
upward inclined pipe flow. However, the transition between them is ambiguous. Coherent and

IP
distinctive structures have been observed within the transition. These structures have been
classified as pseudo-slug flow. The nature of pseudo-slug flow is not well understood due to the

CR
complexity of the structure. At low liquid loading conditions, this flow pattern can occupy a
large operating region and cannot be neglected. This paper presents a detailed experimental
work conducted in a facility with a valley configuration, focusing on the transition region
between segregated and slug flows. Wire-mesh sensors were employed to investigate the liquid
phase distribution within the flow structure.
US
The current paper investigates the flow patterns and their transition in upward inclined pipe
from different perspectives. The investigation includes analysis of images from high-speed
AN
videos, evaluation of 2-D liquid holdup axial evolution, 2-D liquid holdup distribution at pipe
cross-section, 3-D interfacial structure evolution, and analysis of pressure gradient and liquid
holdup measurement, flow characteristics, etc. Differences between slug and pseudo-slug flows
are also presented.
M

Three superficial liquid velocities (0.001, 0.005 and 0.01 m/s) and five inclination angles (2°,
5°, 10°, 15°, and 20°) are studied. Simplified correlations are proposed for pseudo-slug
structural velocity, showing fair agreement.
ED

Keywords: Pseudo-slug Flow, Flow Pattern Transition, Two-phase Flow in Inclined Pipe, Wire-
mesh Sensor, Low Liquid Loading
PT

1. Introduction
Segregated (SEG) and slug (SL) flows are the two most common flow patterns encountered in
the oil and gas industry. Segregated flow includes stratified and annular flows, in which gas and
CE

liquid phases flow separately. Slug flow is a flow pattern, in which, sequences of long gas
bubbles are successively followed by liquid slugs that may contain small bubbles (Fabre and
Line, 1992). These two flow patterns neighbor each other in commonly known flow pattern
maps for two-phase upward inclined pipe flow (Barnea 1987; Zhang et al., 2003). However, the
AC

transition from segregated flow to slug flow does not occur abruptly. There is a transition region
between them, which has been named differently amongst the researchers.
The transition was recognized by Nicholson et al. (1977) and it was named as “proto-slug”.
Barnea et al. (1980) defined it as “wavy-annular” flow. Lin and Hanratty (1987b) later provided
a detailed description of this flow pattern, and named it as “pseudo-slug” flow, which is the term
adopted in the current study. They also stated that the pseudo-slugs observed in the 2.54 cm pipe
resemble the “disturbance wave” described by Butterworth and Pulling (1972) and the picture of
a “slug” presented in the paper by Alves (1954). The pseudo-slug flow was classified into

2
 ACCEPTED MANUSCRIPT

“intermittent flow” together with slug and plug flows in Taitel and Dukler (1976) flow pattern
transition model.
Lin and Hanratty (1987b) conducted an experimental study of air-water flow in horizontal
pipes with two different diameters. They defined the pseudo-slugs as: “The disturbances, which
have the appearance of slugs, but which do not give the identifying pressure pattern and do not
travel at the gas velocity. The pseudo-slug flow pattern resembles: annular flow in that a
continuous liquid film is formed on the pipe circumference; wavy-stratified flow in that a thick
layer of liquid is present at the bottom of the pipe; and slug flow in that large slug-like structures
capable of reaching the top of the pipe are present.” In their experiment, they observed large, ill-

T
defined waves in the pseudo-slug body, which could touch the top wall shortly, but did not block
the entire cross-sectional area of the pipe as slugs did. A large void was observed in the upper

IP
portion of the pseudo-slugs.
There were many other studies reported the observation of pseudo-slug flow in horizontal

CR
pipes under high liquid flow rate conditions such as Lee (1993), Soleimani and Hanratty (2003)
and Loh et al. (2016). Pseudo-slug flow was also reported near the transition from segregated to
slug flows in upward inclined pipes with θ < 30° (Wilkens, 1997; Maley, 1997; Alsaadi, 2013,
2015; Ekinci, 2015; Brito, 2015, 2016, and Soedarmo et al., 2018). Similar with the

US
observations of Lin and Hanratty (1987a, b), the translational velocity of pseudo-slugs was
smaller than that expected for slugs in all of these studies. Wilkens (1997) and Maley (1997)
also showed that the slug flow pattern became dominated by pseudo-slug flow as the pressure
AN
was increased. Alsaadi (2013, 2015) conducted an experimental study in 3-in low pressure
upward inclined pipes (2° ~ 30°) using air and tap water. He observed a significant occurrence
of pseudo-slug flow in his experiments. Moreover, Ekinci (2015) conducted experiments on slug
flow characterization for high-viscosity oil and gas flow in ±2° inclined pipe, and reported
M

pseudo-slug flow at high gas flow rates.

It can be concluded from the previous studies that pseudo-slug flow pattern exists near the
transition from segregated to slug flow for horizontal and upward inclined pipe, and its
ED

topological structure is quite different from that of slug flow. This flow pattern can occupy a
large region on a flow pattern map at low liquid flow conditions or high-viscosity oil and air
flow in an upward inclined pipe. One of the important differences between slug and pseudo-slug
flow is their structural velocities. Since the current study employs many concepts of velocities
PT

for flow pattern characterization, the following sub-section provides a general review of these
classical concepts used in the analysis. The second sub-section summarizes previous theories for
the initiation of intermittent flow.
CE

1.1 Classical concepts of structure velocities

For conventional slug flow, the structure velocity refers to the translational velocity, vT,
which is the propagation or the front velocity of the slug, or the velocity of the interface between
AC

the gas and the liquid phases (Dukler and Hubbard, 1975). Obviously, the entire slug unit moves
at translational velocity for a stable fully-developed slug flow. For Taylor bubble flow in
upward vertical pipes, it is equivalent to long bubble velocity as mentioned by Fabre and Line
(1992), or Taylor-bubble-rise velocity by Sylvester (1987).
Slug translational velocity has presented a clear relationship with the mixture velocity
according to previous studies such as Ekinci (2015), Soedarmo et al. (2018) and Felizola (1992)
to name a few. Nicklin et al. (1962) proposed to decompose the translational velocity in two
terms, as given by:

3
 ACCEPTED MANUSCRIPT

vT  c0vm  vD . (1)
Where co is the flow distribution coefficient, which is related to the contribution of the mixture
velocity to the translational velocity. For turbulent flow, co is 1.2, while for laminar flow, co is
close to 2 (Shoham, 2006).
In the equation, vD is the drift velocity, which represents the bubble velocity in a stagnant
liquid. Bendiksen (1984) proposed a simple correlation for vD as a function of inclination angle,
θ, as given by:
vD  vD, H cos   vD,V sin  . (2)

T
Where vD,H is the drift velocity for horizontal pipe flow, and vD,V for vertical pipe flow. Based on
the studies of Davies and Taylor (1950) and Dumitrescu (1943), the drift velocity in vertical

IP
pipes can be expressed as:
vD,V  0.35 gd . (3)

CR
Benjamin (1968) proposed a correlation for the drift velocity in horizontal pipes, as given by:
vD, H  0.54 gd . (4)
With that, the final expression for the slug translational velocity can be expressed by:

US
vT  c0vm  0.54 gd cos   0.35 gd sin  .
Detailed discussion on the bubble motion in slug flow can be found in Fabre and Line (1992).
(5)

For segregated flow, the structural velocity refers to the wave celerity, which represents the
AN
velocity of waves at gas-liquid interface.

1.2 Previous theory on the initiation of intermittent flow

The transition from segregated flow to the other flow pattern has been extensively studied in
M

the past decades due to its importance in industrial operational application, such as onset of slug
flow, which can induce potential pipeline corrosion, pipeline fatigue, and reduce of production in
oil and gas industry.
ED

The onset of slug flow in horizontal and near-horizontal pipe flow was based on the
instability of stratified gas-liquid flow in earlier studies, and models were proposed utilizing the
classical Kelvin-Helmholtz (KH) instability theory, in which transition occurs when long
wavelength disturbances becomes unstable and grows into slugs (Taitel and Dukler, 1976).
PT

Numerous studies were conducted afterwards to improve the flow pattern transition criteria, by
including viscous effect (Lin and Hanratty 1986; Andritsos et al. 1989; Barnea and Taitel 1992),
arbitrary wavenumber or all-wavelength disturbances (Barmak et al. 2016; Picchi et al. 2018),
CE

and non-linear interfacial instability (Campbell and Liu 2016), to name a few.
For upward inclined pipe, Barnea (1987) proposed a unified mechanics model based on the
instability of the annular-flow configuration, in which the flow pattern transition was associated
AC

with the minimum interfacial shear stress. Another theory stated that the onset of intermittent
flow was associated with the left boundary of the multiple solution region of the classical two-
fluid model (Biberg et al. 2015; Ullmann et al. 2003). Barnea and Taitel (1992) conducted
structural and interfacial stability analysis of these triple solutions, stating that the lowest
solution was the only stable one, while the middle solution is always structurally unstable and the
highest solution is structurally unstable to large disturbances. Based on this theory, a sudden
holdup jump should be observed at the transition from the lowest to the highest solutions, i.e.,
left boundary of the multiple solution region. This hypothesis was experimentally proven by
Kjølaas et al. (2015), Staff and Holm (2016), and Ullmann et al. (2003) in a slightly upward

4
 ACCEPTED MANUSCRIPT

inclined gas-liquid, gas-liquid-liquid, and liquid-liquid system, respectively. Ullmann et al.

