Professional Documents
Culture Documents
For oil wells, the main component of pressure loss is the gravity or hydrostatic term.
Calculation of the hydrostatic pressure loss requires knowledge of the proportion of the
pipe occupied by liquid (holdup) and the densities of the liquid and gas phases. Accurate
modeling of fluid PVT properties is essential to obtain in-situ gas/liquid proportions, phase
densities and viscosities.
Calculation of holdup is complicated by the phenomenon of gas/liquid slip. Gas, being less
dense than liquid flows with a greater vertical velocity than liquid. The difference in
velocity between the gas and liquid is termed the slip velocity. The effect of slip is to
increase the mixture density and hence the gravity pressure gradient.
Multi-phase flow correlations are used to predict the liquid holdup and frictional pressure
gradient. Correlations in common use consider liquid/gas interactions - the oil and water
are lumped together as one equivalent fluid. They are therefore more correctly termed 2-
phase flow correlations. Depending on the particular correlation, flow regimes are
identified and specialised holdup and friction gradient calculations are applied for each
flow regime.
As yet, no single correlation performs better than others for all flow conditions.
Fancher Brown is a no-slip hold-up correlation that is provided for use as a quality
control. It gives the lowest possible value of VLP since it neglects
gas/liquid slip it should always predict a pressure which is less than
than the measured value. Even if it gives a good match to measured
downhole pressures, Fancher Brown should not be used for
quantitative work. Measured data falling to the left of Fancher
Brown on the correlation comparison plot indicates a problem with
fluid density (i.e PVT) or field pressure data.
Hagedorn Brown performs well in oil wells for slug flow at moderate to high
production rates (well loading is poorly predicted). Hagedorn Brown
should not be used for condensates and whenever mist flow is the
main flow regime. It under predicts VLP at low rates and should not
be used for predicting minimum stable rates.
Duns and Ros usually performs well in mist flow cases and should be used in high
GOR oil and condensate wells. It tends to over-predict VLP in oil
wells. Despite this, the minimum stable rate indicated by the
minimum of the VLP curve is often a good estimate.
Duns and Ros is the original published method, without the enhancements applied in
Original the primary Duns and Ros correlation. The primary Duns and Ros
correlation in PROSPER has been enhanced and optimised for use
with condensates.
Petroleum correlation combines the best features of existing correlations. It uses
Experts the Gould et al flow map and the Hagedorn Brown correlation in slug
flow, and Duns and Ros for mist flow. In the transition regime, a
combination of slug and mist results are used.
Petroleum includes the features of the PE correlation plus original work on
Experts 2 predicting low-rate VLPs and well stability.
Petroleum includes the features of the PE2 correlation plus original work for
Experts 3 viscous, volatile and foamy oils.
Petroleum is an advanced mechanistic model suitable for any angled wells
Experts 4 (including downhill flow) suitable for any fluid (including Retrograde
Condensate). Especially good for pipeline pressure drop calculations
and instability calculations (detecting the conditions at which
instability will occur).
Petroleum The PE5 mechanistic correlation is an advancement on the PE4
Experts 5 mechanistic correlation. PE4 showed some instabilities (just like other
mechanistic models) that limited its use accross the board. PE5 reduces
the instabilities through a calculation that does not use flow regime
maps as a starting point.
PE5 is capable of modelling any fluid type over any well or pipe
trajectory. This correlation accounts for fluid density changes for
incline and decline trajectories.
The stability of the well can also be verified with the use of PE5 when
calculating the gradient traverse, allowing for liquid loading, slug
frequency, etc. to be modelled.
Orkiszewski correlation often gives a good match to measured data. However, its
formulation includes a discontinuity in its calculation method. The
discontinuity can cause instability during the pressure matching
process, therefore we do not encourage its use.
Beggs and Brill is primarily a pipeline correlation. It generally over-predicts pressure
drops in vertical and deviated wells.
Gray correlation gives good results in gas wells for condensate ratios up to
around 50 bbl/MMscf and high produced water ratios. Gray contains
its own internal PVT model which over-rides PROSPER's normal
PVT calculations. For very high liquid dropout wells, use a
Retrograde Condensate PVT and the Duns and Ros correlation.
Barnea Criteria
Slugging Calculations
Note
There is no universal rule for selecting the best flow correlation for a given application. It
is recommended that a Correlation Comparison always be carried out. By inspecting the
predicted flow regimes and pressure results, the User can select the correlation that best
models the physical situation.
