You are on page 1of 16

SPE-175056-MS

A CFD Compositional Wellbore Model for Thermal Reservoir Simulators


A. Shahamiri, M. Heidari, W. L. Buchanan, and L. X. Nghiem, Computer Modelling Group Ltd.

Copyright 2015, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Annual Technical Conference and Exhibition held in Houston, Texas, USA, 28 –30 September 2015.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
A robust, fast, and accurate wellbore model is essential for a reliable reservoir simulation. There has been
significant interest to improve the wellbore models as the operations and behavior of advanced wells have
become more complicated, in particular for thermal Steam-Assisted Gravity Drainage (SAGD) process.
The non-linear and dynamic nature of the processes involved, as well as the necessity for prediction and
control of such wells under rigorous operating conditions, makes the development of such a model
complicated. The majority of current models use simplifying assumptions, especially with respect to the
fluid dynamics involved. This work uses CFD techniques to solve the complete form of Navier-Stokes
equation along with the mass and energy conservation equations. The objective is to develop a robust and
efficient CFD numerical procedure for flow in wellbores where the slippage between phases, phase
change, and radial mass/heat transfer between the wellbore and reservoir, are taken into account. A finite
volume approach is employed to discretize the conservative form of the balance equations using a
staggered grid arrangement. A Drift Flux model estimates the physical phase velocities, and the modified
Beggs and Brill pressure-drop model is used to estimate the frictional pressure term in the momentum
equation. Both of these models are valid for all multiphase flow regimes, at upward and downward flow
conditions with different inclination angles from -90˚ to 90˚. The phase fractions, component mole
fractions, and temperature distribution are obtained using isenthalpic flash calculations. The system of
equations is solved implicitly using Newton’s method. The model was validated for a wide range of
operating conditions against available data for different single and multi-phase flow conditions. The
results of CFD wellbore model is also compared with the current thermal wellbore model FlexWell of
CMG’s STARS™ reservoir simulator.

Introduction
Advanced wells including horizontal, undulated, multi-lateral and smart (involving flow control devices)
wells are used in modern reservoirs in order to improve the performance and the production rate. A large
volume of detailed information from the wellbore behavior is required in order to design the wells, as well
as to control and optimize their operation. Therefore, an advanced wellbore model, capable of accurately
modeling the detailed local flow parameters in the wellbore, is necessary. Complicated transport phe-
nomena can take place in the fluid flow of an advanced well, e.g., presence of multiple multiphase flow
regimes, different mechanisms of heat transfer between the fluid and formation, and mass transfer through
2 SPE-175056-MS

