See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/322385944

Using CFD Star-CCM+ to Investigate multiphase flow across a 90 degree elbow

bend flowline: monitoring pressure drop, velocity profile and flow regime

Thesis · August 2016

CITATIONS READS

0 500

1 author:

Abdulrauf Bello
American University of Nigeria
1 PUBLICATION   0 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Abdulrauf Bello on 20 January 2020.

The user has requested enhancement of the downloaded file.

Investigating multiphase flow across a 900 elbow bend flowline: monitoring pressure
drop, velocity profile and flow regime

By

Abdulra’uf Hamid Bello

6740746

A project submitted in partial fulfillment of the requirements for the degree of

Master of Science (Oil and Gas Engineering)

COVENTRY UNIVERSITY

FACULTY OF MECHANICAL, AUTOMOTIVE AND MANUFACTURING ENGINEERING

Project Supervisor: Dr Erdal Turkbeyler

August 2016
ABSTRACT

Multiphase slug flow is going to be analysed while focusing on how pressure is lost and
monitoring the velocity profiles. The problem involved in a multiphase flow is by far
more complex than that which could be observed in a single phase flow. The fluid
coming from the wellbore and onto the first part of the separator is conventionally a
two phase flow of liquid/gas. Using large diameter pipe and industrial based fluids,
focus would be laid on flow directive change and also putting into consideration what
type of flow to be dealt with. Why this results to a more complex problem is because
of the simultaneous flow of both liquid and vapor (water and condensate or oil). The
basic method of estimating pressure drop for multiphase flow is by using an equation.
Furthermore, more complications occur when the fluid is flowing to the surface and
due to pressure change, vapor to liquid fraction also change, velocity changes and also
fluid properties changes. With recent advancements that have been done on CFD codes,
a whole new set of equations for multiphase and a complete fluid dynamic set can be
solved numerically. A worthy CFD code to be used in conducting this analysis is the
Star-CCM+. Unstructured mesh (three dimensional) could be used for the elbow
sections and the pipe using known methods to calculate the energy, mass and
momentum equations. At the pipe’s inlet boundary upstream of the 90 degrees bend is
where the phase velocities and mixture composition is defined. For solving complex
mixture problems, one model is often used due to its accuracy and that is к-ε turbulence
model having standard wall functions. Assumption for gravitational acceleration on
upward flow is taken as 9.81 m/s2.
Keywords: Multiphase, flow, elbow, pressure drop, slug

i
DEDICATION

“And we have enjoined upon man care towards his parents. His mother carried him,
thereby increasing her weakness on weakness and then weaning of him takes two years.
Be grateful to me and your parents; to me is the final abode” (Qur’an 31:14)
This project is dedicated to my brilliant role models, My Parents.

ii
ACKNOWLEDGEMENT

I thank the almighty Allah for providing me with good health all through the course of
this Project, then my parents Alhaji and Hajia Aisha Bello Hamid Jada for literally
everything in making it possible, I don’t suppose I could ever comprehend the abundant
love shown by those two.

I am most grateful to my entire friends and family who have one way or the other helped
this research, either motivationally or financially. People worthy of mention for their
support towards seeing this through include: my siblings: Asma’u, Abdulrasheed, Zara,
Zainab, Amina, Khadija and Aisha Bello; my wonderful grandparents: Alhaji Abubakar
Ahmed Song, Hajia Fa’izah Abubakar & Asma’u Suleiman; the cute nephews: Ahmed
and Muhammed Abubakar Aliyu; and then my step-mom. An idol that cannot be left
out is D.C. Musa M.B.A

I thank Coventry University for providing me with the necessary facilities and materials needed
to successfully carry out this study. I would also thank my supervisor Dr. Erdal Turkbeyler for
guiding and aiding me with information regarding this Project.

iii
TABLE OF CONTENTS
ABSTRACT ............................................................................................................................... i
DEDICATION .......................................................................................................................... ii
ACKNOWLEDGEMENT........................................................................................................ iii
TABLE OF CONTENTS ......................................................................................................... iv
CHAPTER 1: INTRODUCTION ............................................................................................ 1
1.1 Overview.......................................................................................................................... 1
1.2 Background of the research ............................................................................................. 2
1.2.1 Slug flow in upward vertical pipe ............................................................................. 3
1.2.2 Two phase liquid-gas flow for bends at 900 angle .................................................... 5
1.3 Problem Statement ........................................................................................................... 7
1.4 Aims and Objectives ........................................................................................................ 7
1.5 Methodology .................................................................................................................... 7
1.5.1 Computational method .............................................................................................. 8
CHAPTER 2: LITERATURE REVIEW ................................................................................. 9
2.1 Flows in bend ................................................................................................................... 9
2.1.2 Multiphase flows in 900 bends .................................................................................. 9
CHAPTER 3 CFD SOLUTION METHODOLOGY AND MODELLING .......................... 16
3.1 CFD (Computational Fluid Dynamic) Overview .......................................................... 17
3.1.1 Computational Fluid dynamic (CFD) ..................................................................... 17
3.1.2 Fluid Flow Governing Equations ............................................................................ 18
3.2 CFD Multiphase Modelling ........................................................................................... 19
3.2.1 Pre-Processing ......................................................................................................... 19
3.2.2 Solution/Post-processing ......................................................................................... 20
3.3 Geometry Problem Description ..................................................................................... 20
3.3.1 Domain of Computation .......................................................................................... 21
3.3.2 Boundary Conditions .............................................................................................. 22
3.3.3 Initial Conditions ..................................................................................................... 24
3.4 CFD Modelling Assumptions: ....................................................................................... 24
3.4.1 Governing Equations ............................................................................................... 24
3.4.2 Turbulence Model ................................................................................................... 27
3.4.1 Independent Mesh Study ......................................................................................... 29
3.5 Physics Models selection ............................................................................................... 31
CHAPTER 4 ANALYSIS AND RESULTS OF CFD ........................................................... 34
4.1 Solution and Discretization Algorithm .......................................................................... 34
4.2 Pressure Distribution...................................................................................................... 34
4.3 CFD Results ................................................................................................................... 35
4.2.1 CFD results for pressure profiles ............................................................................ 35
4.2.2 CFD results for velocity profiles ............................................................................. 38

iv
 4.2.3 CFD results for flow regime.................................................................................... 39
4.2.4 Convergence............................................................................................................ 40
4.3 Empirical Models to Calculate pressure drop ................................................................ 41
4.3.1 Chisholm Model ...................................................................................................... 41
4.3.2Azzi-Friedel Model .................................................................................................. 42
CHAPTER 5: CONCLUSION AND FUTURE WORK ....................................................... 45
5.1 Conclusion ..................................................................................................................... 45
5.2 Recommendations for Further work .............................................................................. 45
REFERENCES ........................................................................................................................ 46

Table Of Figures
Figure 1 Oil and Gas Offshore installations .............................................................................. 2
Figure 2 Flow regimes in vertical pipe configuration ............................................................... 5
Figure 3 Flow regimes for horizontal pipe configuration .......................................................... 5
Figure 4 3D geometry of the elbow bend pipe flow system.................................................... 21
Figure 5 3D Depiction of the mesh ......................................................................................... 30
Figure 6 Plot for the Multiphase static pressure ...................................................................... 35
Figure 7 Total pressure plot for the multiphase flow .............................................................. 36
Figure 8 Pressure loss in the two-phase flow .......................................................................... 36
Figure 9 Total pressure plot for single phase flow .................................................................. 37
Figure 10 Static pressure for the single phase (Silicone Oil) .................................................. 37
Figure 11 Single phase pressure loss ....................................................................................... 37
Figure 12 Velocity profile for the Oil/Gas mixture ................................................................. 38
Figure 13 Velocity depiction for the single phase oil .............................................................. 38
Figure 14 Void Fraction Flow movement along the pipe ........................................................ 39
Figure 15 Residuals obtained for single phase flow ................................................................ 40
Figure 16 Residuals obtained for multiphase flow .................................................................. 40
Figure 17 Tangents of the pipe bend labelled as ‘C’ ............................................................... 44

v
CHAPTER 1: INTRODUCTION

1.1 Overview

As advancements are made in the oil and gas sector, it has gotten to the point where it
is essential to know the form and also measure the feature of the multiphase oil and gas
flowing within processing and oil production pipelines. This knowledge or information
of the existing flow regimes helps the engineer in optimising the pipeline configuration
and also processes downstream to obtain the most reliable and cheapest design. Flow
assurance is the term used in the oil industry to describe this circumstance. Analysing
multiphase flows around angular bends under varying conditions is a topic which has
been paid very minimal attention to in the oil and sector. To predict or analyse the flow
regime of multiphase flow on a more precise detail, experimentation and
instrumentations which are capable of characterising and measuring the flows in a pipe.
Therefore, this study focuses on the of a CFD package software Star-CCM+ to analyse
liquid-gas flows at a 900 pipe bend with focus on applications in oil and gas Industries.

1
Figure 1 Oil and Gas Offshore installations

1.2 Background of the research

For better accuracy, a real life scenario needs to be setup. Parameters from proficient
experimental work done by Abdulkadir (2011) using industrial based fluids were
adopted.
For subsequent industrial scale processes, the fluids serve as a form of transportation
means for materials. An in depth knowledge regarding the principles governing this
scenario of transporting fluids helps to create a secure and efficient system. Worth
mention, most of the present industries like Chemical, petroleum, oil and gas industries;
we see multiphase kind of flow. Multiphase flow can be defined as different phase fluid
flowing together simultaneously, the simplest of the case is the two-phase flow. The
equations governing multiphase flows are very complex as compared to those for single
phase because various patterns of flow exist in a liquid-gas system. Wallis (1969)
describes gives detailed information on the behaviour phenomenon of flow in two-
phase. For horizontal flows, the usual flow pattern seen are stratified wavy, stratified,

2
bubble, annular, and slug. Whereas for observed vertical flows are plug, annular, plug
(churn), wispy-annular and bubble flow could all be present. Few researchers have
made effort to investigate the friction factor and the pressure drop in both horizontal
and vertical flows for multiphase two-phase flow. The mere existence of two-phase
kind of flow in pipe systems bring about higher unwanted pressure drops. Installation
of elbow fittings in most of the industrial applications is to aid system flexibility and
also used in directing flow. It is significant that pressure drop in theses elbows are
observed since we know that they could also be used in installing some instruments that
helps in monitoring the industrial process main parameters. As these fluids pass through
the elbow, the bends curvature produces a centrifugal force; this force directs towards
the pipe’s outer wall from the curvature’s center which is momentary. The boundary
layer combined with the centrifugal force at the pipes wall creates an ideally organised
secondary flow in two similar eddies. The secondary flow produced is overlaid on the
flow mainstream through the axis of the tube, which leads to a streamline which is
helical in shape, passing through the elbow bend.
A main problem faced by unwanted high pressure drop is how difficult it could be to
find a model of flow in multiphase pipe systems. Although so many proposals have
been brought forward to model accurately flowing multiphase through pipe systems but
all unsuccessful, Chisholm (1983) developed a fundamental model to predict
multiphase flows across bends respect to a two-phase liquid multiplier for various
diameter of pipes, flow rates and also r/D values. Most studies done for pressure drop
has been on horizontal two-phase planes. Martin and Chenoweth (1955) in their work,
showed that although pressure drop in multiphase is higher than that of single phase
flow, it can be associated by adopting the Lockhart-Martinelli model (1949), the model
was developed initially for straight pipes. The association assesses to predict the drops
in pipe fitting e.g. elbow bends. The homogeneous model agreed also at higher mass
velocities.