The energy associated with a unit mass of fluid entering the element plus any shaft work
done by the fluid plus any heat exchanged with the surroundings plus change of energy
with time must equal the energy leaving the pipe element. The energy balance can be
expressed as:
The energy per unit mass entering the element is the sum of the internal and kinetic energy
terms at that point. This must equal the sum of the internal and kinetic leaving the element
plus the heat loss (or gain) from the element plus any work done by the fluid.
d(gCosL) = gCos.dL
For MultiPhase flow of oil, water and gas, the energy balance is derived in a similar
manner.
The energy per unit mass leaving the pipe element is the sum of internal and kinetic energy
at point L+L:
Heat exchange to or from the pipe is proportional to the temperature difference between the
well fluids and the surroundings, the heat transfer area and the overall heat transfer
coefficient Uc:
The rate of work done by the system is the change in potential energy plus the change of
enthalpy (h = U + /). Defining the mass flow rate in the pipe as:
Summing the enthalpies for all phases, this can be simplified to:
The pressure traverse for a well or pipeline can be computed by numerical integration of
the pressure gradient equation over the length of the pipe. This requires knowledge of the
proportion of each phase present and its density for the pressure and temperature of each
incremental pipe segment. The densities and gas/liquid proportions are found from a PVT
model. To find the total pressure gradient requires the above equation to be simultaneously
solved for pressure and temperature.
If the flowing temperature profile is available, then the equation can be evaluated directly
for pressure. In the majority of applications, the flowing temperature has a minor effect on
the calculated pressure drop. If the flowing temperature profile is available, the pressure
gradient can be evaluated directly.
Gravity
The gravity term reflects the fluid density and the angle of deviation from vertical.
Evaluation of the fluid density is straightforward for single-phase applications. For
MultiPhase flow, the mixture density is required. This requires knowledge of the density of
each phase and the proportion of the pipe occupied by each (holdup).
Friction
The frictional pressure gradient is proportional to the friction factor f and the square of the
flow velocity. The frictional pressure gradient comprises a greater proportion of the total
pressure drop as the flow velocity increases.
Acceleration
The magnitude of the acceleration component of the total pressure gradient is insignificant
except when the produced fluid is rapidly changing in velocity e.g. in a gas well operating
with low well head pressure.
In a properly designed oil well completion, the gravity component comprises around 75%
of the total pressure gradient. The accuracy of liquid density and holdup values are the
main factors affecting the accuracy of the total pressure gradient calculations.
Multi-Phase Flow
The density difference between gas and either water and oil is far greater than the density
difference between oil and water. So-called multi-phase flow correlations lump oil and
water together as liquid and calculations are based on liquid/gas interactions. Such flow
correlations are more accurately described as 2-phase methods.
The calculation errors resulting from lumping the water and oil together have been found to
be insignificant for the majority of oil well pressure calculations. Errors have been
observed for large diameter tubing in deviated wells.
Holdup
Holdup is the proportion of the pipe cross-section occupied by liquid as shown in the
following sketch:
Since the sum of the area occupied by liquid plus that occupied by gas must equal the total
pipe cross-section, the following relation must always be obeyed:
The mixture density is calculated from the phase densities and the liquid holdup.
The liquid density is that of the mixture of oil and water.
Estimation of the liquid holdup (and hence the mixture density) for a wide range of flow
conditions is a main goal of flow correlations.
Superficial Velocity
The superficial velocity is the velocity of each phase if it were flowing alone in the pipe.
For the prevailing pressure and temperature, the superficial velocities are calculated from
the volumetric flow rate and pipe cross-sectional area as follows:
Defining the mixture velocity as the sum of the phase superficial velocities (Vm = Vsupl +
Vsupg), the 'no-slip' holdup can be calculated:
Slip
The density difference between gas and liquid results in a buoyancy force between them.
In inclined flow, the gas will rise up through the liquid. This increases the gas flow
velocity relative to the liquid. Since the gas is travelling faster than the oil, a smaller area
of the pipe will be required to produce the same mass flow of gas. The effect of slip is
therefore to increase the liquid holdup, and hence the mixture density.
The amount of gas/liquid slip depends on factors such as density difference, the relative
proportions of oil and gas, flow velocities and pipe diameter. A major factor is the flow
regime.
At low flow rates, the increase in mixture density due to gas/liquid slip can be significant.
