Professional Documents
Culture Documents
SPE 62916
The Green Element Method for Numerical Well Test Analysis
Rosalind A. Archer, SPE, Texas A&M University, and Roland N. Horne, SPE, Stanford University
Abstract
To use pressure transient test data in computerised methods for integrated reservoir charcterisation numerical simulations of the well tests are typically required. Numerical
artifacts occurring in the simulation must be avoided as
much as possible so that they do not adversely affect the
reservoir characterisation.
This work explores the advantages of a hybrid boundary element method known as the Green element method for
modeling pressure transient tests. Boundary element methods are a natural choice for the problem because they are
based on Greens functions, which are an established part of
well test analysis. The classical boundary element method
is limited to single phase flow in homogeneous media. This
works presents formulations which give computationally efficient means to handle heterogeneity. The accuracy of the
scheme is further enhanced by incorporating singularity programming.
Comparisons of the proposed Green element approach to
standard finite difference simulation show that both methods are able to model the pressure change in the well over
time. When pressure derivative is considered however the
finite difference method produces very poor results which
would give misleading interpretations. The Green element
method in conjunction with singularity programming reproduces the derivative curve very accurately.
Introduction
The boundary element method (BEM) was applied by Numbere and Tiab1 to generate steady-state streamlines in sectionally homogenous two-dimensional reservoirs. Masukawa
and Horne2 considered immiscible displacement problems
using BEM. Kikani and Horne3,4 applied BEM to generate
pressure transients in arbitrarily shaped homogeneous reservoirs. The problem of flow in heterogeneous reservoirs was
addressed by Sato and Horne5,6 who developed a perturbation based approach. This approach became increasingly
computationally intensive as the reservoir hetereogeneity
became more pronounced.
The standard form of BEM is not applicable to flow
in heterogeneous media so hybrid boundary element based
schemes were considered. This study focused on the application of the Green Element method (GEM). Taigbenu7,8
first presented GEM in 1990 and described it as an
element-by-element implementation of the boundary element method. Taigbenu considered the Laplace, diffusion, nonlinear Boussinesq and convection-diffusion equations. Taigbenu and Onyejekwe applied GEM to groundwater flow in the unsaturated zone9 .
Archer and Horne10 discussed the application of the
GEM and the Dual Reciprocity Boundary Element Method
to one-dimensional pressure diffusion and tracer flow in heterogeneous media. Archer and Horne11 and Archer12 extended the analysis to two dimensions. Singularity programming was introduced to study well tests. This work compares the performance of the proposed GEM/singularity
programming approach to finite difference simulation of well
tests.
Theory
Treatment of Heterogeneity To use the classical boundary element method the differential equation being considered must include a 2 operator. The single-phase flow
equation:
p
(1)
(kp) = ct
t
is therefore not in a form suitable for solution by a boundary
element method.
Taigbenu and Onyejekwe9 demonstrated how the singlephase flow equation could be solved by rewriting it as:
2 p = lnk p +
ct p
k t
(2)
(5)
Out of regard for numerical stability dimensionless variables were used in the implementation of the singularity
programming. The choice of dimensionless variables in this
work follows that of Sato13 :
pD
pi p
=
pi
x
xD =
A
y
yD =
A
q
qD =
4k0 hpi
k0 t
tD =
ct A
Incorporation of Singularity Programming Masukawa and Horne2 and Sato13 applied singularity programming
in conjunction with boundary element methods to compute
pressure transients. Unlike the current study Satos solution
was performed in Laplace space. Singularity programming
decomposes the solution into singular and nonsingular components:
s
(3)
pD = pns
D + pD
SPE 62916
2 pD 2 psD =
(8)
(9)
(10)
(14)
s
Noting that pD = pns
D + pD Equation (14) can be simplifed
to:
s
pD
ko
k0 pns
2 ns
s
D
1
+
lnk (pns
pD =
D + pD )
k tD
k
tD
(15)
The right hand side of Equation (15) includes both the
singular and nonsingular solutions. This does not present a
problem however since the singular solution is known and
can be incorporated easily into the boundary element solution.
where
f=
(6)
(7)
ko pD
ps
D lnk pD
k tD
tD
k0 pns
D
lnk (pns
D)+f
k tD
ko
1
k
psD
lnk (psD )
tD
(16)
(17)
K2 (r, ri ) D
D
2
n
ZZ
k0 pns
D
G
lnk pns
+
f
d
(18)
=
D
k tD
where
The integrals in Equation (18) are then written as summations over the elements:
pD
k0 pD
=
lnk pD
k tD
(12)
G = K1 = ln(r ri )
K2 =
G(r, ri )
n
M Z
X
i ns
pD (ri ) +
K1 (r, ri )pns
D (r)d
2
e
e=1
M Z
X
e=1
K2 (r, ri )
pns
D (r)
d
n
(19)
(20)
SPE 62916
M ZZ
X
e=1
k0 pns
D
lnk pns
D +f
k tD
d
(21)
ZZ
Uijl =
e
=
Vijl
ZZ
e
G(r, ri )
G(r, ri )Nj Nl d
Nj Nl
Nj Nl
+
d
x x
y y
(25)
G(r, ri )Nj d
(27)
+ Tij fjm+1 = 0
Vijl lnkj pm+1
l
(26)
ZZ
Tij =
(28)
Results
Example 1: Comparison to an Analytical Solution
Example 1: Well Test in a Closed Reservoir To
test the combination of singularity programming and GEM
a pressure transient was computed and compared to a transient generated analytically. The reservoir properties are
given in Table 1. The reservoir dimensions are 2500ft by
3000ft.