(2003) presented PIV experimental results illustrating the velocity profile of oil-water flow in an
upward inclined pipe (0.6°). The data showed a sudden increase of water holdup when
increasing water flow rate while keeping oil flow rate constant, which was due to the liquid film
reversal at pipe bottom. In the current experimental study, the onset of liquid film reversal is
investigated for a gas-liquid system in an upward inclined pipe with a wider range (up to 20°)
using a specially designed dye injection system. The associated flow pattern is also discussed in
this paper.
To better understand the flow pattern transition from segregated to intermittent flow in

T
upward inclined pipe, the current study employed wire-mesh sensors to investigate the liquid
phase distribution at pipe cross-sectional area. This paper provides a detailed analysis of images

IP
from high-speed videos, evaluation of 2-D liquid holdup axial evolution, 2-D liquid holdup
distribution at pipe cross-section and 3-D interfacial structure evolution, and analysis of pressure

CR
gradient and liquid holdup measurement, and structure velocities.
Low liquid loading is crucial in many circumstances, such as a gas condensate pipeline with
a large diameter (Holm, 2015). Literature review shows that few of previous studies investigated
low liquid-flow rate (Fan et al., 2015). In the current study, three superficial liquid velocities

US
(0.001, 0.005 and 0.01 m/s) and five inclination angles (2°, 5°, 10°, 15°, and 20°) are studied.
Simplified correlations are proposed for pseudo-slug structural velocity, which can be easily
applied in pseudo-slug mechanics model such as Soedarmo et al. (2018).
AN
2. Experimental Facility
The Tulsa University Fluid Flow Projects (TUFFP) low-pressure air/water 0.0762 m ID facility
was used in this experimental study. The schematic of the facility and the instrumentation in the
M

test section are presented in Figs. 1 and 2, respectively. Tulsa city municipal water and
compressed air were used as the test fluids. The air was compressed by a Garden Denver
compressor through a 0.0508 m air line with a capacity of 1050 SCFM at 0.69 MPa. The air
ED

flow rate was measured by a Micro MotionTM DL 200 mass flow rate meter. Water was pumped
from the water tank by a progressive cavity pump and measured by a Micro MotionTM DS 150
mass flow meter. Both liquid and gas mass flow meters were connected to a PID controller to
control the mass flow rate. The measurement and control of the instruments were automated
PT

with LabviewTM data acquisition system. All of the tests were conducted under atmospheric
pressure and temperature conditions.
CE
AC

5
 ACCEPTED MANUSCRIPT

T
IP
CR
US
Fig. 1. Schematic Diagram of the experimental facility (dark blue line represents water feed line;
orange line refers to air feed line; light blue line is the test section; red arrow shows the flow
direction of the gas and liquid mixture; air and water were injected at the top of the downhill
AN
section separately)
M
ED
PT
CE
AC

Fig. 2. Test Section with Instrumentation (TT: Temperature Transmitter; PT: Pressure
Transmitter; DP: Differential Pressure Transducer; QCV: Quick Closing Valve; WMS: Wire-
mesh Sensor)

6
 ACCEPTED MANUSCRIPT

The valley configuration of the test section consisted of downhill, horizontal and uphill
sections, as shown in Figs. 1 and 2. The horizontal and uphill sections were made of acrylic
glass for better visualization. The outlet of the pipe was open to the atmosphere. As a result,
there were no back pressure or syphon effects on the flow in the test section. The downhill
section was mounted on the same boom structure as the uphill section, making the inclination
angles always the same but with opposite sign. A U-shaped acrylic pipe was utilized at the
middle of the horizontal section, as shown in Fig.1. The instrumentation of the valley test
section and the corresponding dimensions are shown in Fig. 2. Overall, there were three
differential pressure transducers, located in horizontal, horizontal-uphill elbow and uphill section,

T
respectively. Three liquid holdups were measured for each test point, including HL1 (uphill
section, between QCV#1 and QCV#2), HL2 (horizontal-uphill elbow, between QCV#4 and

IP
QCV#1) and HL3 (horizontal section, between QCV#3 and QCV#4). The current study mainly
presents the pressure gradient and liquid holdup in the uphill section (dp/dL1 and HL1). There

CR
were four surveillance cameras located along the uphill section for flow pattern observation.
Wire-mesh sensors were employed in the current study to study the complex behavior of
pseudo-slug flow. The experimental study was conducted in two phases. Pressure gradient and
liquid holdup were measured during the first phase without wire-mesh sensor while the second

US
phase focused only on wire-mesh sensor data acquisition. Dye injection system was employed in
the first phase to study the onset of liquid film reversal.
The experimental test matrix of the current study is given in Table 1. Five inclination
AN
angles were studied while superficial liquid velocity varied from 0.001 to 0.01 m/s. Superficial
gas velocity was reduced approximately from 32 m/s to 2 m/s with a step of 0.5 m/s near the
onset of liquid accumulation region. The water has a viscosity around 1.0 cp, anda density of
998 kg/m3. The air viscosity was approximately 0.0185 cp, and air density was 1.2 kg/m3 at
M

atmospheric condition. The average operating temperature is around 28°C.

Table 1. Experimental Test Matrix for Current Study
ED

Operating Parameters Range

Pipe Geometry Valley Test Section
Water superficial velocity, vSL 0.001, 0.005 and 0.01 m/s
Air superficial velocity, vSg 32 to 2 m/s
PT

Inclination angle, θ 2°, 5°, 10°, 15°, 20°

Operating pressure 0.1 MPa
CE

2.1 Liquid Film Reversal Measurement

Previous experimental studies showed that liquid film reversal was the main source for liquid
accumulation (Zabaras et al., 1986; Guner, 2012, 2015; Alsaadi, 2013, 2015; and Brito, 2015,
AC

2016).
A special dye injection system was employed in the current study to detect the flow
direction of the liquid film at pipe bottom for inclined pipe. The detailed description of the dye
injection system can be found in Fan et al. (2018). If there is no liquid film reversal, the dye
flows upward with the liquid film without any disturbance of the upstream liquid film. Once the
liquid film starts to reverse, the blue dye clearly contaminates the upstream liquid film. The
experimental results are presented in the next section.

2.2 Wire-mesh sensor

7
 ACCEPTED MANUSCRIPT

Since the pseudo-slug body shows a very frothy structure as compared with regular slug body,
the traditional measurement techniques for slug flow, such as conductivity probe or capacitance
sensor, are not reliable in term of precise measurement of the characteristics parameters. A pair
of conductivity wire-mesh sensors (WMS), constructed and supported by HZDR Dresden from
Germany, was employed in this study to better quantify the pseudo-slug characteristics.
WMS enables the investigation of the liquid phase distribution across the cross-sectional
area of the pipe with a very high spatial and temporal resolution (Prasser et al., 1998). The
schematic of the valley test section with WMSs is shown in Fig. 3. Also shown is a picture of
WMS used in the current study.

T
IP
CR
US
Fig. 3. Wire-mesh Sensor installation position in test section (left) and photograph (right)
AN
The WMS measurement principle is based on a local measurement of electrical conductance
of the fluid within the cross-section of a pipe by means of a mesh of crossing electrodes. Each
WMS has two planes, the emitter electrodes in the first plane provide bipolar constant voltage
pulses and the electrodes of the other plane arranged orthogonal to the emitter plane electrodes
M

serve as the current receivers. There is a tiny gap of few millimeters between the emitter and
receiver electrodes, where the conductance of the fluids is measured in the crossing points of the
electrodes as the quotient of the received current and the supplied voltage. The wires are
ED

connected with corresponding transmitting or receiving modules, the maximum connectable

transmitting electrodes of which is 16. A 24x24 WMS was specifically designed for the current
study. The transmitter electrodes are activated sequentially while the receiver electrodes are
sampled in parallel (Kipping et al., 2016). The small receiver current is amplified and converted
PT

to a voltage by means of trans-impedance amplifier circuits and eventually converted to a digital

signal that is transferred to the measurement PC by means of a fast digital signal processing
electronics.
CE

As the conductance of the fluid varies with temperature, the WMSs were calibrated by fully
filling the pipe with water twice each day, in the morning and afternoon, in order to minimize the
temperature effect on the measurement. The final converted output of the WMS is a 3-D void
fraction, ε (i, j, n), where i and j represent the coordinates across the cross-sectional area of the
AC

pipe and n refers to the frame number.

