Você está na página 1de 10

SPE 113651

Multiscale Compositional Simulation of Naturally Fractured Resevoirs


B. Ramirez, SPE, Colorado School of Mines; Safian Atan, SPE, Marathon Oil Company; and Hossein Kazemi,
SPE, Colorado School of Mines

Copyright 2008, Society of Petroleum Engineers

This paper was prepared for presentation at the 2008 SPE Europec/EAGE Annual Conference and Exhibition held in Rome, Italy, 9–12 June 2008.

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
To obtain accurate descriptions of fluid flow in the reservoir, it is necessary to include detailed geological information on the
scale of geocellular models. However, computation resources are often overwhelmed by the vast amount of information that
geocellular models contain.
The multiscale approach presented in this paper is designed to include the detailed geological information into the flow
calculations while making computations less expensive. The distinguishing characteristic of this paper lies on the procedure
used to carry out the computations. The procedure splits the computation into pressure solution on a coarser grid and the
saturations and compostions calculations on a very fine scale. This is based the physical and theoretical evidence that
pressure effects travel at much higher velocities than saturation and composition fronts – thus, different scales on
computation. Splitting up the computations through appropriate physical rules, the multiscale approach divides the
computation of pressure propagation and convective processes like the saturation and composition fronts. The pressure
propagation is diffusive in nature, while the saturation and composition computation being transport processes are convective
in nature. These characteristics allow the computation of the solution at different scale levels. The multiscale approach of this
paper is a sequential formulation based on a volume balance model using partial molar volumes. It starts with a
compositional pressure equation based on the overall components balance. Once the pressure solution is obtained, the total
phase and component velocities are calculated on a fine scale to track the saturation and composition fronts.
The multiscale method begins with the use of coarse grid blocks, which contain numerous finer grid blocks (the
geocellular scale). Flux calculations based on the total face velocities at the boundaries of the coarse grid blocks are used to
set up the fine grid calculations for both the pressure and saturations within each coarse grid block for accurate tracking of the
displacement fronts. The finer scale solution is subsequently employed to improve the accuracy of the coarse grid block
boundary conditions. The scheme is intended to capture the fine scale physics of gravity, capillarity, and phase behavior
interactions as well as the global effect of rock heterogeneity (the dual-porosity effects), pore structure, stratigraphy, and
reservoir architecture.

Introduction
The current practice to evaluate reservoir performance is using reservoir simulation. There is an increasing need for adequate
reservoir descriptions in order to obtain reliable estimations of reservoir performance. In fact, when dealing with highly
heterogeneous formations, it is crucial to conform and include all available reservoir data coming from different sources. As
reservoir characterization techniques improve, vast amounts of information are generated to form reservoir models. Typical
reservoir models at geocellular scale may contain 107 – 108 cells. By contrast, reservoir flow simulations are typically run on
upscaled versions of the reservoir models with less of 106 cells (Christie 1996); these limitations considerably depend on the
kind of simulation being performed (dead oil, black oil, compositional) and the available computing resources (Durlofsky,
2005).
Flow models in reservoir simulation consist of coupled, nonlinear partial differential equations. These equations describe
multicomponent, multiphase flow in hydrocarbon reservoirs. The coupled differential equations are discretized, leading to a
sparse coefficient matrix that is usually solved using optimized sparse linear solvers. As the discretization of the reservoir
model increases, classical reservoir flow models become often extremely slow and impractical. In order to overcome this
issue, more efficient numerical schemes have been devised and are a subject of vigorous research (e.g. domain decomposition
methods and multilevel/multiscale methods).
2 SPE 113651

In multiscale methods, the use of local rules on the finest levels resolves the solution at detailed scales, but always
recognizing that these local rules have broader implications at coarser levels. Using repetitive feedback between levels is
essential to improve the final solution since the calculation on one scale impacts the calculation on the others. In the
multiscale method proposed, the saturation and composition fronts are calculated at the finest scale, while the pressure
propagation is calculated at the coarser scale.

Compositional Model
Compositional models are required when it is necessary to describe reservoir fluid phase behavior in a unified manner.
Through compositional modeling behavior of condensates, volatile oils, and gas injection processes can be studied. In this
chapter a review of compositional modeling is presented. Additionally the compositional model based on the volume balance
technique is discussed (Acs et al., 1982; Watts, 1986; Wong et al, 1990).

