5/28/2024

RESERVOIR SIMULATION
MULTIDIMENSIONAL RESERVOIR
DOMAINS, THE CONTROL VOLUME
APPROACH, AND HETEROGENEITIES

GRID BLOCK INDEXING IN 2D AND 3D

Dr. Khoozan 2

1
 5/28/2024

GRID BLOCK INDEXING IN 2D AND 3D

Domain is discretized into 𝑁𝑥 grids in the

𝑥-direction, 𝑁𝑦 grids in the 𝑦-direction,
and 𝑁𝑧 grids in the 𝑧-direction (for 3D).

The total number of grids is 𝑁 = 𝑁𝑥 ×

𝑁𝑦 × 𝑁𝑧 .

The reservoir has a length, 𝐿, width, 𝑤,

and thickness, ℎ.

Dr. Khoozan 3

GRID BLOCK INDEXING IN 2D AND 3D

The block can be defined with a 𝑗, 𝑘, 𝑙 integer index, where
𝑗 represents the grid index in the 𝑥-direction, 𝑘 the grid
index in the 𝑦-direction, and 𝑙 the grid index in the 𝑧-
direction.

It is also useful to have a single index defining the block and

here the integer 𝑖 is used.

𝑖 = 1 is defined at the bottom left corner (𝑗 = 1, 𝑘 = 1, 𝑙 =

1) and increases from left to right, then from bottom to
top, and finally from front to back.

In the right, upper, back corner, 𝑖 = 𝑁 = 𝑁𝑥 × 𝑁𝑦 × 𝑁𝑧.

Dr. Khoozan 4

2
 5/28/2024

GRID BLOCK INDEXING IN 2D AND 3D

The 𝑖-indexing can be mapped from 𝑗, 𝑘, 𝑙-
indexing using

It is often convenient to refer to a block

using 𝑖-indexing, but refer to the interface
between two blocks using 𝑗, 𝑘, 𝑙 indexing.

For example, the interface between blocks

𝑗 − 1, 𝑘, 𝑙 and 𝑗, 𝑘, 𝑙 is 𝑗 − 1/2, 𝑘, 𝑙.

Dr. Khoozan 5

GRID DIMENSIONS

The dimensions of a grid, 𝑖, is given as ∆𝑥𝑖 , ∆𝑦𝑖 , and ∆𝑧𝑖 and the bulk volume of the block as
𝑉𝑖 = ∆𝑥𝑖 × ∆𝑦𝑖 × ∆𝑧𝑖 .

If the reservoir can be approximated as a 2D areal problem, ∆𝑧𝑖 = ℎ , the thickness, and for 1D
problems, ∆𝑧𝑖 = ℎ and ∆𝑦𝑖 = 𝑤 , the width.

A simple approach to gridding uses uniform. However, it is often desired to use nonuniform
grid block sizes to optimize accuracy and computational speed.

Relatively small grid blocks are desired where changes in variables, e.g., pressure, are
significant over short distances, e.g., near a well, and larger grid blocks where pressure does
not change much with distance, e.g., far from a well.
Dr. Khoozan 6

3
 5/28/2024

GRID DIMENSIONS
The figure shows that fine (small) grids are employed near the wells and coarse grids far
from the wells as desired. However, fine grids are also used in some places where not
needed. Therefore, this gridding algorithm is not ideal, but advanced techniques are
beyond the scope of this text.

Dr. Khoozan 7

IRREGULAR GEOMETRY AND INACTIVE GRIDS

The geometry of naturally occurring reservoirs is irregular and therefore the
domain cannot, in general, be accurately described by cuboids. Instead, grids
are employed to approximate the reservoir geometry.

Dr. Khoozan 8

4
 5/28/2024

IRREGULAR GEOMETRY AND INACTIVE GRIDS

Two methods for implementing orthogonal grids in a noncuboid geometry include: (a) inactive
grids and (b) reindexing.