For the current study, the WMS measurement frequency was set to 5 kHz. Considering the
low pseudo-slug frequency, each measurement was recorded for 300 s. The distance between the
two WMSs was 0.06 m. The first sensor, the one close to inlet, was used for detailed data
analysis of liquid/gas holdup distributions while the second one was utilized for the purpose of
velocity measurement only. The structure velocity was obtained by cross-correlating the signals
from two adjacent wire-mesh sensors.

8
 ACCEPTED MANUSCRIPT

3. Flow Pattern Classification

It is commonly recognized that in segregated flow the gas and liquid are transported separately in
a stable manner; while in intermittent flow gas and liquid are transported alternately. Generally,
segregate flow is consisted of stratified (ST) and annular (AN) flow, while intermittent flow
includes slug (SL) and elongated-bubble (EB) flows (Shoham, 2006). It is widely accepted that
slug flow is characterized by alternating flow of gas pocket and liquid slugs, in which the liquid
slug body fully bocks the entire cross-sectional area of the pipe.
The transition region between segregated and regular slug flows has been widely recognized
and named differently by various authors. This flow pattern has been classified as pseudo-slug

T
whose nature is not well understood as referred in previous studies (Lin and Hanratty, 1987a, b).
In the current study, pseudo-slug flow has been classified into intermittent flow due to its

IP
intermittent nature as compared with segregated flow.
An example of the observed flow pattern is shown in Fig. 4 as function of inclination angle

CR
and superficial gas velocity for a constant liquid flow rate. The Eötvös number of the current
experimental study is approximately 800, indicating that the surface tension effect is negligible.
In general, the flow pattern can be classified into three categories: segregated (SEG), intermittent
(INT) and the transition between segregated and intermittent flow (TRA). These flow patterns

US
are further divided into several sub-categories based on the liquid film reversal, liquid
entrainment in gas core, and the flow characteristic. Detailed description is provided in the
following sub-sections.
AN
M
ED
PT
CE

Fig. 4. Variation of Flow Pattern with vSg and θ (vSL = 0.005 m/s)
AC

(SEG: Segregated flow with no film reversal; INT: Intermittent Flow; TRA: Transition between
SEG and INT; SW&ED: Stratified wavy with entrained liquid droplet in gas core; SW: Stratified
wavy with no liquid droplet entrainment; SW&LFR: stratified wavy with liquid film reversal at
pipe bottom; LW&LFR: Large wave with liquid film reversal; PSL&ED: Pseudo-slug flow with
entrained liquid droplet in film region; PSL: pseudo-slug flow with no liquid entrainment in the
film region; SL&PSL: combination of slug and pseudo-slug flow)

3.1 Flow Pattern Transition between SEG, TRA and INT

9
 ACCEPTED MANUSCRIPT

The black-solid line in Fig. 4 represents the experimental measurement of the onset of liquid film
reversal. To the left of the black-solid line, the liquid phase starts to accumulate at the bottom of
inclined pipe, increasing the risk of pipeline internal corrosion, and increasing pressure losses.
The superficial gas velocity corresponding to the onset of liquid film reversal at pipe bottom is
referred as critical gas velocity, vSgc (Fan et al., 2018). In the current study, the onset of liquid
film reversal is measured by a dye injection system at the bottom of inclined pipe, as shown in
Fig. 5. The blue dye can easily contaminate the upstream liquid film if there is any film reversal
at the bottom. Theoretically, the liquid film reversal is caused by the insufficient interfacial
shear stress exposed at the gas liquid interface, which cannot overcome the gravity force exerted

T
on the liquid film region. The liquid film reversal induces the increase of liquid holdup
(explained later in Fig. 6), which is consistent with the observation from previous studies

IP
(Ullmann et al., 2003). Also shown in Fig. 4 is the model predictions by Fan et al. (2018) (red-
dashed line) and TUFFP Unified Model (Zhang et al. 2003) (black-dashed line). Detailed

CR
description about the model development and evaluation can be found in Fan et al. (2018). The
variation of flow pattern with gas flow rate and inclination angle for the other two liquid flow
rates is shown in Appendix A. It can be noticed that the liquid flow rate effect on the onset of
liquid film reversal is more evident for a higher inclination angle. This is mainly due to the

US
increase of liquid holdup with increasing liquid flow rate, thus increasing of the gravitational
component in the axial direction opposite to the flow direction, and leading to an easier film
reversal at pipe bottom. This phenomenon can be captured well by Fan et al. (2018)’s model as
AN
can be seen from Fig. 4, A-1, and A-2. In the current study, the onset of liquid film reversal is
used as the transition from segregated flow to the other flow patterns on the left.
M
ED
PT
CE

(a). vSg = 11.3 m/s (b). vSg = 10.2 m/s

Fig. 5. Flow Direction Test with Dye in Upward Inclined Pipe (vSL = 0.005 m/s; θ = 2°)

The flow pattern transition to INT flow can be seen from the pressure gradient and liquid
AC

holdup measurement, as presented in Fig. 6. Before the onset of liquid film reversal, both phases
flow separately and the frictional pressure loss dominates. The pressure gradient decreases with
decreasing gas velocity due to the decreasing frictional loss in the gas phase. Once the liquid
starts accumulating in the inclined pipe, the decrease of frictional pressure gradient is balanced
by the increase of gravitational pressure gradient due to the increase of liquid holdup. The flow
changes from frictional to gravitational dominated, and the variation of the total pressure
gradient with decreasing superficial gas velocity changes from decrease to increase. After the
flow changes to INT where regular periodical behavior persists, the variation of pressure gradient
and liquid holdup with superficial gas velocity becomes smoother.

10
 ACCEPTED MANUSCRIPT

T
IP
(a). Pressure Gradient vs. vSg (b). Liquid Holdup vs. vSg

CR
Fig. 6. Variation of Pressure Gradient and Liquid Holdup with Inclination Angle and vSg at
vSL = 0.005 m/s

3.2 Flow Pattern Description

US
As aforementioned, the three major flow patterns are divided into smaller groups depending on
the flow characteristics, which are described below.
AN
3.2.1 Segregated Flow
Segregated flow occurs when the interfacial shear stress is sufficient to drag the liquid phase in
the same direction as the gas phase. This flow pattern was observed at high gas flow rates (RHS
M

of the red lines in Fig. 4).

Based on the liquid wetted perimeter and liquid entrainment, the segregated flow is
subdivided into annular flow (AN), stratified wavy flow with entrained droplet (SW&ED) and
ED

stratified wavy flow without entrained liquid droplet (SW). Annular flow was only observed at
vSL = 0.01 m/s at high superficial gas velocities, which is shown in Fig. A-2. The schematics and
the corresponding pictures are shown in Fig. 7, with the numbering marked in Fig. 4 and Fig. A-
2. Stratified smooth flow (SS), which was observed in horizontal and near-horizontal pipe
PT

(Taitel and Dukler 1976 and Fan 2005), was not observed in the uphill section in the current
study.
CE
AC

11
 ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
Fig. 7. Schematics and Pictures of the Sub-classified Flow Pattern in
Segregated Flow

In annular flow, the gas phase flows as a core with entrained liquid droplets, while most of
M

the liquid phase stays at the pipe bottom due to gravity. There is a continuous thin liquid film
around the pipe wall, which mainly comes from the deposition of entrained liquid droplets.
When the liquid or gas flow rate is not high enough, the deposition of the liquid entrainment
ED

is not sufficient to form a continuous moving liquid film around the entire pipe wall. This flow
pattern is called stratified flow (Shoham, 2006). Based on the liquid entrainment, the stratified
flow is sub-divided into stratified wavy flow with entrained droplet (SW&ED) and without
entrained droplet (SW). For SW&ED, the entrained liquid droplets have been observed hitting
PT

the pipe wall intermittently due to the high turbulence of the gas core, as illustrated in Fig. 7. As
a comparison, there is no liquid droplet deposition observed for SW. SW occurs at lower liquid
and gas flow rates as compared with SW&ED.
CE

3.2.2 Transition Region

When the interfacial shear stress is not enough to drag all of the liquid in the film in upward
AC

direction, the liquid flow direction at the bottom of the pipe reverses due to gravity as described
in Section 3.1. In the current study, the flow pattern, which occurs after the onset of liquid film
reversal but does not show clear regular intermittent behavior as pseudo-slug/slug flow does, is
named as transition region (TRA). Based on the liquid entrainment and wave behavior at the
interface, this flow pattern is sub-divided into two categories, the pictures of which are shown in
Fig. 8.

