Sunteți pe pagina 1din 47

COMPARISON OF VARIOUS

TECHINQUES FOR COMPUTING


WELL INDEX

A REPORT SUBMITTED TO THE DEPARTMENT OF


PETROLEUM ENGINEERING
OF STANFORD UNIVERSITY
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF MASTER OF SCIENCE

By
Jones Shu
August 2005

I certify that I have read this report and that in my opinion it is fully
adequate, in scope and in quality, as partial fulfillment of the degree
of Master of Science in Petroleum Engineering.

__________________________________
Prof. Khalid Aziz
(Principal Advisor)

iii

Abstract
The Well Index (WI) plays a key role in reservoir simulation as it defines the relationship
between well pressure and flow rate to reservoir properties and pressure. Currently, there
are many models for computing well indices based on different assumptions and well
configurations. A clear and complete understanding of these models and restrictions is
essential as the application of miscalculated well indices lead to erroneous results
rendering the simulation model ineffective as a prediction tool.
This work presents a comparison between three different methods for computing well
indices: Peacemans model, the Projection method and a Semi-Analytical approach. This
evaluation compares numerical simulation results using WI computed by each method
with a semi-analytical reference solution. The well and reservoir models presented in this
work include horizontal and vertical wells, 2D and 3D slanted wells, isolated wells and
wells near boundaries or other wells. The models are all homogeneous (isotropic and
anisotropic) reservoirs with uniform Cartesian grids. All simulations were done with
single-phase flow and have closed (no flow) boundary conditions.
For each of these models, the values of the well indices computed by each method were
compared. The influence of permeability, grid size, spherical flow, interference and
boundary effects was studied.
Peacemans WI led to significant errors for all slanted wells and this was also true where
spherical flow or interference of other wells and boundaries is dominant. The Projection
method, a practical correction to Peacemans model for deviated wells, provides a
reasonable approximation for slanted wells. However, it also leads to significant errors
when spherical flow, interference and boundary effects are present. The Semi-Analytical
approach computes exact well indices for any type of well in any scenario and their use in
simulators exactly reproduce the semi-analytical reference solution.

Acknowledgments
I would like to thank Professor Khalid Aziz, my advisor, who first of all opened his doors
to me and brought me to be his student and part of his team. Throughout my two years in
Stanford, I had the chance to learn and grow a lot, as an engineer, as a professional, and
as a person. I am very thankful to Dr. Aziz for contributing to and being a part of my
education.
I am very thankful to Huanquan Pan, who worked with me and supported me during the
completion of this work. And special thanks to Dr. Jonathan Holmes who not only
contributed to this work but whose enthusiasm has always brought motivation to our
group.
Also, I would like to thank Prof. Lou Durlofsky, who has always been a great reference
for me and has always given me important support and encouragement.
My love goes to Ftima and Isabella Dias, who while physically distant lived this entire
experience with me. For sure my life in Stanford is strongly associated with our history
together.
Many thanks to Anson, Gianluca and Lisa, as their friendship and companionship are the
greatest gifts I received here in Stanford.
Also, I would like to thank all my friends in Brazil who have cheered for me from start to
end, especially Joo Batista Csar Neto and his family, Prof. Denis Schiozer, Dionysio
Moriconi, Fome, Marcos Borges and Marina Parahyba. And, of course, all my dear
friends in Stanford who walked all the way along with me!
I am grateful to companies supporting Stanford University Petroleum Research Institute
program in Reservoir Simulation (SUPRI-B), and Advanced Wells (SUPRI-HW), which
made my graduate studies at Stanford possible.
Finally, I would send my love to my parents and sister Brenda, as they for 28 years have
stood besides me, loved and cheered for me.
And mostly, I would like to thank God for his love, blessings, care and mercy to me.

vii

Contents
Abstract ............................................................................................................................... v
Acknowledgments............................................................................................................. vii
Contents ............................................................................................................................. ix
List of Figures .................................................................................................................... xi
1.

Introduction................................................................................................................. 1

2.

Current Methods for Computing Well Index.............................................................. 3


2.1.
2.2.
2.3.

3.

Simulation Results Using Different Well Indices....................................................... 9


3.1.
3.2.
3.3.
3.4.

4.

Vertical and Horizontal Wells ............................................................................ 9


Slanted Wells .................................................................................................... 13
Partial Penetration............................................................................................. 19
Interference Among Wells and Boundary Effects ............................................ 20

Comparison of Different Well Indices...................................................................... 23


4.1.
4.2.
4.3.
4.4.
4.5.
4.6.

5.

Peacemans Well Index....................................................................................... 3


Projection Well Index ......................................................................................... 4
Semi-Analytical Well Index................................................................................ 6

Spherical Flow .................................................................................................. 23


Normalization of Well Indices.......................................................................... 24
Different Permeabilities .................................................................................... 25
Different Grid Sizes .......................................................................................... 27
Partial Penetration............................................................................................. 28
Boundary Effects............................................................................................... 30

Conclusions and Future Work .................................................................................. 31


5.1.
5.2.

Conclusions....................................................................................................... 31
Future Work ...................................................................................................... 32

Nomenclature.................................................................................................................... 33
References......................................................................................................................... 34

ix

List of Figures
Figure 2-1: Peaceman WI Assumptions: Single Isolated Well, Fully Penetrating Grid
Block, Aligned with Grid, Single Phase Radial Flow......................................................... 3
Figure 2-2: (a) Well Trajectory Projected into the Axis, (b) Projection of Well Segments 5
Figure 2-3: Analytical Solution (Well Inflow and Pressure) Coupled with Numerical Grid
(Block Pressures) ................................................................................................................ 6
Figure 3-1: Well Flow Rate: Isolated Vertical Well, Aligned with Grid, Fully Penetrating
Isotropic Reservoir............................................................................................................ 10
Figure 3-2: Well Pressure Distribution: Isolated Vertical Well, Aligned with Grid, Fully
Penetrating Isotropic Reservoir......................................................................................... 10
Figure 3-3: Well Flow Rate: Isolated Vertical Well, Aligned with Grid, Partially
Penetrating Isotropic Reservoir......................................................................................... 11
Figure 3-4: Well Pressure Distribution: Isolated Vertical Well, Aligned with Grid,
Partially Penetrating Isotropic Reservoir .......................................................................... 11
Figure 3-5: Flow Rate per Unit Length: Isolated Vertical Well, Aligned with Grid,
Partially Penetrating Anisotropic Reservoir...................................................................... 12
Figure 3-6: Well Pressure Distribution: Isolated Vertical Well, Aligned with Grid,
Partially Penetrating Anisotropic Reservoir...................................................................... 12
Figure 3-7: Flow Rate per Unit Length: Isolated Horizontal Well, Aligned with Grid,
Partially Penetrating Anisotropic Reservoir...................................................................... 13
Figure 3-8: Well Configuration: Horizontally Slanted Well (Deviated in 2D), Cutting 7
Grid Blocks ....................................................................................................................... 14
Figure 3-9: Flow Rate: Isolated Horizontally Slanted (2D) Well, Partially Penetrating
Isotropic Reservoir............................................................................................................ 14
Figure 3-10: Flow Rate per Unit Length: Isolated Horizontally Slanted (2D) Well,
Partially Penetrating Isotropic Reservoir .......................................................................... 15
Figure 3-11: Flow Rate: Isolated Horizontally Slanted (2D) Well, Partially Penetrating
Anisotropic Reservoir ....................................................................................................... 15
xi