In inactive grids approach, the domain is modeled by a cuboid geometry, but blocks outside the
reservoir are considered “inactive”; they are assigned nonphysical parameters such as zero
porosity and/or permeability.

Dr. Khoozan 9

IRREGULAR GEOMETRY AND INACTIVE GRIDS

Numerically, matrices of size 𝑁 × 𝑁 are still generated, but rows and columns corresponding to the
inactive grid contain the nonphysical values.

The system of equations may be singular or ill-conditioned when using inactive grids. However,
the numerical value of pressure in the inactive grids is not meaningful and iterative solvers should
still converge to the solution in the active grid blocks.

Dr. Khoozan 10

5
 5/28/2024

IRREGULAR GEOMETRY AND INACTIVE GRIDS

The alternative is to reindex the grids in such a way that only active blocks are assigned a block number.

As a result, the number of equations is smaller than the inactive blocks method; however, the connectivity
between blocks is not as simple. The connectivity of blocks must be determined and saved in a reference
list.

Dr. Khoozan 11

SINGLE-PHASE FLOW IN MULTIDIMENSIONS AND THE

CONTROL VOLUME APPROACH
The general diffusivity equation for single-phase (𝛼) flow in 3D (Cartesian),
with anisotropic, heterogeneous permeability, gravitational effects, and
sources/sinks can be expanded from the general form:

Dr. Khoozan 12

6
 5/28/2024

SINGLE-PHASE FLOW IN MULTIDIMENSIONS AND THE

CONTROL VOLUME APPROACH

Endpoint relative permeability is included because only one flowing

phase is assumed here.

This equation can be discretized using finite differences using the

approaches in Chapter 3.

An alternative is to use the control-volume approach to develop mass

balance equations for each block. The same 𝑁 algebraic equations
result in both approaches.
Dr. Khoozan 13

SINGLE-PHASE FLOW IN MULTIDIMENSIONS AND THE

CONTROL VOLUME APPROACH
In the control volume approach, the reservoir domain is first discretized into
grid blocks of finite size and then mass balance equations are imposed on
each grid.

Dr. Khoozan 14

7
 5/28/2024

SINGLE-PHASE FLOW IN MULTIDIMENSIONS AND THE

CONTROL VOLUME APPROACH

A mass balance (in − out + generation − consumption = accumulation) on each

phase a can be imposed on each control volume (grid) as was done in Chapter 2.

Mass flux, 𝜌𝛼 𝑢𝛼 , enters or exits the block from six faces (in 3D).

Mass is generated/consumed by wells.

Mass is accumulated by changes in density, porosity, and saturation.

Dr. Khoozan 15

SINGLE-PHASE FLOW IN MULTIDIMENSIONS AND THE

CONTROL VOLUME APPROACH

Dr. Khoozan 16

8
 5/28/2024

ACCUMULATION
The accumulation term can be simplified by first recognizing that the time
difference can be expanded:

The porosity and density (for slightly compressible fluids) differences in time
can be simplified by using a first-order Taylor Series approximation:

Dr. Khoozan 17

ACCUMULATION
Analogous equations for the density change for a compressible gas can be derived using a
real gas law. These Eqs can be substituted to give:

In this equation, an approximation was made that, for a slightly compressible fluid, the
porosity and density could be treated as constants and the superscripts for time index
were removed. The pressure/time-dependence on these variables was accounted for in
the previous equation.

Dr. Khoozan 18

9
 5/28/2024

FLUX TERMS
There are six flux terms (one for each face of the block) in a three-
dimensional balance. Darcy’s law can be substituted for velocities in Eq. (4.3).
For example, in the 𝑥-direction at the interface of blocks 𝑗 − 1, 𝑘, 𝑙 and 𝑗, 𝑘, 𝑙:

Dr. Khoozan 19

FLUX TERMS
In this equation an interblock phase (Ta) transmissibility is defined, which includes an
interblock formation volume factor in the denominator. The units of this interblock
transmissibility are standard conditions (e.g., scf/psi-day).