12
 ACCEPTED MANUSCRIPT

T
IP
CR
Fig. 8. Pictures of the Sub-classified Flow Pattern in Transition Region

US
For low inclination angles (2° and 5° in the current study), this transition region is simply
stratified wavy with liquid film reversal (SW&LFR). Roll waves were observed at the gas liquid
interface (Fig. 8). This flow pattern is marked as black crosses in Fig. 4.
AN
For higher inclination angles (10°, 15° and 20° in the current study), large backward flowing
waves were observed as shown in Fig. 8. The gas phase breaks the large waves and creates a
large amount of liquid droplets, which are entrained in the gas phase. The entrained liquid
droplets then deposit on the pipe wall generating a liquid film surrounding the pipe wall. The
M

flow behavior is chaotic and shows no significant intermittent behavior as compared with
intermittent flow. The difference between the transition region and intermittent flow can also be
reflected from the pressure gradient and liquid holdup data as shown before in Fig. 6. This flow
ED

pattern is also different from regular annular flow in that the large liquid waves move backward.
This flow pattern is named as large wave with liquid film reversal (LW&LFR) in the current
study, which is marked as red crosses in Fig. 4.
The transition region is very narrow as compared with other well-defined flow patterns,
PT

such as segregated, pseudo-slug (explained later), or slug flow. It may not attract too much
attention from modeling point of view, but it serves as a bridge between segregated and
intermittent flow.
CE

3.2.3 Intermittent Flow

With decreasing vSg, the intermittent behavior of the liquid phase is more evident, similar with
slug flow but with a frothy liquid body. This flow pattern has been classified as pseudo-slug
AC

flow (PSL). The current study sub-divides the pseudo-slug flow pattern into two sub-
classifications depending on the liquid entrainment in the film region. The flow patterns are
pseudo-slug flow with entrained liquid droplet in the film region (PSL&ED), and pseudo-slug
flow with no entrainment in the film region (PSL). The corresponding pictures are shown in
Fig. 9.

13
 ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M

Fig. 9. Pictures of the Sub-classified Flow Pattern in Intermittent Flow (PSL&ED: pseudo-slug
flow with entrained liquid droplets in the film region; PSL: pseudo-slug flow with no liquid
droplets entrained in the film region; SL&PSL: combination of conventional slug and pseudo-
ED

slug flow – only a picture of pseudo-slug is shown in this figure)

PSL&ED is marked as orange diamond in Fig. 4. At the gas/liquid interface in the film
PT

region, the gas phase breaks the roll waves and entrains the created liquid droplets into the gas
phase. The deposition of the entrained liquid droplets then forms a thin liquid film surrounding
the pipe wall as shown in Fig. 9 (#6). For this flow pattern, the liquid phase is transported by
three flow structures: pseudo-slugs, liquid film and liquid droplets entrained in the gas phase in
CE

the film region, the amount of which depends on inclination angle and the gas flow rate. This
flow pattern was observed at high inclination angles (θ ≥ 10°) and high superficial gas velocities.
For lower inclination angles (2° and 5° in the current study), the gas phase is unable to break
AC

the waves due to the higher gravitational force perpendicular to the liquid film. Consequently,
there is no entrained liquid droplet observed in the liquid film region. For high superficial gas
velocities, roll waves are still observed at the interface as shown in Fig. 9 (#7). At lower
superficial gas velocities, smooth interface or backward flowing tiny ripple waves were observed
at the gas/liquid interface in the liquid film region (#8 in Fig. 9).
A graduate and smooth transition from pseudo-slug to conventional slug flow was observed
at the lowest gas flow rates. The flow pattern, which includes both pseudo-slug and slug flows,
is named as Slug&Pseudo-slug (SL&PSL). A detailed description of the comparison between

14
 ACCEPTED MANUSCRIPT

conventional slug and pseudo-slug is provided in the next section, followed by the transition
between PSL and PSL&SL.

3.3 Comparison between pseudo-slug and slug

Figure 10 shows pictures of a slug (left) and a pseudo-slug (right) respectively. It can be seen in
Fig. 10 (a) that the liquid phase within the slug body can fully block the entire cross-sectional
area of the pipe with some gas bubbles entrained in the slug body. However, the pseudo-slug
body shows a much more frothy structure as compared with slug flow as shown in Fig. 10 (b).

T
IP
t1 Liquid Film at Slug Front t1 Liquid Film at Pseudo-Slug Front

CR
Slug Front Pseudo-Slug Front
t2 t2

t3

Slug Tail
Slug Body
US t3
Pseudo-Slug Body
AN
t4 t4 Pseudo-Slug Tail

Liquid Film after Slug Liquid Film after Pseudo-Slug

t5 t5
M

(a). Slug (b). Pseudo-slug

(θ = 10°; vSL = 0.005 m/s; vSg = 3 m/s) (θ = 10°; vSL = 0.005 m/s; vSg = 6 m/s)
ED

Fig. 10. Images of Regular Slug (a) and Pseudo-slug (b)

To better understand the liquid phase distribution within pseudo-slug body, wire-mesh
PT

sensor data were analyzed from different perspectives. Figures 11 and 12 show examples of a
slug body and pseudo-slug body, respectively. The 3-D plots on top illustrate the iso-surface
evolution for slug and pseudo-slug flows. The iso-surfaces were generated by TecplotTM by
CE

connecting the points having the same liquid holdup (0.3 for the current study) within the
volumetric data field (Tecplot 360TM User’s Manual Version 2013). The 2-D axially sliced
images in Figs. 11 and 12 show the liquid holdup distribution evolution at the center of the cross-
AC

sectional area of the pipe, while the figures at bottom illustrate the spontaneous liquid phase
distribution at the pipe cross-sectional area at different time.
Figure 11 shows clearly that, the liquid phase can fully block the entire cross-sectional area
of the pipe in slug liquid body. The 3-D images also illustrate the entrained gas bubbles within
the continuous liquid phase in slug body. Since the passage of the continuous gas phase is fully
blocked by the continuous liquid slug body, the slug body is pushed forward by the upstream gas
pocket efficiently, leading to a greater slug translational velocity than the mixture velocity. The
translational velocity of this particular slug is 3.9 m/s, which is comparable with that calculated
from slug translational velocity closure relationship (3.6 m/s from Eqn. (5) with co = 1.2).

15
 ACCEPTED MANUSCRIPT

t1 t2 t 3 t4 t5 t6 t7 t8 t9 t10 t11 t12 t13 t14

T
IP
CR
t= 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 s

0 10 20 30 40 50 60 70 80 90 100

US
AN
t1 = 0 t2 = 0.100 s t3 = 0.180 s t4 = 0.181 s t5 = 0.207 s t6 = 0.235 s t7 = 0.293 s
M

t8 = 0.409 s t9 = 0.430 s t10 = 0.496 s t11 = 0.548 s t12 = 0.605 s t13 = 0.700 s t14 = 0.800 s
ED

Fig. 11. WMS Data Visualization of a Slug Body (θ = 10°; vSL = 0.005 m/s; vSg = 2.6 m/s; vT =
3.9 m/s) (Top: iso-surface created by TecplotTM 360; Middle: liquid holdup evolution at pipe
PT

center; Bottom: liquid phase distribution at different time)

However, the pseudo-slug body shows a different topological structure within the body as
CE

shown in Fig. 12. It can be seen that the liquid phase cannot fully block the entire cross-sectional
area of the pipe in the pseudo-slug body, consequently there is continuous gas flowing through
the pseudo-slug body. It has also been observed that there is a large amount of liquid droplet
entrained in the gas phase passing through the pseudo-slug body. The deposition of these
AC

droplets forms a liquid film surrounding the pipe wall (t5 – t10 in Fig. 12). The translational
velocity of this particular pseudo-slug is 2.54 m/s, much less than the gas superficial velocity, 6.1
m/s. This observation is consistent with the findings from previous literatures (Lin and Hanratty
1987a; Ekinci 2015; Fan et al. 2015; and Fan 2017).