Compositional Modeling
The molar continuity equations are written for nc hydrocarbon components and one water component. Three phases are
considered, a hydrocarbon gas phase, a hydrocarbon liquid phase and a water phase. The molar continuity equation for each
component is given by

∂ ⎡
uc =
∂t ⎣
( )
φ ξ g S g + ξ o S o + ξ w S w zc ⎤

(1)

where

( )
uc = ∇ ⋅ ycξ g λg k ∇Φ g + xcξo λo k ∇Φ o + wcξ w λw k ∇Φ w + qc (2)

and

qc = ycξ g qg + xcξ o qo + wcξ w qw . (3)

In this work, it is assumed that the water component exists solely in the water phase; while the hydrocarbon components
exist in the hydrocarbon gas phase and hydrocarbon liquid phase only. This can also be expressed as

r
x = ( x1 , x2 ,..., xnc , 0 ) ,
T
(4)

r
y = ( y1 , y2 ,..., ync , 0 ) ,
T
(5)

and

r
w = ( 0, 0,..., 0,1) .
T
(6)

The sum of the mole fractions in each phase satisfies the following relations.

nc

∑y
c =1
c = 1, (7)

nc

∑x
c =1
c =1, (8)

nc

∑w
c =1
c = wnc +1 = 1 , (9)

as well as the sum of the mole fraction in the three phases,


SPE 113651 3

nc

∑z
c =1
c = 1. (10)

The equilibrium constraints are expressed through equality between the fugacities of each component in the hydrocarbon
liquid phase and hydrocarbon gas phase,

f c ,o = f c , g . (11)

Additionally, the phase saturations are constraint by

S g + So + S w = 1 . (12)

Volume Balance Method


The volume balance method reduces the component flow equations to a single flow equation using partial molar volumes
as appropriate weighting factors. There are two ways of implementing the volume balance formulation, IMPES and
sequential approach.

Compositional Pressure Equation


The method starts with the compositional pressure equation based on the overall components balance,

nc +1 ⎡ 1 ∂φ 1 ∂v ⎤ ∂p
∑v u tc c =φ ⎢
⎢ φ ∂p

v ∂p
t ⎥
zc ⎥
∂t
. (13)
c =1 ⎣ t ⎦

In Eq. 13, vtc is the partial molar volume of the multiphase system with respect to component c, ( −1/ vt )( ∂vt / ∂p ) is
zc
the isothermal fluid compressibility, and um is the net molar flux of component c per block volume. Eq. 13 can be solved for
pressure when all other terms are used at the previous time step condition.
The discreatized form of the compositional pressure equation is

⎡ ⎤
1 ⎢⎛⎜ n ⎞
nc +1 n
∂V
∑( ) ( )
⎟ pin +1 − pin − (Vt − V p ) ⎥ .
n
vtcU cn +1 = V p cφ − t (14)
i ∆t ⎢ ⎜ ∂p ⎟ i ⎥
c =1 ⎣⎝ Nc ⎠i ⎦

The term (Vt − V p )i in Eq. 14 serves as a correction term, Watts (Watts, 1986) refers to it as the volumetric error. It
n

merely reflects the fact that the pressures obtained at the previous time level may have not been exactly correct.

Volume Balance Implementation – IMPES


The implementation of the volume balance method starts with the implicit solution of the pressure distribution. Once the
pressure distribution is been solved, the net molar flux of component c is used to compute the change in the number of moles
in a specific block during a time step (Eq. 15),

1 ∂N c
uc = . (15)
Vt ∂t

Using the amount of mass at the new pressure level, a flash calculation provides the saturations and the phase
compositions at the new time level.

Volume Balance Implementation – Sequential Approach


The second approach uses a compositional saturation equation that is analogous to the pressure equation, Eq. 16. This
equation is based on the reservoir volume occupied by an individual phase in instead of the volume occupied by the total
mixture.
4 SPE 113651

nc +1 ⎡ 1 ∂φ 1 ∂v ⎤ ∂p ∂S p
∑ v pc uc = φ ⎢ S p
⎢ φ ∂p

v ∂p
p

⎥ ∂t

∂t
. (16)
c =1 ⎣ t zc ⎦

In Eq. 16, v pc is the partial molar volume of phase p with respect to component c and ( −1/ vt ) ( ∂v p / ∂p ) z is the
c

isothermal phase compressibility. Analogous to the IMPES implementation, the sequential approach starts with the implicit
solution of the pressure distribution. Using the pressure distribution at the new time level and the saturations at the previous
time level, interblock total velocities are computed using Eq. 17.

nc +1 ⎡ 1 ∂φ 1 ∂v ⎤ ∂S
∑ v pc uc = φ ⎢ S p −
p
⎥ ∂p + φ p . (17)
⎢ φ ∂p v ∂p ⎥ ∂t ∂t
c =1 ⎣ t zc ⎦

The calculation of individual phase velocities follows; these can be obtained with the following equations.