Figure 3-12: Flow Rate per Unit Length: Isolated Horizontally Slanted (2D) Well,
Partially Penetrating Anisotropic Reservoir...................................................................... 16
Figure 3-13: Flow Rate: Isolated Horizontally and Vertically Slanted (3D) Well, Partially
Penetrating Isotropic Reservoir......................................................................................... 17
Figure 3-14: Flow Rate per Unit Length: Isolated Horizontally and Vertically Slanted
(3D) Well, Partially Penetrating Isotropic Reservoir ........................................................ 17
Figure 3-15: Flow Rate: Isolated Horizontally and Vertically Slanted (3D) Well, Partially
Penetrating Anisotropic Reservoir .................................................................................... 18
Figure 3-16: Flow Rate per Unit Length: Isolated Horizontally and Vertically (3D)
Slanted Well, Partially Penetrating Anisotropic Reservoir............................................... 18
Figure 3-17: Well Pressure: Vertical Well Completed at Single Block, Fading Penetration
........................................................................................................................................... 19
Figure 3-18: Configuration of Horizontal Well (longer well) and Neighboring Vertical
Well (shorter well) ............................................................................................................ 20
Figure 3-19: Flow Rate: Horizontal Well under Influence of another Vertical Well
(Producers)........................................................................................................................ 21
Figure 3-20: Well Pressure: Horizontal Well under Influence of No-Flow Boundary..... 22
Figure 4-1: Spherical Flow Effect in the Semi-Analytical Well Index............................. 23
Figure 4-2: Well Indices Distributed by Respective Well Trajectory Length................... 24
Figure 4-3: Well Indices: Different Permeabilities Change Magnitude but Keeps Same
Pattern ............................................................................................................................... 25
Figure 4-4: Normalized WI of Wells in Different Isotropic Media .................................. 26
Figure 4-5: Normalized WI of Wells in Different Anisotropic Media ............................. 26
Figure 4-6: Normalized WI of Wells in Different Grids Sizes ......................................... 27
Figure 4-7: WI Sensitivity for Different Parameters for Partial Penetration .................... 28
Figure 4-8: Well Indices Relation Regarding Partial Penetration with Varying Parameters
........................................................................................................................................... 29
Figure 4-9: Well Indices Relation Regarding Partial Penetration with Varying Anisotropy
........................................................................................................................................... 29
Figure 4-10: Boundary Effect in the Well Index Value .................................................... 30

xii

Chapter 1
1. Introduction
In reservoir simulation, flow models that define the relation between wells and reservoirs
play a key role. A thorough understanding of the equations that govern these models is
essential to compute correct results, such as inflow and pressure distributions along the
wells.
In numerical models, the well pressure is different from the pressure of the grid block
containing the well, due to large difference in scale of the well and the grid block. In
order to define this relationship, a coefficient known as Well Index or Well
Transmissibility is used. This coefficient accounts for the geometric characteristics of the
well and the surrounding reservoir properties, as well as any interaction with other wells
and boundaries. The Well Index is defined as the ratio of the well flow rate and the
difference between the reservoir block and wellbore pressures (Eq. 1-1):

WI i =

qiw
pi piw

(1-1)

Because of its importance, many techniques for computing the well index have been
developed. In this study, three methods have been considered; the classic approach
known as Peacemans model [1,2,3], the Projection technique developed by J. Holmes
(Schlumberger)[4] and the Semi-Analytical procedure [6,7,8,9,10] developed by the
Department of Petroleum Engineering at Stanford University.
This study evaluates the well indices computed by the different methods regarding their
performance and applicability in different well configurations and reservoir models. This
analysis is done by comparing numerical simulation results using the three different well
indices to an analytical reference solution.
This report proceeds as follow. In Chapter 2, different methods for calculating well
indices are reviewed. In Chapter 3, the models are compared based on simulation results
through sample cases that illustrate the WIs performances in diverse well and reservoir
property scenarios. In Chapter 4, an evaluation of the values of the WI themselves is
presented and their correlations and main discrepancies outlined. Finally, conclusions
and recommendations for future work are listed in Chapter 5.

Chapter 2
2. Current Methods for Computing Well Index
2.1. Peacemans Well Index
The Peaceman WI [1,2,3] is the classical technique. All well index calculation
techniques, including Peacemans method, are based on single phase flow.
The main assumption of this model is that it is derived for a vertical well in a uniform
Cartesian grid, fully penetrating the grid block, with single-phase radial flow and no
interaction with boundaries or other wells.

Figure 2-1: Peaceman WI Assumptions: Single Isolated Well, Fully Penetrating Grid Block,
Aligned with Grid, Single Phase Radial Flow

For computing the well index, based on the single-phase steady-state radial flow equation
(Eq. 2-1), Peaceman [1] introduced the equivalent well block radius ro, defined as the
radial position at which the computed block pressure is equal to the pressure obtained
from the analytical radial solution (Eq. 2-2). The equivalent radius is not a physical
quantity [3], but rather an intermediate variable that makes the well model (Eq. 2-2)
work.

p(r ) = p w +

qw
r
ln
2kh
rw

(2-1)

q =
w

2kh ( p o p w )

(2-2)

r
ln o
rw

To obtain the values of ro, both analytical and numerical solutions were used. For nonsquare grid blocks and anisotropic permeability, Peaceman [2] defined the WI as:

2 k x k y z

WI =

(2-3)

r
ln o + s
rw

Where,

ky
kx

1
2

x 2 +

ro = 0.28
ky
kx

1
4

kx
ky
kx
ky

1
2

1
2

y 2

1
4

(2-4)

This model is the most common and is the standard in commercial simulators.

