Journal of Petroleum Science and Engineering 69 (2009) 40–52

Contents lists available at ScienceDirect

Journal of Petroleum Science and Engineering

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / p e t r o l

A compositional wellbore/reservoir simulator to model multiphase flow and

temperature distribution
P. Pourafshary 1, A. Varavei 2, K. Sepehrnoori ⁎, A. Podio 3
University of Texas at Austin, 1 University Station, C0300, Austin, TX, 78712, USA

a r t i c l e i n f o a b s t r a c t

Article history: Production of hydrocarbon often involves gas and liquid (oil/water) concurrent flow in the wellbore. As a
Received 26 July 2007 multi-phase/multi-component gas–oil mixture flows from the reservoir to the surface, pressure,
Accepted 22 February 2009 temperature, composition and liquid holdup distributions are interrelated. However, nearly all two-phase
wellbore simulations are currently performed using “black oil” simulators. In this paper, a compositional-
Keywords:
wellbore model coupled with a reservoir simulator to compute pressure and temperature distribution is
compositional simulator
presented. In this work, compositions of liquid and gaseous phases in the wellbore can be determined by
wellbore/reservoir simulator
flash calculation two-phase equilibrium flash calculations and by considering the slip between phases. Our simulator has the
multiphase capability of predicting the temperature profile in the wellbore, which helps to predict multiphase flow
flow physics such as liquid holdup and pressure drop more accurately. As the wellbore model is coupled with a
reservoir simulator, it can be used as a tool to calculate fluid-flow compositions between reservoir and
wellbore. The simulated results of our compositional model were compared to the equivalent blackoil model
for pressure and temperature distribution. Although the input requirements and computing expenses are
higher for compositional calculations than for blackoil, our simulations show that in some cases, such as
those involving highly-volatile oil and retrograde condensate gas, ignoring compositional effects may lead to
errors in pressure profile prediction for the wellbore.
© 2009 Elsevier B.V. All rights reserved.

1. Introduction different components such as methane, ethane, propane and other

As hydrocarbon is produced from a wellbore, pressure drop occurs
due to gravity, friction and acceleration. Correct estimation of pressure
drop is essential in artificial lift design calculations and well productivity
computations. Different empirical correlations are available to model
pressure drop in wellbores, such correlations have been published by
Duns and Ros (1963), Hagedorn and Brown (1965), Orkiszewski (1967)
and Beggs and Brill (1973). The range of applicability of these cor-
relations depends on several factors, such as tubing size, oil gravity, gas–
liquid ratio and water cut. Solution methods have recently shifted to the
mechanistic modeling approach due to the inability of correlations to
simulate various cases. The main procedure of mechanistic modeling
consists of the determination of flow regimes and the prediction of flow
characteristics in the wellbore. Different mechanistic models such as
Hasan and Kabir (1988), Ansari et al. (1994) and Aziz and Petalas (1998)
are available in the literature. Produced gas and oil phases consist of
⁎ Corresponding author. Tel.: +1 512 471 0231.
E-mail addresses: pourafshary@mail.utexas.edu (P. Pourafshary),
varavei@mail.utexas.edu (A. Varavei), kamys@mail.utexas.edu (K. Sepehrnoori),
apodio@mail.utexas.edu (A. Podio).
1
Tel.: +1 512 471 6947.
2
Tel.: +1 512 471 1240.
3
Tel.: +1 512 471 3260.
hydrocarbons. However, in most of the available pressure loss models,
pressure calculations are performed based on the simplified blackoil
equations. The basic assumption in the blackoil approach is to consider
three distinct phases: gas, oil and water phase. Oil and gas phases are
recognized with oil specific gravity and gas specific gravity, respectively,
which are assumed to remain constant in the wellbore. In a blackoil
model, the gas can be dissolved in the oil phase. A blackoil model usually
treats PVT properties of hydrocarbon phases as single functions of
pressure and temperature. Hence, oil and gas properties such as density,
viscosity and specific volume are computed by experimental correla-
tions at each pressure and temperature. Empirical correlations are
applied to calculate dissolved gas in the oil phase. With the blackoil
approach, the effect of compositions on pressure and temperature
changes is neglected.
The main question in using blackoil approximation is its validity.
When the flowing liquid and gas are composed of more than one
component, the effect of compositions on pressure profile, tempera-
ture distribution and fluid flow properties should be considered. One
approach is to use a more complicated compositional method instead
of blackoil approximation. The term “compositional” implies that the
in-situ fluid composition may vary point by point in the wellbore as
functions of pressure, temperature and slip between the phases.
When the compositions are known, fluid-flow properties are obtained
from the phase behavior calculations. Several authors such as Gregory

0920-4105/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.petrol.2009.02.012
 P. Pourafshary et al. / Journal of Petroleum Science and Engineering 69 (2009) 40–52 41