the completions. The fluid flow in the wellbore becomes even more dynamic if these phenomena are
accompanied by short time scale processes, such as phase change processes or sudden changes to the well
conditions. Generally, the wellbore operation is strongly related to the distributions of phase velocities,
pressure, temperature, phase fractions, and phase compositions.
Phase velocities significantly influence the rate of transports in the axial direction of the wellbore, and
several methods are used to estimate these velocities. In a multi-fluid multiphase model, a momentum
equation is solved for each phase in order to determine the velocity of the phase. The exchange of
momentum and forces between the phases must be known to bring closure to the system of momentum
equations. The available relationships for estimating these forces are limited to specific flow regimes and
usually introduce a highly nonlinear behavior to the system of equations. The multi-fluid technique was
used by Bonizzi et al. (2003) to solve the slug flow in a pipe. Their isothermal black oil model used a
sequential CFD procedure to solve the system of equations explicitly. Early use of the multi-fluid method
for a fully compositional wellbore model was performed by Stone et al. (1989) and Stone et al. (2002).
Their model included the frictional pressure drop and accelerational terms in the momentum equation, as
well as non-isothermal effects via the energy balance equation. They used empirical correlations proposed
by Winterfeld et al. (1989) to calculate the interphase forces. It was concluded that the nonlinear nature
of the interphase forces in the momentum equations, and the discontinuity of the holdup calculated from
the flow regime maps used, led to convergence difficulties, and many time cuts were required to tackle
these problems. Pourafshary et al. (2009) also used the same method to develop a fully implicit thermal
compositional model for multiphase flow in wellbores. They used a more continuous flow regime map,
developed by Ansari et al. (1994), to add to the model robustness.
In addition to the multi-fluid model, the phase velocities can be estimated using the Drift Flux
approach. The concept of Drift Flux model was proposed by Zuber et al. (1965). In this model, the
slippage between the phases is related to two underlying physics of the flow: first, non-uniform velocity
distribution across the flow that tends to carry more amount of gas through the core of the flow, away from
the walls; second, the effect of buoyancy forces. These two effects are integrated into the empirical
coefficients of the Drift Flux models. Initially the model correlations were tuned differently for each flow
regime. Hibiki et al. (2003) followed this approach and developed a model for vertical upward two-phase
flows. Shi et al. (2005) developed a unified set of Drift Flux correlations for co-current upward multiphase
flow regimes at different inclination angles. These correlations were used by Soprano et al. (2010) for
developing an isothermal black oil multiphase flow model in wellbores. They solved one momentum
equation to determine the mixture velocity and estimated the physical phase velocities using the Drift Flux
correlations. Livescu et al. (2010) used the same Drift Flux correlations to develop a fully coupled thermal
black oil model for a coupled wellbore reservoir simulator. Recently, a unified set of correlations, for
upward and downward concurrent multiphase flows with -90° to 90° inclination angles, was developed by
Bhagwat et al. (2014). The interest for using the unified Drift Flux models is because they provide
continuous and differentiable relations for the entire multiphase flow regime map. The unified models
have not been extended to counter-current flow applications yet, although the available concurrent models
have been used successfully as an approximation for counter-current flow applications (Livescu et al.
(2010)).
The estimation of frictional pressure drop is very important for the accuracy of a wellbore model,
especially for horizontal or near horizontal wells where the effect of hydrostatic pressure gradient is
negligible. Mechanistic models are known to calculate the pressure drop accurately because they use a
flow regime map to identify the correct multiphase flow regime and take a pressure drop relation
accordingly. The pressure drop model developed by Taitel & Dukler (1976) is one of the initial
Mechanistic models for horizontal and near horizontal upward and downward multiphase flows. Their
model was used by Stone et al. (1989) to calculate the pressure drop in their fully coupled thermal
wellbore model. Pourafshary et al. (2009) compared the Mechanistic models developed by Ansari et al.
SPE-175056-MS 3

(1994), Petalas & Aziz (1996) and Hasan & Kabir (1999) in a thermal compositional wellbore/reservoir
simulator. The mechanistic model of Ansari et al. (1994) was found to be the most efficient and accurate
model, followed by that of Petalas & Aziz (1996) and Hasan & Kabir (1999). Usually, the interpolation
techniques are required for making the flow regime transition regions continuous in the Mechanistic
models and this adds to the computations. In addition to the Mechanistic models, the empirical models can
be used for estimation of pressure drop (Beggs & Brill (1967), Aziz & Govier (1972), etc.) The available
empirical correlations have been extensively studied, and Beggs & Brill’s correlation has been frequently
reported as a reliable correlation for two and three phase multiphase flows as studied by Ahmed et al.
(2014) and Spedding et al. (2006). This model is valid for concurrent upward and downward multiphase
flows. Siu et al. (1991) developed a fully coupled thermal wellbore model and used Beggs & Brill
correlations to calculate the pressure drop and holdup. They were able to reproduce highly transient field
data of steam injection wells.
The majority of the multiphase wellbore models use the black oil approach (Soprano et al. (2010),
Livescu et al. (2010)). In this approach, three distinct phases represent the entire behavior of the
components involved. The significance of using a compositional approach depends on the complexity of
the phase behavior involved. It has been shown that neglecting compositional effects results in erroneous
pressure and temperature distributions in the wellbore, Pourafshary et al. (2009). Thus in the present study
a compositional model has been used for modeling multiphase flow.
The objective of this work is to develop a thermal compositional model that uses complete forms of
mass, energy and momentum equations, along with a Drift Flux modeling technique for the calculation
of physical phase velocities. The phase behavior of each component is tracked via isenthalpic flash
calculations. The model has been verified extensively against available experimental and numerical data,
and is designed to be coupled with the CMG’s STARS™ reservoir simulator. This paper describes the
main assumptions, methodologies and some of the verification steps towards developing the CFD model.