As rates increase, the flowing velocity increases and the liquid 'catches up' with the gas.
At very high flow rates, slip is insignificant since both phases are travelling with virtually
the same velocity. On the other hand, the frictional loss increases rapidly with flow
velocity.
The total gradient is the sum of gravity, friction and acceleration terms. The flowing
bottom hole pressure displays a minimum when the rate of change of the gravity gradient
equals the rate of change of friction gradient.
The rate corresponding to the minimum flowing bottom hole pressure is termed the
minimum stable rate. Rates less than this are termed 'head dominated'. Higher rates are
termed 'friction dominated'. Wells cannot sustain steady state flow for rates less than the
minimum stable rate. The unstable production is termed 'heading' where alternate slugs of
oil and gas are produced. While some wells may be able to continue unstable production
for rates below the minimum stable rate, VLP correlations cannot be used, since they apply
only to steady state flow conditions.
Flow Regimes
Observation of flow loop experiments has identified characteristic patterns that occur under
varying conditions of multi-phase flow. Different flow patterns occur depending on many
factors including the proportion of phases present, flow velocity, viscosities and interfacial
tension. Most methods use 'flow maps' to identify flow regimes. A flow map is a
presentation of flow regime as a function of dimensionless liquid and gas velocity
parameters. The correlating parameters vary between authors.
Frictional losses and liquid holdup can vary widely between flow regimes. Most pressure
loss correlations require accurate flow regime prediction so that the appropriate calculation
methods can be applied.
The major flow regimes encountered in producing wells are discussed below:
An undersaturated reservoir producing above bubble point pressure will have single phase
oil production at the sand face. As the oil is produced towards surface, the wellbore
pressure will decline in response to the reduction in hydrostatic head and frictional losses
until the bubble point pressure is reached.
Bubble Flow
When the bubble point is reached, small evenly dispersed gas bubbles within the liquid will
form. As both the liquid and gas expand, the flowing velocity will increase. This relatively
well ordered flow regime can be modelled with a good degree of precision.
Slug Flow
As the pressure reduced further, more gas is evolved. The gas bubbles interact and coalesce
to form slugs of gas that can occupy most of the pipe cross-section. Production is by
alternate slugs of liquid and gas. This moderately chaotic flow regime can be modelled
with acceptable precision.
Annular Mist
As the pressure decreases, both the volumetric proportion and the flow velocity of the gas
increases greatly. A liquid film forms on the walls of the pipe and gas production occurs
through the central pipe core. Some fine liquid droplets can be entrained in the gas as a
mist.
For conditions that fall between the main flow regimes described above, transitional flow
regimes may be observed. The onset of highly chaotic flow regimes can be difficult to
predict. Pressure drop correlations generally handle transition flow regimes by averaging
results calculated using the adjacent fully developed flow regime methods.
Although many methods have been proposed, the prediction of flow regimes is imprecise
and remains the subject of on-going research.
PVT
Pressure loss correlations require values for the in-situ gas/liquid ratio, flow velocities, fluid
densities and viscosity for the entire range of temperatures and pressures in the calculations.
The accuracy of pressure gradient calculations critically depends on the accuracy of fluid
density values used.
The following example (in field units) illustrates how surface flow rates and PVT data are
used to calculate densities and velocities at downhole conditions.
Densities
If the surface densities of oil, water and gas are o ,w and g. at standard conditions, the
corresponding densities at downhole conditions are:
Velocities
Superficial velocities are calculated from the surface oil production at standard conditions,
solution GOR and the water cut. Defining:
and Ap = d2 / 4
These basic parameters at downhole conditions, together with gas and liquid viscosities are
passed to the flow correlation for calculation of the pressure gradient terms.
Black oil and compositional methods may be used to predict PVT properties. For oil
wells, the liquid density is the most important PVT parameter. Prediction of liquid
density is a weakness of compositional PVT methods. The use of compositional PVT
cannot be recommended for the majority of oil well pressure loss calculations.
Historical Development
Methods of modeling vertical two-phase flow can be divided into 3 classes in order of
complexity:
1. Liquid holdup and friction losses are expressed an equivalent single phase energy
loss factor. The flowing mixture density is used in place of the in-situ mixture density to
calculate the gravity component. No attempt is made to identify flow regimes.
2. The effect of two-phase holdup and friction losses are evaluated separately. No
attempt is made to identify flow regimes.