2.2. Projection Well Index


The Projection WI is based on Peacemans model and therefore is limited by the same
assumptions.
However, the Projection WI, developed by Jonathan Holmes
(Schlumberger) [4], corrects the model for slanted (not aligned with the grid) wells.
In this method, the well trajectory is projected onto three orthogonal axis, as shown in
Figure 2-2a. Using the three projected lengths and using Peacemans equation for WI and
ro, WI values are calculated for each direction (Eqs. 2-5 and 2-6). The WI for the well
segment in the block is the square root of the sum of the squares of these partial well
indices (Eq. 2-7)

Figure 2-2: (a) Well Trajectory Projected into the Axis, (b) Projection of Well Segments

2 k y k z Lx

WI x =

ln

ky
kz

1
2

ro , x
+s
rw

k
z + z
ky

ro , x = 0.28

ky
kz

1
4

WI y =

k
+ z
ky

1
2

ln
i
1
2

1
4

2 k x k z L y

ro , y = 0.28

kz
kx

1
2

ro , y

kz
kx

+s

rw

x 2 +
1
4

2 k x k z L y

WI y =

ro , y

ln

rw

kx
kz
kx
kz

1
2

1
2

ky

z 2

1
4

kx

1
2

k
x + x
ky

ro , z = 0.28
ky
kx

WI pj = WI x2 + WI y2 + WI z2

+s

1
4

k
+ x
ky

i
1
2

1
2

(2-5)

y 2

1
4

(2-6)

(2-7)

In case of segmented wells where there are more than one segments within the same grid
block, the projected length for the well index is the sum of the projections of all segments
in that direction (Eq. 2-8, Fig. 2-2b).

Ldirection _ k =

L j ,k

segment _ j

(2-8)

The Projection WI approach is part of Schlumberger Schedule tool, a pre-processor of


GeoQuest ECLIPSE.

2.3. Semi-Analytical Well Index


The Semi-Analytical WI is obtained using well pressure and inflow distributions
calculated semi-analytically and well block pressures obtained from a numerical
simulator. The WI can be computed directly using:

WI i =

qiw
pi piw

(2-9)

The general approach used in this work was developed by many previous researchers at
Stanford University: Valvatne [8], Serve [9], Wolfsteiner et al. [6,7] and others. A brief
description is given here.
The first step of this framework is to compute accurately well inflows and well pressures
for each segment representing the well by applying a semi-analytical procedure based on
Greens functions to solve the single phase flow problem (Eq. 2-10). To obtain this
reference solution, only the well, not the reservoir, is discretized into well segments.
After this solution is obtained for each segment, the reference well flow rate for each
block that is intercepted by the well is determined by an intersection algorithm. From
these rates, using a single-phase numerical simulator, block pressures for each well block
are obtained.

.(k ) = c

(2-10)

"#
$

q wj , p wj

q wj , p wj qiw , piw
qiw , piw piblock
%

Figure 2-3: Analytical Solution (Well Inflow and Pressure) Coupled with Numerical Grid (Block
Pressures)

The Semi-Analytical WI is finally obtained by combining the well rates and pressures,
from the semi-analytical solution, and the respective block pressures from the single
phase numerical simulation (Eq.2-9). In order to obtain one single WI value for each
block, the rates and pressures must be obtained under steady-state or pseudo-steady-state
conditions.
In order to include near-well permeability heterogeneities, this framework also includes
the s-k* model, creating a representation of heterogeneity through a constant background
permeability k* obtained by power averaging or some other upscaling technique and a
local skin s.
This allows accurate modeling of wells of any trajectory that intersect any grid arbitrarily.
It also accounts for spherical flow at well ends and interaction among wells and
boundaries.
This entire framework is implemented in AdWell 2.1, a research simulator developed by
the SUPRI-HW team.

Chapter 3
3. Simulation Results Using Different Well Indices
In this chapter, the applicability of the previously introduced methods for computing well
indices is discussed. Each of these comparisons shows different well configurations and
reservoir models, with the idea of analyzing the applicability of each method in each
scenario. This analysis is done by comparing the analytical reference solution and
numerical simulation results using the three different well index calculation procedures.
The analytic reference solution is provided by AdWell 2.1, a computer program
developed by SUPRI-HW for research purpose. It was used to solve the well model
semi-analytically (well inflow and pressure distribution). Also, making use of the
extended features implemented in AdWell, the Semi-Analytical WI was computed in
pseudo-steady state (rate control) conditions. The Projection WI was also calculated by
this application. Both of these indices were calculated using the techniques presented in
the previous chapter. The numerical simulation results using the different WIs were
obtained using GeoQuest ECLIPSE 100 - 2004a, a commercial reservoir simulator.
For each well and reservoir model, four simulation results are compared: ECLIPSE using
its default WI (Peacemans WI), ECLIPSE using the Semi-Analytical WI, ECLIPSE
using the Projection WI and AdWells analytical reference solution. The well and
reservoir models included in this work are horizontal, vertical, 2D and 3D slanted wells,
in homogeneous (isotropic and anisotropic) reservoirs with uniform Cartesian grids. All
simulations were with single-phase flow and closed boundaries.

3.1. Vertical and Horizontal Wells


In this example, the condition in which Peacemans WI was derived is reproduced. A
single isolated vertical well is inserted in an isotropic reservoir, which is discretized by a
uniform Cartesian grid. The well is aligned with the grid, fully penetrates the entire
reservoir (from top to bottom) and is located away from other boundaries.
The reservoir is 1300 ft x 1300 ft x 60 ft, with a grid of 13 x 13 x 6, permeability equals
200 md. The well radius is 0.25 ft and accounts only for pressure drop due to hydrostatic
head. Well is located at the center of the grid in blocks i = 7, j = 7, k = 1-6. The well
operates under constant rate control, producing 1500 bbl/day with an initial reservoir
pressure of 3000 psi.

As Figures 3-1 and 3-2 show, as expected, the results of all WI match perfectly. Because
the well fully penetrates the reservoir, only radial flow is present. In this case, the inflow
is distributed homogeneously throughout the six blocks (Figure 3-1) while the well
pressure varies due to the hydrostatic head only (Figure 3-2). These values were taken on
day 100 of the simulation run.