Methodology
Conservation equations
The coupled system of equations consists of material, energy, and momentum balance equations that
are solved fully implicitly. The material balance equations are solved for all the components present in the
mixture,
(1)

For the phase j, Sj is the phase saturation, xij is the molar fraction of component i, vj is the physical
velocity and q̇j is the inflow or outflow rate through the completions. Different forms of energy equation
have been used in the literature to satisfy the system energy balance. Equation (2) is obtained by
subtracting the momentum equation from the original form of energy equation. In this form, the work due
to gravity and frictional forces, as well as the kinetic energy of the flow are accounted for, although these
terms do not appear explicitly in the equation terms,
(2)
4 SPE-175056-MS

It should be noted that in this form of energy equation, the terms on the left involve the derivatives of
internal energy uj rather than enthalpy. The term Pⵜ · (冱j Sj vj) implicitly accounts for the work of
pressure, gravity, friction forces. Heat loss in radial direction, axial thermal conduction and the energy
transport by the flow through the perforations also appear in this equation. This energy equation enables
the model to capture non-isothermal effects such as Joule-Thompson. With respect to the balance of
momentum, equation (3), a complete form of momentum equation including transient and accelerational
terms (terms on the right), gravity force, friction force, and the momentum change due to the flow rate
from/to the reservoir, (the last term on the right) are taken into account.
(3)

The above system of equations is solved for pressure, global mole fractions, mixture enthalpy, and
volumetric mixture velocity. Pressure, global mole fractions, and mixture enthalpy are used to determine
the phase mole fractions, component mole fractions in the phases, and the temperature via isenthalpic flash
calculations. The phase velocities are also calculated from the Drift Flux model.

Multiphase flow model


The Mechanistic model of Petalas & Aziz (2000) and the Drift Flux model of Shi et al. (2005) were
used and compared in this work. The two models predicted very close phase velocities, especially in
distributed and intermittent flow regime regions. However, the solutions of Drift Flux model were found
to be more robust, experiencing much less numerical difficulties in the flow regime transition regions.
Ultimately, a recently developed Drift Flux model Bhagwatet al. (2014) suitable for both upward and
downward flows was incorporated into the model. In a Drift Flux model, the physical velocity of the gas
phase is estimated from equation (4) and the liquid velocity can be calculated from the definition of
volumetric mixture velocity, equation (5), and the physical phase velocity.
(4)

(5)

The effect of non-uniform cross-sectional velocity profile in a wellbore cross section is included in C0.
The minimum value of this parameter is 1.0, which corresponds to a uniform velocity profile. This
parameter can take the maximum value of 1.2 to 2.0 depending on the model empirical coefficients and
their validity range. Vgm is the Drift velocity and accounts for the effects of buoyancy and phase densities.
In the present CFD model, the homogeneous flow assumption is made if the flow parameters are outside
the validity range, in order to prevent unphysical velocity values. For such cases, an interpolation scheme
is used to ensure a smooth transition from the Drift Flux and Homogenous flow models.

Heat loss
Heat loss from fluid inside the wellbore is estimated from equation (6),
(6)

where Rt is the overall thermal resistance between the moving fluid and the adjacent reservoir layer,
and is calculated as follows,
SPE-175056-MS 5

(7)

The convective thermal resistance between the moving fluid and the inner wall, RNu, as well as the
thermal conduction resistances of the casing cement and the formation, respectively shown by Rcas, Rcem,
Rfrom contribute to the overall thermal resistance. The Nusselt number, Nu, is obtained from the empirical
correlations for the pipe flow. In equation (7), rci, rco and rh are respectively inner casing, outer casing and
cement/earth interface radiuses. The effective radius, re, is obtained from an analogy between the pressure
and radial temperature distributions in the formation around the wellbore, and is taken from the relations
suggested by Peaceman (1983) for flow through perforations,
(8)

Here ⌬x and ⌬y are the reservoir grid block sizes in x and y directions, f depends on the location of
the wellbore in the reservoir grid block and CC depends on the geometry of the grid block. The equivalent
radius is used to estimate the effective thermal resistance of the formation adjacent to the wellbore and
accounts for the effect of temperature distribution between the cement/earth interface and the formation
according to Figure 1.