{ ⎣ ( ⎦ ) ( )
u g = f g ut − ⎡ λw γ w − λg + λo γ o − γ g ⎤ k ∇D + λw k ∇pcwof − ( λw + λo ) k ∇pcog , } (18)

{ ( )
uo = f o ut − ⎡λw ( γ w − γ o ) + λg γ g − γ o ⎤ k ∇D − λw k ∇pcwo + λg k ∇pcog ,
⎣ ⎦ } (19)

and,

{
uw = f w vt − ⎡⎣ λo ( γ o − γ w ) + λg ( γ g − γ w ) ⎤⎦ k ∇D + ( λo + λgf ) k ∇pcwo + λg k ∇pcog . } (20)

Additionally, the phase velocities can be related to the net molar flux of component c using the following equation.

(
uc = ∇ ⋅ ycξ g u g + xcξ o uo + wcξ wuw + qc . ) (21)

Finally, Eq. 21 is substituted in the compositional saturation equation, Eq. 17, and solved for saturations at the new time
level. It is important to note that using the sequential method would allow to solve for saturations implicitly and obtain
additional stability if needed.
Once the new saturations have been obtained, new time level phase velocities are computed and finally substituted in the
following equation to compute the change in the number of moles in a specific block during a time step.

1 ∂N c
(
∇ ⋅ ycξ g vg + xcξo vo + wcξ w vw + qc = ) Vt ∂t
. (22)

Using the amount of mass at the new pressure level, a flash calculation provides the saturations and the phase
compositions at the new time level.

Volume Balance Method in Naturally Fracture Reservoirs


The following equations are an extension of the volume balance method to dual porosity systems. The fracture and matrix
compositional pressure equations are respectively,

nc +1 nc +1 ⎡ 1 ∂φ 1 ∂v ⎤ ∂p f
∑ (v u ) + ∑ ( vtcτ tc ) f / m = φ ⎢ −
⎢⎣ φ ∂p vt ∂p
t
⎥ (23)
⎦ f ∂t
tc c f
c =1 c =1 zc ⎥

and

nc +1 ⎡ 1 ∂φ 1 ∂v ⎤ ∂p
∑ (v τ ) tc tc f /m
=φ ⎢ −
⎢⎣ φ ∂p vt ∂p
t

⎥ ∂t
m
. (24)
c =1 zc ⎦ m
SPE 113651 5

where,
( uc ) f (
= ∇ ⋅ ycξ g λg k ∇Φ g ) f + ∇ ⋅ ( xcξo λo k∇Φo ) f + ∇ ⋅ ( wcξ wλw k∇Φ w ) f + ( qc ) f , (25)

(τ tc ) f / m = ycξ gτ g + xcξoτ o + wcξ wτ w , (26)


(
⎡ pof − pom ) ⎤
⎥ ⎡ ∂pgm ∂S gm ⎤
τ g ≡ σ km λgf / m ⎢ σ ⎥ = φm ⎢ S gm cg + cφ ( )m + ⎥, (27)
⎢−
⎣ σ
z
(
γ g hgf − hgm + pcgof − pcgom) ( ) ⎥

⎣ ∂t ∂t ⎦


(
⎧ pof − pom ) ⎫
⎪ ⎡ ∂p ∂S ⎤
τ o ≡ σ km λof / m ⎨ σ ⎬ = φm ⎢ Som ( co + cφ )m + om ⎥ ,
om
(28)
⎪+ z
γ o ⎡( hwf − hwm ) − ( hgf − hgm ) ⎤ ⎪ ⎣ ∂t ∂t ⎦
⎩ σ ⎣ ⎦⎭

and

(
⎡ pof − pom ) ⎤

τ w ≡ σ km λwf / m ⎢ σ
⎥ ⎡
⎥ = φm ⎢ S wm cw + cφ ( )m ∂p∂wm ∂S ⎤
+ wm ⎥ . (29)
⎢+
⎣ σ
z
(
γ w hwf − hwm − pcwof − pcwom ) ( ) ⎥

⎣ t ∂t ⎦

The derivation of the transfer function is presented in the Appendix.

