Sunteți pe pagina 1din 390

NTNU

Faculty of Engineering And Technology


Well Production
Decline
Branimir Cvetkovic
December 2008
Supervisor:
Assessor:
ii
i
ii
Abstract
Eective rate-time analysis during a declining production in an oil or gas wells is
an important tool for establishing a successful management. The reason behind
the production decline include reservoir, fracture and well conditions. A wells
decline rate is transient, signifying that the pressure wave propagates freely from
the wellbore, leading to depletion when the outer boundary for the well is reached
and to the wave propagation coming to a halt. This thesis studies the transient
decline, with emphasis on a horizontal well with fracture wellbore responses. It
also deals with depletion decline, investigating the wellbore pressure responses
for a vertical well producing under variable rate conditions of Arps decline.
The well decline model solutions are analytical, and the modelling itself is
carried out in two steps. The rst step involves modelling the transient well
responses of a multi-fractured horizontal well. These responses originate from
an innitive reservoir and are considered as full-time rate-time responses. Multi-
fractured horizontal well rate-time responses represent the solutions to a diu-
sion equation with varying boundary conditions and dierent fracture options
(i.e., with or without fracture, a variety of fracture orientations, various frac-
ture lengths, etc.). The transient model calculates individual fracture rates,
productivity indexes and an equivalent wellbore radius for the multi-fractured
well. For the transient decline of a fractured-horizontal well model, well data
is matched and the reservoir diagnosis and production prognosis are improved
through the individual fracture production, with a model screening ability, and
novel model features that can handle wellbore conditions changing from rate-
to-pressure. Screening analyses can generate valuable information for fracture
diagnosis in addition to a well and fracture production prognosis. Further model
runs are carried out to match the real well data. The model solution is comple-
mentary to the reservoir simulation. More geology features should be considered
to fully take advantage of the modelling ndings. The starting point of the sec-
ond modelling step concerns late time vertical-well responses or decline curves
involving empirical solution of Arps type. This includes an investigation of well
pressure responses for a rate decline of an Arps-type variable-rate of a wellbore
for selected exponents, /. The modelling explores pressure wellbore responses
during a variable-rate production, and the approach introduces a no-ow speci-
ed boundary that moves outwards from a wellbore axis with a predened speed.
For the specic speed of a no-ow moving boundary, the model generates pres-
sure proles causing the decline in production. In the depletion decline, pressure
proles were generated for various decline exponents, b. Known b values were se-
lected and each of them was empirically derived through the inow performance
relationships to a drive mechanism. This modelling approach with analytically
derived pressure solutions can be extended to a horizontal well. Furthermore,
the continuously measured well rates and pressure models can be calibrated and
veried.Eective rate-time analysis during a declining production in an oil or gas
iii
wells is an important tool for establishing a successful management. The reason
behind the production decline include reservoir, fracture and well conditions. A
wells decline rate is transient, signifying that the pressure wave propagates freely
from the wellbore, leading to depletion when the outer boundary for the well is
reached and to the wave propagation coming to a halt. This thesis studies the
transient decline, with emphasis on a horizontal well with fracture wellbore re-
sponses. It also deals with depletion decline, investigating the wellbore pressure
responses for a vertical well producing under variable rate conditions of Arps
decline.
The well decline model solutions are analytical, and the modelling itself is
carried out in two steps. The rst step involves modelling the transient well
responses of a multi-fractured horizontal well. These responses originate from
an innitive reservoir and are considered as full-time rate-time responses. Multi-
fractured horizontal well rate-time responses represent the solutions to a diu-
sion equation with varying boundary conditions and dierent fracture options
(i.e., with or without fracture, a variety of fracture orientations, various frac-
ture lengths, etc.). The transient model calculates individual fracture rates,
productivity indexes and an equivalent wellbore radius for the multi-fractured
well. For the transient decline of a fractured-horizontal well model, well data
is matched and the reservoir diagnosis and production prognosis are improved
through the individual fracture production, with a model screening ability, and
novel model features that can handle wellbore conditions changing from rate-
to-pressure. Screening analyses can generate valuable information for fracture
diagnosis in addition to a well and fracture production prognosis. Further model
runs are carried out to match the real well data. The model solution is comple-
mentary to the reservoir simulation. More geology features should be considered
to fully take advantage of the modelling ndings. The starting point of the sec-
ond modelling step concerns late time vertical-well responses or decline curves
involving empirical solution of Arps type. This includes an investigation of well
pressure responses for a rate decline of an Arps-type variable-rate of a wellbore
for selected exponents, /. The modelling explores pressure wellbore responses
during a variable-rate production, and the approach introduces a no-ow speci-
ed boundary that moves outwards from a wellbore axis with a predened speed.
