14 Multiphase Flow
14 Multiphase Flow
Contents
1 INTRODUCTION .......................................................................................................... 3
2 FUNDAMENTAL DEFINITIONS ...................................................................................... 4
2.1 HOMOGENEOUS .............................................................................................................. 4
2.2 PRESSURE DROP .............................................................................................................. 5
2.3 KEY DIMENSIONLESS GROUPS ............................................................................................ 6
2.3.1 Reynolds number ................................................................................................. 6
2.3.2 Froude Number .................................................................................................... 6
2.4 HETEROGENEOUS ............................................................................................................. 6
3 FLOW REGIMES AND FLOW PATTERNS ........................................................................ 8
3.1 HORIZONTAL PIPELINES ..................................................................................................... 8
3.2 VERTICAL PIPELINES ........................................................................................................ 10
4 DEFINITIONS USED IN TWO PHASE GAS-LIQUID FLOW ............................................... 13
4.1 QUALITY ....................................................................................................................... 13
4.2 VOID FRACTION ............................................................................................................. 13
4.3 TWO PHASE DENSITY ...................................................................................................... 13
4.4 TWO PHASE VISCOSITY .................................................................................................... 14
4.5 INTERFACIAL TENSION ..................................................................................................... 14
4.6 VELOCITY ...................................................................................................................... 14
5 FRICTIONAL PRESSURE DROP IN TWO PHASE GAS-LIQUID FLOW ................................ 16
5.1 REVIEW OF METHODS ..................................................................................................... 16
5.2 HOMOGENEOUS FLOW – DARCY EQUATION ....................................................................... 19
5.3 LOCKHART AND MARTINELLI METHOD ............................................................................... 20
5.3.1 Chisholm’s Method ............................................................................................ 22
5.4 CHAWLA METHOD FOR ANNULAR AND STRATIFIED FLOW ..................................................... 23
5.5 WALLIS EQUATION FOR SMOOTH PIPES ............................................................................. 24
5.6 SUN AND MISHIAM’S METHOD ........................................................................................ 24
5.7 BEGGS AND BRILL HOMOGENEOUS FLUID FLOW METHOD .................................................... 24
5.7.1 Extension to Non-Isothermal Flow..................................................................... 33
6 EROSIONAL VELOCITY ............................................................................................... 34
7 TWO PHASE LIQUID-LIQUID FLOW............................................................................. 36
7.1 PRESSURE DROP FOR SLUG, ANNULAR AND ANNULAR DROPLET FLOWS .................................. 38
7.2 ANNULAR PLUG FLOW .................................................................................................... 41
8 REFERENCES ............................................................................................................. 45
1 Introduction
This topic covers the flow of a fluid that has two distinct phases gas/liquid. Most of engineering
fluid mechanics is devoted to flow of homogeneous fluids where we use mixture properties
and averaged velocities. But with two phase flow we need to consider the flow patterns inside
the pipeline more than before. We may also have to bring in our thermodynamics more than
before especially if the fluids inside the pipeline are on the saturation point and changes in
volumes of gas or liquid are caused by changes in pressure and temperature.
For the oil and gas industry, two phase gas-liquid flows are part of everyday operations,
especially those operations taking fluid from the well into the well bore and through production
tubing to the processing facility on the surface.
Many of the theories and governing equations were however developed for the power
generation industry, where water and steam are used to generate work.
2 Fundamental Definitions
Defining terms and understanding what these terms mean is quite important especially when
you realise that there are many different ways to write down the mathematics for fluid flow.
In dealing with multiphase flow, we are faced with equations that usually relate to one phase
or the other phase, or the entire region where all phases are flowing. This section covers basic
definitions of terms that are particular to pressure drop and for fluids in general.
2.1 Homogeneous
This is a volume of fluid where one phase is completely mixed into another and we cannot point
to a distinct region where one phase exists on its own. The classical understanding of this phase
is that we have one phase “dissolved” in the other.
If we take a volume of fluid, then the physical properties of the fluid volume are a true mixture
property. Properties such as the density, viscosity, specific heat capacity are mixture properties
that sometimes can be derived as functions of the molar composition of the individual
components in the mixture.
n
Density : rmix = åw r i i
Mixture density is the weight (mass) fraction x
i=1
pure component density
n
Specific Heat Capacity : cp = åw i c p Mass specific heat capacity (eg kJ/kgK) is the mass
mix i
i=1
fraction x pure component specific heat capacity. Remember that
heat capacity can be expressed as a molar term as well and the mole
fraction is used in that case.