Multiscale Implementation
The multiscale implementation of the volume balance formulation is described in this section. We start by defining the
basic multiscale concepts and then build towards the details of flow calculations.

Multiscale Volume Balance formulation


The basic premise of this thesis lies on the fact that the pressure calculation can be decoupled from the transport
calculations, i.e. saturation and composition fronts. This decoupling is possible by the use of the volume balance formulation
described above, where a compositional pressure equation is derived from appropriate mathematical manipulations that honor
the physics of compositional flow modeling. By observing the behavior of pressure propagation and saturation/composition
propagation, it can be argued that pressure propagates (several hundred ft/day in the well's drainage area) a lot faster than
saturation/composition fronts do (few ft/day). Fig. 1 illustrates the two grid levels used in this work.

uif−1, j +3/ 2 uif, j +3/ 2 uif+1, j +3/ 2

uif−3/ 2, j +1 uif−1/ 2, j +1 uif+1/ 2, j +1 uif+3/ 2, j +1


u Ic, J +1/ 2 f f f
p
i −1, j +1 p i , j +1 p i +1, j +1

uif−1, j +1/ 2 uif, j +1/ 2 uif+1, j +1/ 2

uIc−1/ 2, J pIc, J u Ic+1/ 2, J uif−3/ 2, j uif−1/ 2, j uif+1/ 2, j uif+3/ 2, j


p f
i −1, j
pif, j p f
i +1, j

uif−1, j −1/ 2 uif, j −1/ 2 uif+1, j −1/ 2

uif−3/ 2, j −1 uif−1/ 2, j −1 uif+1/ 2, j −1 uif+3/ 2, j −1


f f f
p
i −1, j −1 pi , j −1 p i +1, j −1
u Ic, J −1/ 2

uif−1, j −3/ 2 uif, j −3/ 2 uif+1, j −3/ 2

Fig. 1: Diagram showing a coarse grid block and its corresponding fine grid blocks.
6 SPE 113651

The coarser grid level is defined using the premise that pressure propagation, which is diffusive in nature, occurs a lot
faster when compared to the transport processes. Using this attribute, it is possible then to calculate the pressure solution at
the center of the coarse grids using upscaled properties.
The finer grid level is defined by the same principle. Saturation/composition fronts are slow in nature and it is of interest
to describe them in a finer scale using most of the underlying geological information as possible. By the nature of the
formulation, it is necessary to define the velocity at the finer scale in order to compute the saturation/compositions fronts.
Most of this section will dwell on the interactions and definitions of the quantities computed at the coarser and finer
levels.

Coarse Grid Level


The basic use of the coarse grid is to establish a pressure field solution. Discretizing the compositional pressure equation
on the coarse grid requires the averaging (upscaling) of rock and fluid properties.
Rock properties are calculated a single time at the beginning of the simulation and they are not considered a function of
time. The calculation is performed using the common averaging techniques. For instance, porosity is averaged using bulk
volume weighted arithmetic average. In the case of absolute permeability, a single phase flow average technique is used. The
flow average technique is usually referred to as a pressure solver method, where a single phase flow simulation is performed
in the fine grid level with specified boundary conditions to obtain an effective value of permeability at the coarse grid level
(Christie, 1996).
Fluid properties are calculated for every time step. The basic premise is that number of moles in every component is
conserved. In other words, the number of moles of component c at a particular coarse grid block equals the sum of the
number of moles of component c in the finer grid blocks, Eq. 30.

nbf

(N ) = ∑( N )
c
c
i
c
f
i
. (30)
i =1

Once the number of moles is been determined, the mixture is flashed to obtain the mole fractions of component c in the
liquid and gas phases and the saturations of each phase at the coarse grid level. Subsequently, the saturations of each phase
are used to compute saturation functions at the coarse grid blocks.
With all the coarse grid properties defined, it is possible to obtain the solution of the pressure field at the coarse grid level
using Eq. 14.