For the specic speed of a no-ow moving boundary, the model generates pres-
sure proles causing the decline in production. In the depletion decline, pressure
proles were generated for various decline exponents, b. Known b values were se-
lected and each of them was empirically derived through the inow performance
relationships to a drive mechanism. This modelling approach with analytically
derived pressure solutions can be extended to a horizontal well. Furthermore,
the continuously measured well rates and pressure models can be calibrated and
veried.
ii
Contents
Abstract iii
Contents iii
List of Tables vii
List of Figures ix
Acknowledgements xxiii
1 INTRODUCTION 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Scope of the Work . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Organisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 TRANSIENT RATE DECLINE REVIEW 9
2.1 Oil Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.1 Vertical Well . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1.2 Horizontal Well . . . . . . . . . . . . . . . . . . . . . . . . 28
2.1.3 Vertical-Fractured Well . . . . . . . . . . . . . . . . . . . . 30
2.1.4 Horizontal-Fractured Well . . . . . . . . . . . . . . . . . . 35
2.2 Gas Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.2.1 Vertical Well . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.2.2 Vertical-Fractured Well . . . . . . . . . . . . . . . . . . . . 41
2.2.3 Horizontal-Well . . . . . . . . . . . . . . . . . . . . . . . . 42
2.2.4 Horizontal-Fractured Well . . . . . . . . . . . . . . . . . . 43
2.3 Multiphase Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.4 Flow Under Variable Rate and Pressure . . . . . . . . . . . . . . . 46
2.5 Other Transient Models . . . . . . . . . . . . . . . . . . . . . . . 49
2.5.1 Multilateral Model . . . . . . . . . . . . . . . . . . . . . . 49
2.5.2 Multiple Wells Model . . . . . . . . . . . . . . . . . . . . . 51
iii
iv CONTENTS
3 DEPLETION RATE DECLINE REVIEW 53
3.1 Empirical Models (Arps) . . . . . . . . . . . . . . . . . . . . . . . 53
3.2 Analytical-Numerical Models . . . . . . . . . . . . . . . . . . . . . 56
3.2.1 Oil Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
3.2.2 Gas Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.2.3 Multiphase Flow . . . . . . . . . . . . . . . . . . . . . . . 60
3.3 Type-Curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.3.1 Vertical Well . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.3.2 Vertical Fractured Well . . . . . . . . . . . . . . . . . . . . 81
3.3.3 Horizontal Well . . . . . . . . . . . . . . . . . . . . . . . . 84
3.3.4 Horizontal Fractured Well . . . . . . . . . . . . . . . . . . 86
3.3.5 Multilateral Well . . . . . . . . . . . . . . . . . . . . . . . 89
3.3.6 Multi Wells . . . . . . . . . . . . . . . . . . . . . . . . . . 93
3.4 Decline Curve Analysis Physics . . . . . . . . . . . . . . . . . . . 96
3.4.1 Solution Gas Drive Decline . . . . . . . . . . . . . . . . . . 98
3.4.2 Solution Gas Drive and Gravity Drainage Decline . . . . . 107
3.5 Analysis of Well Production data . . . . . . . . . . . . . . . . . . 110
3.5.1 Type Curves and Decline Curve Analysis . . . . . . . . . . 110
4 RATE DECLINE OF A FRACTURED WELL 127
4.1 Transient Oil Flow . . . . . . . . . . . . . . . . . . . . . . . . . . 128
4.1.1 Fractured-Vertical Well Model . . . . . . . . . . . . . . . . 128
4.1.2 Horizontal Well with Transversal Fractures . . . . . . . . . 131
4.1.3 Horizontal Well with Longitudinal Fractures . . . . . . . . 136
4.2 Depletion Oil Flow . . . . . . . . . . . . . . . . . . . . . . . . . . 141