å (m x )
n
i i
Mwti
mmix = i=1
å( x )
n
i
Mwti
i=1
Where the molar mass (Mwt) of each component is included.
For many situations, we instinctively look at mixtures as homogeneous and use our standard
engineering tools in all aspects of our work. This is not wrong but there are often consequences
in making an assumption that a mixture is homogeneous or not. In all our fluid mechanics
studies so far, we have been dealing with homogeneous mixtures.
Note of caution – as you progress through this course, you may find alternative methods to
work out mixture properties, especially when we look at two phase gas/liquid flows.
The static pressure drop is due to the elevation change in the pipeline and would normally be
given by:
DPstatic = rghsin q ()
where hsin (q ) will be the vertical distance relative from the start of the pipeline
DPkinetic =
(u 2
2
- u12 ) is the change in momentum for fluid at one end of the pipe to the other.
2g
You may recall that when dealing with pumps, that we sometimes forget to deal with this term
properly. This term must be included if we are transferring fluid from say a reservoir where
there is a pool of liquid into a pipeline. Normally, we would be pumping from a relatively smaller
storage tank into a pipeline and we can often neglect the kinetic change.
The main reason for looking at the kinetic pressure drop term is to deal with situations where
the velocity inside the pipe changes. In a two phase pipeline where we have flow of a saturated
gas and liquid (that is a liquid which is thermodynamically in equilibrium with the gas) due to
the drop in pressure with friction, the you can get liquid flashing off generating more vapour.
This increases the volume of vapour, which will be significantly larger than the liquid it came
from. The result is a rapid increase in the vapour velocity and depending on the flow pattern a
rise in the liquid velocity inside the pipe. What this means is that there could be a change in the
velocity and hence the kinetic pressure loss.
For a moving fluid where gravity exerts influence, Froude number is defined as:
u
Fr =
gd
The Froude number normally appears in flows which are in open channels, rivers and for ship
design and therefore anywhere which we have a “free surface flow”. Free surfaces are the
interfaces between liquid and gas or between two immiscible liquids.
2.4 Heterogeneous
This will be the normal description for a multiphase mixture in this chapter. In this mixture
there are discrete parts of the volume of fluid that are one type, and other parts which are
another type. If we take a sample in a beaker, then it is likely the phases making up this mixture
will separate out into distinct layers provided the density difference between these phases
allows it and the chemistry between the phases.
The physical properties in this case cannot be combined – we are after all going to analyse each
phase which makes up this mixture. However, there are properties which are used to describe
the mixture, especially for two phase mixtures.
There are two, two-phase mixtures that as petroleum engineers, we are likely to encounter:
Gas/Liquid Mixtures – where we have a gas phase and liquid phase together.
Oil/Water – these are two liquids, one of which may form small droplets that are
dispersed into the other. Or we could have a layer of water below a layer of lighter
oil.
No matter what type of system (gas/liquid or liquid/liquid) we are looking at, we have a set of
common definitions:
Flow Pattern –these are simply the picture of the distribution of one phase relative to
others in our pipeline, or vessel. We often refer to profile inside pipelines but there are
equivalent for mixtures in storage tanks, reservoirs.
Slip Velocity – we can imagine a situation of a spherical mass falling through a column
of liquid. The presence of the sphere means that we have a two phase mixture. What
is apparent even from our simple model is that there will be two velocities, one for the
sphere and the other for the liquid. The difference between these two is the slip
velocity.
Residence Time – the time spent by one phase inside a vessel or pipeline is called the
residence time or retention time. For a liquid flowing into and out of a vessel, the
theoretical residence time is defined as Volume of Liquid/Volumetric Flowrate. For
multiphase flow mixtures, different phases can have different residence times
depending on what the slip velocity is.
Hold-Up – the term Hold-Up is similar to residence time in that it describes the quantity
of one phase held back inside the pipeline for example.
This section covers the basic patterns of two phase gas/liquid flow that we normally expect see
in pipelines
The transition between each of these flow patterns is not exact and depends on the ability of
the observer as to what they decide the regime is. Another issue is that a complete flow map
will need a good range of vapour and liquid flowrates. Workers in this area often compile a
composite flow map, plotting the vapour flux against the liquid flux as per the example below.
Figure 2 : Flow map for air/water mixture in a horizontal 2.5cm pipe at 28°C, 1 bar
(after Baker 1954)
Figure 2 is a fairly standard flow map which plots the “modified gas mass flux” against the
“modified liquid mass flux”.