3. Flow regimes are identified and correlations specialised to flow regimes are used to
estimate holdup and friction losses.
The first experimental work in the field of multi-phase flow is generally credited to Davis
and Weidner. Most pressure loss correlations have been developed from laboratory
experimental work with air-water mixtures in short lengths of small diameter tubing. Thus
the effects of large scale inter-phase mass transfer are not included. The results attempt to
correlate liquid holdup and tubing wall friction with measured fluid properties and flowing
phase fraction. Table 1 shows those correlations that have made some contribution to the
problem of vertical MultiPhase flow.
Poettmann and Carpenter developed a simple practical model, but its simplicity limits its
accuracy. They related the irreversible energy losses of 49 well tests with a Fanning type
friction factor term which was in turn related to the Reynolds number of the fluid mixture.
This correlation is typical of a Category 1 method as defined above. Baxendall extended
the Poettmann and Carpenter method to higher flow rates and Fancher and Brown applied
the same method to 94 tests from an experimental well . Fancher and Brown introduced
produced gas-liquid ratio as an additional parameter in the friction factor correlation.
Much of the early work in developing both flow maps and correlations have been attributed
to Griffith and Wallis and Duns and Ros. They concentrated on the definition of the flow
boundaries of the flow regimes and presented the results in the form of a flow pattern map.
Ros and also Duns and Ros gathered laboratory data on pressure drop for two-phase flow
and they considered the dependency of the flow regimes on dimensionless parameters.
Hagedorn and Brown developed a correlation from 475 tests in a 1,500 foot experimental
well using fluids with viscosities up to 110 centipoise. An average mixture density
corrected for downhole conditions was used for calculating friction and acceleration
pressure losses. Liquid holdup was not measured directly - it was calculated from the total
measured pressure loss. Further work by Brill and Hagedorn has improved the correlation
prediction of holdup and slip and included the Griffith bubble flow correlation.
Orkiszewski developed a pressure drop prediction method based on a new flow pattern
map and a combination of features from existing correlations. He combined the work of
Griffith for bubble flow and that of Griffith and Wallis for slug flow together with the Duns
and Ros correlation for mist flow. In addition, new friction and density correlations for
slug flow based on a liquid distribution parameter were developed. The data of Hagedorn
was used to develop a correlation with tubing size, superficial mixture velocity and liquid
viscosity.
Aziz et al developed a mechanistically based model and concentrated their research on the
bubble and slug flow regimes. Predictions for 48 wells were compared to field data and
with the predictions from Orkiszewski, Duns and Ros and Hagedorn and Brown with
favourable results.
Beggs and Brill developed a correlation for prediction of pressure gradient and holdup
from experiments carried out in 1 inch and 1 1/2 inch smooth circular pipes at any angle of
inclination. They found that in inclined flow, the liquid holdup reaches a maximum about
50º from horizontal and a minimum at around -50º from horizontal. The result that the
holdup is approximately equal at 90 and 20 confirms the observation that vertical flow
correlations can often be successfully used in horizontal flow. The Beggs and Brill
correlation is the first method to explicitly account for the effect of deviation on holdup.
Cornish developed a simple homogeneous flow model which assumes no slippage occurs
in high flow rate wells. However for two-phase Reynolds numbers <105, slip will be
significant and the method may not apply.
VLP Correlations
The primary purpose of a flow correlation is to estimate the liquid holdup (and hence the
flowing mixture density) and the frictional pressure gradient. This section details the
methods used by some of the most widely applied correlations.
Hagedorn Brown
The Hagedorn-Brown correlation is probably the most widely applied of all oil well VLP
correlations. It works well for bubble and slug flow regimes in a wide range of
applications. At low flow rates it under-predicts flowing pressures. This can result in
optimistic predictions for minimum stable flow rates.
The Hagedorn-Brown correlation was developed from data obtained from a 1500 ft test
well using fluids with a wide range of viscosities. The liquid holdup was not measured, but
was back-calculated to satisfy the measured pressure gradient after the pressure drop due to
friction and acceleration were accounted for. The liquid holdup is not a true indicator of the
proportion of the pipe occupied by fluid, it is merely a correlating parameter.