1.2.1 Slug flow in upward vertical pipe

Flows coming up in subsea pipelines mostly contains different phases which include
water, oil or/and gas, in which the prior composition is never known. This various
combination of fluid flow in an upward subsea pipeline network could lead to severe
operational problems, starting from either a shut-down or hindered flow production to

3
damaging equipment. Gas which has or consists of a high water quantity or a mixture
of water-oil may simultaneously be produced, which gives rise to some multiphase type
of flow conditions in the system of the pipe transporting from the source to the
production platform. The longer a field grows, the multiphase which is produced would
contain more water quantity resulting to multiple compositions of mixture; which
would in turn affect flow characteristics and flow pattern. For pipes which are vertically
inclined upwards, the dominating flow type or pattern is said to be slug flow,
Hernandez-Perez (2008); which would lead to an enhanced corrosion rate studied by
Kaul (1996). He studied that when its slug flow pattern, the rate of corrosion tends to
always be accelerated. This type of flow pattern has characteristics of alternation
between liquid slugs flowing with gas pockets known as the Taylor bubbles. They are
gas pockets flowing in the form of bullet-shaped which are often large and separated
by liquid slugs in between them with sometimes having little gas entrained bubbles.
One major slug flow characteristic is that they have intrinsic unsteadiness. It is of
concern for Oil industries or any other large process that involves pipeline
transportation system because this type of flow happens between gas and liquid of also
vast range. For a system, liquid slug flow present in that system would lead to a liquid
and gas flow output which is uneven at the system’s outlet or if not then would be at
the succeeding processing stage. For an engineer designing a two-phase or multiphase
flow system, it could be a serious problem and for the operator as well. In terms of
pressure drop, that which is observed in slug flow is considerably higher as compared
to the other types of flow regimes, one needs to know the greatest possible length of
the liquid slug flow in a system. The device mostly used to catch liquid slug flow while
avoiding damaging equipment downstream is the slug catcher.
One question regarding slug flow which has received a lot of attention by researchers
is the thought of how, when are the slug flows formed. Though, there is a limited
information of the studies done on the slug flow behaviour in industries. It is because
of that it is advised for detail focus be done on slug flow while designing a safe, efficient
two-phase liquid-gas flow system operation and design.

4
Figure 2 Flow regimes in vertical pipe configuration

Figure 3 Flow regimes for horizontal pipe configuration

1.2.2 Two phase liquid-gas flow for bends at 900 angle

Engineers dealing with transportation and production on offshore and onshore oil and
gas sectors face problems arising as regards multiphase flow could be environmental

5
and technical challenges. An instance is economical, it is cost effective to transport the
multiphase fluids through one pipe line and then make separation at onshore. By doing
this, a significant amount is saved as compared to transporting trough phase separators
and separate flow lines on the Floating Production and Storage Production (FPSO)
offshore platform. Up until now, the multiphase flow in petroleum industry application
has received some considerable amount of focus by research workers due to its
application. But due to the problems this kind of flow cause like instability and damage
of pipeline system, it makes it difficult for it to be accepted. The geometry of a pipeline
is not always straight, most times it has junctions, valves, bends fittings and others
which in turn increases the complexity of the liquid and gas mixture flow. Examples of
such complexities these fittings bring include unsteady phenomena of flow, pressure
pulsations, flow separation, flow excursions, void fractions which highly fluctuates,
secondary flow and other instable types of flow. The above mentioned
phenomenas may lead to tube failure, corrosion, and burn-out which thereby results to
replacements early, expensive repairs, outages that affects the safety and reliability of
the plant. The most encountered of the above fittings in oil and gas related application
are the bends; they play a major part at combinations between riser and at slow lines,
terrain undulation, and also at production facilities delivery points. At downstream, the
mere presence of a bend could completely alter flow characteristics immediately.
For this reason, it has become of great interest to systems making use of pipeline
transportation. Furthermore, when dealing with angular flows, in depth description and
understanding must be attained on how to optimize the conditions for operation, what
is needed economically for the design, and putting into consideration details on safety
factors. Accurate and precise information is needed especially when reliability
circumstance becomes very vital and also operating/capital cost come to be a challenge.
Although in preceding peer literature reviews, the issue of two-phase flow has gotten a
very minimal consideration, this is because researchers have taken into consideration
more of single phase flow (Spedding (2004); Eustice (1910); Jayanti (1990); Dean
(1927 & 1928); and Dewhurst (1990)). Some few multiphase papers: Carver (1984),
Abdulkadir (2011), Salcudean and Carver (1986), Issa and Ellul (1987), Charles and
Oshinowo (1974), Neller and Gardner (1969), Benbella (2009), Benard and Spedding
(2007), Azzi (2002 & 2005), Van der Akker and Legius (1997), discuss this type of
flow but most focused on pipes that have smaller diameters than what is actually used

6
in or at industries. Also the fluid’s physical properties vary from that used at industries.
This is significant because this paper deals with comparison to industry experimental.

1.3 Problem Statement

CFD simulations using Star-CCM+ to investigate liquid-gas mixture behaviour,

confluent flow at 900 will be carried out to analyse obtained new data and to further
understanding on the flow regime the angular bend promotes. Void fraction
measurements, Pressure drop, velocity profile (liquid holdup, and liquid film thickness
could also be used) for a flow and a pipe configuration would be used in validating
CFD (Computational fluid dynamics) models that eventually would be used for studies
regarding related problems but on industrial scale.

1.4 Aims and Objectives

A good acquaintance with data on flow characteristics like pressure drop, void fraction
and flow regime, is needed while attempting to design multiphase flow lines, separation
facilities and fluid treatment. Hitherto, the prediction or determination of the
characteristics of oil and gas flow has been based duly on mathematical models which
are habitually incomplete or data obtained from experimental done with a very small
pipe diameter by using the typical operating fluid (water and air). Consequently, there
is a need to simulate two-phase flows using larger pipe diameters and a liquid having
more viscosity than water. Thus, the main objective/aim of this particular study is to:
 Develop a very reliable CFD model methodology to analyse pressure drop and
slug flow experienced in a vertical angular 900 bend and then to compare the
CFD models with data obtained by experiments.

1.5 Methodology

For two-phase flow studies, two peculiar approaches are used for studies:
computational/numerical and experimental method. Most of all researchers done before
the 1990’s were experimental based approach. Though, with the creation of the CFD
(Computational fluid dynamics) and computer advances which backs the software, it
has been used even more by research workers in their studies and validated their

7
developed CFD models using the previously done researches by experimental methods.
The use of CFD methods have also been driven and motivated by the high cost of
running an experimental analysis and also by the limited experimental techniques
needed for investigating complex flow. Therefore, in this research, studies is made
systematically using a computational method to accomplish the above mentioned
objective.

1.5.1 Computational method

A commercial fluid dynamic Star-CCM+ software was used to develop the

computational models used. To investigate the different multiphase effect of a flow
regime carried out in an internal diameter pipe of 67mm, formulation of computational
fluid models were done. Using VOF (Volume of fluid) method found in the software,
the slug flow pattern observed flowing along the 900 vertical bend was modelled.
Chapter 3 and 4 discuss respectively how the computational model is constructed and
its presentation; then the analysis made, result and discussion of the study.

8
CHAPTER 2: LITERATURE REVIEW

Multiphase flow between Oil and gas is mostly experienced at process industries,
nuclear generation, power generation, oil industries, and chemical reactors. Since the
interface between the liquid and gas two-phase flow is deformable, one of the phase is
compressible and the other is non-compressible, it makes analysis complicated. The
densities and the flow rates affects the various different ways the fluid is able to flow.
Knowledge of the system’s hydrodynamics is needed in understanding multiphase mass
& heat characteristics and it’s flow behaviour.
In this chapter, broader emphasis would be made on liquid-gas two-phase flow
fundamentals. Precisely dealing with vertical related flows as this particular study
closely focuses on that. Finally, the chapter will give a good literature review on slug
liquid-gas flow in pipe bends.

2.1 Flows in bend

Fluids passing through any pipe networking is likely to come across pipe fittings like
bends. The mere existence of the bend can completely alter the flow regime conditions
occurring in both the upstream and the downstream bend sections of the pipe. Most of
the researches done on bend pipe bend angles are single-phase, just fewer of the authors
have worked on multiphase flow, below talks about some of the research studies
obtained:

2.1.2 Multiphase flows in 900 bends

The flow pattern of multiphase flows in straight pipes are the same qualitatively as that
seen in bends. The difference being that the bend introduces brings about an alteration
in the flow pattern, it causes a redistribution of the two-phase local velocities and their
positions.