Nomenclature

A Wellbore area, ft2

f Fugacity
F Molar rate of production, lbm mol/(ft3 day)
g Acceleration owing to gravity, ft/s2
h Fluid enthalpy, Btu/lbm
H Liquid holdup
G Gas phase fraction in equilibrium
K Equilibrium ratio
−
K Permeability tensor, Darcy
kr Relative permeability
L Phase mole ratio in mixture
Le Liquid phase fraction in equilibrium
Lne Liquid phase fraction not in equilibrium
mt' Molar flow rate for ith component between wellbore and reservoir, lbm mol/day
M Molecular weight, lbm mol/lbm
nc Number of components
np Number of phases
N Overall concentration of component i
−
N Molar flux vector, lbm mol/ft
P Pressure, psi
q Molar flow rate, lbm mol/day
qH Enthalpy injection rate per unit rock volume, Btu/lbm s
qL Heat loss to the over- and underburdens per unit rock volume, Btu/lbm s
Q Heat transfer rate to the wellbore, Btu/(h ft)
R Gas constant
S Saturation
T Temperature, °F
u Sum of internal energies per unit rock and the fluid contained in the unit rock, Btu/lbm
v Fluid velocity, ft/s
v Mole fraction of gas in absence of water
v̄j Molar volume of phase j
V Volume, ft3
W Overall concentration, lbm mol/ft3
xi Molar fraction of ith component in liquid phase
yi Molar fraction of ith component in gas phase
z Phase compressibility
Z Overall hydrocarbon composition

Greek symbols
α Phase volumetric fraction
ρ Density, lbm/ft3
θ Wellbore angle, radian
ϕ Porosity
ϕij Fugacity coefficient of component i in phase j
ξ Molar density, lbm mol/ft3
λ Mobility ratio, Darcy/cp
γj Gravity term for phase j, defined as p j g
μ Viscosity, cp

Subscripts
b Bulk
i Component
j Phase (1 for oil, 2 for gas)
g Gas
l Liquid
o Oil
S Standard condition
T Total
42 P. Pourafshary et al. / Journal of Petroleum Science and Engineering 69 (2009) 40–52
and Aziz (1973), Anis and Buthod (1974), Gould (1975), and Furukawa
et al. (1986) presented compositional approach modeling to simulate
fluid flow in pipelines. To the best of our knowledge, the importance
of compositional modeling of multiphase/multicomponent fluid flow
in vertical wellbores has not been presented in the literature.
Furthermore, temperature gradient occurs in the wellbore due to heat
exchange between the fluid and its surrounding formations. Tempera-
ture variation affects fluid compositions at different depths and makes
the problem more complicated. Hence, to have an accurate composi-
tional multiphase fluid-flow model, the energy equation should also be
solved simultaneously with pressure loss equations to model tempera-
ture distribution. Ramey (1962) was the first to present a theoretical
model to estimate fluid temperature in the wellbore as a function of
well depth. However, his method has a few limitations due to neglecting
the effects of kinetic energy and friction and considering only single-
phase flow. Satter (1965) extended Ramey's model to multiphase flow.
Alves et al. (1990) presented a unified model to predict temperature
distribution in the wellbore for the entire range of inclination angles.
Hasan and Kabir (1994) presented a general method to model wellbore
heat loss and flowing-fluid temperature. It should be noted that the
blackoil approximation is assumed in all of the available methods to
calculate thermal properties of multiphase-flow mixture.
Two limiting models can be applied to compute composition of a
moving multi-component fluid in the wellbore. The first one assumes
equilibrium between vapor and liquid phases at all positions in the
wellbore. With this assumption it is possible to use flash calculation at
each section of the wellbore to calculate compositions. On the other
hand, differential vaporization can be considered, which is based on
the assumption of no mass transfer of vapor components back into the
liquid. We believe that the actual flow condition in the wellbore falls
somewhere between these two extremes of equilibrium and differ-
ential models. Two approaches are widely used to express phase
fractions and pressure distribution in multiphase-flow systems in
wellbores. The simplest approach is to consider the multiphase flow as
a pseudo single phase flow which is called a homogeneous model. In
contrast, another method is the separated approach. In a separated
method we assume that the phases can move with unequal velocity in
the same or different directions. In a homogeneous model for fluid
flow in a wellbore, the assumption of equilibrium between liquid and
gas is reasonable due to the no-slip behavior of the phases. If we use
the separated models, there is a slip velocity between phases. The gas
phase moves faster than the liquid phase at each section and this
makes the equilibrium assumption doubtful. To solve this problem, we
Fig. 1. Schematic of wellbore/reservoir system.
did modifications for flash calculations to consider the effect of the slip
velocity. The same problem occurs in fluid-flow modeling in pipelines.
Anis and Buthod (1974) showed that in a pipeline assuming the
equilibrium model always results in more pressure drop than assuming
the differential model. They also showed that the differential model
predicts a lower vapor velocity than the equilibrium model.
In this paper we present a compositional wellbore simulator to
model pressure profile, phase velocity profile and temperature
distribution in the wellbore. This wellbore simulator is coupled to a
reservoir simulator to accurately compute compositional fluid flow
between the wellbore and reservoir. Different cases are modeled with
this coupled simulator to show when the compositional approach is
important and when the blackoil approximation is accurate enough to
predict pressure and temperature profiles in the wellbore. For a
reservoir simulator, we used a compositional simulator called GPAS
(Wang et al., 1997, 1999; Han et al., 2005) which was developed at the
University of Texas at Austin.
2. Governing equations
To calculate pressure, temperature, and composition profiles in the
wellbore, a coupled wellbore/reservoir governing equations should be
solved simultaneously. The mass, momentum and energy balance
equations govern the multiphase fluid flow in the wellbore. A flash
calculation method and an equation of state are used to calculate
compositions and phase properties. The wellbore is divided into
segments in the z direction as shown in Fig. 1. Wellbore gridblocks are
connected to the reservoir gridblocks in the perforation zones.
2.1. Wellbore governing equations
2.1.1. Continuity equations
In a three phase gas/oil/water system, liquid and gas continuity are
satisfied in each gridblock. For any component, in each block, the
output moles to the wellbore neighbor blocks or reservoir blocks are
equal to the input moles from neighbor blocks or reservoir. For
example, for component in the liquid phase:
A
A fρ x v g + mVi1 = 0; i = 1; N ; nc ð1Þ
Az i1 i 1
In this equation, nc represents the number of components in
liquid phase, it should be noted that the water is assumed as a
component in this phase. ρi1 is molar density of ith component
 P. Pourafshary et al. / Journal of Petroleum Science and Engineering 69 (2009) 40–52 43