4.2.1 Fractured Vertical Well Model . . . . . . . . . . . . . . . . 142
4.2.2 Horizontal Well with Transversal Fractures . . . . . . . . . 144
4.2.3 Model Solutions Comparison . . . . . . . . . . . . . . . . . 147
4.3 Additional Well-Fracture Features . . . . . . . . . . . . . . . . . . 151
4.3.1 Fracture Conductivity . . . . . . . . . . . . . . . . . . . . 151
4.3.2 Well Conductivity . . . . . . . . . . . . . . . . . . . . . . . 152
4.3.3 Fracture-Well Limited Communication (Choking Eect) . 154
4.3.4 Restart Option . . . . . . . . . . . . . . . . . . . . . . . . 154
4.3.5 Late Time Approximations . . . . . . . . . . . . . . . . . . 157
5 RATE DECLINE WITH A MOVING BOUNDARY 165
5.1 Vertical Well Oil Flow . . . . . . . . . . . . . . . . . . . . . . . 167
5.1.1 Introduction to Moving Boundary Problems . . . . . . . . 167
5.1.2 Fixed Boundary . . . . . . . . . . . . . . . . . . . . . . . . 168
5.1.3 Moving Boundary . . . . . . . . . . . . . . . . . . . . . . . 169
5.1.4 Variable Rate Production of Arps Type . . . . . . . . . . . 178
5.2 Vertical Well - Gas Flow . . . . . . . . . . . . . . . . . . . . . . . 184
5.2.1 Pseudo-Pressure Transformation (Intermediate Pressures) . 185
CONTENTS v
5.2.2 Pressure Squared Transformation . . . . . . . . . . . . . . 189
5.2.3 No-Flow Moving Boundary - Gas Flow Solutions . . . . . 190
5.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
6 RATE DECLINE CURVES 193
6.1 Transient Decline of a Fractured-Horizontal Well . . . . . . . . . . 194
6.1.1 Transient-Rate Response of a Well . . . . . . . . . . . . . 195
6.1.2 Well Transient-Pressure Responses . . . . . . . . . . . . . 213
6.1.3 Well Pressure-to-Rate Responses . . . . . . . . . . . . . . 215
6.1.4 A Well with Longitudinal Fractures . . . . . . . . . . . . . 218
6.2 Well Depletion Responses . . . . . . . . . . . . . . . . . . . . . . 220
6.2.1 Closed (BOX) Model . . . . . . . . . . . . . . . . . . . . . 220
6.3 Rate Decline with a No-ow Moving Boundary . . . . . . . . . . . 222
6.3.1 Hyperbolic Decline (/ = 1,3) . . . . . . . . . . . . . . . . 222
6.3.2 Hyperbolic Decline (/ = 0.5) . . . . . . . . . . . . . . . . . 225
6.3.3 Harmonic Decline (b = 1) . . . . . . . . . . . . . . . . . . 232
6.3.4 Decline Exponent (b = 2) . . . . . . . . . . . . . . . . . . 242
6.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
7 CASE STUDIES 255
7.1 Model Comparison and Validation . . . . . . . . . . . . . . . . . . 255
7.1.1 Well Models Comparisons . . . . . . . . . . . . . . . . . . 255
7.1.2 Semi-Analytical versus Numerical Model Validation . . . . 256
7.2 Fractured-Horizontal Well - Oil Production . . . . . . . . . . . . . 259
7.2.1 Fractured-Horizontal Well (Ekosk Oil Field - North Sea) . 259
7.2.2 Fractured-Horizontal Well (North Sea Oil Field) . . . . . . 262
7.2.3 Syd-Arne Oil Field (North Sea) . . . . . . . . . . . . . . . 266
7.2.4 Syd-Arne Oil Field (North Sea) . . . . . . . . . . . . . . . 274
7.2.5 Valhall Oil Field (North Sea) . . . . . . . . . . . . . . . . 282
7.3 Fractured-Horizontal Well - Water Injection . . . . . . . . . . . . 289
7.3.1 Syd-Arne (North Sea) . . . . . . . . . . . . . . . . . . . . 289
7.4 Fractured-Horizontal Well - Gas Production/Injection . . . . . . . 294
7.4.1 Syd-Arne Synthetic Data . . . . . . . . . . . . . . . . . . . 294
7.5 Vertical Well Exponential Decline with the Moving Boundary . 296
7.5.1 Variable Rate IBCs of b Almost Zero . . . . . . . . . . . . 296
8 DISCUSSION, CONCLUSIONS, RECOMMENDATIONS 311
8.1 DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