Hagedorn and Brown found the liquid holdup could be correlated to four dimensionless
parameters:
The above expressions are dimensionless when the parameters are expressed in the
appropriate field units as shown below:
The liquid comprises both oil and water, therefore the issue of calculating mixture
properties arises. In the Hagedorn-Brown correlation a weighted average approach is used:
L = O fO + W fW
L = O fO + W fW
L = O fO + W fW
Where the fractional flow of oil and water are defined as follows:
fO = qO/(qO+qw)
fW = qW/(qO+qw) = 1 - fO
The correlating function is entered with a value of CNL. The corrected liquid number is
read from a plot of CNL vs NL:
The secondary holdup correction factor is determined from a correlation of NGV, NL and
Nd. Once the correction factors have been determined, the holdup can be calculated. The
pressure gradient due to elevation change (gravity term) is calculated from:
where:
The friction gradient can be written in terms of the mass flow rate (w) as follows:
Where:
The two-phase friction factor is correlated with a two-phase Reynolds number using the
standard Moody diagram. The two-phase Reynolds number is defined as follows:
where:
The refinements suggested by Brill and Hagedorn have been implemented in PROSPER:
Some additional refinements have been added to the basic Hagedorn-Brown correlation:
Field experience has shown that Hagedorn Brown gives excellent results for oil wells in
bubble and slug flow away from the gravity dominated flow region. Errors are greatest for
large bore deviated wells in the 35-70º range with moderate water cuts where water/oil slip
may be significant.
Ls = 50 + 36 NLV
Lm = 75 + 84 NLV0.75
Duns and Ros developed a dimensionless slip velocity correlation from which the actual
slip velocity and liquid holdup can be calculated using the following relations:
where:
Solving for liquid holdup yields:
1. Calculate the dimensionless slip velocity S using the appropriate correlation. The
correlation for S is different for each flow regime.
The Bubble Flow regime is defined by a Gas Velocity number falling between zero and an
upper limit (0 <= Ngv <= L1 + L2NLV).
The bubble slip velocity numbers F1 F2 F3 and F4 are correlated with the liquid viscosity
number NL. F3' is obtained from:
The Duns and Ros friction term for bubble flow is given by:
From experimental data, Duns and Ros obtained the following expression for fm:
f1 is obtained from the Moody diagram as a function of the liquid Reynolds number:
Note that for low values of Reynolds number corresponding to laminar flow conditions that
the friction factor becomes independent of pipe roughness.
The factor f3 is an additional correction for both liquid viscosity and in-situ gas-liquid ratio.
It becomes important for viscosities greater than approximately 50 centistokes.
The acceleration term is insignificant for the bubble flow regime and is therefore not
calculated.
For the slug flow regime ( L1 + L2 * Nlv < Ngv < ( 50 + 36nlv )) the dimensionless slip
velocity is calculated as follows:
The slug slip velocity numbers F5, F6 and F7 are found from a plot as a function of liquid
viscosity number NL:
The friction pressure gradient is calculated using the same procedure as for bubble flow.
For the Mist flow regime ( Ngv > ( 75 + 84 Nlv0.75 )), the slip velocity is taken as zero.
This is because with the high gas flow rates, the liquid and gas travel with essentially the
same velocity. With no slip, the mixture density can be calculated directly from:
In the mist flow regime, the friction term is based on the gas phase only:
The friction factor f is read from the Moody diagram as a function of the gas Reynolds
number:
In mist flow, there is a film of liquid on the pipe wall. The ripples of the wall film cause a
drag on the gas. This process is governed by a form of the Weber number:
Liquid viscosity also has an influence which is accounted for by making Nwe a function of
a dimensionless number containing the liquid viscosity:
The value of pipe roughness may be very small, but /d never becomes smaller than the
value for the pipe itself. At the transition to slug flow, /d approaches 0.5. Between these
limits, /d can be calculated from the following equations:
Values of f for the mist flow regime can be found for /d > 0.05 from:
As the wave height on the walls increases, the actual area available for flow of gas is
reduced to d-. Duns and Ros suggested that the prediction of friction loss could be refined
by substitution of (d-) for d and
for Vsg throughout the calculation of friction gradient. In this case, the determination of
roughness is iterative.
For the region (Ls < Ngv < Lm) linear interpolation of the total pressure gradients is used to
determine the total pressure gradient. This means that when Ngv falls between Ls and Lm,
pressure gradients must be calculated using both slug flow and mist flow correlations as
follows:
where:
The Duns and Ros correlation has been found to perform better in mist flow than most
others. It is particularly useful for condensate wells. Although the accuracy of pressure
gradient predictions in slug flow is generally inferior to Hagedorn-Brown, prediction of
minimum stable flow rates using the minimum value of the Duns and Ros flowing bottom
hole pressure is generally accurate.