A “flux” is simply the flowrate per unit cross sectional area but in this case, the word “modified”
means that you have to do something special to the standard flux terms.
where:
1/3
æ r r ö
1/2
s æ m ær ö
2
ö
l=ç g l
÷ Y = water ç l ç water ÷ ÷
è rair rwater ø s çè mwater è rl ø ÷ø
You can see that the reference fluids for this plot are water and air. Most of the experiments
are done with air and water and then corrected for the actual fluids used. But be careful that
these maps may not account for situations where the fluid is in a state of changing
thermodynamics e.g flashing flow.
You might question why the term “velocity” is not used instead of the word “flux”. In theory
the two are the same. The main difference is that the velocity is the “average velocity” and not
a vector property. Coming from the process engineering fluid flow point of view, you should be
familiar with the term “velocity” to mean the average velocity of a fluid.
The horizontal and vertical flow patterns so far have been for fully developed flow where the
volume fraction of gas remains constant. In some situations which occur where vapour is
“flashed” off as a result of a drop in pressure or increase in temperature, then the flow pattern
along the tube will change.
For an oil and gas platform, the change in riser elevation will cause a change to the two phase
flow pattern.
For example, the next two images show the changes occurring inside an evaporator tube where
the flow regime changes due to vaporization of the liquid within the pipe.
http://www.thermalfluidscentral.org/encyclopedia/index.php/Flow_Regimes_in_Horizontal_and_Vertica
l_Tubes
Before launching into any analysis, it is worth spending some time looking at some definitions.
Remember that the definitions here are for a mixture of two different phases, and not different
chemical components. We defined these earlier.
Before we start, it is worth spending a few minutes thinking about these properties and what
they mean. We are going to be using these definitions to discuss two phase flow of a
compressible gas and incompressible liquid. So how do we deal with the fact that gas properties
will change depending on the actual pressure and temperature? The short answer is we cannot.
We can try our best to estimate what the pressure drop will be or more correctly, what
conditions we need to work out the density of the gas. We may need to iterate our calculations
once we have an idea what changes are going to take place.
4.1 Quality
Quality – is defined as the vapour content in terms of mass to the total mass. In other
words, quality is the vapour mass fraction:
Gg
x=
Gl + Gg
Vv
ò dV
Defined as a = = Vv
Vv +Vl ò dV
V
Vv +Vl
Surface tension is a term usually reserved fro the force needs to produce a stable interface
between a liquid and gas. In reality it should be “interfacial tension” to signify that this force is
between two fluids.
There is no “mixture” term per say, but it is a concept that we need to remember.
The other feature of surface tension is that when surfaces are created, it takes energy. So when
a bubble of air in water passes through an orifice plate and it breaks into smaller bubbles, any
additional surface that is created takes energy. This energy must contribute to the loss in
pressure through the orifice.
4.6 Velocity
There are several different velocity values depending which fluid we are talking about. The
reason for the number is down to how velocity is defined when you can have one fluid travelling
along a pipeline, with another fluid in the same pipeline that travels at a different speed. On
average, the only definable quantity that we have is mass flow so how do you get a velocity
from that?
Ql Qg
wl = , wg =
Al Ag
Note that this is the volumetric flowrate for the area occupied by the phase in question.
By analysis of the definition of void fraction, superficial velocity can also be given from:
ul = wl (1- a ) , ug = w ga
Slip Velocity
As indicated earlier, gas and liquid may travel at different velocities from each other;
the density and viscosity are different so it is likely there is more frictional pressure
drop associated with liquid than gas. We define the slip velocity to be:
wg 1- a ug
S= ,S=
wl a ul
Some of the very early work in two phase flow targeted the heat transfer sector where there
was a need to accurately predict the pressure drop in steam generation plants. Water boiling
inside tubes in boilers creates a two phase flow mixture and so it was necessary to come up
with methods to work out pressure losses. Similarly, when steam condenses, a two phase gas-
liquid mixture is created and again there is a need to know what the pressure losses will be.
As a result, much of the work into two phase flow modelling comes from the heat transfer
sector of engineering.
Industry-University partnerships seem to give, or at least have the primary goal, of developing
more robust understanding of problems. The University of Tulsa is one example and the models
coming from that research should be looked at.
In their review article, Shippen & Bailey (Bailey and Shippen 2012) discussed the evolution of
methods to predict multiphase flow with a focus on the oil and gas industry. One of the
interesting comments relates to the development of modelling and predictive capabilities and
how the level of physics has increased:
Single Phase – is the basic flow equations which apply well to single phase fluids. These
are all homogeneous flow type models which use mixture properties. For multiphase
flow, we treat the velocity with a mean value even when one of the phases may have
a higher velocity than another. These models are sometimes called “no slip” models.