Adding Eqs. (4) and (5), a momentum equation is obtained for

mixture. Eq. (6) can be used to calculate steady-state pressure drop in
the wellbore.

X n  
AP A 2X
nc
A 2X
nc c
AP
= v ρ x + v ρ y + g sin θ ðρi1 xi + ρi2 yi Þ+
Az Az 1 i = 1 i1 i Az 2 i = 1 i2 i
i=1
Az Wall
ð6Þ

In this equation the pressure loss due to friction is calculated by the

method presented by Su and Lee (1995).

2.1.3. Energy equation

Fig. 2. Schematic of equilibrium between phases in a block. The general overall energy balance is considered for the mixture as
Eq. (7), where Q is the heat flux from the surrounding  formation to the

P
np Pnc v2
wellbore fluid and the last term, ρij vij d hij + 2ij + gz sin θ ,
j=1 i=1
dz
is the energy convection in the direction due to enthalpy exchange,
in phase 1 (liquid) and xi is fraction of ith component in liquid acceleration and gravity.
′ denotes the molar flow rate of component i in the
phase. Also, mi1
!
liquid phase between wellbore and reservoir. We assume that all X X
np nc
d
2
vij
components in a phase move with the same velocity. For the whole Q + ρij vij hij + + gz sin θ = 0 ð7Þ
j=1 i=1
dz 2
liquid phase, we can add all the component equations together as
follows:
To solve these governing equations, phase compositions are
needed. We use the following procedure to determine the number,
A X
nc X
nc
A v ρ x + mVi1 = 0 ð2Þ amount and composition of all the phases:
Az 1 i = 1 i1 i i=1
1. The number of phases in a gridblock is determined using the phase
stability analysis at each pressure and temperature. The system is
The same procedure can be used for a gas continuity equation. If yi multiphase if the value of Gibbs free energy is lower than a single-
is the ith component fraction in the gas phase and ρ12 is the molar phase mixture of overall hydrocarbon composition, Z (Michelsen,
density of ith component in the gas phase, then the overall continuity 1982).
equation for the gas phase is 2. After the number of phases is determined, flash calculations are
performed to compute the mole fraction and composition of each
A X
nc Xnc phase at the given temperature, pressure and overall composition
A v2 ρi2 yi + mVi2 = 0 ð3Þ of the fluid. The governing equations for the flash require equality
Az i = 1 i=1 of component fugacities and mass balance. Three conditions should
be satisfied for equilibrium:
1. Mass conservation of each component in the mixture
2.1.2. Momentum equation
2. Equality of chemical potentials for each component in all phases
Pressure loss in the wellbore occurs due to gravity, friction and
3. Minimum condition for Gibbs free energy at constant tempera-
acceleration. In a compositional approach we consider all of these
ture and pressure
mechanisms in momentum equations for the gas phase and the liquid
3. The effect of slip velocity on compositions is considered, which will
phase. For liquid, the momentum equation can be written as
be discussed later.

A X
nc
A 2X
nc Xnc
AP Fugacity is calculated using the Peng–Robinson equation-of-state
v1 ρi1 xi + v1 ρi1 xi + g sin θ ρi1 xi + α ð4Þ (Peng and Robinson, 1976). Both the phase composition constraint,
At i = 1 Az i = 1 i=1
Az 1
    which states that the sum of the mole fraction of all the components in
AP AP a phase is equal to one, and the Rachford–Rice equation are used
+ + =0
Az Wall;1 Az interphase

A
P
nc
In this equation the first term, At v1 ρi1 xi is the transient part of
i=1
A 2
P
nc
pressure loss which is zero in steady-state condition. Az v1 ρi1 xi ,
P
nc AP AP i=1
g sin θ ρi1 xi ; Az Wall;1 and Az interphase are pressure loss due to accel-
i=1
eration, gravity, wall friction and friction between phases, respectively.
The term AP
Az α 1 represents the pressure drop due to liquid. α1 is the liquid
volumetric fraction in the mixture.
Similar mechanisms are involved for pressure loss in the gas
phase. The gas momentum equation represents all of these terms as
follows:

A X
nc
A 2X
nc Xnc
AP
v2 ρi2 yi + v2 ρi2 yi + g sin θ ρi2 yi + α ð5Þ
At i = 1 Az i = 1 i=1
Az 2
   