Neller and Gardner (1969) conducted experimental and visual studies slug/bubble flow
of air and water by the use of 76mm internal diameter transparent pipe having a 900
bend orientation with radii of curvature of 305 and 610mm at atmospheric pressure.
Measurements were done for air concentrations across selected cross-sections. They

9
used their experimental analysis result to understand the effects of gravitational
centrifugal force effects on the distribution of flow around the 900 bend. They also
suggested a Froude number modified could represent the gravitational force and
centrifugal force ratio. These researchers also found that depending on the value of the
Froude number, gas can either flow through the bend’s inner part or outer part which is
defined below as

𝑈𝑀 2
𝐹𝑟𝜃 = (Equation 1)
𝑅𝑔 𝑆𝑖𝑛 𝜃

Whereby 𝜃 is the bend’s angle. They attempted to use the Froude number to explain
the situation of the flow at different positions; deducing that if the value of 𝐹𝑟𝜃 number
is more than one, then the gas (air) is said to be flowing through the inner part of the
pipe bend, but if the value is less than one, then the gas (air) would be flowing at the
outer wall of the pipe bend. And when 𝐹𝑟𝜃 is equal to unity, then the fluid is said to
have both it’s phases in a radial equilibrium. But for liquids that however have varied
or different surface tension, density, and viscosity; their conclusion does not apply.

Carver (1984) conducted a 2D numerical modelling to interpret the flow at a bend angle
of 900. He went ahead to make comparisons between the results he obtained from the
numerical simulation he did with that of experimental data carried out by Neller and
Gardner (1969). His concluding result did not really agree with their experiment. Then
he went further later Salcudean and Carver (1986) to discover the limitations involved
with modelling in 2D numerical simulating approach to model a phenomena of flow
which is actually a flow in 3D situation. Therefore, they decided to transform the 2D
model into a 3D model. On doing this, they found out that the result they generated
from that simulation was much closer to that of the experimental results obtained by
Neller and Gardner (1969).

Furthermore, Issa and Ellul (1987) came up with a better 3D simulation model whereby
a completely different algorithm solution was used. In the momentum equations they
used, they added an abridged momentum gas term, and applied that model for oil-gas
mixtures and also for water-air mixtures. The data generated from this gave an even
similar correspondence with that of the experimental analysis initially done by Neller

10
and Gardner (1969) compared with the results Carver (1984) obtained. Although in the
work done by Issa and Ellul (1987), it was observed that there was no analysis report
on grid sensitivity. An ideal mesh could not be chosen for the result obtained by the
simulation. They could validate the oil-gas mixture they simulation they carried out as
at that time there was no experimental data they could use.

Van der Akker and Legius (1997) conducted both a numerical study and experimental
study on water-air mixture at atmospheric pressure using a ninety degree bend angle
with a radius of curvature of 630mm. The materials used at their experimental facility
were transparent horizontal acrylic 9m long pipe horizontally joint with a vertically 4m
high pipe riser having a bend radius of 630mm. 10mm was used as the internal diameter
of the overall pipe. Observing visually the flow regime, a digital camera was used (200
Hz) combined with an auto-regressive kind of modelling method. In the riser, churn
and slug flow was seen while a slug, stratified and geometry enhanced slug (a new
regime) was seen in flowline. Solution package code which is an in-house for
hyperbolic function (SOPHY-2) was used for multiphase flow time dependent
behaviour. There was a decent correlation for both analysis in almost all varied
conditions except when the gas has higher flow rates and the liquids lower. Although,
the information they gave on slug flow characteristics was not very vibrant. They
omitted discussing for instance Taylor bubble and void fraction parameters which are
important. Also, they did not point out the dependency of liquid slugs and Taylor bubble
on the flow rate of gas. The researchers used a frequency sample assumed to be slow
between 50-100Hz which might not give upright signal spatial resolution result.

Azzi (2002) alongside his co-researchers made the observation that pressure drop
differs in single phases as compared to in multiphase phases at a bend. Conventionally,
it is assumed that for flow in single phase, the loss in pressure is caused by generation
of secondary flow, detachment of vortex, and downstream pipe losses related to
recovering the conforming velocity profile at the upstream which is first established
completely. While for a multiphase flow, extra effects were pointed out by them which
included dissipation (caused by separating and mixing again the liquid and the gas
phases) and also that the force of gravity has an influence on the behaviour of the flow
both in the vertical flow and the horizontal flow. So far all the empirical methods of
two-phase flows proposed for a bent pipe, Azzi (2000) and the rest in a broad review

11
came to a conclusion that Chisholm’s (1980) B-method gives the most near prediction
based data previously published.

In another work done was by Azzi (2006) and the rest went on to further lay emphasis
and observation on results obtained from past studies. Based on Azzi (2002) and co
conclusion that Chisholm B-method which is the pressure drop at a bend for a
multiphase flow can also be associated with the pressure loss in a single phase flow a
multiplier means as shown:

𝛥𝛲𝑓
𝜙𝑙𝑜 2 = 𝛥𝛲 (Equation 2)
𝑙𝑜

𝜌
𝜙𝑙𝑜 2 = 1 + [𝜌 𝑙 − 1] [𝐵𝑋𝑔 (1 − 𝑋𝑔 ) + 𝑥𝑔 2 ] (Equation 3)
𝑔

With

2.2
B=1+ 𝑅 (Equation 4)
𝐾𝑙𝑜 (2+ )
𝐷

And

𝜆𝑙𝑜 𝑙𝑒
𝐾𝑙𝑜 = (Equation 5)
𝐷

For angle bend of 900, Klo represents loss coefficient for the complete liquid flowing

Whereas the below represents angle bends more than 900

𝐾90
B = 1 + (B90 – 1) (Equation 6)
𝐾

Where K is actual bend value and K90 represents the 900 bend loss coefficient.
Benard and Spedding (2006) gave an experimental data report using water-air two-
phase flow for a 900 vertical bend to horizontal. In their case, pressure drop was

12
compared between the straight pressure in the vertical before getting to the bend and at
the vertical inlet bend tangent. Where they gave a remark that there was a significant
difference in the total pressure value. They also made another comparison between the
pressure at the horizontal outlet bend tangent and the horizontal straight flow. Where
they gave a remark that the significant pressures agreed between the comparisons.
These researchers came up with a correlation regarding the pressure drop at the bend
angle based on the Reynolds number total. They presented saying that the pressure drop
at the bend angle can best be done with the use of the vertical tangent’s actual pressure
𝑙 𝑙
expressed in the term 𝐷𝑒 . The term 𝐷𝑒 was seen to be a function of the Reynolds number

total,

𝑙𝑒
= 0.001384 ReT – 13.53 (Equation 7)
𝐷

Where

𝑙𝑒
is an equivalent length which is dimensionless
𝐷

ReT is total Reynolds number

Additionally, another method they used to show their data was Lockhart-Martinelli
model

1
𝛥𝛲
𝜙𝑥 = [𝛥𝛲𝑇𝑃 ]2 𝐵 (Equation 8)
𝑆𝑋

1
𝛥𝛲
𝜙𝑥 = [𝛥𝛲𝑆𝐿 ]2 𝐵 (Equation 9)
𝑆𝐺

Here 𝜙 is pressure parameter for Lockhart-Martinelli

Lockhart-Martinelli model (1949)

13
So far from the above literature reviews provided, it is safe to come with a conclusion
that most of the two-phase elbow bend analysis that have been done are based on small
pipe diameters and using the common water-air combination phase as their fluids.
Focus was made on pressure drop determination and distribution of the two phases in
the elbow bends. For the flow pattern that changes after bends, it was primarily known
by visualization methods but the concept behind that phenomena was not discussed.
This encourages future researchers to use fluids which are industry based in studying
the elbow angle effects on multiphase flow as it would aid in better/optimal operations
and more efficient/safer designs of processing and flow systems. One of the aim of the
research analysis result studied by Abdulkadir (2011) using air and silicon oil
experimentally was to understand more the about the fundamental phenomenon of flow
around the elbow bend and help provide new data experimentally. He measured void
fractions using some advanced instrumentation including WMS (Wire Mesh Sensors),
video camera (high speed) and ECT (Electrical Capacitance Tomography). He used
PDF (Probability Density Function) analysis to determine the flow pattern, then
validated it by visualising the flow in the bend using the high speed camera.

Author(s) Modelling Fluids Elbow bend radius Diameter of Pipe

of curvature (mm) (mm)
Neller and Gardner Water - Air 305 76
(1969)
Maddock (1974) Water - Air 127,51, and 254 25.4
Charles and Water - Air 152.4 and 76.2 25.4
Oshinowo (1978)
Ribeiro (2001) Water - Air 166 32
Azzi Water - Air 180, 120, 300 and 30
(2000;2002;2005) 240
Benard and Water - Air 17 26
Spedding (2007)
Shannak (2009) Water - Air 300. 200, 500 and 50
400
Abdulkadir (2011) Silicone oil -Air 154.4 67

14
 Bello (2016) Silicone oil - Air 154.4 67
Table 1 some previous researches, with the fluid used and the pipe parameters

15
CHAPTER 3 CFD SOLUTION METHODOLOGY AND MODELLING

Hitherto, a reasonable number of researchers have put in effort to investigate slug flow
regime for multiphase flows. Some of the researchers workers to have studied this are:
Dumitrescu (1943), Griffith and Moissis (1962), Nicklin (1962), Beardmore and White
(1962), Brown (1965), Sekaguchi anda Akagawa (1966), Collins (1978), Wallis (1969),
Dukler and Mao (1985), Fernandes (1983), DeJesus (1995), Brauner and Barnea
(1993), Issa and Clarke (1997), Campos and Pinto (1996), Cui and Taha (2006),
Ullmann and Brauner (2004), and then Van Houst (2002) among other researchers. Of
the previous works, Line and Fabre (1992) had given the best in depth review on the
topic.
Although despite that, there is still a lot left to be investigated and known regarding the
slug flow pattern. Precisely, a more in depth investigation is required to best understand
the internal slug flow structure. However, reports available on the slug flow
characteristics are mainly not industry based fluids. In the review literature provided,
mechanistic models and empirical relations were shown. They are mainly approaches
and means that do not exactly characterize this type of flow because it is 1-Dimensional.
CFD
(Computational fluid dynamic) serve to solve the issue of the high limitations faced by
the 1-Dimensoinal approach. Depending highly on the regime or pattern of flow being
studied, different CFD models are applied because of the varying closure models for
different patterns of flow. To be able to use those models confidently, they must be
validated. Validating these CFD models need parameters from experimental data which
characterises the flow using different values.