Figure 1—Temperature distribution of the wellbore in the radial direction

Isenthalpic flash calculations


In order to determine the temperature along the wellbore, an isenthalpic flash procedure developed by
Agarwal et al. (1991) is performed. The mixture enthalpy, pressure and global mole fractions are provided
to the isenthalpic flash equations and the phase mole fractions, Lj, component mole fractions in phases xij,
and temperature are obtained. Component mole fractions in a phase are used to relate the mole based
component properties to the corresponding phase properties. The phase saturations are calculated from
phase mole fractions as follows,
(9)

and are used to calculate the mixture properties such as volumetric mixture velocity, mixture viscosity and
mixture thermal conductivity, etc.
6 SPE-175056-MS

Numerical procedure
All the balance equations are in their conservative forms, so the mass fluxes remain conserved when
the convective terms are discretized. Accordingly, the equations are discretized using a finite volume
approach that uses upwind scheme for the flux terms, central difference for diffusive terms and backward
differencing for the temporal terms. Initially, a collocated grid arrangement was used for discretization of
the wellbore. However, this grid arrangement led to spurious oscillations in velocity and pressure profiles
especially when the gravity effects were significant. This was found to be due to the interpolation schemes
used for calculating the flow variables on the grid faces. Thus, to enforce the necessary coupling between
the pressure and velocity variables, a staggered grid arrangement was used. The staggered grid and the
arrangement of cells are shown in Figure 2. The solid-line cells show the control volumes for collocated
variables, i.e. P, Zi, H while the dotted-line cells are the control volumes for the staggered variable, i.e.
V. The number of collocated cells is less than that of staggered cells by one, as shown in Figure 2. Mass
flows through the perforations are taken into account in the collocated cells, which are aligned with
reservoir grid blocks. There are two smaller staggered cells at the end boundaries that are responsible for
momentum balance at these locations.

Figure 2—Staggered grid arrangement for primary variables

The conservation equations are discretized and solved fully implicitly using Newton’s method.
Providing Newton’s method with a suitable set of initial guesses is very critical for a successful solution.
The solution initialization method used in this work is similar to the method used by Stone et al. (2002).
The injector and producer wells are initialized differently. For an injector well, it is assumed that the well
is filled with the injection fluid while for a producer well the composition initial conditions are taken from
the adjacent reservoir grids. Both wells are assumed to be initially at thermal equilibrium with the
reservoir.
If BHP is provided, it is used to determine a hydrostatic pressure along the well and the flow rates
through completions. These flow rates are used to estimate the velocity from mass balance for each
wellbore cell. The frictional and accelerational pressure losses are neglected at this step. The BHP is
considered as total pressure, comprising dynamic and static pressures. This approach makes the applica-
tion of BHP boundary condition significantly more robust, especially when an abrupt change takes place
in the boundary conditions. If the flow rate is provided, a hydrostatic pressure distribution is obtained
using the reservoir pressure distribution and a velocity distribution is obtained by assuming mobility
weighted completion rates. For all conditions, the homogenous flow model is used for initialization of the
phase velocities in the wellbore, as the Drift Flux method is not suitable for initialization purpose
according to Stone et al. (2002).
The maximum changes of the primary variables are bounded at every Newton’s iteration, to prevent
unphysical solutions and make the convergence smoother. This becomes critical when a phase change
takes place or a well boundary condition is changed at an intermediate simulation time.
SPE-175056-MS 7

Results
Our model has been verified extensively for a wide range of conditions. Initially, the single-phase flow
model was tested using the analytical solutions of a single phase pipe flow, as well as using the FlexWell
model in CMG’s STARS™. The isothermal multiphase flow predictions were verified using the exper-
imental data conducted by Oddie et al. (2000) for multiphase flows of gas/water and gas/oil in pipes. They
used an 11 m long pipe with 15 cm diameter for this purpose and measured the holdup in the middle of
the pipe for input water cuts of 0-100% and gas flow rates of 5-100 m3/h. The same conditions were
considered in the CFD model to calculate the holdup using the Drift Flux model of Shi et al. (2003). The
pipe was discretized using 5 computational cells and the holdup at the center of the middle cell was
recorded for comparison. The results were obtained for different inclination angles, as shown in table 1.
There is a close agreement between the model predictions and the experimental values for the holdup. The
same level of agreement is obtained for gas/oil data.