8.1.1 Transient Rate Decline . . . . . . . . . . . . . . . . . . . . 311
8.1.2 Depletion Rate Decline . . . . . . . . . . . . . . . . . . . . 314
8.2 CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
8.3 RECOMMENDATIONS . . . . . . . . . . . . . . . . . . . . . . . 316
vi CONTENTS
9 NOMENCLATURE 319
9.1 Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
9.2 SI Metric Conversion Factors . . . . . . . . . . . . . . . . . . . . 321
10 REFERENCES 323
A No-Flow Moving Boundary Model Solutions 345
B "LAPLACE" Inversion Transforms 347
C Regression Techniques 351
C.1 Linear Regression . . . . . . . . . . . . . . . . . . . . . . . . . . 351
C.2 Linear Multiple Regression . . . . . . . . . . . . . . . . . . . . . 352
C.3 Weighted Residuals Regression . . . . . . . . . . . . . . . . . . . 354
D Relevant Reports and Papers 355
D.1 SPE Papers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
D.2 Presentations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
D.2.1 Conferences/Forums . . . . . . . . . . . . . . . . . . . . . 355
D.2.2 Schlumberger Internal EUREKA Presentations: . . . . . . 356
D.3 Industry Reports . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
D.4 NTNU Faculty Reports . . . . . . . . . . . . . . . . . . . . . . . . 356
D.5 Other Related Presentations and Reports . . . . . . . . . . . . . . 357
List of Tables
1-1 Pressure testing and rate testing comparisons of events, interpre-
tation methods and historical emphasis (Following the 2006 review
by Gringarten and Anderson et al.] . . . . . . . . . . . . . . . . . 7
2-1 Pressure gradients and dimensionless pressure functions for radial
reservoir ow at the well - After Valko and Economides (1995) . . 11
2-2 Variable rate publications the history [After Gringarten (2006)] . 47
3-1 The rate-time equation for a gas well in terms of the back pressure
exponent, n, with constant "p
&)
" of 0 as dened by Fetkovich (1980) 71
3-2 The dimensionless ratio as a function of dimensionless pressure as
dened by Anash et al. (2000) . . . . . . . . . . . . . . . . . . . . 78
3-3 The vertical well, the vertical and the horizontal fracture . . . . . 82
4-1 Solution for a fractured-vertical (longitudinal) and a fractured-
horizontal (transversal) well . . . . . . . . . . . . . . . . . . . . . 148
4-2 Finite and innite conductivity fractures - parameters) . . . . . . 153
4-3 Model solutions for a fractured-vertical well as compared to those
of a fractured-horizontal (single-transversal-fracture) well . . . . 163
5-1 The dimensionless pressure, "p
1
", calculated by the model at time
"t
1
", for a variable-rate production of Arps type, dened by ex-
ponent b (ranging from 0.3; 0.5; 1., and 2) . . . . . . . . . . . . . 183
6-1 Reservoir, Well and Fracture Data . . . . . . . . . . . . . . . . . . 195
6-2 The sensitivity to the initial reservoir pressure, "P
i
", the porosity
and the eective reservoir thickness, h . . . . . . . . . . . . . . . 196
6-3 Sensitivity to fracture half length sizes, Lf . . . . . . . . . . . . . 208
6-4 Sensitivity to the distance bewtween two fractures due to well
length, L changes (L= 2520, 2100, and 1680 ft . . . . . . . . . . . 208
6-5 Sensitivity to the distance bewtween two fractures due to number
of fractures changes sor the same well length, L = 2100 ft . . . . . 208
6-6 Sensitivity to fracture conductivity "F
C
" (mDft) equal to 2500,
1000, qnd 100 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
vii
viii LIST OF TABLES
6-7 The sensitivity to the fracture conductivity ,"F
C
" (mDft), when
it is innite, 1000, and 50 mDft . . . . . . . . . . . . . . . . . . . 211