This chapter describes the CFD model used and the obtained results will also be
discussed and presented. The only way the result obtained could mean something is
when it realistically represents moving/flowing fluids. Thus, the CFD simulations was
done under alike conditions with the experiments run by Abdulkadir (2011). The result
were showed to satisfactorily agree both quantitatively and qualitatively.

It is though very true that modelling in CFD cannot replace or substitute completely for
works done experimentally, but then it gives opportunities and options that cannot
however be done with experimental approach done physically. Using CFD allows, the

16
researcher to alter environments and conditions of flow without actually inquiring the
cost of carrying out and the rigorous efforts. Also, another advantage that is worth the
mention is that for CFD models is that one can be united with another to still give profit.
If the user has simulated a model and verified it with real experimental values of a
particular flow scenario, then he can go ahead and change or alter the conditions the
fluid’s governing parameters to fit other conditions of flow.

3.1 CFD (Computational Fluid Dynamic) Overview

3.1.1 Computational Fluid dynamic (CFD)

Engineering problems most times are almost impossible to be analytically solved,

therefore phenomena related to the flow of fluids can be described or seen by partial
fractions which are non-linear. Although, the use of discretization method allows for
the possibility of replacing the equations by obtaining approximated numerical
solution. Peric and Ferziger (2002) tell us that the method uses algebraic equation
system to approximate the differential equations, then a computer is used in solving the
algebraic equations. It is now the computational fluid dynamic area of focus method
towards tackling problems of engineering flow.

There are basic mathematical equations that govern the motion of fluid and the CFD
technique is a method that lies on the solved numerical of those equations.
Computational fluid dynamics analyses flow of fluid, mass and heat transfer and other
similar phenomenon using simulations that are computer based. The computer offers
an algorithm (numerical) that solves the governing mathematical equations of the
process. So far the solutions obtained from using this computer-based method has been
mostly accurate, while also providing good understanding of the fluid flow physics.
Fluid mechanics has been further investigated and understood since the introduction of
the CFD being supported by the present computer developments.

Computational fluid dynamics as a fluid dynamics branch eases theoretical and

experimental fluid dynamics approach as it provides a cost efficient alternative method
of simulating actual flows. Fletcher (1991) noted that by doing so, it provides a means

17
to test theoretical developments or advances for conditions that cannot be otherwise
done experimentally.

Fletcher (1991) then went on further to say that comparing the Computational and
Experimental fluid dynamics, the computational offers more advantages but five main
ones:

 It reduces lead time in developing and designing

 Using CFD, conditions of flow which that not be simulated by experimental
approach can now be done
 CFD offers more comprehensive and detailed information
 Compared to testing by Wind-tunnel, CFD is considerably cheaper
 CFD consumes lesser energy
There are currently so many CFD software codes that could be used for simulation.
Factors affect the choice of code selection, some of those factors include: usage ease,
cost, speed of computing, how it applies to the problem of flow in study and mainly
availability. Star-CCM+ which is used in this study is one of the most used code within
the industry alongside others like Star-CD. The Star-CCM+ is completely unrivalled
when it comes to it being able to solve complex geometries problems and multi physics
problems. It also has the capability to give optimal results using just one code and
applying lesser effort by the user Star-CCM+ (2016).

3.1.2 Fluid Flow Governing Equations

𝜕(𝑢𝜙) 𝜕(𝜌𝑢𝜙) 𝜕(𝜌𝑢𝜙) 𝜕2 𝜙 𝜕2 𝜙 𝜕2 𝜙

+ + = 𝛤 ( 𝜕𝑥 2 + 𝜕𝑦 2 + ) + 𝑆𝜙 (Equation 10)
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑧 2

In the equation above, parameter 𝜙 stands for property transported, 𝛤 for scalar 𝜙
diffusion coefficient and 𝑆𝜙 for source term.

The equations shown above applied for incompressible fluid are the basic motion
equations known in general as Navier-Stokes (N-S) equations.

18
3.2 CFD Multiphase Modelling

For any CFD simulation, there are three fundamental steps and they include:
 Pre- processing: Here the computational domain geometry is defined, grid is
generated, and then finally the proper boundary conditions are specified.
 Solution: This step is where the equations that govern the flow are solved by
means of iteration series or by discretization method by using a code. Also,
properties of the fluids are defined in this step and flow models are selected.
 Post-processing: It is the final step and it mainly involves the analysis and
interpretation of obtained results from above step 2.

3.2.1 Pre-Processing

For the finite volume method, the first step is defining the domain of computation and
then dividing them into cells or discrete control volumes. The process is called meshing
or grid generation. Then it the next step is specifying the proper cells conditions that
touch or coincide with the boundary domain.
The shapes and the sizes of grids generated play a major role on the CFD predictions
accuracy. If the modelling configuration is really complex, then there might be
challenging problems in trying to make discretization of computational domain. Boysan
and Kim (1999) proposed a means to tackle complex configurations and geometries by
using method of unstructured meshes. They said that by using this method, it eases
economically the problem of modelling complex configurations, geometries and
complex type of flow phase for users. But structured meshes are adopted for geometries
that are fairly simple.

To generate the grid and the geometry, the 3D-CAD software code which is a computer-
aided form of drawing available in most of the CFD codes is used. The CAD model is
available in its in own way in all of the major CFD package codes including
PHONENICS, Star-CD, ANSYS/CFX and FLUENT; If not provides the ability to
import mesh generators and surface modellers which are proprietary Malalasekera and
Versteeg (2007).

19
3.2.2 Solution/Post-processing

No matter what kind of flow being dealt with, be it turbulent or laminar, CFD simulation
can be used. Compared to laminar modelling, turbulent flows are much more difficult
to simulate and analyse because it needs special/extra considerations. In the case of
modelling turbulent flow, there exist some turbulence models in the Star CCM+ code
that solves the equations which govern flow and also helps in simulating the turbulent
flows. The Star-CCM+ give options of turbulent flow kind of modelling some of them
are: k-ω model, k-ε model (ε = turbulence dissipation rate, k = turbulence kinetic energy,
ω = specific dissipation rate. Therefore the turbulence model selection is done based on
the problem under study.

The most used and most common within engineering field application is the k-ε model
for turbulence. This model has been successfully used widely for simulating flows in
engineering practically and they are relative to the RANS (Reynolds Averaged Navier
Stokes) equations Malalasekera and Versteeg (2007). Although, it’s not all the time the
assumptions/predictions it make is accurate. For this reason of the k-ε model predicting
some results more than the satisfied values needed, it has led many researchers to go
after some alternative models like RNG (Renormalization Group) Boysan and Kim
(1999). This issue with the RANS model alongside the advancements made in
computing sector is what have caused more people using CFD to venture into LES
(Large Eddy Simulation) usage. Nevertheless, the RANS model is still very much
adopted and applied while giving some fine results.
Simulations made in CFD gives its results in the form of vectors, contours, other
properties of flow like: turbulence pressure, velocity and temperature. The primary
usage of CFD lies around these, interpreting and analysing the results obtained and
comparing it in contrast with real life conditions of flow.

3.3 Geometry Problem Description

The result would be discussed and presented in the next chapter but before then, this
section tell of a little basic simulation features. Slug pattern of flow was simulated
across an elbow bend of a pipe using the commercial CFD code Star-CCM+

20
3.3.1 Domain of Computation

For the experimental that is going to be used for comparison, it has the following the
geometric arrangement: the first part is a vertical riser having an internal diameter of
67mm with length of 6m and the second part is 2m long horizontal pipe connected with
the vertical arrangement of the riser at an elbow bend radius of curvature of 154mm.
Although equivalent conversions in the values were done to meet the CFD criterion
field of computation. The computational geometry was created as an image of the the
experimental arrangement, this is key because a verified experimental is required for
referral so as to meet real life setting and condition. Therefore, a complete 3D depiction
of this settings was done and based on past conclusions from experiments done by
Hernandez-Perez (2011) and Azzopardi (2010), it was seen that the flow is
axisymmetric. An instrumental method was used to determine the distribution of flow
along the 6m length and 67mm diameter riser, introduction of the WMS (Wire mesh
sensor) was made. A conclusion was made that the typical Tylor bubble was barely
seen in the pipe of that diameter. To investigate the slug flow in the riser, two ECT
(electrical capacitance tomography) planes at position 4.4m and position 4.489m high
and then a WMS (wire mesh sensor) was used at position 4.92m. Then after the elbow
bend another ECT was kept to monitor the flow regime at location 0.21m. The
multiphase fluids silicone oil and air were injected through the pipe’s inlet and let to
flow till it comes out at the outlet which has an atmospheric pressure. The Table 2
shows the properties of the fluids

Figure 4 3D geometry of the elbow bend pipe flow system

21
 Density of Dynamic Surface Thermal
the Fluids Viscosity of the Tension Conductivity
(kgm-3) Fluids (kg-1m-1) (Nm-1) (Wm-1k-1)
Air 1.18 0.000018
Silicone Oil 900 0.00525 0.02
Both 0.1
Table 2 Properties of the multiphase fluids at 1bar (atmospheric) and at a temperature of 200C

3.3.2 Boundary Conditions

The next thing to be done after the generation of mesh was to specify the conditions of
the boundaries in the domain of computation. Defining the boundary specifies the
operational and physical characteristics of what kind of model used at the topology unit
represents the boundaries model. It is of utmost importance that the boundary condition
be handled in carefulness as it depends very much on the physics related to the problem
under study. Therefore, boundary condition selection was made according to the
arrangement done in the experimental as just described above. Below summarizes the
conditions of the boundary:
^

For Inlet; Velocity Inlet was selected n .ῡ = 𝑈𝑜 (Equation 11)

Dirichlet Boundary Condition

For Outlet; Pressure outlet was selected P = 𝑃𝑜 (Equation 12)

Dirichlet Boundary Condition

^ ^

For Wall; ‘Wall’ was selected n .ῡ = t .ῡ = 0 (Equation 13)

^ ^

Where n.ῡ means no penetration and t .ῡ means no slip

22
A phenomenon of non-slip boundary condition was assumed for all the solid wall
boundaries, at ν = 0. For the inlet position at the bottom of the riser, the boundary type
was chosen to be velocity inlet where the void fraction of liquid was specified and the
superficial velocity of the mixture was also specified. The profile of the velocity was
monitored although based on the assumption, it should be uniform. Based on
Hernandez-perez (2008) to setup in the software a numerical model simulation, further
slug flow knowledge based on experiment is not needed in this type of approach. The
experimental study was conducted in similar fashion. Spalding and Launder (1974)
proposed equations to solve for dissipation rate ε and turbulent kinetic energy k values
at the inlet.