Flow Rate (STB/day)

300
250
200

Eclipse with SA WI

150

Eclipse with PJ WI
Eclipse Default (PM)

100

AdWell

50
0

Grid Block

Figure 3-1: Well Flow Rate: Isolated Vertical Well, Aligned with Grid, Fully Penetrating
Isotropic Reservoir

AdWell

Pressure (psia)

690

Eclipse with SA WI

675

Eclipse with PJ WI

660

Eclipse Default (PM)

645
630
615
600

Grid Block

Figure 3-2: Well Pressure Distribution: Isolated Vertical Well, Aligned with Grid, Fully
Penetrating Isotropic Reservoir

In this next example, the isolated vertical well is not fully perforating the reservoir.
Conditions similar to those in the previous case apply: rectangular reservoir grid (3100 ft
x 3100 ft x 165 ft discretized into 31 x 31 x 11 blocks) and isotropic permeability (300
md). The well is aligned with the grid and fully perforates 4 blocks (blocks i = 16, j = 16,
k = 4-8).

10

From Figures 3-3 and 3-4, it is clear that Peacemans WI (ECLIPSE Default) and the
Projection WI, which has the same value as Peacemans for aligned wells, do not give the
correct result since they do not account for spherical flow. The analytical reference
solution, in this case, shows a result 30 bbl/day higher in blocks k = 1 and k = 4 (well
ends) where spherical flow takes place, and 30 bbl/day lower in the two center well
blocks. The Semi-Analytical WI, as expected, matches the analytical solution for both
inflow and well pressure distributions. This is true for all scenarios.
540

Eclipse with SA WI
Eclipse Default (PM)
Eclipse with PJ WI
AdWell

Flow Rate (STB/day)

530
520
510
500
490
480
470
460
450
0

Grid Block

Figure 3-3: Well Flow Rate: Isolated Vertical Well, Aligned with Grid, Partially Penetrating
Isotropic Reservoir
3300
3290
Pressure (psia)

3280
3270
3260
3250

Eclipse with SA WI
Eclipse Default (PM)
Eclipse with PJ WI
AdWell

3240
3230
3220
3210
3200
0

Grid Block

Figure 3-4: Well Pressure Distribution: Isolated Vertical Well, Aligned with Grid, Partially
Penetrating Isotropic Reservoir

Using similar well configuration in an anisotropic reservoir (kx = 200md, ky = 200 md, kz
= 20 md), Figure 3-5 shows that the anisotropy ratio (kv/kh) of 0.1 decreases the
influence of spherical flow. But it still creates a mismatch between Peacemans and
Projection WIs with the reference solution (AdWell) and the Semi-Analytical WI.

11

Flow Rate per Length (STB/day/length)

29

Eclipse with SA WI
Eclipse with PJ WI
Eclipse Default (PM)
AdWell

28

27

26

25

10

20

30

40

50

60

70

80

Location Along Well (ft)

Figure 3-5: Flow Rate per Unit Length: Isolated Vertical Well, Aligned with Grid, Partially
Penetrating Anisotropic Reservoir

3100

Pressure (psia)

3090
3080

AdWell
Eclipse with SA WI
Eclipse with PJ WI
Eclipse Default (PM)

3070
3060
3050

Grid Block

Figure 3-6: Well Pressure Distribution: Isolated Vertical Well, Aligned with Grid, Partially
Penetrating Anisotropic Reservoir

The same behavior as seen in the vertical well examples was also observed for horizontal
wells. Figure 3-7 shows an example. This figure shows the results of an isolated
horizontal well, aligned with the grid, with no friction or acceleration pressure drop taken
into account in the well model, in an anisotropic media (kx = 300 md, ky = 300 md, kz =
20 md) with reservoir dimensions 3100 ft x 3100 ft x 165 ft (i = 31, j = 31, k = 11
blocks). But in this example, the well cuts 10 blocks (i = 16, j = 11-20, k = 6), which is
much more than in the previous vertical examples.
12

Flow Rate per Length (STB/day/length)

3.6
3.4

Eclipse with SA WI
Eclipse with PJ WI
Eclipse Default (PM)
AdWell

3.2
3.0
2.8
2.6

200

400

600

800

1000

Location Along Well (ft)

Figure 3-7: Flow Rate per Unit Length: Isolated Horizontal Well, Aligned with Grid, Partially
Penetrating Anisotropic Reservoir

Once again deviation from the reference model at the ends of the well due to spherical
flow is observed, although the inflow differences are more spread among well blocks.
The well pressures were constant throughout the well since no friction or acceleration was
taken into account, and all methods matched each other.

3.2. Slanted Wells


In this section, simulation results of slanted wells wells not aligned with the grid are
discussed.
The first example is a horizontally slanted (deviated in 2D) well, cutting through 7 grid
blocks (Figure 3-8, i = 10-12, j = 9-13, k = 6). The reservoir is isotropic (k = 200 md)
and its dimensions are 2100 ft x 2100 ft x 550 ft (21 x 21 x 11 blocks).
Evaluating the flow rate per block (Figure 3-9), it is evident that Peacemans WI is not
suitable for modeling this well configuration. The Projection WI also gives a very similar
result and shows that it is a good approximation for slanted wells. The flow rate of each
grid block is also related to the well length within the block: where the flow rate is high,
the well length in that block is also high.

13

Figure 3-8: Well Configuration: Horizontally Slanted Well (Deviated in 2D), Cutting 7 Grid
Blocks

700

Flow Rate (STB/day)

600
500
400
300
200

Eclipse with SA WI
Eclipse with PJ WI
Eclipse default (PM)
Adwell

100
0

Grid Block

Figure 3-9: Flow Rate: Isolated Horizontally Slanted (2D) Well, Partially Penetrating Isotropic
Reservoir

Figure 3-10 shows the flow rate per unit length. This plot is appropriate for visualizing
the correct and expected results for the well inflow distribution. In this figure, the
reference solution and the Semi-Analytical approach give a smooth distribution, while the
Projection method is a very good numerical approximation, but Peacemans method gives
an unphysical (oscillatory) solution.
Again, the well pressures were constant throughout the well, as no friction or acceleration
was taken into account.