In PROSPER additional refinements have been made to the basic Duns and Ros method:
Gray
Gray is a gas and gas condensate multi-phase correlation used in the API 14B Subsurface
Safety Valve sizing program. The Gray correlation contains an internal Black Oil PVT
model. The internal PVT routines override the primary PROSPER PVT model in use.
This correlation gives good results in gas wells for condensate ratios up to around 50
bbl/MMscf and high produced water ratios. For very high liquid dropout wells, use a
Retrograde Condensate PVT and the Duns and Ros correlation.
In condensate wells, the liquid holdup is small or zero at the sand face, and increases as the
pressure reduces towards the surface. Depending on the condensate PVT properties, the
liquid holdup can reduce to zero as the pressure is further reduced. This is in contrast to oil
wells where the gas fraction always increases towards the surface.
The effect of holdup on the friction gradient can be adequately expressed as a change in
pipe wall roughness rather than rely on a Moody friction factor. The Gray correlation
utilises the following parameters:
where:
m = Mixture density
fg = (1 - ec) / (R + 1)
where:
C = -2.314 AB
A = Nv (1 + 205.0 / Nd)
The Katz correlations for oil and water surface tensions are as follows:
where:
T = Temperature (ºR)
The Colebrook-White equation is used to calculate the friction factor with roughness
evaluated as follows:
The Aziz et al correlation recognises four flow regimes. The Duns and Ros method is used
for mist flow and new correlations are presented for bubble and slug flow. The slug-mist
transition is handled using the Duns and Ros interpolation method. The Aziz et al
correlation can be classed as a 'mechanistic' model. This tag is applied since some aspects
of the holdup are calculated in a manner that assigns a physical significance to parameters
such as the bubble rise velocity. These parameters are then used to enter empirical
correlations as in other methods.
N2 = 8.6 + 3.8 NY
N3 = 70 (100 NY)-0.152
where:
Bubble Flow
For the bubble flow regime, (NX < N1) liquid holdup is given by:
where:
and
Slug Flow
The slug flow regime is defined by (N1 < NX < N2 for NY < 4; N1 < NX < 2.65 for NY >
4). In the slug flow region, the liquid holdup is given by:
NV m
>250 10
<=18 25
f is obtained from the Moody diagram using the same Reynolds number as for Bubble flow.
The transition regime is defined by (N2 < NX < N3 and NV < 4; transition flow does not
exist for NY > 4). Pressure gradients are calculated using both slug and mist flow
correlations. Linear interpolation is used to determine the total pressure gradient:
where:
The mist flow regime is defined by (NX > N3 and NY < 4; NX > 26.5 and NY > 4). In
mist flow, the Duns and Ros technique is used.
The Beggs and Brill correlation can be used in both horizontal and inclined flow. It was
developed from experimental data in 1 inch and 1 1/2 inch acrylic pipe which could be
inclined at any angle. Air and water were the fluids used. Liquid and gas rates were varied
to enable all flow patterns to be observed with the pipe horizontal.
With a flow rate set up, the pipe inclination was varied so that the effect of angle on holdup
could be observed. Holdup correlations were developed for each of three horizontal flow
regimes. The liquid holdup is first calculated as if the pipe were horizontal and then
corrected for pipe inclination. Beggs and Brill found that the holdup was a maximum at
approximately +50º from the horizontal and a minimum at approximately -50º.
Beggs and Brill modified their flow map from that originally published to include a
transition zone between the segregated and intermittent flow regimes (see below). The
following dimensionless parameters are used to identify the flow regime that would exist if
the pipe were horizontal.