AP AP
+ − =0
Az Wall;2 Az interphase Fig. 3. Schematic for grid blocks in the wellbore.
44 P. Pourafshary et al. / Journal of Petroleum Science and Engineering 69 (2009) 40–52

Fig. 4. Procedure to calculate pressure and temperature at each wellbore gridblock.

implicitly in the solution of the fugacity equation. The Rachford–Rice feed and r(v) is the residual of the Rachford–Rice equation. After
equation evaluates the amount and composition of each equilibrium solving this equation, the component mole fractions in the liquid and
phase in a classical flash calculation. This equation is expressed as gas phases are computed from:

X
nc
ðKi − 1ÞZi Zi
r ðvÞ = =0 ð8Þ xi = ð9Þ
1 + vðKi − 1Þ 1 + vðKi − 1Þ
i=1

where v is the mole fraction of gas in absence of water, Ki is the

Zi Ki
equilibrium ratio, Zi is the overall mole fraction of component i in the yi = ð10Þ
1 + vðKi − 1Þ

Newton iteration can efficiently solve Eqs. (8)–(10).

Table 1
General parameters for wellbore modeling.

Depth (ft) 12,000

Inner tubing radius (ft) 0.125 Table 2
Outer tubing radius (ft) 0.129 Blackoil composition.
Wellbore radius (ft) 0.425
Inner casing radius (ft) 0.2843 Component Mole fraction
Outer casing radius (ft) 0.3154 C1 0.3
Tubing friction coefficient 0.0006 C3 0.12
Formation heat conductivity (Btu/(h ft °F)) 1.4 C4 0.12
Formation density (lbm/ft3) 144 C5 0.12
Formation heat capacity (Btu/(lbm °F) 0.22 C7 0.17
Cementing heat conductivity (Btu/(h ft °F)) 4.021 C8 0.17
 P. Pourafshary et al. / Journal of Petroleum Science and Engineering 69 (2009) 40–52 45

Fig. 5. Pressure profiles obtained from blackoil and compositional calculation.

Fig. 6. Pressure and temperature change in oil production.

−
2.2. Reservoir governing equations N i is the flux vector of component i, and Fi is the molar rate of
injection/production of component i. Each term of the equation
−
We used a compositional reservoir simulator called GPAS (Wang et al., can be written in more detail, for example, N i in the accumulation
1997, 1999; Han et al., 2005) in our coupled wellbore/reservoir model. A term is a function of porosity, ϕ, the molar density of phase j, ξj,
brief description of the simulator is presented below. Multicomponent and saturation of phase j, Sj and the mole fraction of component i in
multiphase flow in a porous medium can be described using four different phase j, xij as
types of equations:
1. Partial differential, component-mass balances describing compo- X
np
Ni = / nj Sj xij ð12Þ
nent flow, in which Darcy's law is used to govern the transport of
j=1
phases from one cell to another
2. Phase equilibrium equations dealing with equilibrium component
mass transfer between phases The flux vector at each gridblock of the reservoir results from a
3. Equations constraining phase saturation and component combination of two mechanisms: convection and dispersion. The
concentrations.
4. Energy balance equations controlling energy flow
Neglecting mutual solubility between water and hydrocarbon phases,
for a system consisting of nc hydrocarbon components and np fluid
Table 3
phases (excluding the aqueous phase), the above four types of equations
A typical composition for a volatile-crude oil.
may be mathematically expressed for a control volume as follows:
Component Mole fraction
Component material balance: The overall material balance for C1 0.55
component i is written in terms of moles per unit time. Nt is the C3 0.1
number of moles of component i per unit bulk volume. C4 0.1
C5 0.1
ANi P P C7 0.075
− j kern5pt N i − Fi = 0 ð11Þ C8 0.075
At
46 P. Pourafshary et al. / Journal of Petroleum Science and Engineering 69 (2009) 40–52

Fig. 7. Pressure profiles in a wellbore producing volatile oil.

Fig. 8. Pressure–temperature change for volatile oil production.

dispersion term is ignored in GPAS equations for simplicity. Hence, Volume constraints: The pore volume in each cell must be filled
the flux vector in Eq. (11) can be expressed as completely by the total fluid volume. This constraint gives rise to:

X
np
X
nc X
np
Ni = nj xij vj ð13Þ Ni Lj v j − / = 0 ð16Þ
j=1 i=1 j=1

Darcy's law is used to govern the transport of phases from one cell where Lj is the ratio of moles in phase to the total number of moles
to another under the local pressure gradient, rock permeability, in the mixture, and vj̄ is the molar volume of phase j.
relative permeability and viscosity. From Eqs. (11)–(13), the over- Molar energy balance: The molar energy balance for the control vol-
all mass balance for each component is ume using internal energy as a primary variable can be expressed as
0 1 2 3
A@ X
np
X
np P  Xnp  
P q Au P P
/ nj Sj xij A − j ●4 nj xij λj j Pj − γj D 5 − i = 0 ð14Þ + j: nj λj hj jPj − γj jD − j :ðλT jT Þ − qH + qL = 0
At j=1 j=1
Vb At j=1