14

Flow Rate per Length (STB/day / ft)

12
10
8
6
4
Eclipse with SA WI
Eclipse with PJ WI
Eclipse default (PM)
AdWell

2
0
0

100

200

300

400

500

600

700

Location along Well

Figure 3-10: Flow Rate per Unit Length: Isolated Horizontally Slanted (2D) Well, Partially
Penetrating Isotropic Reservoir

In an anisotropic case (kx = 600md, ky = 300md, kz = 60 md) with same well and grid
configuration as the previous example, simulation results (Figures 3-11, 3-12) were
similar to the isotropic case but with some important differences.
700

Flow Rate (STB/day)

600
500
400
300
200

Eclipse with SA WI
Eclipse with PJ WI
Eclipse Default (PM)
AdWell

100
0
0

Grid Block

Figure 3-11: Flow Rate: Isolated Horizontally Slanted (2D) Well, Partially Penetrating
Anisotropic Reservoir

15

Peacemans method, although already giving unrealistic results due to the deviated
trajectory, was further impacted by reservoir anisotropy and the fact that the well is not
aligned with the permeability tensor. (Eq. 2-3 and 2-4). On the other hand, the projection
method is not influenced by the anisotropy, as it weights and calculates the different
partial WI for each direction, making it independent of the anisotropy.
It is also apparent that the spherical flow causes a discrepancy in inflow distribution
between the Projection WI and the reference solution. The well blocks at the extremes
have lower flow rates and are compensated for by higher rates from the intermediate well
blocks.

Flow Rate per Length (STB/day / ft)

14

Eclipse with SA WI
Eclipse with PJ WI
Eclipse default (PM)
AdWell

12
10
8
6
4
2
0
0

100

200

300

400

500

600

700

Location Along Well (ft)

Figure 3-12: Flow Rate per Unit Length: Isolated Horizontally Slanted (2D) Well, Partially
Penetrating Anisotropic Reservoir

Slanted wells in 3D (deviated both vertically and horizontally) showed the same behavior
as the ones in 2D (deviated vertically or horizontally). Figure 3-13 shows the flow rate
per well block for a slanted well that cuts 10 blocks (i = 10-12, j = 9-12, k = 9-13) in an
isotropic reservoir (k = 100 md) with a uniform Cartesian grid (2100 ft x 2100 ft x 1050
ft, 21 x 21 x 21 blocks). Figure 3-14 shows the flow rate per unit length for this well. As
observed in the 2D cases, the Projection method provides a good approximation of the
well index for modeling deviated wells, although the end effect is not reproduced.

16

450

Eclipse with SA WI
Eclipse with PJ WI
Eclipse default (PM)
AdWell

Flow Rate (STB/day)

400
350
300
250
200
150
100
50
0
0

10

12

Grid Block

Figure 3-13: Flow Rate: Isolated Horizontally and Vertically Slanted (3D) Well, Partially
Penetrating Isotropic Reservoir

Flow Rate per Length(STB/day / ft)

7.0
6.0
5.0
4.0
3.0
Eclipse with SA WI
Eclipse with PJ WI
Eclipse default (PM)
AdWell

2.0
1.0
0

10

Location at Well (ft)

Figure 3-14: Flow Rate per Unit Length: Isolated Horizontally and Vertically Slanted (3D) Well,
Partially Penetrating Isotropic Reservoir

Furthermore, 3D deviated wells in anisotropic reservoirs also gave similar results to 2D


cases, with Peacemans WI being dependent on the well and permeability tensor
orientation while the other two techniques are not. Figure 3-15 and 3-16 show flow rate

17

distributions of an isolated well with same configuration as the previous example and
permeability field of kx = 100 md, ky = 100 md, kz = 50 md.

Flow Rate (STB/day)

450
400

Eclipse with SA WI

350

Eclipse with PJ WI

300

Eclipse default (PM)


AdWell

250
200
150
100
50
0
0

10

Grid Block

Figure 3-15: Flow Rate: Isolated Horizontally and Vertically Slanted (3D) Well, Partially
Penetrating Anisotropic Reservoir

Flow Rate per Length (STB/day/ft)

7.0
6.0
5.0
4.0
3.0
2.0

Eclipse with SA WI
Eclipse default (PM)
Eclipse with Proj WI
Adwell

1.0
0.0
0

200

400
600
Location at Well (ft)

800

1000

Figure 3-16: Flow Rate per Unit Length: Isolated Horizontally and Vertically (3D) Slanted Well,
Partially Penetrating Anisotropic Reservoir

18

3.3. Partial Penetration


Several simulation runs were performed in order to investigate the performance of
different well indices with respect to partial penetration of wells in grid blocks. One
example is presented in this section to illustrate the results.
In this example, an anisotropic, single-layered reservoir was used. Its dimensions are
2100ft x 2100ft x 20ft (21 x 21 x 1 blocks) and permeability field kx = 400 md, ky = 400
md and kz = 100 md. An isolated vertical well completed in a single block (at center of
the grid, i = 11, j = 11, k = 1) was simulated for a range of penetration of 0% to 100%.
The well operated under flow rate control (1500 bbl/day) and the different output well
pressures are compared (day 100, Figure 3-17).
It is observed that for the Semi-Analytical WI, as the well penetration is extrapolated to
zero, the well pressure also approaches zero, just as expected. On the other hand, the
Projection WI shows the same behavior as the Semi-Analytical WI, but results in a faster
pressure drop as the penetration decreases. The differences between the methods increase
as the penetration goes lower than 50%. For example, at 10% penetration the well
pressure difference is 650 psi, which is significant. It is also noticed that the well
pressure, given by the Projection WI decreases faster than expected as it extrapolates to a
well pressure of 0 psi at a penetration of 5% and not 0%.
These differences between the results of the Projection method and the Semi-Analytical
method are due to the fact that spherical flow is taken into account by the SemiAnalytical WI but not by the Projection method. This effect becomes larger as the
penetration becomes smaller.

Well Pressure (psia)

2000
1600

Eclipse with PM WI
Eclipse with PJ WI
Eclipse with SA WI
AdWell

1200
800
400
0
0%

20%

40%

60%

80%

100%

Penetration

Figure 3-17: Well Pressure: Vertical Well Completed at Single Block, Fading Penetration

19

Peacemans WI was kept constant for all partial penetrations. No multipliers were used
to weight it. By weighting it with the penetration length, it would give the same value as
the Projection WI
These simulation results show that, while the Semi-Analytical WI gives the expected
solution, the Projection WI gives a faster pressure because it neglects spherical flow.