3
𝑘𝑖𝑛 = 2 𝐼 2 𝑈𝑖𝑛 2 (Equation 14)

3
𝜀𝑖𝑛 = 2𝑘𝑖𝑛 2 /𝐷 (Equation 15)

0.16
𝐼= 1 (Equation 16)
𝑅𝑒 8

I here standards for the turbulence intensity in the pipe flow which is fully developed.

The pipe walls are presumed to be impermeable and rigid, and for the roughness of the
wall, it was selected as smooth wall. Boundary condition of no-slip is applied for
velocities at any point in time there was a contact with solid walls. Again according to
Spalding and Launder (1974) the close to wall approach was used in predicting the
flows near/around the wall based on their standard function approach. To satisfy the
balance in mass flow at the of the horizontal outlet which is the downstream of the
elbow, all the other variables were taken out of the domain of computation having a
static pressure averaging zero. Throughout the operation, the operating pressure was set
to be standard (i.e 101.3 kPa atmospheric pressure) and using a 200C temperature. Also
the effects of gravity was considered having a value of -9.81 ms-2 acceleration due to
gravity due to the vertical flow upwards and assuming an industrial standard. Gravity
would have been ignored if it was just the domain of computation.

23
3.3.3 Initial Conditions

When the time is 0s, the general component of velocity all over is also set to 0 ms-1
meaning
^

( n. ῡ = Vo and t = to). For the volume fraction of liquid, the initial condition specified
was due to the case at hand for the vertical pipe and the elbow flow situations, either a
pipe full of silicone oil or air was set as the initial condition. Convergence process is
eased by specification of initial conditions. Furthermore, for the dissipation rate and the
turbulent kinetic energy, a guess at the initial stage was used throughout the simulation.

3.4 CFD Modelling Assumptions:

Star-CCM+ which is a code from the CFD family package was applied for this study
in simulating Taylor bubble which rises within the flowing liquid and also monitoring
the pressure drop along a vertical 67 mm internal diameter pipe through to the elbow
bend angle. The domain of flow computation was discretized numerically by using the
Finite Volume Method. A method from Nicholas and Hirt (1981) for tracking the
motion of the liquid-gas interface modelled was by using a 𝛼𝐺1 distribution, which is
based on the gas volume fraction in the cell of computation; where it is liquid phase
cell if 𝛼𝐺 = 0 and a gas cell if 𝛼𝐺 = 1. Which means that for a cell having a liquid-gas
interface, it should exist when 𝛼𝐺 is between the values of 0 and 1.

3.4.1 Governing Equations

For the numerical simulation, the VOF (volume of fluid) method which can be found
under the Eulerian model approach in the commercial Star-CCM+ was used. The
scenario of flow being dealt with in this study is a multiphase incompressible slug
having isothermal condition of flow. Assuming Isothermal conditions for pipes having
broad surface area compared to volume ratio at constant temperature is reasonable.
The VOF method is a type of technique that tracks the interface between different
phases by capturing it. Kvicinsky (1999) describes it as a strong tool that permits any
kind of free surface complex flow to be simulated at any situation including that of

24
break-up or folding shape. The VOF method has the capability of capturing important
features of flow surrounding the complex free surface, but has to have a proper
confirmed numerical technique for conservation of energy, momentum and mass; and
also for interface resolution Rhee (2004). Heat and mass transfer can be handled
according to Hernandez-Perez (2008) over the interface by using VOF method. As
similar to other tools, the VOF method also has its own limitations which is important
to note. For situations whereby the velocities between the multiphase significantly
differ, it might not be suitable to use the VOF method since we know that it is used for
cases where similar momentum equations are shared between the multiphase fluids.
But apart from that, it is still a very robust and useful tool used for tracking and since it
was developed, it has become a popular algorithm which has deserved the frequency of
choice under the multiphase flows Eulerian models DeJesus (1997). Ubbink (1997)
introduced a scheme and then Peric and Muzaferija (1999) enhanced it, it is the HRIC
(High Resolution Interface Capturing Scheme) which operates in the Star-CCM+ and
it is a subset of Compressive Interface Capturing Scheme for Arbitrary Meshes
(CISCAM). When using the VOF method, the values are represented as volume-
averaged and the phases share properties/variables fields so far each phase’s volume
fraction is known at every location, solving the continuity equation is done in order for
the system’s mass balance to be maintained. Surface tension and gravitational forces
are the forces acting on the body involved in the momentum equation. Molecules along
a fluid undergo forces of attraction and this can cause surface tension within the
interface. To introduce the surface tension under the VOF model, a momentum source
is added. The equation (18) below is a momentum equation which depends on all the
phases volume fraction by their properties μ and ρ.
For the multiphase domain flow, below are the conservation equations for momentum,
volume fraction, and mass:

𝜕𝜌 𝜕𝜌𝑢𝚒
+ =0 (Equation 17)
𝜕𝑡 𝜕𝜌𝑖

𝜕𝜌𝚞𝑗 𝜕𝜌𝚞𝑖 𝑢𝑗 𝜕𝑃 𝜕 𝜕𝑢 𝜕𝑢
+ = − 𝜕𝑥 + 𝜕𝑥 𝜇 (𝜕𝑥 𝑖 + 𝜕𝑥𝑗) + 𝜌𝑔𝑗 + 𝐹𝑗
𝜕𝑡 𝜕𝑥𝑖 𝑗 𝑖 𝑗 𝑖

(Equation 18)

25
Where F, P and G respectively stand for Force external per unit of volume, pressure
and acceleration due to gravity. But the volume fraction of each of the phase based on
their 𝜇 (Viscosity) and their 𝜌 (density) determines the above shown momentum
equation. If a system of multiphase flow has 1 and 2 as subscripts as representation for
its phases, where phase 2 volume fraction is known, then for the individual cells the
equation below for the 𝜇 and 𝜌.

𝜌 = α2𝜌2 + (1 – α2) 𝜌1, 𝜇 = α2𝜇2 + (1- α2) 𝜇1 (Equation 19)

For function of volume fraction, when the continuity equation below is solved, it traces
the interface between the phases.

𝜕𝛼𝑞 𝑢𝑖 𝜕(𝛼𝑞 )
+ =0 (Equation 20)
𝜕𝑡 𝜕𝑥𝑖

Where ui denotes velocity component, t stands for time and then xi for co-ordinates in
i (either 1,2,3) direction; and across the momentum equation’s resolution the two fluids
in context share.

The equation for the volume fraction of the primary phase which is the equation (19)
above is obtained from the equation below. Then the primary phase volume fraction
obtained is computed.

∑𝑛𝑞−1 𝛼𝑞 = 1 (Equation 21)

Surface Tension was modelled using Brackbill (1992) proposed CSF (Continuum
surface force). In this model, surface tension is not seen as a boundary condition free
surface but rather as volume force over the interface. A source term is added to the
equation of momentum when the CSF model is used.

Star-CCM+ software carries out the numerical solution involved with the equations.
Since the VOF model needs the fluxes which are at the faces of the control volume to
be determined, it was done by using a discretizing scheme of the second order.

26
3.4.2 Turbulence Model

As long as the Taylor bubble is rising along the liquid flow even if the rate is low, what
is created on the wall of the bubble is a liquid developing film and also at the tail a wake
is created. Due to this, turbulence must be put into consideration while simulating
numerical. The turbulence model alongside complexities in flow type can affect how
accurate a turbulent flows solutions done in CFD could be. One basic uncertainty found
in predictions made in CFD is trying to choose the appropriate model of turbulence for
turbulent flow effects in time-averaged equations mean-flow noted by Malalasekera
and Versteeg (2007) and also Pope (2000). Pope (2000) went further to document that
the effects of turbulence in discretised equation is the main non-linearity source. Using
Spalding and Launder (1974) k-ε model which is standard for simulating turbulence
was applied in this work, Behnia and Cook (2001) and Ramos-Banderas (2005) have
suggested it for studying two-phase flows. Since its introduction in 1974, this standard
model grew to become the most popularly applied model for turbulence in not just oil
and gas industry application but overall engineering DeJesus (1997).

An assumption the k-ε model makes is that energy transfer is at near-balance with
dissipation rate and production rate of turbulent flows; thereby ε the dissipation rate is
as given below:

𝑘 3⁄2
ε= (Equation 22)
𝐿

Where L in this equation stands for the involved length and k represents the flow’s
kinetic energy. Based on model by Prandtl for mixing length, 𝜇t turbulent viscosity can
then be varied with ε dissipation rate as shown:

𝑘2
𝜇t = 𝜌𝐶𝜇 𝜀
(Equation 23)

𝜌 represents the flow’s density and then 𝐶𝜇 being a constant

27
The elliptic equation below as requirement for RANS (Reynolds Averaged Navier Stokes)
closure is used to describe the model:

k- equation:

𝜕(𝜌𝑘) 𝜕(𝜌𝑢𝑗 𝑘) 𝜕 𝜇𝑡 𝜕𝑘 𝜕𝑢𝑖 𝜕𝑢𝑖 𝜕𝑢𝑗 𝜕𝑢𝑡

+ 𝑢𝑖 = ( )+ 𝜇𝑡 ( + ) − 𝜌𝜀 (Equation 24)
𝜕𝑡 𝜕𝑥𝑗 𝜕𝑥𝑗 𝜎𝑧 𝜕𝑥𝑗 𝜕𝑥𝑗 𝜕𝑥𝑖 𝜕𝑥𝑗 𝜕𝑥𝑗

Convection Term Diffusion Term Production Term Destruction Term

𝜀 – equation:

𝜕(𝜌𝜀) 𝜕(𝜌𝑢𝑗 𝜀 𝜕 𝜇 𝜕𝜀 𝜀 𝜕𝑢 𝜕𝑢 𝜕𝑢 𝜀2
+ 𝑢𝑖 = 𝜕𝑥 (𝜎𝑡 𝜕𝑥 ) + 𝐶𝜀1 𝑘 𝜇𝑡 𝜕𝑥 𝑖 (𝜕𝑥 𝑖 + 𝜕𝑥𝑗) − 𝜌𝐶𝜀2 (Equation 25)
𝜕𝑡 𝜕𝑥𝑗 𝑗 𝜀 𝑗 𝑗 𝑗 𝑖 𝑘

Convection Term Diffusion Term Production Term Destruction Term

Subsequent ‘production term’ of turbulence is the reason for energy transfer to

turbulence from mean flow, gradient of mean velocity and Reynolds stresses interaction
counterbalances it. While ‘Destruction term’ signifies energy dissipation to become
heat because of the flow’s viscous nature. Fokeer (2006)

Some of the values above are constants and their respective values are 𝜎k = 1.0, 𝜎𝜀 =
1.3, C𝜇 = 0.09, 𝐶𝜀1 = 1.44 and 𝐶𝜀2 = 1.92. Velocity of the fluid’s i component is
denoted by 𝑢𝑖 and then spatial coordinate for j is represented by 𝑥𝑗 and u. For the
equations of Navier stokes, corrections on the viscosity of the fluid must be made when
dealing with turbulence; this is done when a viscosity which is effective is employed

28
𝜇𝑒𝑓𝑓 = 𝜇 + 𝜇𝑡
(Equation 26)

𝜇𝑡 denotes turbulent viscosity

𝜇 denotes dynamic viscosity

3.4.1 Independent Mesh Study

Computationally, simulations done with CFD’s are expensive. One primary factor
worthy of mention that influences time of computation is computational grid sixe the
user chooses. Independent study for mesh is carried out to make sure that the minimum
density of mesh needed to ensure obtained converged solution from the CFD
calculations is sovereign of the overall grid density. Therefore, if the mesh cells amount
is increased in the flow simulated would still not affect the integrated quantities and
flow solution. In application, indication of mesh independence happens when mesh
refinements are further done and yield just insignificant or small changes within the
obtained numerical solution. The meshing process can be used to analyse flow of fluid
that cannot be solved by the use of the method of analytical solutions. The problem is
further simplified by cells a subdivision of the flow. The general elements collection is
known as “Grid or Mesh” and they could either be in the form of 2D or 3D.
Classifications of meshes are as follows: hybrid, structured, unstructured or element
based, however for this study the structured element was selected as it gives fluid
simulations are better solution mechanism. The selection of the structured element was
also due to its automated meshing and connectivity, also to speed up the rate of the
process simulation while giving accurate results, the prism layer process was utilized.
For the mesh, 45 was the amount of prism layer selected and a cell size of 100,000.
To make an appropriate selection of mesh for a particular duct, the mesh quality is to
be assessed. In carrying this out, the mesh convergence rate would be greater in order
to give a good quality mesh which would thus give a CFD result which is more accurate.
CFD-Online (2015). While selecting models for the mesh, the surface repairer which is
automatized was selected to improve the quality of mesh by fixing the poor ones so as
to improve the overall flow simulation

29
Figure 5 3D Depiction of the mesh

Modelling Procedures and Selections:

The first step taken was to create the geometry of the elbow bend pipe, to do that:
 Star-CCM+ was started and a new simulation was opened and saved
 Using the in software 3D-CAD model which serves as the solid parametric
modeller, the geometry was created
The pipes dimensions were as follows:
 67mm internal diameter was used
 The vertical length of the pipe before the elbow bend was measured 0.5m
 The horizontal length of the pipe downstream of the elbow bend was measured
1m
The sweep function was used to generate the body, it needs two made sketches; the first
sketch act as the main profile to be swept while the second one act as the sweep path.
Inlet face and Outlet face. A new type of body was created and renamed to be the Fluid.
To finalise the geometry modelling step, specification of the inlet face and the outlet
face is done when the face names are set.
Next assigning of the geometry created to a region. Since both the Inlet face and the
Outlet faces in the 3D-CAD were renamed, it would get added automatically to
different boundaries.
The case was then set-up; the case was ran at a turbulent type of flow

30
All the boundaries are by default set as walls. The Inlet and Outlet boundary type was
modified. For the properties of the inlet node, it was set to velocity inlet and at the
properties of the outlet boundary, it was set to pressure outlet.
After that had been done, a mesh was generated. A polyhedral mesh was generated with
the use of the generalized cylindrical mesher.
The options chosen in the meshing selection model where:
 From Surface Mesh box, Surface Remesher
 From Volume Mesh box, Polyhedral Mesher
 From Optional Models box, Generalised Cylinder
 From Optional Models box, Prism Layer Mesher

After making these selections, the settings of the mesh was defined as below:
 The property of the base size was set to a value of 0.005m
 The property for number of prism layer from the Prism layer option was set to
2
 Prism layer stretching value from the prism layer option was set to 2.0
 The value for base property percentage from the Prism layer thickness was set
to 45
Before the use of Generalised cylinder mesher to generate the volume mesh, the surface
mesh was created and analysed.
The boundaries forming the cylindrical geometry were defined before moving forward
to create the volume mesh.
The Generalised cylinder extrusion type was selected from the mesh conditions option
and the property was set to constant. The properties window under the mesh values
shows 235 as the number of layers, that is automatically calculated by Star-CCM+.

3.5 Physics Models selection

While creating the mesh process, a physics continuum option was made automatically.
The physics models were selected to define the flow
The selections made are from the model options are as below:
 From Time box, Implicit Unsteady
 From Space box, Two Dimensional

31
  From Material box, Eulerian Multiphase
 From Eulerian Multiphase Model box, Volume of Fluid (VOF)
 From Viscous Regime box, Turbulent
 From the Reynolds-Averaged Turbulence, K-Epsilon Turbulence
 From Optional Model box, Gravity
The layer of the boundary near the wall was captured accurately using the y+ treatment

Two Eulerian phases are created, renamed, phase models were selected, and the existing
phase materials replaced with required, then the material properties were customised.
As, Silicone oil did not exist on the Star-CCM+ database, it had to be created. Due to
the fact that surface tension modelling was introduced, the liquid phase was made the
first phase as expected.
 The model for the first phase was set to be Liquid (Silicone Oil), Constant
density & Turbulent.
 The model for the second phase was set to be Gas (Air), Constant Density &
Turbulent.
 The properties for air was automatic from the Star-CCM+ software (Dynamic
Viscosity = 1.85508E-5 Pa-s & Density = 1.18415 kg/m3)
 Properties for Silicone Oil were manually inputted (Dynamic Viscosity =
0.00525 pa-s & Density = 900.0 kg/m3)
 For Material properties for the mixture, Prandtl number was set to be 0.9
 Surface Tension option was applied under the multiphase interaction and had a
property constant value of 0.02N/m

For Reference Values

 Gravity was used at -9.81 m/s2
 Reference Density of 1.0 kg/m3
 Minimum allowable wall distance of 1.0E-6 m
 The reference pressure was standard at 101325.0 Pa

For Initial Conditions

 Again the pressure was atmospheric pressure of 101325.0 Pa
 Turbulence Intensity value of 0.01

32
  Turbulent Velocity Scale of 1.0 m/s
 A turbulent Viscosity ratio of 10.0
 Velocity of Silicone Oil and Air were 0.05 and 0.34 respectively
 Liquid Volume fraction of 0.2

Volume of fraction constant option defines uniform distribution, for instance a mixture
of initially liquid is (1.0,0.0), composite option describes the initial distribution of a
phase.
Volume fractions at flow boundaries was not set for phase 1. Setting the option constant
at the inlet boundary tells the specific percentage distribution of both phases.

To run it, a scalar scene was created to show the section plane through the elbow bend
displaying the velocity magnitude. This aids in visualizing of the solution as the
simulation is still running. Other scalar scenes were created to display static pressure
and total pressure. A vector Scene was also created to display the velocity magnitude.
A stopping criteria of 1000 maximum steps was applied.

33
CHAPTER 4 ANALYSIS AND RESULTS OF CFD

To make the investigation of the effect of the elbow bend on the multiphase flow across
it, another simulation is conducted in single phase with the same conditions as the
multiphase using Silicone Oil. Comparisons is going to bed between the results
obtained from the multiphase model and the Single phase model: focusing on pressure
drop. Flow regime plot was also simulated.

4.1 Solution and Discretization Algorithm

For solving numerically the differential partial equations governing the system, the
equations were discretized by the use of a FVM (Finite Volume Method) available in
Star-CCM+ along with also using a solver segregated algebraically and a grid co-
located arrangement just as applied in the Star-CCM+ software. In the grid co-located
arrangement. The FVM discretization method are not in depth discussed here but is
seen other works like that of Malalasekera and Versteeg (2007). The VOF model in the
Star-CCM+ makes use of a solver which is segregated; hence, both the momentum and
continuity equations should be connected. For this connection to be done, some
techniques in the literature discuss it. The equation from Spalding and Patankar (1972)
is used since it gives a solution which is convergent and fast, the algorithm is called
SIMPLE (Semi-Implicit Method for Pressure-Linked Equations). Furthermore, the
technique AMG (Algebraic Multigrid) helps in improving the present iterative solver
by yielding a convergence rate which is better.

4.2 Pressure Distribution

For a pipeline design, one key parameter is pressure. A system’s pressure drop is a
variable which is very important while determining a particular flow’s pumping energy.
Compared to single phase flow, the pressure drop at a multiphase flow would be higher
with both having mass flow which is the same. A smaller density results to higher
velocity of fluid. For density of fluid which is the same, multiphase flow would have
higher turbulences than that of single flow, which leads to more dissipative losses as
compared with a single phase flow Al-Hashimy et al. (2015). The basic method of
estimating pressure drop for multiphase flow is by using an equation. Although a prime

34
problem is while determining the density, friction factor and velocity while carrying
out these calculations. The drop in pressure of a horizontal/vertical pipe would not be
as much as that from the inlet of a vertical riser to the downstream horizontal outlet
caused by a pipe’s elbow end which provides some kind of choke in the flow which
eventually results to a liquid hold-up and pressure build-up in the tangent leg of the
vertical riser as would be shown in the CFD pressure simulation below.