ð17Þ
In this molar equation qi is the molar flow rate for component i
Eqs. (14)–(17) give (2n c + 3) independent equations and
from the wellbore gridblocks. is calculated by well model
unknowns for each cell. These equations are discretized on a
equations presented by Peaceman (1983).
Phase-equilibrium relationship: The component distribution among
the various phases is determined by the phase-equilibrium
calculation. This requires that the molar-balance constraint be Table 4
Composition for a typical gas–condensate system.
preserved, the chemical potentials of each component be the same
for all phases, and the Gibbs free energy at constant temperature Component Mole fraction
and pressure be minimized, which can be described as C1 0.8
C3 0.04
j r
fi − fi = 0 i = 1; 2; N ; nc j = 1; 2; :::; np − 1 ð15Þ C4 0.04
C5 0.04
j C7 0.04
where fi = ln(xijϕij) and ϕij is the fugacity coefficient of component
C8 0.04
i in phase j. The r superscript denotes a reference phase.
 P. Pourafshary et al. / Journal of Petroleum Science and Engineering 69 (2009) 40–52 47

Fig. 9. Pressure profile in a wellbore producing gas condensate.

Fig. 10. Pressure–temperature change during condensate gas production.

rectangular grid using finite differences with one-point upstream The holdup value is defined as volume fraction of liquid in the
weighting. In the fully implicit solution, this discretization results block. Hence,
in a system of nonlinear equations that are solved using Newton's
method. The independent variables used are P, hT, Ni and ln Ki, Vl zl n l
H= = ð20Þ
i = 1,2,…,nc. Vl + Vg zl nl + zg ng

3. Effect of slip velocity We assume that both Le and Lne sections in Fig. 2 have the same
composition and same compressibility factor. Dividing Eq. (20) by the
The general method used to calculate composition of each phase is total mole value in the block, we obtain:
equilibrium flash calculation. With this method we assume that liquid
phase and gas phase are in equilibrium, but this is not true when these zl n nl zl ðLe + Lne Þ
H= total
ng = ð21Þ
two phases are moving at different velocities. To consider the effect of zl n nl
+ zg n zl ðLe + Lne Þ + zg G
total total
slip velocity, we assume that at each block in the wellbore the gas phase
is in equilibrium with only a portion of the liquid phase. Hence, the zl ðLe + Lne Þ
=
liquid phase is divided into two parts as shown in Fig. 2: Le that is in zl ðLe + Lne Þ + zg ð1 − Le − Lne Þ
equilibrium with the gas phase and Lne, which appears because of slip
velocity and is not in equilibrium with the gas phase (G). Hence,
Performing a flash calculation at each temperature and pressure,
G and Le are computed, because only these two portions are in zg H
Le + Lne = ð22Þ
equilibrium. zl ð1 − H Þ + zg H
In this block, liquid volume is
For multiphase flow, holdup can be estimated by different
zl nl RT mechanistic or experimental methods. In this paper we used Ansari
Vl = ð18Þ
P et al. (1994) and Hasan and Kabir (2007) models to compute holdup
where zl is liquid compressibility factor, and nl is liquid moles in the at different flow regimes. Computing Lne, we can calculate Zi′, the in-
block. Similarly for gas volume: situ overall fraction of ith component, by

zg ng RT yi G + xi Le + xi Lne
Vg = ð19Þ Z Vi = = Zi + xi Lne ð23Þ
P G + Le + Lne
48 P. Pourafshary et al. / Journal of Petroleum Science and Engineering 69 (2009) 40–52