Empirical A-B – where models rely on experimental data. Majority of these models will
be based on air/water flows under laboratory conditions and valid for a limited range
of pipe diameters In category A, the no slip condition is modified by looking at the hold
up of liquid and gas inside the pipeline. What these models do is to modify the friction
factor obtained from some homogenous mixture model. Empirical B uses the liquid
hold up to directly work out the slip velocity. Here an attempt is made to look at each
phase on its own
Mechanistic – With access to computer modelling, we can now approach the governing
equations in more detail combining fluids with phase changes. Computer modules
don't always mean tools like CFD.
Most of the models today fit with the 10% rule, that they are accurate to within +/- 10% for
holdup and pressure gradients when applied with velocity and fluid properties that give:
Flow type – is from stable flows to just below erosion limits.
Inclination angles ate between +10 to -10° from the horizontal or from the vertical
position
Liquid hold up is more than 10%
Water/oil hold ups are less than 10%
Internal pipe diameters are less than 10 inches
Oil viscosities are less than 10 cP
At the time of writing, the CAPD software tools such as AspenTech HYSYS (version 9.0) offer
the user a range of different models. Only by making sure that the situation you are modelling
fits with a particular model can you stand a reasonable chance of being as accurate as you can.
With two phase flow modelling, there are going to be differences and it is down to the engineer
which values to take to the next step in any design or analysis.
A list of the models available and under what conditions they are valid are shown in Table 1. In
the table, the reference to “liquid hold up” and “flow map” signifies that some attempt is made
to refer to the actual mass of liquid and gas occupying the pipeline and to the type of flow
pattern. You should realise that especially for frictional loss along pipe walls, the overall loss
will depend which fluid is in contact with the pipe wall at any point.
In the next few sections, we will look at a few of these methods, especially the ones where
hand calculation is possible.
The key property in homogenous flow is the “quality” or “void fraction”. Both of these were
given previously. Using this, you can find the density and viscosity
rmix = (1- a ) rl + arg
mmix = (1- x) ml + xmg
The next quantity that you need is the velocity or flowrate in order to work out a Reynolds
number. For the velocity, homogeneous two-phase flow equations normally use the mass
velocity value. Mass velocity if you remember is
Mass Flowrate G
M= =
Cross Sectional Area Atotal
The Reynolds number is used and the problem here is the mixture velocity. Often equations
will replace the velocity with the mass velocity:
Md
Remix =
mmix
2
where “M” is the mass velocity (kg/m s)
Exercise 9.1
Using the homogenous flow pressure drop model, calculate the frictional pressure drop in a
horizontal tube of 10mm internal diameter which is 2m long. Assume the flow is adiabatic and
that vapour and liquid velocities are the same value. Mass flowrate is 0.02 kg/s, vapour quality
is 0.05. At the pipeline conditions the liquid density is 1518 kg/m3, vapour density 2.6 kg/m3,
liquid viscosity 0.0005856 kg/ms, vapour viscosity 0.0000126 kg/ms.
Exercise 9.2
Using the homogeneous flow pressure drop method, calculate the two phase pressure drop for
upflow in a vertical tube of 10mm diameter that is 2m long. The flow is adiabatic, the mass flow
is 0.02 kg/s and the vapour quality is 0.05. The fluid liquid density is 1518 kg/m3, vapour density
is 2.6 kg/m3. Liquid viscosity is 5.856x10-4 kg/ms, vapour viscosity is 1.26x10-5 kg/ms.
Note – this is the same problem as previously, but now we are asked for the pressure drop not
the frictional pressure drop.
Lockhart and Martinelli measured pressure drops using air with a range of liquids including
water, benzene, kerosene, diesel and oils in pipelines raging from 1.5mm to 25mm in diameter.
Some later workers claim that the equations and data produced do not extend well to larger
pipe diameters and can over estimate the pressure drop by several factors ( up to 250%).
The pressure drop along a pipeline carrying the two phases, is given by:
æ DPf ö æ
2 DPf
ö
ç ÷ = fL ç ÷
è L ø2 p è L øL
And
æ DPf ö æ
2 DPf
ö
ç ÷ = fV ç ÷
è L ø2 p è L øV
æ DPf ö
where ç ÷ is the pressure drop along the pipeline assuming that the vapour occupies the
è L øV
æ DPf ö
entire cross sectional area. Likewise, ç ÷ is the pressure drop for the liquid.