4.3 CFD Results

4.2.1 CFD results for pressure profiles

Figure 6 Plot for the Multiphase static pressure

35
Figure 7 Total pressure plot for the multiphase flow

Figure 8 Pressure loss in the two-phase flow

The pressure plots above for the multiphase flows show us that the pressure is really
high at the inlet but on getting to the curvature of the pipe we can see it begins to drop
gradually until where the static pressure at the outlet is zero. The total pressure drop
solved numerically by the Star-CCM+ code is also shown, where it is understood to be
very high.

36
Figure 9 Total pressure plot for single phase flow

Figure 10 Static pressure for the single phase (Silicone Oil)

Figure 11 Single phase pressure loss

From all the above pressure profiles, we can see from the results above obtained using
the in-built pressure loss report calculation in the software that the total pressure loss

37
from the Inlet to the Outlet for the single phase was less as compared with that of the
two-phase flow.

4.2.2 CFD results for velocity profiles

Figure 12 Velocity profile for the Oil/Gas mixture

Figure 13 Velocity depiction for the single phase oil

Shown above are the velocity profiles obtained for both the single phase flow and the
multiphase flow. At the same inlet velocity of 0.05ms-1 for the silicone oil in the single
phase flow, it remained constant all through pipe components. While for the multiphase

38
flow comprising of the silicone oil entering at the same velocity and the air coming in
at 0.34 ms-1 made the overall velocity increase all through the system of the pipe.

4.2.3 CFD results for flow regime

Figure 14 Void Fraction Flow movement along the pipe

A diiferent simulation was done with the aim of capturing each phase interface using
the VOF model. At the upstream of the elbow bend, a slug flow movement is observed
but as it moves past the elbow bend we see a change in flow regime to a stratified wavy
flow. From the above figure; ‘A’ represent the first and the second Taylor bubbles
observed from the inlet vertical riser inlet, ‘B’ shows the movement of the Taylor buble
at the outside of the bend, and ‘C’ depictss the beginning movement of the stratified
wavy flow along the horizontal downstream of the elbow.

39
4.2.4 Convergence

Figure 15 Residuals obtained for single phase flow

Figure 16 Residuals obtained for multiphase flow

The above graphs represent the convergence of both simulations at an iteration of 3000
before complete convergence.

40
4.3 Empirical models to calculate pressure drop

4.3.1 Chisholm Model

Chisholm made a proposal which involves correlating dimensionless parameters gotten

from correlation experimental data from multiphase flow Chisholm (1983). Le which
is the pipe length depending on the radius of curvature value for the elbow, it angle and
its internal diameter. Le increment as a function of the internal diameter is primarily
because of secondary flow existing in the elbow, centrifugal force and friction. Hence,
pressure drop for single phase in the elbow as expressed by:

𝐺 2 𝐿𝑒
∆𝛲1𝑝ℎ = 𝑓 2𝜌 ( 𝐷 ). (Equation 27)

Chisholm made a suggestion to evaluate the coefficient k1 of the pressure drop by

making an assumption that throughout the pipe fitting, the multiphase mixture flowing
is in liquid form
only is evaluated by:

𝐿𝑒
𝑘1 = 𝑓1 ( 𝐷 ). (Equation 28)

Putting into consideration that the pipe is filled up with liquid phase only, the pressure
drop for the multiphase mixture is given by:

𝑘1 𝐺 2 𝑇
∆𝛲1𝑝ℎ,1 = 2𝜌𝑙
. (Equation 29)

Thus, the equation for calculating the pressure drop for a multiphase flowing across an
elbow bend already has mass quality in it and also a correlation factor, considering
multiphase properties:

∆𝛲2𝑝ℎ = ∆𝛲1𝑝ℎ,1 [1 + {𝐸(𝑥(1 − 𝑥)) + 𝑥 2 }], (Equation 30)

Where E stands for elbow 900 coefficient, of which the relative elbow radius is included

41
 2.2
𝘌=1+𝑘 . (Equation 31)
1 (2+𝑅𝑒 ⁄𝐷

4.3.2Azzi-Friedel Model

According to the model by Azzi and Friedel (2005) the multiphase flow multiplier
determines the pressure drop defining it as a ratio of multiphase flow in elbow bend
pressure drop to that of the single phase flow in liquid having same overall mass flow
rate shown below:

∆𝛲
Ф2 = ∆𝛲 2𝑝ℎ , (Equation 32)
1𝑝ℎ,𝐼

Where ∆𝛲1𝑝ℎ,I denotes bend pressure drop for liquid fluid flowing in single phase at
same bend as in:

𝐺𝑙 2
∆𝛲1𝑝ℎ,1 = 𝑘𝑖 , (Equation 33)
2𝜌𝑙

Where

𝐿𝑒
𝑘𝑖 = 𝑓𝑖 ( 𝐷 ), (Equation 34)

Where the equivalent length for single phase flow which is dimensionless is Le/D, while
the single phase friction factor for the pipe is …. The equation below could be used to
calculate that factor based on Churchill (1977)

8
𝑓𝑖 = 8((𝑅𝑒)12 + (𝐴𝑖 + 𝐵𝑖 )−1.5 )1⁄12 (Equation 35)

Where

42
 7 0.9 𝜀
𝐴𝑖 = (2.457𝑙𝑛(( ) + 0.27 )−1 )16 , (Equation 36)
𝑅𝑒𝑖 𝐷

37530 16
𝐵𝑖 = ( 𝑅𝑒𝑖
) , (Equation 37)

𝜌𝑖 𝑉𝑖 𝐷
𝑅𝑒𝑖 = 𝜇𝑖
. (Equation 38)

The i subscript in the equations above could be to represent both gas or liquid
Azzi and Friedel’s multiphase flow multiplier is expressed below:

𝑟 0.502 0.7
Ф2 = 𝐶 + 7.42𝐹𝑟𝑙 0.125 𝐷
𝑥 (1 − 𝑥)0.1 (Equation 39)

(𝜌𝑙 −𝜌𝑔 ) 0.14 (𝜇𝑙 −𝜇𝑔 )

𝘹( ) ( , (Equation 40)
𝜌𝑙 𝜇𝑙

𝜌𝑘
𝐶 = (1 − 𝑥) + (𝜌𝑙 𝑘𝑔) 𝑥 2 . (Equation 41)
𝑔 𝑙

Where below is the Froude number (Frl)

(1−𝑥 2 )𝐺 2 𝑇
𝐹𝑟𝑙 = . (Equation 42)
𝜌𝑙 2 𝑟𝑔

The conditions around each tangent on either side of the elbow bend for a multiphase
vertical flow to horizontal flow would be different since the effect of uplift forces and
gravity at the inlet of the vertical tangent would be absent at the outlet of the horizontal
tangent. Furthermore, most times the flow phenomena and regime at the vertical tangent
would be different from that at the horizontal tangent thus the pressure drop calculations
using models across the 900 bend is more difficult and complex as compared to that of
the flow in single phase that has significant same phase density all through and also
frictional pressure of the straight pipe could be used in calculating the pressure drop in
an elbow no matter the structure or orientation of the elbow’s plane. But it is a different

43
case for a flowing multiphase which would require each tangent’s pressure drop in
calculating the overall as shown in Figure 17. In the figure, A is the centreline radius,
B is the degree of bend and both tangents of the pipe labelled as ‘C’ where the upper C
is the downstream tangent and the lower C the upstream tangent of the pipe. The
pressure drop obtained from the vertical tangent was compared with that of the
horizontal tangent to determine if at all the elbow bend had an effect on the pressure of
the two-phase flow at each tangent leg.

Figure 17 Tangents of the pipe bend labelled as ‘C’

44
CHAPTER 5: CONCLUSION AND FUTURE WORK

5.1 Conclusion

A general investigation on multiphase flow across a vertical riser through a 900 elbow
bend was carried out. From observation of results, the pressure loss experienced in a
multiphase flow is seen to be really high and greater as compared to that in a single
phase flow. To verify that there was an actual pressure drop between the vertical tangent
leg and the horizontal tangent leg, a calculation was made which confirmed it. For a
single phase flow and a multi-phase flow, the presence of an extra flow increases the
velocity from the inlet to the outlet for the overall flow. Considering a two-phase
vertical to horizontal flow, it would almost always change its regime from a slug flow
regime upstream of the bend to a stratified way flow downstream of the bend as it passes
through the elbow, a simulation was conducted to verify that.

5.2 Recommendations for Further work

 Since the introduction of CFD, experimental 1D method has been steadily

abandoned and only researched by a few. If available, further experimental
should be done altering conditions

 Unlike almost all other previous studies done by researchers, future works
should focus on larger pipe diameters.

 Further work using industrial fluids rather than the conventional air/water flows

 More analysis should be done using and focusing on slug characteristics in two-
phase flow; how it is formed, when are they formed and peculiar effects the lay
on the flow as compared to other types of flow

45
  Other form of investigations for elbow bend estimating corrosion-erosion on
multiphase flow including sand

 Different orientation of pipes should also be taken into consideration while focusing
on the effects they play on flow regimes and flow properties.

REFERENCES

Abdulkadir, M. (2011) Experimental And Computational Fluid Dynamics (CFD)

Studies Of Gas-Liquid Flow In Bends. Ph.D Thesis. University of Nottingham

Akagawa, K. and Sakaguchi, T. (1966) "Fluctuation Of Void Fraction In Gas-Liquid

Two-Phase Flow". Bulletin JSME, 9, 104-110

Azzi, A., Belaadi, S. and Friedel, L. (2000) "Two-Phase Gas/Liquid Flow Pressure Loss
In Bends". Forschung im Ingenieurwesen 65 (10), 309-318

Azzi, A. and Friedel, L. (2005) "Two-Phase Upward Flow 90� Bend Pressure Loss
Model". Forschung im Ingenieurwesen 69 (2), 120-130

Azzi, A., Friedel, L., Kibboua, R. and Shannak, B. (2002) "Reproductive Accuracy Of
Two-Phase Flow Pressure Loss Correlations For Vertical 90° Bends". Forschung im
Ingenieurwesen 67 (3), 109-116

46
Benbella, S., Al-Shannag, M. and Al-Anber, Z. (2009) "Gas–Liquid Pressure Drop In
Vertical Internally Wavy 90° Bend". Experimental Thermal and Fluid Science 33 (2),
340-347

Brackbill, J., Kothe, D. and Zemach, C. (1992) "A Continuum Method For Modeling
Surface Tension". Journal of Computational Physics 100 (2), 335-354

Brauner, N. and Barnea, D. (1986) "Slug/Churn Transition In Upward Gas-Liquid

Flow". Chemical Engineering Science 41 (1), 159-163

Brauner, N. and Ullmann, A. (2004) "Modelling Of Gas Entrainment From Taylor

Bubbles. Part A: Slug Flow". International Journal of Multiphase Flow 30 (3), 239-
272

Brown, R. (1965) "The Mechanics Of Large Gas Bubbles In Tubes: I. Bubble Velocities
In Stagnant Liquids". Can. J. Chem. Eng. 43 (5), 217-223

Carver, M. (1984) "Numerical Computation Of Phase Separation In Two Fluid Flow".