3.4. Interference Among Wells and Boundary Effects


Another relevant effect for well modeling is interference of other wells and boundaries.
In this section, examples of each are presented.
Figure 3-18 illustrates a horizontal well, which fully penetrates the reservoir (no end
effect at extremes) and is influenced by a vertical producer which is located 3 blocks
away from its center.

Figure 3-18: Configuration of Horizontal Well (longer well) and Neighboring Vertical Well
(shorter well)

The reservoir grid is 700 ft x 2100 ft x 100 ft (7 x 21 x 5 blocks) and its permeability field
is kx = 200 md, ky = 500 md, kz = 30 md. The horizontal well is aligned with the grid and
is completed in 7 blocks (i = 1-7, j = 11, k = 3) while the vertical well is completed in 5
blocks (i = 4, j = 14, k = 1-5).
The simulation results for the horizontal well (Figure 3-19) show that, although
Peacemans method does not take account for the interference between these wells, the
block pressures most affected by the depletion of both producers do influence inflow
distribution. There is a mismatch between the Projection WI results and the analytical
reference solution, mostly where the wells are closer (central block, 300 ft, 3 blocks
apart).

20

Inflow per Length (STB/day/ft)

450

Eclipse with PM WI
Eclipse with PJ WI
Eclipse with SA WI
AdWell

440
430
420

410
400
390
0

100

200

300

400

500

600

700

Location Along Well (ft)

Figure 3-19: Flow Rate: Horizontal Well under Influence of another Vertical Well (Producers)

Another example of interference is regarding boundaries. In the next example, a


horizontal well (600ft long, cuts 6 blocks) is placed close (150 ft, i = 2, j = 1-6, k = 8) to a
no-flow boundary. The reservoir is 3100 ft x 600 ft x 300 ft (31 x 6 x 15 blocks) and its
permeability field is kx = 600 md, ky = 300 md and kz = 30 md. Figure 3-20 shows the
well pressure changes as the well is moved closer and closer to the boundary (moves from
i = 2 to i = 1, from 150ft progressively to 5 ft from the boundary) .
There are significant differences between Peacemans WI and the analytical solution
when the well gets very close to the boundary (in this case, less than 50 ft). Peacemans
and the Projection methods overestimate the well pressures for these scenarios.

21

Well Pressure (psia)

2498

2494

2490

Eclipse with PM WI
Eclipse with PJ WI
Eclipse with SA WI
AdWell

2486

2482
0

10

20

30
40
50
Distance from Boundary (ft)

60

Figure 3-20: Well Pressure: Horizontal Well under Influence of No-Flow Boundary

22

70

Chapter 4
4. Comparison of Different Well Indices
In this chapter, the values of well indices computed by the different methods are
compared. The purpose is to investigate how these values are related to each other, rather
than looking at their simulation results as in the previous chapter. The objective is to see
whether there is any correlation between the Semi-Analytical and Projection WI.
To do this, first a reference to normalize the well indices was selected. After
normalizing, well indices were computed and compared for various scenarios. The
influence of parameters and other effects, such as permeability, grid size and spherical
flow were studied and are presented here.

4.1. Spherical Flow


Spherical flow plays a key role when modeling wells as seen in Chapter 3. In this
example, well indices of an isolated horizontal well aligned with the grid (550 ft x 1550 ft
x 110 ft, 11 x 31 x 11 blocks) are presented (Figure 4-1).
11.5
1 block
3 blocks
5 blocks

11.0

SA WI

7 blocks
9 blocks
11 blocks

10.5

10.0

10

11

12

Grid Block

Figure 4-1: Spherical Flow Effect in the Semi-Analytical Well Index

The well length faded from penetrating the entire reservoir (11 blocks, i = 1-11, j = 16, k
= 6) to only penetrating the center (1 block, i = 6, j = 16, k = 6). In the fully penetrating
23

case (11 blocks), it is seen that only radial flow is present. The Semi-Analytical WI is the
same as the Projection WI along the entire well. For wells partially penetrating the
reservoir, Figure 4-1 shows how the spherical flow impacts the well index at the well
extremes, giving it higher values. In this example, it increased the WI values by 4% to
12%.
From this comparison, it is clear that spherical flow should be considered for the correct
modeling of wells, mostly when short wells are completed in just few grid blocks.
Moreover, while the Projection and Peacemans WI ignore spherical flow, the SemiAnalytical WI correctly captures this effect.

4.2. Normalization of Well Indices


A normalization procedure is used to better compare the well indices for diverse
scenarios and parameters, such as permeability and grid sizes. The idea is to create a
scale that gives results of the same magnitude, independent of the well and reservoir
parameter values.
A pattern was observed in plots of the value of well indices vs. the well trajectory length
in the respective block. Figure 4-2 shows this pattern for a long horizontal deviated well
(4500ft long, cuts 61 grid blocks, i = 4-41, j = 13-34, k = 17)) in an anisotropic reservoir
(4500 ft x 4500 ft x 175 ft, 45 x 45 x 33 blocks, kx = 200 md, ky = 200 md and kz = 50
md).

Well Index

20

15

10

PM WI
PJ WI

SA WI

0
0

20

40

60
80
Completion length

100

120

Figure 4-2: Well Indices Distributed by Respective Well Trajectory Length

Peacemans WI is constant and independent of the well trajectory length as it is always


calculated considering a fully penetrated block, even when it is not. The Projection WI,
on the other hand, shows a linear relationship, that would start from the origin, cross
24

Peacemans value at exactly 100% and continue until the biggest trajectory length case
(the possible maximum being the 3D diagonal). The Semi-Analytical WI, however,
shows a distinct slightly concave curve, resulting in values lower than those calculated for
the Projection WI. In this case, the exception comes from the WI of the blocks at the well
ends, where spherical flow is dominant and the WI value is higher than the pattern for
radial flow (outlier point in SA WI curve). It is important to emphasize that the points in
Figures 4-2 and 4-3 are not ordered by location in the well, but rather by the well
trajectory length in each grid block.

80
70

Well Index

60
50
40
30
20
10
0
0

20

40

60

80

100

120

Completion length in cell


kx200 ky200 kz50

kx200 ky200 kz200

kx400 ky400 kz400

kx1000 ky1000 kz50

Figure 4-3: Well Indices: Different Permeabilities Change Magnitude but Keeps Same Pattern

Based on these observations, it was decided to normalize the WIs by taking ratios of the
Semi-Analytical WIs to Peacemans WI (SAWI / PMWI) and the Projection WI to
Peacemans WI (PJWI / PMWI).