6-8 Wellbore inner boundary conditions, IBCs of variable pressure for
the innite conductivity and nite conductivity for fractures with
"F
C
" = 50 mDft . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
6-9 The constant A as a function of the coecient of the no-ow
moving boundary . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
6-10 A as a function of the coecient of the no-ow moving boundary 231
6-11 A as a function of the coecient of the no-ow moving boundary 239
6-12 The constant A as the function of the coecient of the no-ow
moving boundary . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
6-13 The dimensionless pressure, "p
1
", calculated by the model for
"t
1
" going to 0,and the "t
1
" going to innity, for a variable rate
production of Arps type. The decline exponent, b (b with values
of 0.333; 0.5; 1., and 2) denes the Arps type production decline . 252
7-1 The input data for fractured-horizontal well (MF), fractured-vertical
well (VFW) and partially perforated horizontal well (PP) model . 257
7-2 Input data from the Ekosk eld - North Sea (for a horizontal
well with 8 transversal-fractures) . . . . . . . . . . . . . . . . . . 260
7-3 Some general parameters of the Syd Arne North Sea eld (SPE
103282) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
7-4 Wellbore IBCs of variable pressure for the innite conductivity
and nite conductivity, for "F
C
" = 50 (mDft) fractures . . . . . . 285
7-5 Input parameters (n and "r
1
") to the semi-exponential decline in
the moving boundary model . . . . . . . . . . . . . . . . . . . . . 297
B-1 The inverse "Laplace" transform methods [after Davies and Mar-
tin (1979)] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
B-2 The comparrison of the inverse "Laplace" transform methods . . . 350
List of Figures
2.1 The domain in which the pseudo-pressure, . varies linearly with
j and j
2
[After Bourdarot (1998)]. . . . . . . . . . . . . . . . . . . 38
3.1 Dimensionless Arps curves (Decline / = 0.0; 0.5, and 1.0) [After
Cvetkovic (1992)]. . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.2 Semi-analytical dimensionless rate-time type curves (for various
dimensionless radii, :
1
) [After Cvetkovic (1992)]. . . . . . . . . . 64
3.3 Combined transient-depletion dimensionless Fetkovich (1973) rate-
time type curves [After Cvetkovic (1992)]. . . . . . . . . . . . . . 65
3.4 Transient dimensionless rate-time curves (for two values of :
1
)
[After Cvetkovic (1992)]. . . . . . . . . . . . . . . . . . . . . . . . 66
3.5 Transformed depletion dimensionless rate-time curves (for two di-
mensionless r
1
) [After Cvetkovic (1992)]. . . . . . . . . . . . . . . 66
3.6 Transient dimensionless rate-time curves (for various dimension-
less :
1
values) [After Cvetkovic (1992)]. . . . . . . . . . . . . . . . 67
3.7 Arps dimensionless rate-time curves [After Cvetkovic (1992)]. . . . 67
3.8 A type-curve match for a constant- pressure drawdown test with
variable property solutions [After Samaniego and Cinco (1980)]. . 69
3.9 The dimensionless ow rate compared to the Arps decline rates
[After Samaniego and Cinco (1980)]. . . . . . . . . . . . . . . . . 70
3.10 Radial-linear gas reservoir type curves [After Carter (1985)]. . . . 73
3.11 The linear and the radial ow geometry [After Chen and Teufel
(2000)]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
3.12 The dimensionless rate,
1
, and the cumulative production, Q
1
,
versus the dimensionless time, t
1
[After Chen and Teufel (2000)]. 75
3.13 The composite type-curves: (A) The ow rate vs. time; (B) The