è L øL
The terms fv and fL are the “drag ratios which are functions of physical properties (density,
viscosity) and mass flux terms. Lockhart & Martinelli produced a plot in order to find f but
several authors have correlated this.
Note that Lockart and Martinelli decided that the transition region between turbulent and
laminar flow is from Re from 1000 to 2000. For single phase flow, the standard practice is to
assume a transition from abut Re=2000 to Re=3000.
For the “drag ratio” term, Lockhart and Martinelli defined different expressions depending if
the flow was bubbly, slug etc:
( )
fG = 4 -12ln ( X ) + 28 ln ( X )
2
Dispersed bubbly flow:
= 4 -15ln ( X ) + 26 (ln ( X ))
2
fG
Elongated bubble flow:
(
fG = 2 + 4.5ln ( X ) + 3.6 ln ( X ) )
2
Smooth surface stratified flow:
(
fG = 3 +1.65ln ( X ) + 0.45 ln ( X ) )
2
Stratified wavy flow:
(
fG = 4 + 2.5ln ( X ) + 0.5 ln ( X ) )
2
Annular mist flow:
The method can be represented graphically to find the liquid and gas “drag ratio” terms.
(Note that the “X” here is not the “x” that is the quality number), X is defined as:
DPL
X=
DPG
L
and just for completeness, DPi = 4f ri ui2 , where “I” refers to a phase i=L for liquid, i=G for
d
vapour phase.
Warning Note: For heat exchanger applications such as flow of two phases in reboilers,
æ 1- x ö æ rG ö æ mL ö
0.9 0.5 0.1
However, to use this expression for non-heat transfer applications we have to make sure that
any assumptions apply to our particular case. Using the X defined as the ratio of the two single
phase pressure drops is the normal one for two phase flow.
“d” is therefore the diameter of the entire pipe or the equivalent diameter of the
duct/section.
In Chisholm’s definitions, he uses the terms laminar and turbulent. These are, as is normal in
fluid mechanics, defined by Reynolds numbers.
Exercise 9.3
Gas and liquid flows along a 3 inch diameter horizontal pipeline with the gas velocity at 10 m/s,
and liquid velocity 2 m/s. Properties of each phase are given below. Calculate the following:
a) The quality value and the void fraction
b) The mixture density and viscosity
c) The frictional pressure drop for the gas and liquid phase if they flowed along the pipe
separately
d) The two phase flow frictional pressure drop – using the Chisholm method.
Liquid density is 1200 kg/m2, viscosity 2.4 cP. Gas density is 60 kg/m3, viscosity 0.5 cP
1 1 1
where = +
z z z 23
3 3
1
B=ç ÷ç ÷ çç g ÷÷ çç g ÷÷
è x ø è Rel Frl ø è rl ø è ml ø
Gtp2 (1- x )
2
Frl =
gD rl2
l = liquid
g = gas
x = quality
D = pipe diameter
k = absolute roughness
Csm 1
fl2 = 1+ +
X 1.19
X2
æ Re ö æ 1- x ö0.5
0.4
Csm = 1.79 çç g ÷÷ ç ÷
è Rel ø è x ø
Example
For the example, we will calculate the two phase pressure drop for 11.84 m3/hr water and 127.6
m3/hr of air flowing in a 76.2mm diameter pipeline, 10m long. The end of the pipe is 1m above
the start. We will assume an average pressure of 200 bar and temperature 20°C along this
pipeline and that the properties and fraction of water and air do not change along the 10m
length.
Step 1
Collect fluid conditions and properties, at the average pipeline conditions. Remember that gas
properties in particular will depend on the pipeline pressure and temperature (see the
comment at the start of this section).
One comment to make here is the role of “surface tension”. More correctly we should be using
the term “interfacial tension”. By its very nature, we are dealing with a system that has discrete
droplets of water-in-air, bubbles of air-in-water and a surface interface between the two
phases. We may correctly assume that surface tension must play a part in the formation and
disruption of new surfaces and therefore feature in the analysis.
Step 2
Liquid Hold up assuming a horizontal line
We work out the liquid and gas superficial velocities, the hold up of liquid and then the mixture
velocity.
Step 3
Mixture Froude Number and Flow Pattern
The parameters L1 to L4, and the value of Fr are then used to determine the type of flow pattern
within the pipeline.
They also classed a 4th type “Transitional” as something between Segregated and Intermittent.
lL ³ 0.01and Fr < L2
Transitional lL ³ 0.01and L2 £ Fr £ L3
Intermittent 0.01£ lL < 0.4 and
L3 < Fr £ L1 or
lL ³ 0.4 and L3 < Fr £ L4
Distributed lL ³ 0.4 and L3 < Fr < L4 or
lL ³ 0.4 and Fr > L4
As you can see, the decision matrix is usable but not straightforward.