Journal of Fluids Engineering 106 (2), 147

Carver, M. and Salcudean, M. (1986) "THREE-DIMENSIONAL NUMERICAL

MODELING OF PHASE DISTRIBUTION OF TWO-FLUID FLOW IN ELBOWS
AND RETURN BENDS". Numerical Heat Transfer 10 (3), 229-251

Collins, R., Moraes, F., Davidson, J. and Harrison, D. (1978) "The Motion Of A Large
Gas Bubble Rising Through Liquid Flowing In A Tube". J. Fluid Mech. 89 (03), 497

Chenoweth, J.M. and Meritt W.M. (1955) “Turbulent two phase flow” Petroleum
refiner 34.10 151-155

Chisholm, D. (1983) Two-Phase Flow In Pipelines And Heat Exchangers. London: G.

Godwin in association with Institution of Chemical Engineers

47
Churchill, S. W. (1977). “Friction-factor equation spans all fluid-flow
regimes”.Chemical engineering, 84(24), 91-92.

Clarke, A. and Issa, R. (1997) "A Numerical Model Of Slug Flow In Vertical Tubes".
Computers & Fluids 26 (4), 395-415

Cook, M. and Behnia, M. (2001) "Bubble Motion During Inclined Intermittent Flow".
International Journal of Heat and Fluid Flow 22 (5), 543-551

Davidson, J., Wilkes, J. and Nicklin, D. (1962) "Two-Phase Flow In Vertical Tubes".
Transaction of Institution of Chemical Engineers 40, 61-68

Dean, W. (1927) "XVI. Note On The Motion Of Fluid In A Curved Pipe". The London,
Edinburgh, and Dublin Philosophical Magazine and Journal of Science 4 (20), 208-
223

Dean, W. (1928) "LXXII. The Stream-Line Motion Of Fluid In A Curved Pipe (Second
Paper)". The London, Edinburgh, and Dublin Philosophical Magazine and Journal of
Science 5 (30), 673-695

DeJesus, J. (1997) An Experimental And Numerical Investigation Of Two-Phase Slug

Flow In A Vertical Tube. PhD thesis. University of Toronto, Canada

DeJesus, J., Ahmad, W. and Kawaji, M. (1995) "Experimental Study Of Flow Structure
In Vertical Slug Flow". Advances in Multiphase Flow 31, 105-118

Dewhurst, S., Martin, S., Jayanti, S. and Costigan, G. (1990) Flow Measurements Using
3-D LDA System In A Square Section 90 Degree Bend.. Report AEA-In Tech-0078

Dumitrescu, D. (1943) "Strömung An Einer Luftblase Im Senkrechten Rohr". ZAMM -

Zeitschrift für Angewandte Mathematik und Mechanik 23 (3), 139-149

Eustice, J. (1910) "Flow Of Water In Curved Pipes". Proceedings of the Royal Society
A: Mathematical, Physical and Engineering Sciences 84 (568), 107-118

48
Fabre, J. (1992) "Modeling Of Two-Phase Slug Flow". Annual Review of Fluid
Mechanics 24 (1), 21-46

Fernandes, R., Semiat, R. and Dukler, A. (1983) "Hydrodynamic Model For Gas-Liquid
Slug Flow In Vertical Tubes". AIChE Journal 29 (6), 981-989

Ferziger, J. and Perić, M. (2002) Computational Methods For Fluid Dynamics. Berlin:
Springer

Fokeer, S. (2006) An Investigation Of Geometrically Induced Swirl Applied To Lean

Phase Pneumatic Flows. PhD thesis. University of Nottingham.

Gardner, G. and Neller, P. (1969) "Phase Distributions In Flow Of An Air—Water

Mixture Round Bends And Past Obstructions At The Wall Of A 76‐Mm Bore Tube".
ARCHIVE: Proceedings of the Institution of Mechanical Engineers, Conference
Proceedings 1964-1970 (vols 178-184), Various titles labelled Volumes A to S 184 (33),
93-101

Hernandez-Perez, V. (2008) Gas-Liquid Two-Phase Flow In Inclined Pipes. Ph.D.

University of Nottingham

Hirt, C. and Nichols, B. (1981) "Volume Of Fluid (VOF) Method For The Dynamics
Of Free Boundaries". Journal of Computational Physics 39 (1), 201-225

Issa, R. and Ellul, I. (1987) "Prediction Of The Flow Of Interspersed Gas And Liquid
Phases Through Pipe Bends". Transaction of Institution of Chemical Engineers 65, 84-
96
Jayanti, S. (1990) Contribution To The Study Of Non-Axisymmetric Flows. PhD Thesis.
Imperial College London

Kaul, A. (1996) Study Of Slug Flow Characteristics And Performance Of Corrosion

Inhibitors In Multiphase Flow In Horizontal Oil And Gas Pipelines. PhD Thesis. Ohio
University, USA

49
Kim, S. and Boysan, F. (1999) "Application Of CFD To Environmental Flows".
Journal of Wind Engineering and Industrial Aerodynamics 81 (1-3), 145-158

Kvicinsky, S., Longatte, F., Kueny, J. and Avellan, F. (1999) "Free Surface Flows:
Experimental Validation Of Volume Of Fluid (VOF) Method In The Plane Wall Case".
Proceedings of the 3rd ASME, JSME Joint Fluids Engineering Conference, San
Francisco, California

Launder, B. and Spalding, D. (1974) "The Numerical Computation Of Turbulent

Flows". Computer Methods in Applied Mechanics and Engineering 3 (2), 269-289

Legius, H. and van den Akker, H. (1997) "Numerical And Experimental Analysis Of
Translational Gas-Liquid Pipe Flow Through A Vertical Bend". Proceedings of the 8th
International Conference, BHR Group, Cannes, France

Lockhart, R. W., and R. C. Martinelli (1949) "Proposed correlation of data for

isothermal two-phase, two-component flow in pipes." Chem. Eng. Prog 45.1: 39-48.

Mao, Z. and Dukler, A. (1985) "Rise Velocity Of A Taylor Bubble In A Train Of Such
Bubbles In A Flowing Liquid". Chemical Engineering Science 40 (11), 2158-2160

Mohmmed, A., Nasif, M., Al-Kayiem, H. and Al-Hashimy, Z. (2016) "Numerical Study
On Fluid Structure Interaction Of Slug Flow In A Horizontal Pipeline". AMM 819, 319-
325

Moissis, R. and Griffith, P. (1962) "Entrance Effects In A Two-Phase Slug Flow". J.

Heat Transfer 84 (1), 29

Muzaferija, S. and Peric, M. (1999) "Computation Of Free Surface Flows Using

Interface-Tracking And Interface-Capturing Methods Chap.2 In O. Mahrenholtz And
M. Markiewicz (Eds.)". Nonlinear Water Wave Interaction. Computational Mechanics
Publication, WIT Press, Southampton

50
Oshinowo, T. and Charles, M. (1974) "Vertical Two-Phase Flow Part I. Flow Pattern
Correlations". Can. J. Chem. Eng. 52 (1), 25-35

Patankar, S. and Spalding, D. (1972) "A Calculation Procedure For Heat, Mass And
Momentum Transfer In Three-Dimensional Parabolic Flows". International Journal of
Heat and Mass Transfer 15 (10), 1787-1806

Pinto, A. and Campos, J. (1996) "Coalescence Of Two Gas Slugs Rising In A Vertical
Column Of Liquid". Chemical Engineering Science 51 (1), 45-54

Pope, S. (2000) Turbulent Flow. Cambridge: Cambridge University Press

Ramos-Banderas, A., Morales, R., Sánchez-Pérez, R., García-Demedices, L. and

Solorio-Diaz, G. (2005) "Dynamics Of Two-Phase Downwards Flows In Submerged
Entry Nozzles And Its Influence On The Two-Phase Flow In The Mold". International
Journal of Multiphase Flow 31 (5), 643-665

Rhee, S., Makarov, B., Krishinan, H. and Ivanov, V. (2005) "Assessment Of The
Volume Of Fluid Method For Free-Surface Wave Flow". J Mar Sci Technol 10 (4),
173-180

Spedding, P. and Benard, E. (2007) "Gas–Liquid Two Phase Flow Through A Vertical
90° Elbow Bend". Experimental Thermal and Fluid Science 31 (7), 761-769

Spedding, P., Benard, E. and McNally, G. (2004) "Fluid Flow Through 90 Degree
Bends". Development of Chemical Engineering Mineral Process 12, 107-128

Taha, T. and Cui, Z. (2006) "CFD Modelling Of Slug Flow In Vertical Tubes".
Chemical Engineering Science 61 (2), 676-687

Ubbink, O. (1997) Numerical Prediction Of Two Fluid Systems With Sharp Interfaces.
PhD thesis. University of London

51
Van Hout, R., Barnea, D. and Shemer, L. (2002) "Translational Velocities Of Elongated
Bubbles In Continuous Slug Flow". International Journal of Multiphase Flow 28 (8),
1333-1350

Versteeg, H. and Malalasekera, W. (2007) An Introduction To Computational Fluid

Dynamics. Harlow, England: Pearson Education Ltd.

Wallis, G. (1969) One-Dimensional Two-Phase Flow. New York: McGraw-Hill

White, E. and Beardmore, R. (1962) "The Velocity Of Rise Of Single Cylindrical Air
Bubbles Through Liquids Contained In Vertical Tubes". Chemical Engineering Science
17 (5), 351-361

52
NOMENCLATURE

53
54
 55

View publication stats