The grid used for the GEM solution of the nonsingular
component of the problem was a 10 by 10 mesh. The well
was located in the center of the reservoir. Constant pressure
boundaries were imposed on all sides. The resulting drawdowns are compared to the analytical solution in Figure 1.
The agreement between the GEM and analytical solutions
is excellent.
Examples 2 and 3: Comparison to Finite Difference
Solutions
before sensing the closed reservoir boundary. The nonuniformly gridded cases performed worse than the uniformly
gridded cases in their reproduction of the pressure derivative. The infinite-acting radial flow period did not appear.
Instead of remaining level during the time corresponding to
infinite-acting radial flow the derivative maintains a steady
downward slope.
The effects of the errors in the pressure derivative curve
in this suite of finite difference simulations were quantified
by treating each one as data in a well test analysis. The
interpretation was performed using a standard regression
procedure. The estimates for skin and permeability from
the regression are highly correlated so the skin was set to
zero in each case to ensure the treatment of the permeability
was consistent. The initial pressure was also fixed at its
true value of 2000psi. The wellbore storage coefficient was
set to zero. Regression was used to determine the reservoir
permeability and the location of the boundaries. The results
are shown in Table 3.
Recall that the true permeability is 150md and the true
boundary location is 1500ft. The uniformly gridded cases
all predict permeability values of 154-161md, which is due
to the regression trying to fit the early time derivative data
which appear to show wellbore storage. If the fit is performed manually ignoring this data the true permeability
of 150md is predicted.
Example 4: Reservoir with Constant Pressure
Boundaries A well test was simulated in a reservoir with
the properties given in Table 1. The well was located in
the center of the reservoir and constant pressure boundary
conditions were applied on all sides. Finite difference simulation was used to simulate the well test. The constant
presure boundary was simulated by putting injection wells
in every cell along the boundary. The resulting well test
response is shown in Figure 11. The effect of the constant
pressure boundary was not well matched on the derivative
curve. The same well test was simulated using GEM combined with singularity programming. The transient is shown
in Figure 12. This simulation captures the effect of the constant pressure boundary accurately also.
Examples 5 and 6: Well Test Simulation without
Singularity Programming Two wells tests were modeled
to study the performance of the GEM simulation scheme
without singularity programming. The set of reservoir properties given in Table 4 were common to both models. The
boundary conditions applied were constant pressure boundary conditions.
To assess how the strength of the singularity affects the
computed results two cases were considered. In the first case
(Example 5), k = 100md and qw = 100ST B/d and in the
second case (Example 6) k = 150md and qw = 50ST B/d.
The singularity in Example 5 is three times stronger than
in Example 6. The pressure and pressure derivative curves
computed when Example 5 and Example 6 were simulated
SPE 62916
SPE 62916
Nomenclature
c
compressibility, psi1
G Greens function
h
thickness, f t
k
permeability, md
L GEM element matrix
M total number of elements
r
distance (radial), f t
p
pressure, psi
q
flowrate, ST B/d
R GEM element matrix
T GEM element matrix
t
time, hours
U GEM element matrix
V GEM element matrix
x
distance, f t
y
distance, f t
Greek
, porosity
, viscosity, cp
Subscripts
D dimensionless
i
initial
o reference
Superscripts
ns nonsingular
s
singular
References
1. Numbere, D. T. and Tiab, D.: An Improved StreamlineGenerating Technique that uses the Boundary (Integral) Element Method, SPE Reservoir Engineering, August 1988,