16
 ACCEPTED MANUSCRIPT

t1 t2 t3 t4 t5 t6 t7 t8 t9 t10 t11 t12 t13 t14

T
IP
t= 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 s

CR
0 10 20 30 40 50 60 70 80 90 100

t1 = 0.100 s t2 = 0.200 s t3 = 0.300 s

US
t4 = 0.323 s t5 = 0.343 s t6 = 0.378 s t7 = 0.396 s
AN
M

t8 = 0.439 s t9 = 0.474 s t10 = 0.493 s t11 = 0.518 s t12 = 0.562 s t13 = 0.700 s t14 = 0.800 s
ED

Fig. 12. WMS Data Visualization of a Pseudo-slug Body (θ = 10°; vSL = 0.005 m/s; vSg = 6.4 m/s;
vPSL = 2.54 m/s) (Top: iso-surface created by TecplotTM 360; Middle: liquid holdup evolution at
pipe center; Bottom: liquid phase distribution at different time)
PT

3.4 Transition from PSL to SL

As aforementioned, there is no sudden change from pseudo-slug to conventional slug flow.
CE

There is a region in which both pseudo-slug and conventional slug coexist, the number fraction
of which depends on the inclination angle, gas and liquid flow rates, fluid properties, etc.
To illustrate the superficial gas velocity effects on flow pattern transition, a series of 1-D
average HL data and 2-D liquid holdup distribution at pipe center are shown in Figs. 14 – 16 for
AC

different gas velocities (2.1, 4.1 and 6.1 m/s, respectively). These points are marked in the flow
pattern map in Fig. 13. The plots on top represent the variation of spatial average liquid holdup
from wire-mesh sensor with time. It is worth to mention that the x-axis of the 3-D and 2-D axial
images in Figs. 14 – 16 represents time, not length. The time duration of the slug/pseudo-slug
body does not stand for the actual slug/pseudo-slug body length. It can be seen that the flow
structure varies even at one constant superficial gas velocity. At vSg = 2.1 m/s (the lowest vSg that
the current study can reach at), both slug and pseudo-slug are observed (Fig. 14), while slug flow
dominates. Figure 15 shows that the flow structure gradually becomes pseudo-slugs dominated

17
 ACCEPTED MANUSCRIPT

with increasing vSg. Only pseudo-slugs are observed when superficial gas velocity is increased to
6.1 m/s (Fig. 16).
Figure 17 shows the variation of the pseudo-slug and slug number fraction with gas and
liquid flow rates and inclination angle. The differentiation of slug and pseudo-slug is based on
the liquid phase distribution shown in the axial and cross-sectional 2-D plots. If the liquid phase
blocks the entire pipe cross-section without a passageway for gas within the liquid slug body, it
is a slug. Otherwise it is a pseudo-slug. It can be seen that conventional slug starts to occur
when the superficial gas velocity decreases to a critical velocity, the value of which depends on
inclination angle and liquid flow rate for the current experimental condition. Then, the number

T
fraction of conventional slug body increases with decreasing gas flow rate or increasing liquid
flow rate. It can be concluded from Fig. 17 that the transition from pseudo-slug to conventional

IP
slug flow does not occur abruptly. There is a smooth transition between these two flow patterns.
Figure 17 also shows that the slug fraction increases with increasing inclination angle (for ≤ 20°).

CR
This trend can also be illustrated from the pseudo-slug velocity plot, as discussed below.

US
AN
M
ED
PT

Fig. 13. Flow Pattern Map (θ = 10°)

CE
AC

18
 ACCEPTED MANUSCRIPT

1
1 2 3 4 5 6 7 8 9 10
HL

0
t= 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 s
1 6

T
2 7

IP
3 8

CR
4 9

5
US 10
AN
t= 0 0.2 0.4 0.6 0.8 1s t= 0 0.2 0.4 0.6 0.8 1s

0 10 20 30 40 50 60 70 80 90 100

Fig. 14. 1-D and 2-D Liquid Holdup Evolution (θ = 10°; vSL = 0.005 m/s; vSg = 2.1 m/s) (Point 1
M

in Fig. 13)
ED
PT
CE
AC

19
 ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
0 10 20 30 40 50 60 70 80 90 100

Fig. 15. 1-D and 2-D Liquid Holdup Evolution (θ = 10°; vSL = 0.005 m/s; vSg = 4.1 m/s) (Point 2
M

in Fig. 13)
ED
PT
CE
AC

20
 ACCEPTED MANUSCRIPT

1
1
2
5 7 8 10
6
HL

3 4 9

0
t= 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 s

1 6

T
2 7

IP
3 8

CR
4 9

5 US 10
AN
t= 0 0.2 0.4 0.6 0.8 1s t= 0 0.2 0.4 0.6 0.8 1s

0 10 20 30 40 50 60 70 80 90 100
M

Fig. 16. 1-D and 2-D Liquid Holdup Evolution (θ = 10°; vSL = 0.005 m/s; vSg = 6.1 m/s) (Point 3
in Fig. 13)
ED
PT
CE
AC

21
 ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M
ED
PT
CE
AC

Fig. 17. Fraction of Number of Slug and Pseudo-slug over the Total Number of Structure

It has been recognized that one of the distinguished difference between conventional slug
flow and pseudo-slug flow is their structure velocity (Lin and Hanratty 1987a; Fan et al. 2015;

22
 ACCEPTED MANUSCRIPT

Ekinci 2015). Pseudo-slug flow shows a smaller structure velocity as compared with slug flow
due to the insufficient blockage. Figure 18 shows the variation of the structure velocity with
superficial gas velocity for the five different inclination angles, which also indicates the flow
pattern transition. In SEG, the structure velocity represents wave celerity (CW), which decreases
with decreasing gas flow rate. This is consistent with previous studies for segregated flow such
as Gawas et al. (2014). When intermittent flow initiates, the structure velocity corresponds to the
pseudo-slug body velocity for PSL (vPSL), or slug translational velocity for SL (vT). The blue
diamond represents the slug translational velocity from Felizola (1992) for air and low-viscosity-
oil pipe flow with inclination angle from 0° to 30°. The dash line in Fig. 18 represents the

T
correlation for slug translational velocity proposed by Bendiksen (1984) given by Eqn. (5), with
co equals to 1.2. Since the θ effect on vD is small, the predicted vT from Eqn. (5) for different θs

IP
were overlaid on each other. To make the plot easier to read, only the correlation for 2° is shown
in Fig. 18.

CR
It is not new that the translational velocity follows Eqn. (1) for conventional slug flow
(Ekinci, 2015; Soedarmo et al., 2018; Felizola, 1992). Interestingly, the pseudo-slug flow
structure velocity falls in between the traditional slug translational velocity and the wave celerity.
Combining with Fig. 17, it can be seen that when the fraction of number of SL is higher than 0.2,

US
the structure velocity matches with the traditional slug translational velocity well indicating the
transition from PSL dominated to SL dominated, keeping in mind that the slug body carries more
liquid than pseudo-slug body. In another words, the structure velocity can be used as an
AN
indicator for the transition to PSL. When the average structure velocity deviates from Nicklin et
al. (1962), the flow pattern becomes pseudo-slug flow. In the current study, the transition from
PSL to SL&PSL occurs when the structure velocity collapses with the traditional slug
translational velocity, as shown in Fig. 18.
M

The inclination angle effect on the structure velocity is more evident in pseudo-slug flow
region. To better understand the inclination angle effect on structure velocity, the liquid phase
distribution within pseudo-slug body is analyzed. Figure 19 compares the liquid phase
ED

distribution in pseudo-slug body for two different inclination angles (2° and 15°). It is
speculated that the pseudo-slug body velocity is a function of the degree of mixing of gas and
liquid phase in pseudo-slug body, although the liquid holdup in pseudo-slug body does not show
significant difference (Fan, 2017). The liquid phase distribution at the pipe cross-sectional area
PT

in Fig. 19 shows that the gas phase tends to stay at the upper part of the pseudo-slug body for the
lower inclination angle, owing to the higher gravity component perpendicular to the flow
direction which facilitates separation. More mixing is observed for the higher inclination angle
CE

resulting in a more efficient blockage of the gas passage, and a higher structure velocity as a
consequence. This may also relate with the transition to SL&PSL in terms of inclination angle
effects as shown in Fig. 18.
AC

23
 ACCEPTED MANUSCRIPT

T
IP
CR
US
Fig. 18. Variation of Structure Velocity with Mixture Velocity and Inclination Angle
(Diamond: structure velocity for pseudo-slug or slug flow; square: wave celerity for segregated
flow)
AN
M
ED

(a). θ = 2°; vSL = 0.005m/s; vSg = 5.3 m/s (#1 in Fig. 16)
PT
CE

(b). θ = 15°; vSL = 0.005m/s; vSg = 5.1 m/s (#2 in Fig. 16)
AC

Fig. 19. Inclination Angle Effect on Liquid Phase Distribution in Pseudo-slug Body

3.5 Initiation of Pseudo-slug and Slug

The pseudo-slug and slug initiation phenomena were observed for varying inclination angle, gas
flow rate, and the distance from the inlet of the upward inclination angle. In general, two
initiation mechanisms, wave coalescence along the inclined pipe, and the wave growth at pipe
inlet were observed. Figures. 20 and 21 depict the two initiation mechanisms respectively.