4.3. Different Permeabilities


In this comparison, several wells were modeled with different reservoir permeability
values.
For isotropic reservoirs, it was observed that the normalized WIs for various
permeabilities collapse to a single curve. Figure 4-4 shows results of 33 horizontally
slanted wells in 5 different grids and different isotropic media (permeabilities varying
from 30 md to 1000 md). From these results, it is apparent that the Semi-Analytical WI
and the Projection WI keep the same relation, independent of the permeability values for
isotropic cases.

25

1.2

(SA WI / PM WI )

1
0.8
0.6
0.4
0.2
0
0

0.2

0.4
0.6
0.8
( PJ WI / PM WI)

1.2

Figure 4-4: Normalized WI of Wells in Different Isotropic Media

Moreover, if the reservoir is anisotropic, the ratio of the well indices changes. Figure 4-5
shows a horizontally deviated well, cutting 36 grid blocks, in different anisotropic media.
The results show that although the same pattern is maintained, the values of the
normalized WI change.
1.8

k 100 x 300 x 30
k 100 x 200 x 50
k 200 x 400 x 50
k 200 x 600 x 50
k 200 x 900 x 30
k 300 x 100 x 30
k 400 x 200 x 50
k 500 x 300 x 50
k 1000 x 500 x 50

1.6
1.4

(SA WI / PM WI )

1.2
1
0.8
0.6
0.4
0.2
0
0

0.2

0.4

0.6

0.8

1.2

1.4

( PJ WI / PM WI )

Figure 4-5: Normalized WI of Wells in Different Anisotropic Media

26

1.6

1.8

It is also observed that for scenarios where the anisotropic ratio for the wells direction
(horizontal well, x-y direction in this example) is the same (i.e. k = 100 x 200 x 50 md
and k = 200 x 400 x 50 md), SAWI/PMWI and PJWI/PMWI keep the same relationship.
This indicates that the anisotropic ratio impacts the relation between the Semi-Analytical
WI and the Projection WI. This influence increases for PJWI/PMWI larger than 0.7.

4.4. Different Grid Sizes


Another important geometric property is the grid size. Here, well indices were computed
for several wells and reservoirs under diverse course and fine grids. The block sizes
varied from 20 to 200 ft, in uniform Cartesian grids. Figure 4-6 shows results from 16
different horizontally deviated wells in different reservoir grids.
From Figure 4-6 it is observed that the grid does not influence the relation between the
Semi-Analytical WI and the Projection WI. The SAWI/PMWI and the PJWI/PMWI keep
similar patterns for the cases studied, except that there is more scatter among the
SAWI/PMWI results for PJWI/PMWI greater than 0.7.

1.4

(SA WI / PM WI )

1.2
1
0.8
0.6
0.4
0.2
0
0

0.2

0.4
0.6
0.8
1
( PJ WI / PM WI)

1.2

1.4

Figure 4-6: Normalized WI of Wells in Different Grids Sizes

27

4.5. Partial Penetration


The influence of partial penetration in the relationship between different WI values was
also studied. For this, well indices were calculated by using different methods for vertical
wells aligned with the grid, where its grid block penetration varied from 0 to 100%.
It is observed that the semi-analytical technique gives a higher WI value than the
projection method for all cases (Figure 4-7). This is expected since the Semi-Analytical
WI reproduces the well end effect of spherical flow, while the Projection WI does not.
20
18

'
" ' ((
((
' ((
' ((
)
# '(*+

&

16
14
Well Index

Nx 15
Ny 35
rw = 0.2
k= 600

(
' ((

rw = 0.35

12

SA WI

10

PJ WI

base case

6
4
2

k= 100

0
0%

20%

40%

60%

80%

100%

Penetration

Figure 4-7: WI Sensitivity for Different Parameters for Partial Penetration

Also, as shown in Figure 4-7, even when geometric parameters are changed such as well
radius, grid size or permeability, the absolute WI value varies but not the pattern. This
suggests that a normalization of the WI values would be appropriate for this comparison.
Comparing normalized WI values as before (SAWI/PMWI and PJWI/PMWI), it becomes
clear that there is a strong relation between these two well indices with respect to partial
penetration. Figure 4-8 shows a comparison of the SAWI/PMWI and PJWI/PMWI for
varying parameters, with all cases aligning along the same curve. The largest difference
is around 30%-70% penetration, where the SAWI/PMWI and PJWI/PMWI are about
20% different.
One property, however, made a significant difference: the anisotropy ratio. Figure 4-9
shows some normalized well indices and how they differ for various scenarios. The
variation comes from the influence of spherical flow on the model.
Anisotropic ratios that promote spherical rather than radial flow give a larger difference
between the Semi-Analytical WI to Projection WI.

28

1.0

(SA WI / PM WI)

0.8
k 100
k 500
k 1000
N 21 x 15
N 15 x 15
N 15 x 35
rw 0.2
rw 0.35
N 15 x 35, rw 0.2, k 600
k 200, N 21 x 21, rw 0.25

0.6
0.4
0.2
0.0
0.0

0.2

0.4

0.6

0.8

1.0

(PJ WI / PM WI)

Figure 4-8: Well Indices Relation Regarding Partial Penetration with Varying Parameters

1.0

(SA WI / PM WI)

0.8
0.6
k 50 x 200 x 200
k 200 x 100 x 200

0.4

k 500 x 500 x 50
k 1000 x 50 x 300

0.2

k 1000 x 300 x 300

0.0
0.0

0.2

0.4

0.6

0.8

1.0

(PJ WI / PM WI)

Figure 4-9: Well Indices Relation Regarding Partial Penetration with Varying Anisotropy

29

4.6. Boundary Effects


The Semi-Analytical WI, as shown in Chapter 3, captures the influence of boundary
effects. Here, different well indices are compared for cases where the distance of the well
to the boundary varies.
Figure 4-10 shows the well indices of a vertical well in a reservoir 1050ft x 1050ft x 50ft
(21 x 21 x 3 blocks) with a permeability field of 400 md x 400 md x 20 md. As the well
was moved from the center blocks (i = 11, j = 11, k = 1-3) towards a corner (i = 1, j = 21,
k = 1-3), the Semi-Analytical WI only started to change as it got half-block (25ft) from
the boundary.
The projection and Peacemans methods are not influenced by the boundary effects and
therefore their values do not vary with distance to the boundary. However, the SemiAnalytical WI decreases significantly as it gets closer to the boundary.