Table 1—Comparison between the CFD wellbore model and experimental data Oddie et al. (2003)
No. 1 2 3 4 5 6 7 8 9

Input water cut % 9 44.5 28 78 58 14 83.5 44.5 2.2


Input gas rate m3/h 12 30 61 12 30 61 12 30 61
Inclination angle ° 0 0 0 45 45 45 70 70 70
Water hold up % , experiment 83 78 60.5 90 81 59 93.5 82 56
Water hold up % , present work 83 75.8 59.3 90.2 81.87 62.46 90.75 78.56 57.41
Difference from experiment % 0.0 2.8 2.0 -0.2 -1.1 -5.9 2.9 4.2 -2.5

Table 2 shows the predicted flow regimes using our CFD wellbore model and the Mechanistic model
developed by Petals & Aziz (2000). The multiphase flow regime in their model depends primarily on the
superficial velocities, fluid properties, flow inclination angle, and pipe diameter. Air/water flow in a
vertical pipe of 51 mm in diameter under standard conditions was simulated.

Table 2—Flow regime predicted using the CFD wellbore model and the Mechanistic model of Petalas & Aziz (2000)
8 SPE-175056-MS

All possible flow regime transitions were studied to verify the CFD model. For a given condition close
to a regime transition zone on the flow map, the superficial velocity of one phase is changed gradually
until the model predicts different flow regime. The velocity for which the transition occurs is compared
with the corresponding data reported by Petalas et al. (2000) under the same conditions. As can be seen,
the CFD model is capable of predicting the flow regimes according to the criteria suggested by the
Mechanistic model.
As a more complicated example of the non-isothermal flow with phase change, a steam injection
process was simulated using the CFD and FlexWell models. A 500 m horizontal injector well was taken
into account for this purpose. The well contains liquid water, initially at rest, at 25 °C and reservoir
pressure. Water steam at 198 °C and quality of 0.75 is injected into the well that operates based on a BHP
of 1500 kPa. A simple sink/source model was used for modeling the producer well. Figure 3 shows the
configuration of the two horizontal wells as well as the temperature propagation in the reservoir after 30,
120 and 270 days, and the reservoir properties used in the simulations are shown in Table 3.

Figure 3—Horizontal injector well


SPE-175056-MS 9

Table 3—Properties of reservoir and wellbores


Reservoir
permeability I, J, K 2200, 2200, 440 md
porosity 0.35
grid block size ⌬x, ⌬y, ⌬z 5, 5, 2 m
number of grid blocks , x, y, z 110, 11, 11
initial temperature and pressure 25 °C and 1000 kPa (top)
initial water, oil, gas saturations 0.35, 0.65, 0.00
initial time step size 0.005 day
Injector
length 500 m
inner casing, outer casing, outer cement diameters 0.16, 0.17, 0.18 m
number of grid blocks 100
well constraint Bottom hole pressure at 1500kPa
relative wall roughness 0.0001
Producer
length 500 m
diameter 0.3 m
number of grid blocks 100
well constraint Bottom hole pressure at 1400kPa
relative wall roughness 0.0001

Both models predicted that the steam front reaches the well toe after 4 minutes, and the entire reservoir
and wells reach to fully steady state condition after approximately 270 days. The corresponding results to
these two specific times are presented in Figures 4–7. The distribution of flow parameters along the well
in the transient period shows that predictions resulted from the two well models are very close. For the
steady state solutions, the pressure has a decreasing trend due to the pressure loss along the wellbore, and
the temperature is expected to have a similar trend because of the Joule Thompson effect. It is seen that
the both models predicted the same pressure profiles with slight differences closer to the heel. This
difference is due to the use of different pressure boundary conditions applied at the heel. Correspondingly,
the gas saturation distributions are also predicted to be relatively different closer to the heel as shown in
Figure 6. Another factor influencing the prediction of phase saturation is the estimation of physical phase
velocities that is not identical in the two well models. In the CFD model, the phase velocities are obtained
from the Drift Flux equations, while FlexWell uses a simplified form of phase momentum equation to
estimate these velocities.