Step 4
Liquid Hold Up
The no-slip liquid holdup lL plays a key part in finding the two-phase pressure drop and will
depend on the type of flow pattern. Beggs and Bill modifies the hold up calculated in step 2 as:
C1l
C2
HL0 = L
C3
Fr
For non-transitional flow, C1..C3 are terms found in:
C1 C2 C3
Segregated 0.98 0.4546 0.0868
Intermittent 0.845 0.5351 0.0173
Distributed 1.065 0.5824 0.0609
If the flow is transitional, that is somewhere between segregated and intermittent, then we
calculate two hold up terms – HLSegregated and HLTransitional . We then work out how far from
one flow type we are by:
L3 - Fr
A=
L3 - L2
The corrected hold up is then:
HLCorrection = ( A) HLSegregated + (1- A) HLTransitional
One of the benefits of the Beggs and Brill methods, is to include inclination angles for pipes.
The liquid hold up calculated in step 3 and modified in step 4 is for horizontal pipes. We now
correct that figure, if needed, for pipes in an incline or decline.
C4 C5 C6 C7
Segregated Uphill 0.011 -3.768 3.539 -1.614
Intermittent Uphill 2.96 0.3050 -0.4473 0.0978
Downhill (all types) 4.7 -0.3692 0.214 -0.5056
Distributed Uphill yc = 0 , and y = 1
The mixture properties are based on the no-slip hold up factor and are defined as:
The Reynolds number and the friction factor, for smooth bore pipe is
The two phase friction factor is then obtained from: ftp = eSfn , where
ln ( y )
S=
-0.0523 + 3.182ln ( y ) - 0.8725ln ( y ) + 0.-1853ln ( y )
2 4
lL
and y =
HL2
(
when 1< y < 1.2 , then S = ln 2.2y -1.2 )
Out MathCAD solution is then:
The pressure drop is therefore about 10.1 kPa, for this example.
Note the Beggs and Brill 1973 paper makes several things clear that are not so apparent in other
papers:
1. it clear that some terms related to the change in density of the gas with height are so
small compared to the liquid that they can be neglected.
2. Gas compressibility can be neglected.
3. That the friction loss is significantly affected by liquid hold up and so an accurate
calculation of this is essential.
4. The accuracy of the liquid hold up can be improved by referring to actual flow patterns
rather than the 3 generalised ones (segregated, dispersed etc) that the method uses.
Such a scheme is very common in solving calculations where some estimation is needed to start
the procedure off. The trick of course is to ensure the scheme has a proper termination point
and is not divergent or worse, oscillatory.
6 Erosional Velocity
With gas and liquid, the presence of small droplets of liquid in high speed flows, and on a similar
argument the presence of small gas bubbles, can cause considerable erosion on pipe fittings,
bends etc. It is well known that solid particles in liquid can cause erosion but so can these fluid
droplets.
Usually, the velocity at which erosion can become unsustainable is given by a moment type
expression using the mixture density and mixture velocity:
rmum2 = 15000
where the mixture density is in kg/m3, and the mixture velocity is in m/s
API RP 14E is the American Petroleum Institute recommended practice document for piping
used in offshore oil and gas operations. It contains recommendations on topics such as pressure
drop calculation methods and velocity limits. It also states methods to find the maximum
allowable velocity in gas and in two phase gas-liquid flow lines, to avoid excessive erosion.
Erosion of a pipeline is enhanced in the presence of corrosive materials such as CO2, H2S.
The velocity of a stream at above which erosion is likely to occur can be obtained from:
C
Vc =
rm
where C is a constant, the velocity Vc is in ft/sec, and the density at the flowing conditions in
lb/ft3
Calculation of maximum velocities are relatively straight forward if you are following the API
recommendation.
Exercise 9.4
A gas pipeline 12 inches in diameter is considered for a new gas system the gas has a molar
mass of 34 kg/kmol. The inlet pressure of the pipeline is 20 barg, and the discharge pressure
has been calculated as 5 barg. Take the compressibility factor for the gas to be 0.8
a) Calculate the density of the gas at inlet and outlet conditions. Assume the temperature along
the pipeline remains constant at 20°C
b) Calculate the maximum possible velocity at both inlet and outlet conditions
c) Explain why there are two different numbers for part b
Oglesby (1979) documented the flow patterns in oil and water pipelines, and characterized the
different patterns in a flow map
Semi-Segregated – some
mixing at the interface B
Semi-Mixed – segregated
flow of a dispersion of the C K
free phase
Figure 12: Flow pattern map for two phase liquid-liquid flow
With two liquids, if one liquid is in excess of the other in terms of volume, then a “phase
inversion” can occur where the dispersion changes from a dispersion of say the oil phase in
water, to a mixture where the water is dispersed in the oil.