The total (T) transmissibility is the interblock phase transmissibility multiplied by the block
formation volume factor, summed over all phases. If only one phase is flowing the
summation is not required. The units of the total transmissibility are reservoir conditions
(e.g., ft3/psi-day).

Dr. Khoozan 20

10
 5/28/2024

FLUX TERMS

Formation volume factor of the phase appears in both the numerator and
denominator of total compressibility, but the block and interblock formation
volume factors are not, in general, equal because they are pressure dependent.

for slightly compressible fluids they can be approximated as equal.

For compressible fluids, such as gases, the difference between the two
formation volume factors may not be negligible.

Dr. Khoozan 21

SOURCES AND SINKS (WELLS)

Mass can be generated or consumed in a grid block through wells and,
therefore, are implemented as sources (injection wells) or sinks (producer
wells). Usually, the well rate is measured at surface conditions (𝑠𝑐) instead of
reservoir conditions (𝑟𝑐) and formation volume factor is needed for the
conversion:

Dr. Khoozan 22

11
 5/28/2024

SINGLE-PHASE FLOW
The derived accumulation, flux and source/sink terms can be substituted into the phase
balance equation to give:

where 𝐴𝑖 = 𝑉𝑖 𝜙𝑖 ΤΔ𝑡. The shorthand notation is defined as follows where where 𝑈 is any
spatially or time-dependent variable, e.g., 𝑃, 𝐷 or 𝑆𝛼 :

Dr. Khoozan 23

SINGLE-PHASE FLOW
All phases can be summed together to give:

and 𝑐𝑡 is the saturated-weighted total compressibility. Time differences in saturation are

eliminated in the pressure equation because saturations of all phases must sum to unity.

If there is only one flowing phase, the summation of phases on the left-hand side is not
necessary because production rates are only finite for the flowing phase.

For simplicity, the phase density of the flowing phase is assumed constant in the gravity
terms.

Dr. Khoozan 24

12
 5/28/2024

SINGLE-PHASE FLOW
In Eq. (4.12), the time level for pressure in the flow terms was not specified pressures are
evaluated at the 𝑛 time level for the explicit method, 𝑛 + 1 time level for the implicit
method, and an intermediate time level for mixed methods:

Dr. Khoozan 25

SINGLE-PHASE FLOW

The matrix 𝐽 is used for boundary conditions and constant

bottomhole pressure wells.

The transmissibility matrix is tridiagonal in 1D, pentadiagonal in 2D,

and heptadiagonal in 3D.

Importantly, the phase transmissibility, 𝑇𝛼 , is symmetric but 𝑇 is only

strictly symmetric for the case that formation volume factor is
uniform.
Dr. Khoozan 26

13
 5/28/2024

CONSTANT RATE WELLS

Here, we consider only wells, injectors or producers, in which the rate is specified (in
standard conditions).

The grid block, 𝑖, that contains the well is included as a source term, 𝑄𝑖 = 𝐵𝛼,𝑖 𝑄𝑖𝑠𝑐 , where
the specified standard rate is converted to reservoir conditions using the formation
volume factor.

The convention used here is that injector wells have a positive rate and producers a
negative rate.

For our purposes, the location of the well in the grid does not affect the value of 𝑄𝑖 ;
however, good practice is to place the grids in such a way that the wells are near the grid
centers.
Dr. Khoozan 27

NEUMANN BOUNDARY CONDITIONS

Neumann conditions are those of constant specified rate on a boundary and can be
positive (into the reservoir), negative (out of the reservoir), or zero (a no-flux condition).

Therefore, Neumann boundary conditions can be treated similarly to constant rate wells.

The specified rate at the boundary can be directly included into the source vector, 𝑄𝑖 ,
where 𝑖 refers to the block with the specified boundary condition.

Dr. Khoozan 28

14
 5/28/2024

NEUMANN BOUNDARY CONDITIONS

Since boundary fluxes are generally specified at reservoir conditions, they do

not need to be converted from standard conditions using formation volume
factors.