24
 ACCEPTED MANUSCRIPT

Wave Coalescence:

T
3

IP
Fig. 20. Schematics of Pseudo-slug Initiation Mechanism – Wave coalescence (pseudo-slugs are
initiated along the inclined pipe)

CR
Wave Growth:

1
US
AN
2

3
M

Fig. 21. Schematics of Pseudo-slug Initiation Mechanism – Wave growth at pipe inlet (pseudo-
ED

slugs/slugs are initiated at the inlet of the upward inclined pipe)

In the wave coalescence mechanism, the pseudo-slug is initiated due to the coalescence of
large waves, which occurs downstream in the inclined pipe. Stratified flow was always observed
PT

in the downhill and horizontal section of the test section. When the liquid holdup of the merged
large wave is high enough, it partially blocks the passage of the gas phase, increasing the local
gas velocity dramatically. The gas phase can shear the liquid phase at the crest of the large wave,
CE

creating a large amount of liquid droplet, which can touch the top of the pipe, forming a liquid
film around the pipe wall as droplets deposit. The initiated pseudo-slug body is then transported
downstream by the gas phase with a higher velocity as compared with the large waves and roll
AC

waves in the film region.

For low superficial gas velocities or high inclination angles (θ > 5°), the liquid phase has a
higher potential to flow back, blocking the gas passage at the bottom of the inclined pipe. In this
mechanism, waves at the bottom of the upward inclined pipe grows faster due to the higher
backward liquid flow rate (from downstream section) and liquid supply (from upstream section,
which is constant). When the local liquid height at the bottom of the upward inclined pipe is
sufficiently large, the gas phase can initiate pseudo-slugs. This initiation is called wave growth
for the current study. Most of the pseudo-slugs/slugs are initiated at the entrance of the inclined

25
 ACCEPTED MANUSCRIPT

pipe. Stratified flow was always observed in the downhill and horizontal section in the current
experimental range.
Figure 22 shows the pseudo-slug initiation mechanism as a function of superficial gas
velocity and inclination angle for a liquid velocity of 0.005 m/s. The wave coalescence
mechanism was mainly observed at the lowest inclination angle (2°) with high superficial gas
velocities, while wave growth at inlet was the dominated one for the other experimental
conditions in the current study.

T
SEG

IP
vSgc

TRA

CR
INT
Wave Growth

US
AN
Wave Coalescence
M

Fig. 22. Pseudo-slug Initiation Mechanisms (vSL = 0.005 m/s)

ED

4. Simple correlation for pseudo-slug structure velocity

The current experimental data shows that the structure velocity in PSL region shows a nearly
"linear" relationship between vT and CW (Fig. 18). Based on this phenomenon, we propose two
PT

simplified approaches to estimate the pseudo-slug velocity, which can be easily implemented in
any pseudo-slug point model. Since the data for pseudo-slug flow is very limited in literature,
only the data from the current study was used for the development of the closure relationship.
CE

4.1. First Approach

The first approach assumes a linear relationship between SL and SEG flow with a constant slope
for all inclination angles, CPSL, which can be a function of pressure, pipe diameter, liquid flow
AC

rate and fluid properties. The right boundary is combined with the wave celerity at the critical
gas velocity for the onset of liquid film reversal. Based on the current experimental data, the
following correlation is proposed:
vPSL  CPSL  vM  vSgc  vSL   CWvSgc , (6)

where vSgc is the critical gas velocity for the onset of liquid film reversal, and CWvSgc is the
corresponding wave celerity at vSgc. In this paper, Fan et al. (2018) model is used to calculate the
critical gas velocity; while the wave celerity is calculated from Gawas et al. (2014). The

26
 ACCEPTED MANUSCRIPT

comparison between the measured and predicted structure velocity and the performance curve
are shown in Figs. 23 – 25 for different liquid flow rates. For these cases, CPSL was obtained by
fitting the current experimental data. It can be seen from Figs. 23-25 that, the magnitude of the
slope decreases with increasing liquid flow rate, which is mainly due to the vSL effect on the right
boundary (onset of liquid film reversal). However, the transition from SL&PSL to PSL is not
sensitive to the liquid flow rate for the current experimental conditions.

T
IP
CR
US
AN

Fig. 23. Comparison between measured and predicted pseudo-slug velocity from the first
M

approach for vSL = 0.001 m/s, (CPSL = -0.2962)

ED
PT
CE
AC

Fig. 24. Comparison between measured and predicted pseudo-slug velocity from the first
approach for vSL = 0.005 m/s, (CPSL = -0.2297)

27
 ACCEPTED MANUSCRIPT

T
IP
CR
4.2. Second Approach
US
Fig. 25. Comparison between measured and predicted pseudo-slug velocity from the first
approach for vSL = 0.01 m/s, (CPSL = -0.1820)
AN
To apply the first approach, CPSL has to be determined, which requires further experimental
investigation. This brings the second approach, which employs the flow pattern transition from
SL&PSL to PSL as the left boundary. The right boundary is still the onset of liquid
accumulation, same as in the first approach. We analyzed the existing models for the SL to
M

CH/PSL transition as reviewed by Soedarmo et al. (2018), showing that the wave effect of
Taylor bubble concept proposed by Chen and Brill (1997) gives the best prediction of inclination
angle effect on the flow pattern transition. As a result, the extension of the slug to churn
ED

transition in upward vertical two-phase flow proposed by Chen and Brill (1997) in inclined pipe
is employed as the transition boundary in the second approach. The criteria for the transition
from slug to churn flow is given by:
PT

LS LU  0.15 or H LS  0.48 , (7)

where LS is the slug body length, LU is slug unit length, HLS is the liquid holdup in slug body. As
CE

discussed in Section 3.4, the transition from conventional slug to pseudo-slug is gradual. From
modeling point view, it is more essential to identify the transition from SL&PSL to PSL instead
of SL to SL&PSL, where the structural velocity begins to show different behavior. In the current
study, 0.1 was used as the criteria instead of 0.15, while the LS and LU was calculated from
AC

TUFFP Unified Model (Zhang et al., 2003). The pseudo-slug velocity, vPSL can be expressed as:
vSgt  vSL

vPSL  CPSL vM  vTvSgt  CWvSgc vSgt  vSgc
 vTvSgt , (8)

where,
vTvSgt  CWvSgc
CPSL  , (9)
vSgt  vSgc

28
 ACCEPTED MANUSCRIPT

vTvSgt  c0  vSgt  vSL   vD , (10)

In above equations, vSgt is the superficial gas velocity corresponding to the transition from
SL&PSL to PSL from modified Chen and Brill (1997) model, and vTvSgt is the corresponding slug
translational velocity at vSgt from Eqn. (5). c0 is 1.2 in the current study. However, for a higher
viscous fluid (or small Reynolds number conditions), a value of 2.0 is recommended (Fabre,
1992).
The comparisons of experimental data and the model predictions from the second approach

T
are shown in Figs. 26 - 28. The red diamonds represent the model prediction for the flow pattern
transition from SL&PSL to PSL, showing good agreement. It can also be seen that the second

IP
approach captures well the inclination angle and liquid flow rate effects on vPSL. Another
advantage of the second approach is that it links the pseudo-slug flow with conventional slug
flow from the modeling point of view.

CR
The model performance curve for the above two methods were shown in Fig. 29, while the
statistical parameter is listed in Table 2. In general, both methods give fair prediction with the
average absolute error less than 20%.