For pressure drop, then we are in more of a difficult area than with gas-liquid flow. For
segregated flow (as in the figures above), we can easily appreciate that there will be a different
drag term on the upper pipe wall for the lighter phase than on the bottom pipe wall for the
heavier liquid phase. We have also to consider the wetting characteristics of the pipe wall itself.
For dispersed flow then we have a more challenging problem to solve.
7.1 Pressure drop for Slug, Annular and Annular Droplet Flows
This analysis comes from Hooman and Kawaji (2011) who looked at pressure drops in small
capillary tubes. Although the application was directed towards non-petroleum applications, it
is still worth reviewing the work as it combines what is known about two-phase gas-liquid flow
with oil-water flows.
Figure 13: Flow patterns for viscous oil-water flow in a micro channel flow; variation in oil/water
flowrates.
The images in the figure were generated by initially filling the pipe with oil and pumping water
through a very low flowrates, typically from 2 to 2000 micro litres/min. As the oil and water
flowrate increases, droplets of water-in-oil are seen to become longer so that the flow pattern
changes from droplet flow to larger droplet flow and then annular flow. Since the flow is
laminar there will be a velocity profile from the pipe wall to the pipe centre line which has to
cross the interface between the two liquids. The presence of different viscosities will result in
different drag.
In two phase liquid flows, there are several important dimensionless groups that we can use to
define the flow patterns, pressure drops etc.
r ouoD r uD
Reynolds number for the oil Reo = and water Re w = w w
mo mw
mouo m u
Capillary number for oil Cao = and for water Ca w = w w
s ow s ow
The capillary number is a ratio of the viscous forces to the interfacial forces.
rouo2D r u2 D
Weber number for oil Weo = and for water We w = w w
s ow s ow
The weber number is a ratio of the inertial forces to the interfacial forces.
Where the density, viscosity of each phase should be apparent. The term s ow is the interfacial
tension between the oil and water phases. Note that the interfacial tension is similar to the
“surface tension” although we would normally use the term “surface tension” to be the
interfacial tension between the liquid and a gas about the liquid.
The velocity of course is defined by the flowrate for each phase that is calculated for the entire
diameter of the pipe.
Qo Qw
uo = and uw =
p p
D2 D2
4 4
There is another dimensionless group, called the BOND number that indicates the impact of
gravity on the flow pattern.
DrgD2
Bo =
s ow
If the Bond number is small, this means that gravity has a smaller effect than the interfacial
tension so that any changes in elevation of the pipeline can be ignored.
In their work, Hooman and Kawaji found that the pressure drop along the pipe where two
phase flow occurs follows a straight line, which suggests that the pressure drop follows a similar
equation as the Hagen-Poiseuille equation for single phase flow:
æ DP ö 32mouo æ DP ö 32mwuw
çè L ÷ø = D2 and çè L ÷ø = D2
o w
The overall pressure drop is assumed to be the summation of the two single phase pressure
drops, using some constants that depend on the flow pattern.
æ DP ö æ DP ö æ DP ö
çè L ÷ø = C1 çè L ÷ø + C2 çè L ÷ø
ow w o
The simple additive nature of the pressure drop suggests that a function like that from Lockhart
and Martinelli can be used:
æ DP ö 2 æ DP ö æ DP ö 2 æ DP ö
çè L ÷ø = f wç
and = f
ow
è L ÷ø w çè L ÷ø
ow
o ç
è L ÷ø o
C1
fw2 = C1 + C2 X2 and fo2 = + C2
X2
The Lockhart-Martinelli parameter X is defined as:
æ DP ö
çè L ÷ø
X =
2 o
æ DP ö
çè L ÷ø
w
Constants C1 and C2 were defiend in the table above and have been validated for X2 values
from 20 to 140.
His analysis follows a different path and is looked at here to indicate that there can be still
debate over the exact mechanism for predicting pressure drops in two phase liquid-liquid
systems.
Consider a pipe segment of overall diameter D, where we have oil and water flow along this
pipe.