Table 5 node is known. The goal is to calculate the pressure and temperature
Production parameters for three-phase flow case. distribution in the wellbore. In our compositional model, we assume
Depth (ft) 10,000 that the fluid flow in the wellbore is in steady state condition and the
Inner tubing radius (ft) 0.125 overall mole fraction of each component does not change from the
Outer tubing radius (ft) 0.129 sandface to the surface. Fig. 3 shows the schematic for gridblocks in the
Wellbore radius (ft) 0.525
wellbore. If pressure and temperature are known at block i − 1, the
Inner casing radius (ft) 0.28
Outer casing radius (ft) 0.315 procedure described in Fig. 4 is applied to calculate pressure and
Tubing friction coefficient 0.0006 temperature at block i.
Water flow rate (STB/day) 300
Water gravity 1
5. The necessity of compositional modeling
Total hydrocarbon molar rate (lbm mol/day) 2500
Surface formation temperature (°F) 90
Formation temperature gradient (°F/ft) 0.006 Using the compositional approach requires more data than using
Formation heat conductivity (Btu/(h ft °F)) 1.4 the blackoil approximation. Using the compositional approach also
Formation density (lbm/ft3) 144 involves more computations. The main question is the necessity of
Formation heat capacity (Btu/(lbm °F) 0.22
applying this more complicated approach. In the following section, we
Cementing heat conductivity (Btu/(h ft °F)) 4.02
Annulus brine salinity (ppm) 35000 compare the results obtained from both approaches and discuss the
Bottomhole wellbore temperature (°F) 150 differences by comparing three different cases: a well producing oil, a
Surface pressure (psi) 500 well producing volatile oil and a well producing condensate gas.
Number of nodes 40 Different methods are available to predict pressure distribution in the
Number of phases 3
Number of components 6
wellbore by blackoil approach. As previously mentioned, , we used a
Total mole fraction (c1,c3,c4,c5,c7,c8) 0.78,0.08,0.05, 0.05,0.02,0.02 blackoil method based on Ansari et al.'s work (1994) to simulate two-
phase flow in the wellbore. For the compositional method, we used
the approach discussed in this paper. The general parameters for
wellbores used in these simulations are shown in Table 1.
The necessity for using the compositional approach is discussed by
where xi is the fraction of ith component in liquid phase. yi represents comparing the pressure profiles obtained from the blackoil approach
the fraction of ith component in the gas phase and Zi is the overall and the compositional approach. In order to make a comparison, we
fraction of ith component measures at equilibrium. lump the compositions for each phase and calculate phase properties
Eq. (23) shows the new overall component fractions, so the total in- with the following procedure:
situ composition in the wellbore will vary point-by-point. Holdup calcu-
1) We use the flash calculation equilibrium at a standard condition to
lation is strongly dependent on composition of mixture so an iterative
calculate each phase composition.
method should be used to update composition and holdup accurately.
2) The density of the liquid phase and the gas phase are calculated at a
standard condition obtained from Eqs. (24) and (25).
4. Solution procedure
P
nc
PS yi Mi
Our wellbore simulator is coupled with a compositional reservoir i=1
ρg = ð24Þ
simulator by a well model. Well models are general equations that relate Rzy TS
fluid flow between the reservoir and the wellbore, and pressure. In P
nc
general, a functional relation between the well rates and flowing PS xi Mi
i=1
bottomhole pressures is required to couple both reservoir and wellbore ρo = ð25Þ
Rzx TS
models. In our simulator, the well models based on Peaceman (1983) are
provided to relate the controlled variables for the reservoir to the In these equations, Mi is the molecular weight for component i. xi
wellbore. In this coupled simulator, different options are developed to and yi are the mole fraction of component i in the liquid phase and
control a well. For example, with a producer, constant bottomhole the gas phase, respectively. zx and zy are the compressibility for oil
pressure and constant production rate can be assumed. For constant and gas phase obtained from the flash calculation. ρo, ρg are oil
bottomhole pressure conditions, the pressure at the deepest wellbore density and gas density in lbm/ft3.

Fig. 11. Pressure profile in the wellbore for three-phase flow production.
 P. Pourafshary et al. / Journal of Petroleum Science and Engineering 69 (2009) 40–52 49

Fig. 12. Holdup profile in the wellbore for three-phase flow production.

Fig. 13. Liquid velocity profile in three-phase flow production.

Fig. 14. Gas velocity profile in three-phase flow production.

3) The main parameters to define hydrocarbon mixture with a blackoil 4) In the compositional approach, the production is expressed by the
approach are oil specific gravity (γAPI) and gas gravity (γg). These molar production rate, nt. This value is converted into oil flow rate
parameters are calculated in Eqs. (26) and (27). and gas flow rate in Eqs. (28) and (29). In these equations Vl is the
P
nc liquid mole fraction in the mixture.
yi Mi
γg = i=1
ð26Þ nt Vl Rzx TS
29zy qo = ð28Þ
PS

141:5 nt ð1 − Vl ÞRzy TS
γAPI = − 131:5 ð27Þ qg = ð29Þ
ρo = 62:37 PS
50 P. Pourafshary et al. / Journal of Petroleum Science and Engineering 69 (2009) 40–52

Fig. 15. Temperature distribution in the wellbore in three-phase flow production.

If the oil and gas flow rates and specific gravities in a blackoil
system are known, we can simulate the wellbore fluid flow and
compare it with the compositional results. This procedure is
applied to the following cases.
6. Case studies
6.1. Case 1: Oil production
The oil gravity for liquid with the composition shown in Table 2 is
38°API. This oil consists of six components and is classified as a
blackoil. (Ahmed, 1989).
Pressure profile in the wellbore is calculated with compositional
approach and equivalent blackoil approximation. Fig. 5 shows the
analogy between these two approaches. The arrows on Fig. 6 and
other two-phase diagrams show the phase behavior from the sandface
to the surface.
The difference in pressure calculations is less than 1%, so this
shows that using a blackoil approximation is valid for this kind of oil
production. The temperature profiles simulated by these two
approaches show no significant difference. The pressure and tem-
perature changes in the wellbore are shown on the phase equilibrium
curve in Fig. 6.
6.2. Case 2: Volatile oil production
The same procedure is applied for a high-shrinkage oil (volatile
oil). Table 3 shows the composition that represents oil with gravity
equal to 50°API, which is classified as volatile oil (Ahmed, 1989).
With the same wellbore configuration, we simulate the fluid flow
when the pressure at the surface is 1000 psi, and the wellbore is
producing 1500 lbm mol/day of the hydrocarbon mixture. Fig. 7
shows the pressure profiles for both blackoil and compositional
approaches.
The average difference between the two curves is about 2%, but it
should be noted that in the bottomhole, the estimated pressure by
compositional approach is 70 psi less than the simulated value with
blackoil approach which is a noticeable difference. It seems that in the
case of volatile oil using a compositional approach improves our
estimation.
The pressure–temperature change is shown in Fig. 8 on the phase
behavior diagram. At the bottomhole, most of the produced hydro-
carbon is in the liquid phase. After pressure declines, gas comes out of
the liquid phase. In high-shrinkage oil, quality lines are close together
near the bubble point and more widely spaced at lower pressure.
Hence, a high liquid shrinkage occurs immediately below the bubble
point because of a rapid decrease in the quality.