Transfer Betwen Levels


The calculated pressures at the coarse grid are prolonged to the finer grid through the following procedure:
1. Coarse intergrid molar fluxes for each component c are calculated using the current pressure solution.
2. The coarse intergrid molar fluxes for each component c are distributed along each phase on the fine grid level using
the transmissibilities as the weighting function.
3. The fine intergrid molar fluxes are used as a boundary condition to calculate the pressure solution within each coarse
grid.
In practice, each coarse grid is composed of 5×5×Kmax or 7×7×Kmax fine grid blocks. Kmax is the number of reservoir
layers in the vertical direction, the reason to this approach is to account for the gravity and capillarity interactions in the
vertical direction which often dominates the recovery mechanisms in natural fractured reservoirs.

Fine Grid Level


The transport calculations are performed at this level. Having obtained the fine grid pressure approximation at the current
time level, it is possible then to compute the new number of moles at every fine grid block. With the number of moles, the
saturation/composition fronts can be calculated at the new time level by using either the IMPES or the sequential
implementation explained above.

Nomenclature
c Compressibility [psi-1]
D Depth [ft]
k Absolute permeability [md]
nc Pseudo components considered in the hydrocarbon description [ ]
np Number of phases [ ]
Np Number of moles of hydrocarbon phase p [lb-mole]
Nc,p Moles of component c in phase p[lb-mole]
Nt Total Moles [lb-mole]
p Pressure [psi]
SPE 113651 7

q Reservoir flow rate per rock volume [1 /day] or [lb-mole/(ft3⋅day)]


Q Reservoir flow rate [ft3 /day] or [lb-mole/day]
R Gas constant, ft3⋅psi/(lb-mole⋅˚R) (10.731)
T Temperature, ˚R (˚F + 459.67)
uc Net molar flux of component c per pore volume [(lb-mole/day)/ft3]
up Phase Velocity [ft/day]
Uc Net molar flux of component c [lb-mole/day]
V Volume [ft3]
v Specific molar volume [ft3/lb-mole]
vtc Partial molar volume with respect to component c [ft3/lb-mole]
xc Mole fraction of component c in the hydrocarbon liquid phase [ ]
yc Mole fraction of component c in the hydrocarbon gas phase [ ]
zc Mole fraction of component c in the three phases combined [ ]

Greek
φ Porosity [ ]
Φ Potential [psi]
γ Fluid gradient [psi/ft]
µ Viscosity [cp]
λ Fluid relative mobility [cp-1]
π Pi value
ρ Density [lb/ft3]
σ Shape Factor [1/ft2]
ξ Molar density [lbmole/ft3]
∇ Gradient operator
∇⋅ Divergence operator

Subscripts
B Bottom
c cth component
f Fracture
g Gas phase
i x-direction index for the fine grid
I x-direction index for the coarse grid
j y-direction index for the fine grid
J y-direction index for the coarse grid
k z-direction index for the fine grid
K z-direction index for the coarse grid
m matrix
o Oil phase
p Phase or Pore
r Rock
t Total (oil + gas + water)
T Top
w Water phase
x x-direction
y y-direction
z z-direction
φ Porosity

Superscripts
c Coarse level
f Fine level
(l) Current iteration level
( l + 1 ) New iteration level
n Current time step
n + 1 New time step
8 SPE 113651

n Current time step


±½ Boundary of a block

Acknowledgments
We would like to aknowledge the continuing support by Marathon Oil Company to conduct this research.