1061-68
2. Masukawa, J. and Horne, R. N.: Application of the
Boundary Integral Method to Immiscible Displacement Problems, SPE Reservoir Engineering, August 1988, 1069-77
3. Kikani, J. and Horne, R. N.: Application of Boundary
Element Method to Reservoir Engineering Problems, Journal of Petroleum Science and Engineering, 1989, 229-241
4. Kikani, J. and Horne, R. N.: Pressure-Transient Analysis of Arbitrarily Shaped Reservoirs With the Boundary
Element Method, SPE Formation Evaluation, 1992, 53-60
5. Sato, K. and Horne, R. N.: Perturbation Boundary
Element Method for Heterogeneous Reservoirs: Part 1 Steady-State Flow Problems, SPE Formation Evaluation,
1993, 306-314
6. Sato, K. and Horne, R. N.: Perturbation Boundary Element Method for Heterogeneous Reservoirs: Part
2 - Transient Flow Problems, SPE Formation Evaluation,
1993, 315-322
7. Taigbenu, A. E.: A More Efficient Implementation of
the Boundary Element Theory, Proc. 5th International
Conference on Boundary Element Technology (BETECH
90), Newark, Delaware, 1990, 355-366
pi
qw
ct
rw
h
150 md
0.3
1 cp
2000 psi
50 STB/d
10e-06 psi1
0.3 f t
50 f t
SPE 62916
102
11 by 11
21 by 21
41 by 41
101 by 101
37 by 37
45 by 45
35 by 35
272.72
142.85
73.17
29.70
10 to 223.83
5 to 189.68
5 to 215.73
Gridding
style
Uniform
Uniform
Uniform
Uniform
Nonuniform
Nonuniform
Nonuniform
Geometric
Factor
N/A
N/A
N/A
N/A
1.2
1.2
1.3
psi
10
Number of cells
GEM, p
GEM, dp
Exact, p
Exact, dp
10-1
10-2
10-2
10-1
10
103
102
104
time, hours
11 by 11
21 by 21
41 by 41
101 by 101
37 by 37
45 by 45
35 by 35
Gridding
Style
Uniform
Uniform
Uniform
Uniform
Nonuniform
Nonuniform
Nonuniform
Permeability
Estimate, md
154
158
160
161
185
196
203
Boundary
Estimate, ft
2750
2380
2047
2024
3430
4010
4000
10
Case
102
10-1
Finite Difference, p
Finite Difference, dp
Exact, p
Exact, dp
-2
10
10-3
10-3
10-1
10-2
10
10
aa
aa
aaaa b
aaaaaaa
aaa
aaa
aaa
b
aaa
aaaaaa
b
a a a a a a aa
b
b
b
b
b
b
b
b
b
bb
b
b b b b b b bbb bbbbbbbbbbbbbbbbb
10-1
a
b
-2
10
Value
6.46 stb/psi
-7.62
48md
Confidence Interval
6.2e02 %
2.7 %
2.1e02%
103
102
102
0.3
1cp
2000 psi
10e-06
0.3ft
50ft
3000ft
3000ft
pi
ct
rw
h
Length
Width
Time, hours
True Value
0 STB/psi
0
50 md
10-3
10-3
10-2
10-1
10
GEM, p
GEM, dp
Exact, p
Exact, dp
102
103
Time, hours
SPE 62916
102
10
500 ft
Geometric factor = a/b
10-1
Finite Difference, p
Finite Difference, dp
Exact, p
Exact, dp
-2
10
10-2
10-1
10
102
103
Time, hours
10
10-1
Finite Difference, p
Finite Difference, dp
Exact, p
Exact, dp
10-2
10-3
10-3
102
10-2
10-1
10
102
103
Time, hours
102
10
10
10-1
Finite Difference, p
Finite Difference, dp
Exact, p
Exact, dp
10-2
10-1
10-3
10-3
Finite Difference, p
Finite Difference, dp
Exact, p
Exact, dp
-2
10
10-2
10-1
10
102
103
Time, hours
10-2
10-1
10
102
103
Time, hours
102
10
a a aaaaaaa
a a aaaaaaaa
a a aaaaaaaa
SPE 62916
aaaaaaaaaaaaaaaaaaaaaaaaaaaa
aaaaaaaaaaaaaaa a aa
a a aaaa
10
10-1
Finite Difference, p
Finite Difference, dp
Exact, p
Exact, dp
-2
10
10-3
10-3
10-2
10-1
10
102
b
10-1
bb
bbb
b
b
bb
bb
b b
b
a
b
10-2
10-3
10-3
103
b b b bbbbbbbb bbb
bbbbbbb b bbbbbbbbbbbbbbbbbb bb
bb bbbbbbbb
Finite Difference, p
Finite Difference, dp
Exact, p
Exact, dp
bb
1
10-1
10-2
Time, hours
10
103
102
Time, hours
102
10
aaa
a a aaaa
aaa
a a aaaa
aa
a a aaaaa
a a aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
10
10-1
b bbbbbbbb
b b bbbbbbb
10-1
a
b
Finite Difference, p
Finite Difference, dp
Exact, p
Exact, dp
10-2
b b bbbbbbb
GEM, p
GEM, dp
Exact, p
Exact, dp
b b
b
bb
b
b
b
b
b
b
b
b
b
b
b
b
b
10-3
10-3
10-2
10-1
10
102
103
Time, hours
10-2
10-3
b
bbbbbbbbbbbbbbbbbbbbbbbbbbbb
10-2
10-1
10
102
103
Time, hours
SPE 62916
102
10
10-1
GEM, p
GEM, dp
Exact, p
Exact, dp
-2
10
10-3
10-3
10-2
10-1
10
102
103
Time, hours
10
10
10-1
GEM, p
GEM, dp
Regression Match, p
Regression Match, dp
10-2
1
10-3
10-3
10-2
10-1
10
102
103
Time, hours
10-1
GEM, p
GEM, dp
Exact, p
Exact, dp
10-2
10-3
10-3
10-2
10-1
10
102
103
Time, hours