A common Neumann boundary condition is one of zero flux (sealed

boundary), in which case 𝑄𝑖 = 0.

Dr. Khoozan 29

NEUMANN BOUNDARY CONDITIONS

If the specified rate is defined across several blocks or an entire boundary (in
2D or 3D) the flux must be distributed among those grids.

One simple method is to distribute evenly between the blocks, but this can
be inaccurate if the transmissibility varies significantly along the boundary
blocks because fluid preferentially flows through the blocks of highest
transmissibility.
A more accurate approach is to weight the fluxes by the block transmissibility

Dr. Khoozan 30

15
 5/28/2024

NEUMANN BOUNDARY CONDITIONS

The flux at the boundary edge is defined by Darcy’s law; for example, at the
left boundary:

The flux is defined as zero, even for a finite flux boundary condition, because
the specified rate was already included in 𝑄.

Setting the interblock transmissibility to zero forces the expression to be true

and therefore interblock transmissibilities at Neumann boundaries are zero
and included as such in the Transmissibility matrix, 𝑇.

Dr. Khoozan 31

DIRICHLET CONDITIONS
Dirichlet (constant pressure) boundary conditions were implemented in Chapter 3 by
approximating the boundary pressure as the arithmetic average of the pressures of the
imaginary block and boundary block (e.g., 𝑃𝐵 = (𝑃0,𝑘,𝑙 +𝑃1,𝑘,𝑗 )Τ2).

Equivalently, one can rewrite Darcy’s law at the boundary using the boundary pressure:

where the “2” arises because the distance from the boundary edge to the center of the
block equals to half of the length.

Dr. Khoozan 32

16
 5/28/2024

DIRICHLET CONDITIONS

The interblock transmissibility is then equal to twice the

transmissibility of the block, i, and the result is that the
transmissibility matrix includes a “2𝑇𝑖 ” in the main diagonal.

Mathematically, a diagonal matrix, 𝐉, is used which includes +2𝑇𝑖

in the main diagonal for any block that has a constant pressure
boundary condition.

In the source vector, 𝑄𝑖 = +2𝑇𝑖 𝑃𝐵 .

Dr. Khoozan 33

CORNER BLOCKS

2D domains have four boundary faces and 3D domains have six.

Blocks that are adjacent to more than one boundary are referred
to here as “corner blocks.”

It may seem like blocks on corners in 2D and 3D domains need to

be handled differently than other boundary blocks.

However, corner blocks are no different than other boundary

blocks. They are just subject to more than one boundary condition.
Dr. Khoozan 34

17
 5/28/2024

EXAMPLE 4.1. MASS BALANCE EQUATIONS

Dr. Khoozan 35

EXAMPLE 4.1. MASS BALANCE EQUATIONS

Dr. Khoozan 36

18
 5/28/2024

EXAMPLE 4.1. MASS BALANCE EQUATIONS

Dr. Khoozan 37

EXAMPLE 4.1. MASS BALANCE EQUATIONS

Dr. Khoozan 38

19
 5/28/2024

EXAMPLE 4.1. MASS BALANCE EQUATIONS

Dr. Khoozan 39

EXAMPLE 4.1. MASS BALANCE EQUATIONS

Dr. Khoozan 40

20
 5/28/2024

EXAMPLE 4.1. MASS BALANCE EQUATIONS

Dr. Khoozan 41

EXAMPLE 4.1. MASS BALANCE EQUATIONS

Dr. Khoozan 42

21
 5/28/2024

EXAMPLE 4.1. MASS BALANCE EQUATIONS

Dr. Khoozan 43

EXAMPLE 4.1. MASS BALANCE EQUATIONS

Dr. Khoozan 44

22
 5/28/2024

INITIAL CONDITIONS

Initial pressure of reservoir blocks as stated in Chapter 1 is

determined by

• The depth of the grid blocks

• Density of the fluid
• A reference pressure at a reference depth (e.g., surface
or water-oil contact line)

Dr. Khoozan 45

RESERVOIR HETEROGENEITIES
In most practical applications, heterogeneities in permeability, porosity, grid
size, fluid properties, etc., must be accounted for in the reservoir simulator.
This figure depicts a 1D reservoir with variable grid block sizes and
permeability.