References
Acs, G., Doleschall, S., and Farkas, E.: “General Purpose Compositional Model,” SPEJ (Aug. 1985) 543-553.
Christie, M.A.: “Upscaling for Reservoir Simulation,” SPE JPT (Nov. 1996) 1004–1010.
Durlofsky, L. J.: “Upscaling and Gridding of Fine Scale Geological Models for Flow Simulation,” 8th International Forum on Reservoir
Simulation, Iles Borromees, Stresa, Italy, June 20-24, 2005.
Heinemann, Z.E. and M.G. Mittermeir: “Rigorous Derivation of the Kazemi-Gilman-Elsharkawy Generalized Dual-Porosity Shape
Factor,” paper B044, presented at the 10th European Conference on the Mathematics of Oil Recovery, Amsterdam, the Netherlands,
Sept. 4 – 7, 2006.
Kazemi, H., Gilman, J.R.: “Multiphase Flow in Fractured Petroleum Reservoirs,” in: Bear J., Tsang C.F., de Marsily G., editors. Flow and
Contaminant Transport in Fractured Rock, Academic Press, San Diego, (1993) 267-323.
Kazemi, H., Gilman, J.R.: “Improved Calculations for Viscous and Gravity Displacement in Matrix Blocks in Dual-Porosity Simulation,”
JPT (Jan. 1988) 60-70.
Kazemi, H., Merrill, L. S. Jr., Porterfield, K. L., Zeman, P. R.: “Numerical Simulation of Water-Oil Flow in Naturally Fractured
Reservoirs”, SPEJ (Dec. 1976) 317-326.
Ramirez, B., Kazemi, H., Al-Kobaisi M., Ozkan, E., and Atan, S.: “A Critical Review for Proper Use of Water-Oil-Gas Transfer Functions
in Dual-Porosity Naturally Fractured Reservoirs – Part I,” SPE 109821, 2007 SPE Annual Technical Conference and Exhibition held
in Anaheim, California, U.S.A., 11–14 November 2007.
Watts, J. W.: “A Compositional Formulation of the Pressure and Saturation Equations,” SPE Res. Eng. Jour., May 1986, 243-252.
Whitson C. and Brule M.: Phase Behavior, SPE Monograph Volume 20, Henry L Doherty Series, Richardson, Texas (2000).
Wong, T. W., Firoozabadi, A., and Khalid Aziz: “Relationship of the Volume-Balance Methods of Compositional Simulation to the
Newton-Raphson Method,” SPE Res. Engr. Jour., Aug. 1990, 415-422.

Appendix – Potential Difference between the Fracture and Matrix Media


The transfer functions for a three phase system are defined as:


τ g ≡ ρ gmσ km λgf / m ⎡⎣Φ gf − Φ gm ⎤⎦ =
∂t
(
φρ g S g )m , (1)


τ o ≡ ρomσ km λof / m ⎡⎣Φ of − Φ om ⎤⎦ = (φρo So )m , (2)
∂t

and,


τ w ≡ ρ wmσ km λwf / m ⎡⎣Φ wf − Φ wm ⎤⎦ = (φρ w Sw )m . (3)
∂t

To define the potential difference between the matrix and the fracture for each phase, the following ideal segregation of
fluids is assumed.
SPE 113651 9

DT
Gm Gm
hgm
Gf hgf

Om
hom Om Of hof

Wf Wm hwf
Wm
hwf
DB

Fig. 2: Idealized segregation of fluids between fracture and matrix.

Using Fig. 2, the following average phase potentials are defined:

(
Φ wf = pwf − γ w DB − hwf , ) (4)

Φ wm = pwm − γ w ( DB − hwm ) , (5)

(
Φ gf = pgf − γ g DT + hgf , ) (6)

(
Φ gm = pgm − γ g DT + hgm . ) (7)

Additionally, the potentials in the oil phase are created by difference,

( ) (
Φ of = pof − γ o ⎡ DB − hwf + DT + hgf ⎤ ,
⎣ ⎦ ) (8)

⎣ ⎦(
Φ om = pom − γ o ⎡( DB − hwm ) + DT + hgm ⎤ . ) (9)

Using Eq. 35 through Eq. 38, the average potential difference between the fracture and the matrix are defined as,

( )
Φ gf − Φ gm = pgf − pgm − γ g hgf − hgm , ( ) (10)

( ) (
Φ of − Φ om = pof − pom + γ o ⎡ hwf − hwm − hgf − hgm ⎤ ,
⎣ ⎦ ) ( ) (11)

( )
Φ wf − Φ wm = pwf − pwm + γ w hwf − hwm . ( ) (12)

Substituting back the flow potential difference definitions into the transfer function equations and introducing a correction
for the gravity term, we obtain

⎡ σz ⎤ ∂
τ g ≡ ρ gmσ km λgf / m ⎢( pgf − pgm ) − γ g ( hgf − hgm ) ⎥ = (φρ g S g ) , (13)
⎣ σ ⎦ ∂t m
10 SPE 113651

(⎧ pof − pom

) ⎫
⎪ ∂
τ o ≡ ρomσ km λof / m ⎨ σ ⎬ = (φρo So )m , (14)
⎪+
⎩ σ
z
(
⎣ ) (
γ o hwf − hwm − hgf − hgm ⎪ ∂t
⎡ ⎤
⎦⎭ )
⎡ σz ⎤ ∂
τ w ≡ ρ wmσ km λwf / m ⎢( pwf − pwm ) + γ w ( hwf − hwm ) ⎥ = (φρ w S w ) m . (15)
⎣ σ ⎦ ∂t

Você também pode gostar