8
7

Well Index

6
5
4

SA WI

PJ WI

2
1
0
0

10

15

20

25

Distance from Boundary (ft)

Figure 4-10: Boundary Effect in the Well Index Value

30

30

35

40

Chapter 5
5. Conclusions and Future Work
5.1. Conclusions
In this work, three different methods for computing well index - Peacemans model, the
Projection method and the Semi-Analytical approach are compared for different well
configurations and reservoir models. The well and reservoir models presented in this
work included horizontal, vertical, 2D and 3D slanted wells, in homogeneous (isotropic
and anisotropic) reservoirs with uniform Cartesian grids.
The values of the well indices computed by each method were also compared with the
goal of finding whether there is any correlation between the Semi-Analytical and
Projection WI. The influence of various parameters, such as permeability, grid size and
proximity to the boundaries were studied.
The following main conclusions can be drawn from this work:
- Use of Peacemans method can lead to significant errors for slanted wells and when
spherical flow or interference of other wells and boundaries are dominant.
- The Projection method is a practical correction to Peacemans model for slanted wells.
However, this technique breaks down when spherical flow is present (i.e. block at well
ends). It underestimates the true WI where spherical flow is present and overestimates it
elsewhere. Also, it does not consider interference and boundary effects, which becomes
critical when the well is very close (half block away) to a boundary.
- The Semi-Analytical approach is appropriate for computing exact well indices for any
type of well.
- By comparing normalized values of the Semi-Analytical and Projection WI, it was
observed that there is a correlation between them, unless spherical flow at well ends is
present or the well interferes with other wells or boundaries.

31

5.2. Future Work


This study suggests that there is a correlation between the Semi-Analytical WI and the
Projection well index. It may be possible to derive a correction factor that can be applied
to the Projection WI to further improve this numerical approximation.
Also, further tests should be performed with more complicated and real-field scenarios,
including non-uniform grids, multi-lateral wells and heterogeneous reservoirs. Other
methods for computing the well index for slanted wells or partial penetration could also
be included in the comparison, such as methods proposed by Cinco-Ley and Ramey [17]
and Babu and Odeh [18].

32

Nomenclature
c

Compressibility, 1/psi

Height, ft

Permeability, md

Length, ft

Pressure, psi

pi
p

PJWI

Block Pressure, psi


Well Pressure, psi
Projected Well Index

PMWI Peaceman Well Index


qw

Well Flow Rate, bbl/d

Radius, ft

ro

effective well-block radius, ft

rw

Wellbore Radius, ft

Skin

SAWI

Semi-Analytical Well Index

Time, days

x,y,z

Coordinate, ft

WI

Well Index unit

Viscosity, cp

Potential, psi

Porosity

Subscripts/Superscripts
i
Block
j

Segment

x,y,z

Coordinate

Well

33

References
1. Peaceman, D.W.: Interpretation of Well-Block Pressures in Numerical
Reservoir Simulation, paper SPE 6893, presented at the SPE-AIME 52nd
Annual Fall Technical Conference and Exhibition, Denver, Oct 9-12, 1977
2. Peaceman, D.W.: Interpretation of Well-Block Pressures in Numerical
Reservoir Simulation With Nonsquare Grid Blocks and Anisotropic
Permeability, paper SPE 10528, presented at the 1982 SPE Symposium on
Reservoir Simulation, New Orleans, Jan 31-Feb 3
3. Peaceman, D.W.: Interpretation of Wellblock Pressures in Numerical
Reservoir Simulation: Part3 Off-Center and Multiple Wells Within a
Wellblock, paper SPE 16976, presented at the 1987 SPE Annual Technical
Conference and Exhibition, Dallas, Sept 27-30
4. Schlumberger ECLIPSE: Schedule User Guide 2004A, Chapter 6
Technical Description (2004)
5. Schlumberger ECLIPSE: ECLIPSE Technical Description 2004A, Chapter
64 Well Inflow Performance (2004)
6. Wolfsteiner, C., Aziz, K. and Durlofsky, L. J.: Modeling Conventional and
Non-Conventional Wells, presented at Sixth International Forum on
Reservoir Simulation, Hof/Salzburg, Austria, Sept 3-7 2001
7. Wolfsteiner, C., Durlofsky, L. J. and Aziz, K.: Calculation of Well Index for
Nonconventional Wells on Arbitrary Grids, Computational Geosciences, 7,
61-82, 2003
8. Valvatne, P.H.: A Framework for Modeling Complex Well Configurations,
Masters Report, Stanford University, 2000
9. Serve, J.: An Enhanced Framework for Modeling Complex Well
Configurations, Masters Report, Stanford University, 2002
10. Maizeret, P.D.: Well Indices for Non-Conventional Wells, Masters Report,
Stanford University, 1996
11. Williamson, A.S. and Chappelear, J.E.: Representing Wells in Numerical
Reservoir Simulation: Part 1 Theory, paper SPE 7697, presented at the SPE
Fifth Symposium on Reservoir Simulation, Denver, Jan31-Feb 2, 1979
12. Peaceman, D.W.: Representation of a Horizontal Well in Numerical
Reservoir Simulation, paper SPE 21217 (1991), Advanced Technology
Series, Vol1, No.1
13. Aziz, K., Durlofsky, L.J. and Gerritsen, M.: Notes for Petroleum Reservoir
Simulation, Stanford University, 2004

34

14. Aziz, K. and Settari, A.: Petroleum Reservoir Simulation, Applied Science
Publisher, 1979
15. Pan, H.Q.: AdWell 2.1 User Manual, Stanford University, 2005
16. Schlumberger ECLIPSE: ECLIPSE Reference Manual 2004A, 2004
17. Cinco-Ley, H. and Ramey Jr., H. J.: Pseudo-Skin Factors for PartiallyPenetrating Directionally-Drilled Wells, paper SPE 5589, presented at the
SPE_AIME 50th Annual Fall Meeting, Dallas, Sept 28 Oct 1, 1975
18. Babu, D.K. and Odeh, A.S.: Productivity of a Horizontal Well, paper SPE
18298, presented at 1988 SPE Annual Technical Conference and Exhibition,
Houston, Oct 2-5

35

S-ar putea să vă placă și