US
AN
M
ED
PT

Fig. 26. Comparison between measured and predicted pseudo-slug velocity from the second
CE

approach for vSL = 0.001 m/s

AC

29
 ACCEPTED MANUSCRIPT

T
IP
CR
Fig. 27. Comparison between measured and predicted pseudo-slug velocity from the second

US
approach for vSL = 0.005 m/s
AN
M
ED
PT
CE

Fig. 28. Comparison between measured and predicted pseudo-slug velocity from the second
AC

approach for vSL = 0.01 m/s

30
 ACCEPTED MANUSCRIPT

T
IP
CR
(a). First Approach (b). Second Approach

Fig. 29. Performance of the new model for pseudo-slug velocity prediction in terms of different
liquid flow rates

US
AN
M
ED
PT
CE
AC

31
 ACCEPTED MANUSCRIPT

Table 2. Statistical Parameter for the Model Prediction of Pseudo-slug Velocity

Approach\Error ε1 (%) ε2 (%) ε3 (%) ε4 (m/s) ε5 (m/s) ε6 (m/s)

First Approach (vSL = 0.001 m/s) -8.44 16.11 19.08 -0.15 0.36 0.42
Second Approach (vSL = 0.001 m/s) -18.87 20.31 16.29 -0.40 0.45 0.34
First Approach (vSL = 0.005 m/s) 3.40 14.14 18.67 0.02 0.28 0.35
Second Approach (vSL = 0.005 m/s) 7.83 18.60 27.08 0.06 0.33 0.41
First Approach (vSL = 0.01 m/s) 8.96 18.17 25.28 0.05 0.31 0.37

T
Second Approach (vSL = 0.01 m/s) -2.00 15.79 20.15 -0.16 0.33 0.36
First Approach (Total) 1.31 16.14 21.01 -0.03 0.32 0.38

IP
Second Approach (Total) -4.35 18.23 21.17 -0.17 0.37 0.37

CR
4. Conclusions
A detailed experimental study was conducted in a 3-in low-pressure facility to investigate the
transition region between segregated and conventional slug flow in upward inclined pipes. Tulsa
city municipal water and compressed air were used as the test fluids. Three liquid flow rates

US
(0.001, 0.005 and 0.01 m/s) and five inclination angles (2°, 5°, 10°, 15° and 20°) were studied.
Overall, the general flow patterns observed in the uphill section for the current experimental
test matrix are segregated flow, intermittent flow and the transition between these two flow
AN
patterns. These flow patterns are further divided into several different categories based on the
liquid entrainment, liquid wetted perimeter, and flow characteristic. Pseudo-slug flow has been
observed between segregated and conventional slug flow. It has been classified into intermittent
flow in the current study.
M

This paper compares pseudo-slug flow with normal slug flow in terms of 1-D liquid holdup
evolution, 2-D axial liquid holdup distribution evolution, 2-D cross-sectional liquid holdup
distribution, and 3-D interfacial structure. The experimental data show that the cross-sectional
ED

area of the pipe cannot be fully blocked by the continuous liquid phase in the pseudo-slug body.
There is some gas passing through with a large amount of liquid entrainment. The entrained
liquid droplets deposit on the pipe wall and form a liquid film surrounding the pipe wall. Due to
the inefficient blockage of the gas passage, the pseudo-slug velocity is less than the one supposed
PT

for slug flow.

The current experimental data also show that the flow pattern transition from pseudo-slug to
slug flow is gradual, and the structure velocity can be used as an indicator of flow pattern
CE

transition. When pseudo-slug flow becomes dominant, the structure velocity deviates from the
slug translational velocity due to the insufficient blockage within the pseudo-slug body.
Finally, two simplified correlations for pseudo-slug velocity are proposed. These
AC

correlations can be used as closure relationships in any pseudo-slug mechanics model.

Acknowledgement
The authors would like to acknowledge the company members of Tulsa University Fluid Flow
Project (TUFFP) for their support in the current research, and Helmholtz-Zentrum Dresden-
Rossendorf (HZDR) for their technical support on mire-mesh sensor. The authors also
acknowledge Dr. Carlos F. Torres and Mr. Auzan Soedarmo for their fruitful discussion on this

32
 ACCEPTED MANUSCRIPT

challenging work, and Dr. Tayfun Besim Aydin for his technical support at the beginning of this
work.

References
Alsaadi, Y. 2013. Liquid Loading in Highly Deviated Gas Wells. MS thesis, University of Tulsa,
Tulsa, Oklahoma (December 2013).
Alsaadi, Y., Pereyra, E., Torres, C. and Sarica, C. 2015. Liquid Loading of Highly Deviated Gas

T
Wells from 60° to 88°. Presented at SPE Annual Technical Conference and Exhibition,
Houston, Texas, 28-30 September. SPE-174852-MS.

IP
https://doi.org/10.2118/174852-MS.
Alves, G.E. 1954. Concurrent Liquid-Gas Flow in a Pipeline Contactor. Chem. Engng Prog. 50,

CR
449-456.
Andritsos, N., Williams, L., and Hanratty, T. J. 1989. Effect of Liquid Viscosity on the
Stratified-Slug Transition in Horizontal Pipe Flow. Int. J. Multiphase Flow, 15(6): 877-892.
https://doi.org/10.1016/0301-9322(89)90017-7

US
Barmak, I., Gelfgat, A., Vitoshkin, A., Ullmann, A., and Brauner, N. 2016. Stability of Stratified
Two-phase Flows in Horizontal Channels. Physics of Fluids. 28 (4), 044101-1.
https://doi.org/10.1063/1.4944588
AN
Barnea, D. 1987. A Unified Model for Predicting Flow-Pattern Transitions for the Whole Range
of Pipe Inclinations. Int. J. Multiphase Flow 13 (1): 1-12.
https://doi.org/10.1016/0301-9322(87)90002-4.
Barnea, D., Shoham, O. and Taitel, Y. 1980. Flow Pattern Characterization in Two-Phase Flow
M

by Electrical Conductance Probe. Int. J. Multiphase Flow 6 (5): 387-397.

https://doi.org/10.1016/0301-9322(80)90001-4.
Barnea, D., and Taitel, Y. 1992. Structual and Interfacial Stability of Multiple Solutions for
ED

Stratified Flow. Int. J. Multiphase Flow, 18 (6): 821-830.

https://doi.org/10.1016/0301-9322(92)90061-K
Bendiksen, K. H. 1984. An Experimental Investigation of the Motion of Long Bubbles in
Inclined Tubes. Int. J. Multiphase Flow 10 (4): 467-483.
PT

https://doi.org/10.1016/0301-9322(84)90057-0.
Benjamin, T.B. 1968. Gravity Currents and Related Phenomena. J. Fluid Mech. 31 (2): 209-248.
https://doi.org/10.1017/S0022112068000133
CE

Biberg, D., Staff, G., and Hoyer, N. 2015. Accounting for flow model uncertainties in gas-
condensate field design using the OLGA High Definition Stratified Flow Model. 17th
International Conference on Multiphase Technology, Cannes, France, 10-12 June. 53.
AC

Brito, R. 2015. Effect of Horizontal well trajectory on Two-phase Gas-Liquid Flow Behavior.
Ph.D. Dissertation, University of Tulsa, Tulsa, Oklahoma (December 2015).
Brito, R. Pereyra, E., and Sarica, C. 2016b. Effect of Well Trajectory on Liquid Removal in
Horizontal Gas Wells. SPE Annual Technical Conference and Exhibition. Dubai, UAE, 26-
28 September. SPE-181423-MS.
https://doi.org/10.1016/j.petrol.2017.03.042.
Butterworth, D. and Pulling, D. J. 1972. A Visual Study of Mechanisms in Horizontal, Annular,
Air-Water Flow. WERE Report M2556, Harwell, Oxon.

33
 ACCEPTED MANUSCRIPT

Campbell, B. K., and Liu, Y. 2016. A nonlinear flow-transition criterion for the onset of slugging
in horizontal channels and pipes. Physics of Fluids. 28 (8).
https://doi.org/10.1063/1.4960084
Chen, X.T., Brill, J.P., 1997. Slug to churn transition in upward vertical two-phase flow. Chem.
Eng. Sci. 52 (23): 4269–4272.
https://doi.org/10.1016/S0009-2509(97)00178-4.
Davies, R. M. and Taylor, G.I. 1950. The Mechanics of Large Bubbles Rising through Extended
Liquids and through Liquids in Tubes. Proc., The Royal Soc. London, 200 (1062): 375-390.
https://doi.org/10.1098/rspa.1950.0023

T
Dukler, A. E. and Hubbard, M. G. 1975. A Model for Gas-Liquid Slug Flow in Horizontal and

IP
Near Horizontal Tubes. Ind. Eng. Chem. Fundam. 14 (4), 337-347.
https://doi.org/10.1021/i160056a011
Dumitrescu, D.T. 1943. Strömung An Einer Luftblase Im Senkrechten Rohr. Z. Angew. Math.