Fig. 16. Liquid composition for a three-phase flow production.

 P. Pourafshary et al. / Journal of Petroleum Science and Engineering 69 (2009) 40–52
Fig. 17. Gas composition for a three-phase flow production.
6.3. Case 3: Gas condensate production
In the last case, we model gas condensate production. In this case,
the wellbore fluid temperature lies between the critical temperature
and cricondentherm of the produced fluid. This reservoir is classified
as a retrograde gas condensate reservoir.
We used the composition shown in Table 4 to make a gas
condensate system with gravity equal to 60°API (Ahmed, 1989). As in
the previous two cases, we model pressure and temperature
distribution in the wellbore with both compositional and blackoil
approaches. In this case, we assume that the surface pressure is 1500
psi and the well is producing 1500 lbm mol/day of hydrocarbon
mixture. Fig. 9 shows the pressure profiles in the wellbore.
Fig. 9 also shows that the bottomhole pressure calculated using the
compositional approach differs by about 11% from the result obtained
from blackoil approach, which is about 360 psi and a considerably
larger value. Our simulation shows that due to complex phase be-
havior of the gas condensate system, using blackoil approximation
may cause errors in the pressure profile prediction.
Fig. 10 shows the pressure–temperature profiles during production
in this case. As the reservoir pressure is above the upper dew-point line,
the hydrocarbon system exists as a single vapor phase. This means that
the sandface flow rate consists of only one phase, gas. As the pressure
declines in the wellbore, liquid begins to condense. As the pressure is
further decreased, the hydrocarbon mixture tends to condense instead
of expanding which occurs for gas phase or vaporizing which occurs for
liquid phase. This condensation process continues until a maximum
liquid drop-out takes place. Then vaporization begins again. In the
wellbore, where the pressure drop is high, enough liquid drop-out
occurs to result in two-phase flow of gas and retrograde liquid.
From our simulations it can be concluded that using a blackoil
approximation instead of modeling with a compositional approach
achieves high accuracy in most cases. But in condensate gas, because
of the unique phase behavior, there is a noticeable difference between
the two approaches; hence we recommend using a compositional
simulator when high accuracy is needed. These conclusions are
similar to the results obtained from the simulations performed for the
pipeline by Anis and Buthod (1974) and Gregory and Aziz (1973). We
did not observe any significant difference in temperature profiles
obtained from the two different approaches.
7. Results
In this section we discuss a three-phase production case to show the
ability of our simulator to model pressure profile, temperature
distribution, phase velocity and phase fraction, and compositions in
51
the wellbore. A three-phase water/oil/gas production is assumed in a
reservoir/wellbore coupled system. Table 5 shows the well parameters
that were used. The production consists of 2500 lbm mol/day of
hydrocarbon and 300 STB/day of water in the surface. At the surface all
three phases are produced but based on the pressure and temperature
values in the wellbore we expect liquid flowing only in some parts of the
wellbore. Fig. 11 shows the pressure profile in the wellbore. The pressure
increases with depth due to gravity. It can be observed that in the deeper
sections the gradient of pressure change decreases. This is due to lower
velocity of fluids in deep sections. Fig. 12 presents the liquid fraction or
holdup in the wellbore. At each section the liquid phase consists of water
and oil. The amount of oil is calculated by flash calculation, which is
modified by slip effect, as previously discussed. At the lower portion of
the wellbore there is little gas due to high pressure and temperature, so
the liquid volume does not change significantly anymore. We can see
near constant holdup below 9000 ft depth. Figs. 13 and 14 show velocity
profiles for liquid and gas, respectively. Gas and liquid velocity are higher
near the surface; this is because of phase expansion near the wellbore
due to lower pressure. This expansion is more noticeable for the gas
phase. As previously discussed, we conclude that assuming phase
equilibrium at each point of the wellbore is incorrect. Gas and liquid
velocity profiles are good evidence for our conclusion. It can be seen that
the gas velocity is higher than liquid velocity at each section, so there is
not sufficient time to reach equilibrium, and the effect of slip velocity
should be considered. Fig. 15 shows the temperature profile in the
wellbore. This figure shows that temperature changes significantly in
the wellbore, hence it is important to consider the effect of temperature
on composition. Another advantage of using a compositional simulator
is the ability to calculate the composition change of each phase during
production. The oil phase and gas phase that enter the wellbore from the
reservoir have different compositions from the produced fluid. This
composition can also be completely different from the oil or gas com-