Figure 4 —Pressure distribution along the injector well, a) 4 minutes, b) 270 days
10 SPE-175056-MS

Figure 5—Temperature distribution along the injector well, a) 4 minutes, b) 270 days

Figure 6 —Gas phase saturation distribution along the injector well, predicted by the CFD model, a) 4 minutes, b) 270 days

Figure 7—Distributions of volumetric mixture velocity and physical phase velocities, a) 4 minutes, b) 270 days

Figure 7 shows that the phase and mixture velocities are very close to the phase velocities, with a
slightly higher gas velocity due to very small Drift under these conditions. In other words, since the
Reynolds number is very high, 2.2 ⫻ 106 at the heel, and the buoyancy effect is negligible in a horizontal
flow, the flow behaves similar to a homogeneous flow.
SPE-175056-MS 11

To verify the downward multiphase flow calculations, the field data of Bleakley (1964) was used. This
field data is related to a steam injection process through the tubing inside a vertical annulus. The field data
was used by Fontanilla & Aziz (1982) for validation of their analytical model, developed for steam
injector wells. This model accounts for pressure drop, phase change, and radial heat loss from steam to
the earth. The heat loss occurs through heat convection in the tubing, heat conduction in the tubing wall,
heat radiation and convection in the annulus fluid, and heat conduction in the annulus wall and cement
layer. These heat transfer modes are incorporated into the CFD model for validation purpose. Steam with
a quality of 0.8 at 250 psi is injected through an injector well with a depth of 1600 ft. All thermophysical
and geometrical parameters are identical with those used by Fontanilla & Aziz (1982). Figure 8a shows
the steam temperature distribution, obtained from the CFD and analytical models as well as the field data
after 71 hours. The steam temperature is underestimated when the analytical model is used with Beggs &
Brill pressure drop equations, and overestimated when Aziz & Govier (1972) pressure drop equations are
used. Figure 8b shows the results for the pressure distributions after 117 hours. It is seen that the CFD
model has good agreement with the field data. Generally, the pressure drop model can affect the
predictions significantly. Fontanilla & Aziz (1982) considered several other cases of downward steam
injection flows in their work and concluded that the pressure distributions, predicted by Beggs & Brill
model, are in a better agreement with field data than the other pressure drop models.

Figure 8 —Comparison between the CFD model, field data and analytical results by Fontanilla & Aziz (1982), a) steam temperature
distribution after 71 hours, b) steam pressure distribution after 117 hours

To examine the Drift Flux model for vertical upward flow, the CFD model was used to simulate the
multiphase flow in a vertical producer well as illustrated in Figure 9. The corresponding reservoir and well
properties used for the simulation are shown in Table 4. Under the given conditions, the wells and
reservoir reach steady state after ~300 days. The predicted distributions of flow parameters, such as phase
saturations, phase mole fractions, pressure, temperature and physical velocities are shown in Figure 10.
When the steady state solution is reached, the flow along the producer well can be considered isothermal,
since the temperatures of the reservoir and injection fluid are very close. Therefore, heat loss from the
wellbore to the formation is negligible.
12 SPE-175056-MS

Figure 9 —Vertical producer well

Table 4 —Properties of reservoir and wellbores


Reservoir
permeability I, J, K 5000,5200,500 md
porosity 0.4
grid block size ⌬x, ⌬y, ⌬z 5, 5, 5 m
number of grid blocks , x, y, z 50, 15, 15
initial temperature and pressure 100 °C and 1000 kPa (top)
initial water, oil, gas saturations 0. 5, 0.3, 0.2
initial time step size 0.01
Injector
length 195 m
diameter 0.16
number of grid blocks 40
well constraint Bottom hole pressure at 1500kPa
relative wall roughness 0.0001
Producer
length 195 m
diameter 0.2 m
number of grid blocks 40
well constraint Bottom hole pressure at 800kPa
relative wall roughness 0.0001
SPE-175056-MS 13

Figure 10 —Distribution of flow parameter versus depth in the producer well after 300 days; a) phase saturations, b) phase global mole
fractions c) pressure and temperature d) liquid, gas and mixture velocities.