L1 = 316 L0.302
L2 = 0.0009252 L-2.4682
L3 = 0.5 L-6.738
Segregated flow
Transition flow
Intermittent flow
or
Distributed flow
or
L >= 0.4 and NFR > L4
When the flow falls in the transition regime, the liquid holdup is calculated using both the
segregated and intermittent expressions and linearly interpolated using the following
weighting factors:
B = 1-A
The same equations are used to calculate liquid holdup for all flow regimes. The
coefficients and exponents used in the equations are changed for each flow regime. Liquid
holdup is given by:
where hL() is the holdup which would exist for the same flow conditions in a horizontal
pipe. The equivalent horizontal holdup is given by:
where a, b and c are taken from the following table according to flow regime:
Flow Pattern a b c
The effect of pipe inclination is accounted for using the parameter , the value of which is
defined as follows:
where is the actual angle of the pipe from horizontal. For vertical upward flow, = 90º and
becomes:
= 1 + 0.3 C
where:
where d, e, f and g are determined for each flow regime from the following table:
where:
n=nL+gg
The no-slip friction factor is determined from the smooth pipe curve on the Moody diagram
or calculated using:
using the following Reynolds number:
where:
The ratio of the two phase to no-slip friction factor is calculated from:
where:
and
S becomes unbounded at a point in the interval 1 < y < 1.2. In this region, S is calculated
using:
S = ln ( 2.2 y - 1.2)
Although the acceleration pressure gradient is small, it is included for increased accuracy.
where
The approach of including the acceleration term as an overall correction factor to the total
gradient is convenient and sufficiently accurate when the acceleration term is small. To
improve accuracy when acceleration is large, PROSPER calculates the acceleration term
explicitly and adds it to the Gravity and Friction terms to find the total pressure gradient.
Orkizewski
Orkizewski conducted an extensive study of existing correlations, comparing field data and
calculated results. The Orkizewski correlation combines the Griffith and Wallis method for
bubble flow with a new correlation for slug flow and the Duns and Ros method for mist
flow. The data of Hagedorn and Brown was used as the basis for the slug flow correlation.
Bubble Flow
Vsg / Vm < LB
where:
The value of the bubble slip velocity Vs is taken to be constant at 0.8 ft/sec.
The friction factor f, is read from the Moody diagram using a Reynolds number defined as:
Slug Flow
The slug flow regime is defined by: Vsg / Vm > LB, and Ngv Ls. For slug flow, the two
phase density is given by:
where
and
Vb = C1 C2 (gd)1/2
where:
Vb can be calculated using the following expressions:
where
Since Vb is a function of NReb and NReb is in turn a function of Vb, an iterative solution is
required for Vb.
The value of is calculated using different expressions depending on the mixture velocity
and the continuous liquid phase as shown in the following table:
water <10 1
water >10 2
oil <10 3
oil >10 4
Orkizewski did not define criteria for determining whether oil or water is the continuous
phase. In a water / oil emulsion, water will generally be the continuous phase above a water
cut of approximately 75%.
These constraints are designed to eliminate pressure discontinuities between flow regimes.
However, significant discontinuities still occur at Vm of 10 ft/sec. This can cause
significant problems, especially in large diameter pipes. Although Orkizewski can give
excellent results in many wells, the use of Orkizewski is discouraged due to the danger of
encountering a pressure discontinuity during pressure matching and VLP calculations.
where f is taken from the Moody diagram using the Reynolds number:
The acceleration term is considered to be negligible in the slug flow regime.
Transition Flow
The transition flow regime is defined by Lm > Ngv < Ls. The total pressure gradient is
found by linear interpolation between the slug and mist flow boundaries using the
interpolation scheme of Duns and Ros.
Mist Flow
The mist flow regime is defined by Ngv > Lm. The method of Duns and Ros is used for
mist flow.
Pipeline Correlations
Dukler
The Dukler correlation was based on similarity analysis. The friction factor and liquid
holdup correlations were developed from field data. Dukler's friction factor is given by:
where:
where:
y = -ln (L)
An iterative procedure is required to find the liquid holdup. The holdup is a function of
liquid fraction and Reynolds number. However, the Reynolds number is also a function of
the holdup. NRek is plotted vs hL and L.
or
The Dukler Flanigan correlation calculates the friction gradient using Dukler's technique
and applies Flanigan's inclined flow method.
Flanigan developed a correlation from data taken on a 16 inch diameter pipeline. Liquid
holdup in the uphill sections of the pipeline were correlated with superficial gas velocity.
Pressure recovery in downhill sections is ignored. The pressure drop due to the hills is
calculated from:
where Z is the sum of the vertical heights of all the hills. The holdup factor hL is a function
of the superficial gas velocity and is determined from:
The Dukler-Flanigan total pressure gradient is found by summing the Dukler friction
gradient and the Flanigan elevation pressure gradient then applying an acceleration
correction.
Dukler Flanigan has been found to give good results in a wide range of conditions.