position at standard condition. Figs. 16 and 17 compare composition of
the liquid phase and the gas phase at the sandface, surface and standard
conditions. These two figures show that at lower pressure and
temperature heavier components have higher composition in the liquid
phase and lighter components do not appear in high concentration. In
higher pressure, lighter components also go into the liquid phase. The
same behavior can be observed for the gas phase. At high pressure, only a
very light component remains in the gas phase, but at standard
condition some heavier ones also appears in this phase.
8. Conclusion
1) A compositional coupled wellbore/reservoir simulator is presented
to model multiphase fluid flow in wellbores.
52 P. Pourafshary et al. / Journal of Petroleum Science and Engineering 69 (2009) 40–52
2) This simulator has the ability to model pressure profile, tempera-
ture distribution, phase fractions, phase velocities and composition
in the wellbore for different production and injection cases.
3) Comparison is made between compositional approach and blackoil
approach to study the importance of compositional modeling. Our
simulations showed that in gas condensate wells and volatile oil
cases the difference between bottomhole pressure estimation is
noticeable.
4) Using blackoil approximation does not show significant difference
from compositional approach results for temperature distribution
for the different cases studied.
References
Ahmed, T.H., 1989. Hydrocarbon Phase Behavior. Gulf Publishing Company, Houston, TX,
p. 424.
Alves, I.N., Alhanati, F.J.S., Shoham, O., 1990. A unified model for predicting flowing
temperature distribution in wellbores and pipelines. Paper SPE 20632, Presented at
the SPE Annual Technical Conference and Exhibition, New Orleans, Louisiana, USA,
pp. 363–367.
Anis, M., Buthod, P., 1974. How flashing fluids change phase in pipelines. Oil and Gas J.
150–157 January.
Ansari, A.M., Sylvester, N.D., Sarcia, C., Shoham, O., Brill, J.P., 1994. A comprehensive
mechanistic model for upward two-phase flow in wellbores. 6th Annual SPE
Meeting, New Orleans, Louisiana, USA, pp. 143–152.
Aziz, K., Petalas, N., 1998. A mechanistic model for multiphase flow in pipes. 49th
Annual Technical Meeting of the Petroleum Society of the Canadian Institute of
Mining, Metallurgy and Petroleum, Calgary, Alberta, Canada.
Beggs, H.D., Brill, J.P., 1973. A study of two-phase flow in inclined pipes. J. Pet. Tech.
607–617.
Duns, H., Ros, N.C.J., 1963. Vertical flow of gas and liquid mixtures in wells. Proc. Sixth
World Pet. Congress, Frankfurt, Germany, Section II, Paper 22-PD6.
Furukawa, H., Shoham, O., Brill, J.P., 1986. Predicting compositional two-phase flow
behavior in pipelines. ASME J. Energy Resour. Technol. 108, 207–210.
Gregory, G.A., Aziz, K., 1973. Calculation of pressure and temperature profiles in
multiphase pipeline and simple pipeline networks. J. Petrol. Technol. 373–384.
Gould, T.L., 1975. Compositional two-phase flow in pipelines. Paper SPE 5685 Presented
at the 50th Annual fall meeting of the Society of Petroleum Engineers of AIME,
Dallas, TX, USA.
Hagedorn, A.R., Brown, K.E., 1965. Experimental study of pressure gradients occurring
during continuous two-phase flow in small diameter vertical conduits. J. Pet. Tech.
475–484.
Han, C., Delshad, M., Sepehrnoori, K., Pope, G.A., 2005. A fully implicit, parallel,
compositional chemical flooding simulator. Paper SPE 97217 Presented at the SPE
Annual Technical Conference and Exhibition, Dallas, Texas, 9–12 October 2005.
Hasan, A.R., Kabir, C.S., 1988. Predicting multiphase flow behavior in a deviated well. SPE
Production Engineering, pp. 474–482. Nov.
Hasan, A.R., Kabir, C.S., 1994. Aspects of wellbore heat transfer during two-phase flow.
SPEPF 9, 211–216.
Hasan, A.R., Kabir, C.S., 2007. A simple model for annular two-phase flow in wellbores. J.
SPE Prod. Oper. 22, 168–175.
Michelsen, J.L., 1982. The isothermal flash problem. Part I. Stability. Fluid Phase Equilib.
9, 1–19.
Orkiszewski, J., 1967. Predicting two-phase pressure drops in vertical pipe. J. Pet. Tech.
19, 829–838.
Peaceman, D.W., 1983. Interpretation of well-block pressures in numerical reservoir
simulation with nonsquare grid blocks and anisotropic permeability. SPE J. 531–543
June.
Peng, D.Y., Robinson, D.B., 1976. A new two-constant equation of state. Ind. Eng. Chem.:
Fundamentals 15, 59–64.
Ramey, H., 1962. Wellbore heat transmission. JPT 427–435 April.
Satter, A., 1965. Heat losses during flow of steam down a wellbore. JPT 845–851 July.
Su, H.J., Lee, S.H., 1995. Modeling transient wellbore behavior in horizontal wells.
International Meeting on Petroleum Engineering, Beijing, China, Paper SPE 29961.
Wang, P., Yotov, I., Wheeler, M., Arbogast, T., Dawson, C., Parashar, M., Sepehrnoori, K.,
1997. A new generation EOS compositional reservoir simulator: Part I — formulation
and discretization. Paper SPE 37979 Presented at the SPE Reservoir Simulation
Symposium, Dallas, TX, 8–11 June.
Wang, P., Balay, S., Sepehrnoori, K., Wheeler, J., Abate, J., Smith, B., Pope, G.A., 1999. A
fully implicit parallel EOS compositional simulator for large scale reservoir
simulation. Paper SPE 51885 presented at the SPE Reservoir Simulation Symposium,
Houston, TX, pp. 14–17. February.