Dr. Khoozan 46

23
 5/28/2024

RESERVOIR HETEROGENEITIES
Flow occurs from one block (e.g., 𝑗 + 1, 𝑘, 𝑙) to another block (e.g., 𝑗, 𝑘, 𝑙). Therefore, if
those two blocks have different permeability, grid size, cross-sectional area, fluid
properties, etc., then the interblock phase transmissibility should be some sort of mean
between the two blocks

In this equation, endpoint relatively permeability of the one flowing phase, 𝛼, is used but
in later chapters it is shown the equation can be used for multiphase flow where relative
permeability is a function of saturation.

Dr. Khoozan 47

RESERVOIR HETEROGENEITIES

0
Interblock fluid dependent properties (𝑘𝑟,𝛼 , 𝜇𝛼 ) are generally treated
differently than geometric properties (𝑘, Δ𝑥, Δ𝑦, Δ𝑧) and thus are
separated.

In this equation, transmissibility is in the 𝑥-direction, so 𝑥-directional

permeability is used along with Δ𝑥 as the distance between blocks
and Δ𝑦Δ𝑧 the cross-sectional area.

For flow in the 𝑦- and 𝑧-direction, it must be adjusted accordingly

Dr. Khoozan 48

24
 5/28/2024

FLUID PROPERTIES

Viscosity and formation volume factor may be assumed a constant or be given as a

function of pressure.

For slightly compressible fluids (e.g. aqueous, oleic phases), viscosity and formation
volume factor are weak functions of pressure.

Pressure-dependent properties may vary in both space and time; they can be evaluated in
each grid block using the known pressure in the previous timestep, 𝜇𝑖 = 𝑓(𝑃𝑖𝑛 )

Endpoint relative permeability is usually assumed to be constant, although could vary

spatially if the rock type or wettability varied in the reservoir or with time if the
wettability changed.
Dr. Khoozan 49

FLUID PROPERTIES

For single-phase flow problems, an arithmetic mean for the interblock fluid
properties is often an acceptable approximation. .

For multiphase flow problems, a technique referred to as upwinding is used

to evaluate the interblock fluid properties, but this is left for discussion in
later chapters.

Dr. Khoozan 50

25
 5/28/2024

GEOMETRIC PROPERTIES
Geometric properties (e.g., 𝑘, Δ𝑥, Δ𝑦, Δ𝑧) can vary significantly between two adjacent grid
blocks, even orders of magnitude in the case of permeability.

Equations for interblock geometric properties are derived by assuming steadystate flow
between two adjacent blocks.

Interblock transmissibility can be written as follows by assuming the flow rate, 𝑞, is

constant through the two adjacent blocks,

Dr. Khoozan 51

GEOMETRIC PROPERTIES

where 𝑃𝑗+1Τ2,𝑘,𝑙 is the pressure at the interface between the two

adjacent blocks.

The total pressure drop between the two blocks is the sum of the
pressure drop from the center of block 𝑖 (𝑗, 𝑘, 𝑙) to the interface
and the interface to block 𝑖 + 1 (𝑗 + 1, 𝑘, 𝑙).

This formula is actually a harmonic mean.

Dr. Khoozan 52

26
 5/28/2024

GEOMETRIC PROPERTIES
Assuming the interblock fluid properties can be determined separately, this
equation can be simplified to give a harmonic mean of geometric properties:

Dr. Khoozan 53

GEOMETRIC PROPERTIES

The harmonic mean for interblock transmissibility was derived mathematically but it also
makes sense physically.
Consider an extreme case, where a relatively high permeability block (e.g., a sandstone
with 100 mD) is adjacent to an impermeable block (e.g., shale with 0 mD) and grids are
of uniform size.
A simple arithmetic mean suggests that the effective permeability is (100 + 0)/2 =
50 mD. and fluid flows, relatively easily, between the two blocks.