The distributions of physical phase velocities along the producer well are almost linear. For the given
conditions, the difference between gas and liquid velocities remains ~0.8 m/s along the well. The
Reynolds number in the producer well varies from 6.3 ⫻ 104 at the bottom to 2.7 ⫻ 106 at the top.
Therefore, the flow regime is completely turbulent along the well and the velocity profile across the
wellbore is uniform. For a uniform flow profile the distribution parameter, C0, in the Drift Flux model is
obtained to be approximately 1.0. Thus, according to the Drift Flux model, the only factor that causes the
slippage between the phases is the buoyancy force, which appears in the Vgm term in equation (4). The
Drift Flux model of Bhagwat & Ghajar (2014) involves non-linear coefficients, strongly dependent on the
phase physical velocities. This leads to more computational effort and time step cuts compared to when
the Drift Flux model of Shi et al. (2003) is used. This issue becomes even more critical for injector wells
when phase change takes place. The same conclusion is valid for horizontal and downward flow cases.

Conclusions
A thermal compositional wellbore model was developed by solving a complete form of mass, energy and
momentum balance equations. The slippage between the phases was taken into account using the Drift
Flux model, and the Beggs and Brill correlation was used for calculating the multiphase flow pressure
drop. The model was successfully verified for different single and multiphase flow conditions and was
coupled with CMG’s STARS™ reservoir simulator. Close agreement between the transient and steady
state results of CFD model and those of FlexWell was observed for horizontal steam injection process, and
14 SPE-175056-MS

existing differences were explained. In addition, the CFD model was verified against real case data and
close agreement between the model predictions and the data was achieved.
So far, the model has been developed to handle wells without any tubing. As a future step, model
capabilities will be extended to more complicated situations such as multi-tube injector and multi-lateral
wells.

Nomenclature
C0 Profile parameter
D Wellbore diameter
H Mixture enthalpy
hloss Overall heat loss coefficient
L Length of grid block
Lj Phase mole fraction
Nu Nusselt number for pipe
P Pressure
q̇ Mass or mole perforation rate
Sj Saturation of phase j
R Thermal resistance
re Equivalent radius
rCi Casing inner radius
rCo Casing outer radius/cement inner radius
rh Cement/earth interface radius
T Temperature
uj Internal energy of phase j
V Volumetric mixture velocity
vj Physical velocity of phase j
Vgm Drift velocity
vsj Superficial velocity of phase j
xij Mole fraction of component i in phase j
Zi Global mole fraction of component i
Z Axial direction

Greek
␳j Density of phase j
␳m Density of the mixture
␪ Inclination angle

References
Agarwal, R. K., Li, Y. K., Nghiem, L. X., and Coombe, D. A. 1991. Multiphase Multicomponent
Isenthalpic Flash Calculations. Journal of Canadian Petroleum Technology 30 (03), 69 –75
Ahmed M. M., Ayoub M. A. 2014. Comprehensive Study on the Current Pressure Drop Calculation
in Multiphase Vertical Wells; Current Trends and Future Prospective. Journal of Applied Sciences:
J. of Applied Sciences. http://dx.doi.org/10.3923/jas.2014.3162.3171
Ansari, A. M., Sylvester, N. D., Sarica, C., Shoham O., and Brill, J. P. 1994. A Comprehensive
Mechanistic Model for Upward Two-Phase Flow in Wellbores. SPE Production & Facilities
Journal. 9 (02): 143–151.
SPE-175056-MS 15