CR
Mech. 23, 139.
https://doi.org/10.1002/zamm.19430230303
Ekinci, S. 2015. Pipe Inclination Effects on Slug Flow Characteristics of High Viscosity Oil-Gas

US
Two-Phase Flow. MS thesis, University of Tulsa, Tulsa, Oklahoma.
Fabre, J. and Line, A. 1992. Modeling of Two-phase Slug Flow. Annu. Rev. Fluid Mech. 24: 21-
46.
Fan, Y. 2017. A Study of the Onset of Liquid Accumulation and Pseudo-slug Flow
AN
Characterization. Ph.D. Dissertation. U. of Tulsa, Oklahoma (May 2017).
Fan, Y., Pereyra, E., and Sarica, C. 2018. Onset of Liquid Film Reversal in Upward Inclined
Pipe. SPE J. In Press. SPE-191129-PA.
https://doi.org/10.2118/191120-PA.
M

Fan, Y., Pereyra, E., Torres-Monzon, C. F., Aydin, T. B. and Sarica, C. 2015. Experimental
Study on the Onset of Intermittent Flow and Pseudo-slug Characteristics in Upward Inclined
Pipes. 17th International Conference on Multiphase Technology, June 10 - 12, 2015, Cannes,
ED

France.
Felizola, H. 1992. Slug Flow in Extended Rearch Directional Wells. MS thesis, University of
Tulsa, Tulsa, Oklahoma.
PT

Gawas, K., Karami, H., Pereyra, E., Al-Sarkhi, A., and Sarica, C. 2014. Wave Characteristics in
Gas-Oil Two Phase Flow and Large Pipe Diameter. Int. J. Multiphase Flow 63: 93-104.
Guner, M. 2012. Liquid Loading of Gas Wells with Deviations from 0°to 45°. M.S. thesis,
University of Tulsa, Tulsa, Oklahoma.
CE

Guner, M., Pereyra, E., Sarica, C. and Torres, C. 2015. An Experimental Study of Low Liquid
Loading in Inclined Pipes from 90° to 45°. SPE Production and Operations Symposium,
Oklahoma City, Oklahoma, 1-5 March.
AC

https://doi.org/10.2118/173631-MS.
Holm, H. 2015. Tanzania Gas Development – Flow Assurance Challenges. 17th International
Conference on Multiphase Technology, June 10 - 12, 2015, Cannes, France.
Kipping, R., Brito, R., Scheicher, E. and Hampel, U. 2016. Developments for the application of
the Wire-Mesh Sensor in industries. Int. J. Multiphase Flow 85: 86-95.
https://doi.org/10.1016/j.ijmultiphaseflow.2016.05.017.
Kjølaas, J., Unander, T.E., Wolden, M., Johansson, P.S., and Holm, H. 2015. Experiments for
low liquid loading with liquid holdup discontinuities in two- and three-phase Flow. 17th
International Conference on Multiphase Technology, June 10 - 12, 2015, Cannes, France.

34
 ACCEPTED MANUSCRIPT

Lee, A.H. 1993. A Study of Flow Regime Transition for Oil-Water-Gas Mixtures in Large-
Diameter Horizontal Pipelines. M.S. Thesis, Ohio University, Athens, Ohio.
Lin, P. Y., and Hanratty, T. J. 1986. Prediction of the Initiation of Slugs with Linear Stability
Theory. Int. J. Multiphase Flow, 12 (1): 79-98.
https://doi.org/10.1016/0301-9322(86)90005-4
Lin. P. Y. and Hanratty T. J. 1987a. Detection of Slug Flow from Pressure Measurements. Int. J.
Multiphase Flow 13 (1): 13-21.
https://doi.org/10.1016/0301-9322(87)90003-6.
Lin. P. Y. and Hanratty T. J. 1987b. Effect of Pipe Diameter on Flow Patterns for Air-Water

T
Flow in Horizontal Pipes. Int. J. Multiphase Flow 13 (4): 549-563.

IP
https://doi.org/10.1016/0301-9322(87)90021-8.
Loh, W.L., Perez, V., Tam, N.D., Wan, T.T., Zhao, Y. and Premanadhan, V.K. 2016.
Experimental Study of the Effect of Pressure and Gas Density on the Transition from

CR
Stratified to Slug Flow in a Horizontal Pipe. Int. J. Multiphase Flow 85: 196-208.
https://doi.org/10.1016/j.ijmultiphaseflow.2016.06.005.
Maley, J. 1997. Slug Flow Characteristics and Corrosion Rates in Inclined High Pressure

US
Multiphase Flow Pipes. MS thesis, Ohio University, Athens, Ohio.
Nicklin, D.J., Wilkes, J.O., Davidson, J.F., 1962. Two-phase flow in vertical tubes. Trans. Inst.
Chem. Eng. 40: 61–67.
Nicholson, M. K., Aziz, K. and Gregory, G. A. 1977. Intermittent Two-Phase Flow in Horizontal
AN
Pipes: Predictive Models. 27th Canadian Chemical Engineering Conference, Calgary,
Alberta.
Picchi, D., Barmak, I., Ullmann, A., and Brauner, N. 2018. Stability of Stratified Two-Phase
Channel Flows of Newtonian/Non-newtonian Shear-Thinning Fluids. 99: 111-131.
M

https://doi.org/10.1016/j.ijmultiphaseflow.2017.10.001
Prasser, H. -M., Böttger, A., Zschau, J., 1998. A new electrode-mesh tomography for gas–liquid
flows. Flow Meas. Instrum. 9 (2): 111–119.
ED

Shoham, O. 2006. Mechanistic Modeling of gas-Liquid Two-Phase Flow in Pipes. SPE book.
Soedarmo, A., Fan, Y., Pereyra, E., and Sarica, C. 2018. A Unit Cell Model for Gas-Liquid
Pseudo-slug Flow in Pipes. Journal of Natural Gas Science and Engineering. 60: 125-143.
PT

https://doi.org/10.1016/j.jngse.2018.10.006
Soedarmo, A., Soto-Cortes, G., Pereyra, E., Karami, H. and Sarica, C. 2018. Analogous behavior
of pseudo-slug and churn flows in high viscosity liquid system and upward inclined pipes. Int.
J. Multiphase Flow 103: 61-77.
CE

https://doi.org/10.1016/j.ijmultiphaseflow.2018.02.001.
Soleimani, A. and Hanratty, T.J. 2003. Critical Liquid Flows for the Transition from the Pseudo-
slug and Stratified Patterns to Slug Flow. Int. J. Multiphase Flow 29 (1): 51-67.
AC

https://doi.org/10.1016/S0301-9322(02)00124-6.
Staff, G., and Holm, H. 2016. Liquid Accumulation and Removal in Transient Three-phase Flow
with Low Liquid Loading. 10th North American Conference on Multiphase Technology, June
8 – 10, 2016, Banff, Canada.
Sylvester, N.D. 1987. A Mechanistic Model for Two-Phase Vertical Slug Flow in Pipes. J.
Energy Resour. Technol. 109 (4), 206-213.
Taitel, Y. and Dukler, A.E. 1976. A Model for Predicting Flow Regime Transition in Horizontal
and Near Horizontal Gas-Liquid Flow. AIChE J. 22 (1): 47-55.
Tecplot 360TM User’s Manual Version 2013.

35
 ACCEPTED MANUSCRIPT

Wilkens, R. J. 1997. Prediction of the Flow Regime Transitions in High Pressure, Large
Diameter, Inclined Multiphase Pipelines. Ph.D. Dissertation, Ohio University, Athens, Ohio.
Ullmann, A., Zamir, M., Gat, S., and Brauner, N. 2003. Multi-holdups in Co-current Stratified
Flow in Inclined Tubes. Int. J. Multiphase Flow, 29 (10): 1565-1581.
https://doi.org/10.1016/S0301-9322(03)00143-5
Zabaras, G., Dukler, A. E., and Moalem-Maron, D. 1986. Vertical Upward Cocurrent Gas-Liquid
Annular Flow. AIChE J. 32 (5): 829-843.
Zhang, H.-Q., Wang, Q., Sarica, C., Brill, J.P., 2003b. Unified model for gas-liquid pipe flow via
slug dynamics - part 1: model development. J. Energy Res Technol. 125 (4), 266–273.

T
IP
CR
US
AN
M
ED
PT
CE
AC

36
 ACCEPTED MANUSCRIPT

Appendix A: Flow Pattern Map

T
IP
CR
US
Fig. A-1. Flow Pattern Map for vSL = 0.001 m/s
AN
M
ED
PT
CE

Fig. A-2. Flow Pattern Map for vSL = 0.01 m/s

AC

37