In reality, no flow would occur at all between the blocks because the shale block is
impermeable.

In the example of a block of 100 mD sandstone adjacent to 0 mD shale, the harmonic

mean of permeability is 0 mD, just as one would expect.
Dr. Khoozan 54

27
 5/28/2024

GEOMETRIC PROPERTIES

The harmonic mean of permeability is exact for 1D, steady-state problems.

For 2D and 3D problems, the harmonic mean can underestimate the effective
permeability or transmissibility.

Advanced approaches for upscaling permeability have been developed (Christie, 1996;
Holden & Nielsen, 2000; Durlofsky, 2005), but this is beyond the scope of this course.

Upscaling is also commonly used to coarsen grids when fine grids are provided in
order to improve computational efficiency.

Dr. Khoozan 55

EXAMPLE 4.2. INTERBLOCK TRANSMISSIBILITIES

Dr. Khoozan 56

28
 5/28/2024

EXAMPLE 4.2. INTERBLOCK TRANSMISSIBILITIES

Dr. Khoozan 57

EXAMPLE 4.2. INTERBLOCK TRANSMISSIBILITIES

Dr. Khoozan 58

29
 5/28/2024

EXAMPLE 4.2. INTERBLOCK TRANSMISSIBILITIES

Dr. Khoozan 59

EXAMPLE 4.2. INTERBLOCK TRANSMISSIBILITIES

Dr. Khoozan 60

30
 5/28/2024

EXAMPLE 4.2. INTERBLOCK TRANSMISSIBILITIES

Dr. Khoozan 61

EXAMPLE 4.2. INTERBLOCK TRANSMISSIBILITIES

Dr. Khoozan 62

31
 5/28/2024

EXAMPLE 4.2. INTERBLOCK TRANSMISSIBILITIES

Dr. Khoozan 63

ACCUMULATION TERMS

Porosity, compressibility, and block volume may also vary with grid
blocks.

These variables are included in the accumulation term of the mass

balance (i.e., 𝐀) and interblock values are not needed.

The diagonal elements of the accumulation matrix can be

computed as follows:

Dr. Khoozan 64

32
 5/28/2024

MATRIX ARRAYS: ACCUMULATION AND COMPRESSIBILITY

The accumulation, 𝐀, and compressibility, 𝐜𝐭 , arrays are both 𝑁 × 𝑁 diagonal

arrays with each diagonal element corresponding to the respective block.

𝐴𝑖,𝑖 is given by:

Compressibilities, 𝑐𝑡,𝑖,𝑖 are a saturation-weighted sum of fluid and formation

compressibilities:

Dr. Khoozan 65

MATRIX ARRAYS: TRANSMISSIBILITY

The transmissibility, 𝐓, matrix is tri-diagonal, penta-diagonal, and hepta-diagonal in

1D, 2D, and 3D, respectively.

The 𝑇𝑖,𝑗 element of the matrix is the negative of interblock transmissibility between
blocks 𝑖 and 𝑗 and thus 𝑇𝑖,𝑗 is only nonzero if the two blocks are adjacent (or 𝑖 = 𝑗).

For a sufficiently large number of grids, most entries of 𝐓 are zero and thus the matrix
is sparse.

The main diagonal, 𝑇𝑖,𝑖 , is the absolute value of the sum of the off-diagonals.

Dr. Khoozan 66

33
 5/28/2024

MATRIX ARRAYS: TRANSMISSIBILITY

For homogeneous, isotropic media, all the interblock

transmissibilities are the same and equal to the block
transmissibility.

If the permeability is anisotropic and/or heterogeneous, the

transmissibility would not be constant.