The apparent velocities of each phase (that is taking the entire single phase flow down the pipe)
are given by:
Qw Qo
u0w = for the water and u0o = for the oil
p p
D 2
D 2
4 4
The average velocity of the mixture is then:
Qw +Qo
um = = u0w + u0o
p
D 2
4
So the average velocity is the SUM of the two individual velocities, and not the arithmetic
average of these two.
If the oil plug is of length Lp and the diameter is d, then the volume of the plug is:
p d 2Lp
Vp =
4
There is a volume of water that essentially travels along with the plug, and surrounds the plug
but also extends to the next oil plug in the pipeline. The volume of this water is given by:
p 2 2
Vw =
4
(
D - d Lp + D 2DL )
Next we have to find the velocity of the oil plug itself. We know what the volumetric flowrate
of oil into the pipeline is, and now we apply continuity for the additional distance from one plug
to the other:
Qo Lp + DL
up =
p 2 Lp
d
4
The average velocity of the water in the annulus surrounding the plug comes from:
uw =
(
Qw Lp + DL )
p
4
((D 2
)
- d 2 Lp + D 2DL )
d
If we define the ratio of plug diameter to pipe diameter as: d=
D
Then we get for the mixture velocity:
Lp æ Lp ö
um = upd 2 + uw çç1- d 2 ÷
÷
Lp + DL è Lp
+ DL ø
If we have a situation where the velocity profile in the water follows the Blasius equation: then
the equation for velocity at any radial position r is given by:
æ rö
1/7
ur = umax ç1- ÷
è Rø
where r is the radial distance from the centre line of the pipe
R is the radius of the pipe
æ r ö
1/7
ç 1- ÷
ur = up ç R ÷
çç 1- d ÷÷
è ø
The cross sectional average velocity of the mixture is then found by integrating the velocity
profile:
æ7 ö
um = up ç d (1- d ) + (1- d ) + d 2 ÷
49 2
è4 60 ø
As you will recall, frictional loss occurs at the interface between the fluid and a solid surface. It
also must occur at the interface between two fluids of different viscosity where the fluids are
also immiscible. The Blasius equation does not take into account the laminar sub layer at the
oil-water interface and so we will need to correct for this somehow. We can by introducing a
velocity term v p* which is defined as:
tp
up* = 12
rw
where t p is the shear stress on the interface between the oil plug and the surrounding water.
Boobok then uses these expressions to define the pressure drop for the two phase mixture as:
æ æ u ö ö
ç 0.82 ç 1- d 2 ps ÷ ÷
è um ø
DPow = DPw ç ÷
çæ7 ö 1÷
( 49
) ( ) ( )
2
ç ç 1- d d + 1- d + d 2 ÷ 1- d 7
÷
çè è 4 60 ø ÷ø
ups
For a specific ratio of , we can plot the two phase pressure drop as a function of the plug
um
ups
diameter to the pipe diameter. In figure Y below, plotted for =0.4 shows that as the plug
um
diameter increases, the pressure drop becomes smaller than that of pure water in the pipeline
but only until the drop size is approximately 85% of the pipe diameter then it increases very
quickly.
If we think about what is happening inside the pipeline, this makes some sense especially the
section where the drop size increases to become equal to the pipe diameter. Remember the
viscous oil will have a larger single-phase pressure drop than water and so we would expect the
pressure drop to eventually increase.
From our discussion on shear thinning fluids, you should appreciate that adding a second phase
can sometimes assist in causing the other to flow. If we had plug flow in our pipeline, we could
allow the oil drop diameter to increase and reduce our pumping requirements by some 10-
20%, provided we have plug flow. If we however allow the plugs to become too large, we would
very quickly loose that advantage and we could be in a situation where our pumps would not
cope.
This example, and the one on shear thinning fluids shows that dealing with two phase flow
introduces a range of benefits and difficulties.
8 References
“An integrated approach to Modelling PipelineHydraulics in a Gathering and Production
System.” Aspen Technology, 2015.
Bailey, Willian J, and Mack Shippen. “Steady-State Multiphase Flow - Past, Present and Future
with a Perspective on Flow Assurance.” Energy & Fuels (American Chemical Society), June 2012.
Beggs, H D, and J P Brill. “A Study of Two Phase Flow in Inclined Pipes.” Journal of Petroleum
Technology (SPE), May 1973: 607-617.
Teixeira, R G.D, A R Secchi, and E C Biscaia. “Two-phase Flow in Pipes: Numerical Improvements
and Qualitative Analysis for a Refining Process.” Oil & Gas Science & Technology (IFP Energies
Nouvelles) 70 (2015).