Aziz, K., and Govier, G. W. 1972. Pressure Drop in Wells Producing Oil and Gas. Journal of
Canadian Petroleum Technology, 11 (03) 38 –48
Beggs, D. H., and Brill, J. P. 1973. A Study of Two-Phase Flow in Inclined Pipes. Journal of
Petroleum Technology, 25 (05), 607–617.
Bhagwat, S. M., and Ghajar, A. J. 2014. A Flow Pattern Independent Drift Flux Model Based Void
Fraction Correlation for A Wide Range of Gas–Liquid Two Phase Flow. International Journal of
Multiphase Flow, 59, 186 –205.
Bleakley, W. B. 1964. Here are case histories of two thermal projects. Oil and Gas Journal, 62 (43),
123–130.
Bonizzi, M., and Issa, R. I. 2003. On the Simulation of Three-Phase Slug Flow in Nearly Horizontal
Pipes Using the Multi-Fluid Model. International Journal Of Multiphase Flow, 29 (11), 1719 –
1747.
Fontanilla, J. P., and Aziz, K. 1982. Prediction of bottom-hole conditions for wet steam injection
wells. Journal of Canadian Petroleum Technology, 21 (02), 82–88.
Hasan, A. R., and Kabir, C. S. 1999. A Simplified Model for Oil/Water Flow in Vertical and Deviated
Wellbores. SPE Production & Facilities, 14 (01), 56 –62.
Hibiki, T., and Ishii, M. 2003. One-Dimensional Drift-Flux Model and Constitutive Equations for
Relative Motion Between Phases in Various Two-Phase Flow Regimes. International Journal of
Heat and Mass Transfer, 46 (25), 4935–4948.
Livescu, S., Durlofsky, L. J., Aziz, K., and Ginestra, J. C. 2010. A Fully-Coupled Thermal Multiphase
Wellbore Flow Model for Use in Reservoir Simulation. Journal of Petroleum Science and
Engineering, 71 (3), 138 –146.
Oddie, G., Shi, H., Durlofsky, L. J., Aziz, K., Pfeffer, B., and Holmes, J. A. 2003. Experimental Study
of Two And Three Phase Flows in Large Diameter Inclined Pipes. International Journal of
Multiphase Flow, 29 (4), 527–558.
Peaceman, D. W. 1983. Interpretation Of Well-Block Pressures in Numerical Reservoir Simulation
with Non-square Grid Blocks and Anisotropic Permeability. Society of Petroleum Engineers
Journal, 23 (03), 531–543
Petalas, N., and Aziz, K. 2000. A Mechanistic Model for Multiphase Flow in Pipes. Journal of
Canadian Petroleum Technology, 39 (6), 43–55.
Petalas, N., and Aziz, K. 1996. Development and Testing of A New Mechanistic Model For
Multiphase Flow in Pipes. ASME Fluids Engineering Division Summer Meeting. Part 1(of 2), San
Diego, CA, USA, 7-11 July 1996, 153–159.
Pourafshary, P., Varavei, A., Sepehrnoori, K., and Podio, A. 2009. A Compositional Wellbore/
Reservoir Simulator to Model Multiphase Flow And Temperature Distribution. Journal of Petro-
leum Science and Engineering, 69 (1), 40 –52.
Shi, H., Holmes, J. A., Durlofsky, L. J., Aziz, K., Diaz, L. R., Alkaya, B., and Oddie, G. 2003.
Drift-Flux Modeling Of Multiphase Flow in Wellbores. SPE Annual Technical Conference and
Exhibition (January). Society of Petroleum Engineers.
Shi, H., Holmes, J. A., Durlofsky, L. J., Aziz, K., Diaz, L. R., Alkaya, B., and Oddie, G. 2005.
Drift-Flux Modeling of Two-Phase Flow in Wellbores. SPE Journal, 10 (1), 24 –33.
Siu, A. L., Rozon, B. J., Li, Y. K., Nghiem, L. X., Acteson, W. H., and McCormack, M. E. 1991. A
Fully Implicit Thermal Wellbore Model for Multicomponent Fluid Flows. SPE Reservoir Engi-
neering, 6 (3), 303–310.
Soprano, A. B., Ribeiro, G. G., da Silva, A. F., and Maliska, C. R. 2010. Solution of A One-
Dimensional Three-Phase Flow in Horizontal Wells Using A Drift-Flux Model. Mecánica Com-
putacional, 29, 8767–8779.
16 SPE-175056-MS

Spedding, P. L., Benard, E., and Donnelly, G. F. 2006. Prediction Of Pressure Drop in Multiphase
Horizontal Pipe Flow. International Communications In Heat And Mass Transfer, 33 (9), 1053–
1062.
Stone, T. W., Edmunds, N. R., and Kristoff, B. J. 1989. A Comprehensive Wellbore/Reservoir
Simulator. SPE, 18419, 6 –8.
Taitel, Y., Dukler, A. E. 1976. A Model For Predicting Flow Regime Transitions In Horizontal And
Near Horizontal Gas᎑Liquid Flow. AIChE Journal, 22 (1), 47–55.
Winterfeld, P. H. 1989. Simulation of Pressure Buildup in A Multiphase Wellbore/Reservoir System.
SPE Formation Evaluation, 4 (02), 247–252.
Zuber, N., and Findlay, J. 1965. Average Volumetric Concentration in Two-Phase Flow Systems.
Journal of Heat Transfer, 87 (4), 453–468.

You might also like