On a given row, 𝑖, each off-diagonal term would be unique and the

main diagonal would be negative the sum of the off-diagonal
terms.
Dr. Khoozan 67

MATRIX ARRAYS: TRANSMISSIBILITY

The 𝐓 matrix, shown below, depicts the penta-diagonal matrix for a
homogeneous 9-block (𝑁𝑥 = 3, 𝑁𝑦 = 3) reservoir:

Dr. Khoozan 68

34
 5/28/2024

MATRIX ARRAYS: TRANSMISSIBILITY

The number of columns between the far-left diagonal and the far-right diagonal is referred to
as the bandwidth.

A higher bandwidth can sometimes require additional computer memory and increase the
computation time for the linear solver.

In the indexing system proposed in this text, the bandwidth for 2D models equals the number
of blocks in the 𝑥-direction (𝑁𝑥 ), but different indexing systems can be used to reduce (or
increase) bandwidth.

It is, however, highly recommended to use sparse storage formatting in code development,
which only saves the nonzero entries.

Dr. Khoozan 69

MATRIX ARRAYS: SOURCE TERMS

For constant rate wells or constant rate boundary conditions, the rate (at reservoir conditions) is
directly substituted into the source vector, 𝑄𝑖 , if the well or boundary is associated with that block.

The value of 𝑄𝑖 is positive for injectors (or flow is into the boundary) and negative for producers
(or flow out of the boundary).

For constant pressure (Dirichlet) boundaries, a diagonal 𝐉 matrix is used with 𝐽𝑖,𝑗 = +2𝑇𝑖 , where 𝑇𝑖
is the transmissibility of that boundary block.

The source vector is 𝑄𝑖 = +2𝑇𝑖 𝑃𝐵 where 𝑃𝐵 is the value of the constant pressure at the boundary
of block 𝑖.

If the block is subject to more than one boundary condition (i.e. corner block), 𝑄𝑖 is a summation
of those boundary conditions.

Dr. Khoozan 70

35
 5/28/2024

MATRIX ARRAYS : GRAVITY

Reservoirs are usually not horizontal and therefore gravity effects

must be included. This figure shows a reservoir with varying depth.

Dr. Khoozan 71

MATRIX ARRAYS : GRAVITY

Gravity was included in the balance equations through Darcy’s law and a gravity vector, 𝐺,
was included in the matrix equations,

where 𝐷 is a vector of depths and 𝐷𝑖 is the depth of the center of grid 𝑖. Note that here the
fluid density has been treated as a constant. If the density is taken to be function of
pressure, interblock densities (or formation volume factors) should be used.

Dr. Khoozan 72

36
 5/28/2024

EXAMPLE 4.3. SOLUTION TO PRESSURE FIELD IN 2D

Dr. Khoozan 73

EXAMPLE 4.3. SOLUTION TO PRESSURE FIELD IN 2D

Dr. Khoozan 74

37
 5/28/2024

EXAMPLE 4.3. SOLUTION TO PRESSURE FIELD IN 2D

Dr. Khoozan 75

EXAMPLE 4.3. SOLUTION TO PRESSURE FIELD IN 2D

Dr. Khoozan 76

38
 5/28/2024

EXAMPLE 4.3. SOLUTION TO PRESSURE FIELD IN 2D

Dr. Khoozan 77

EXAMPLE 4.3. SOLUTION TO PRESSURE FIELD IN 2D

Dr. Khoozan 78

39
 5/28/2024

EXAMPLE 4.3. SOLUTION TO PRESSURE FIELD IN 2D

Dr. Khoozan 79

EXAMPLE 4.3. SOLUTION TO PRESSURE FIELD IN 2D

Dr. Khoozan 80

40
 5/28/2024

EXAMPLE 4.3. SOLUTION TO PRESSURE FIELD IN 2D

Dr. Khoozan 81

EXAMPLE 4.4. SOLUTION TO PRESSURE FIELD IN 2D WITH GRAVITY

Dr. Khoozan 82

41
 5/28/2024

EXAMPLE 4.4. SOLUTION TO PRESSURE FIELD IN 2D WITH GRAVITY

Dr. Khoozan 83

EXAMPLE 4.4. SOLUTION TO PRESSURE FIELD IN 2D WITH GRAVITY

Dr. Khoozan 84

42