cumulative production vs. time; (C) The ow rate vs. the cumu-
lative production [After Chen and Teufel (2000)]. . . . . . . . . . 76
3.14 The distribution of the viscosity-compressibility function [After
Ansah et al. 2000]. . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.15 The "rst-order" polynomial solution for real-gas owunder boundary-
dominated ow conditions. A viscosity-permeability, jc
t
, is linear
with dimensionless pressure, j
1
[After Ansah 2000]. . . . . . . . . 79
ix
x LIST OF FIGURES
3.16 The "exponential" solutions for real-gas ow under boundary-
dominated ow conditions [After Ansah (2000)]. . . . . . . . . . 79
3.17 "General polynomial" solution for real-gas ow under boundary-
dominated boundary conditions [after Ansah 2000)]. . . . . . . . 80
3.18 The vertical well, the vertical and the horizontal fracture. . . . . . 83
3.19 The vertical fractured well in a rectangular drainage area [After
Chen et al. (1991)]. . . . . . . . . . . . . . . . . . . . . . . . . . . 83
3.20 Type of ow for a vertcal fractured well . . . . . . . . . . . . . . . 84
3.21 The dimensionless rate, qD versus the dimensionless time, tDXf
for the horizontal well [After Cox et al. (1996)]. . . . . . . . . . . 85
3.22 Decline curve for a horizontal well ina bounded reservoir [After
Poon (1991)]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.23 The eect of the aspect ratio on horizontal well productivity (the
ratio of the length to the width of a rectangular well pattern)
[After Poon (1991)]. . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.24 The fracture orientation along a horizontal well. . . . . . . . . . . 90
3.25 A well in a three layered reservoir with perforated segments re-
placed by uniform-ux fractures. . . . . . . . . . . . . . . . . . . . 91
3.26 The multibranch and multiple-fracture congurations for horizon-
tal wells [After Economides at al. (2001)]. . . . . . . . . . . . . . 91
3.27 The multilateral well types [After Louis J. Durlofsky TAML, 1999
presentation]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
3.28 A vertical and horizontal well with laterals positioning within an
oil reservoir [After Cvetkovic et al., (2007)]. . . . . . . . . . . . . 93
3.29 The multiple vertical, horizontal and deviated completioned wells
in the layered reservoir [After Gilchrist et al. (2007)]. . . . . . . . 95
3.30 The dimensionless rate vs. the dimensionless time Fetkovich type
curves [After Fetkovich (1980)]. . . . . . . . . . . . . . . . . . . . 102
3.31 The rst decline on Fetkovichs type curve, for / 0 [After Padilla
and Camacho (2004)]. . . . . . . . . . . . . . . . . . . . . . . . . 107
3.32 The rst decline on Fetkovichs type curve, for / = 0 [After Padilla
and Camacho (2004)]. . . . . . . . . . . . . . . . . . . . . . . . . 108
3.33 The second decline on Fetkovichs type curve, for / < 0. The
decline exponent is negative and constant [After Padilla and Ca-
macho (2004)]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
3.34 Production decline curves for a nite-conductivity, vertcally frac-
tured well positioned in a closed rectangular reservoir.[after Pos-
ton and Poe (2008)]. . . . . . . . . . . . . . . . . . . . . . . . . . 120
3.35 The production decline curves for a vertical well positioned in a
cylindrical reservoir with a step-rate ux outer-boundary condi-
tion .[After Poston and Poe (2008)]. . . . . . . . . . . . . . . . . . 121
LIST OF FIGURES xi
3.36 The production decline curves for a vertical well positioned in a
cylindrical reservoir with a ramprate ux outer-boundary condi-
tion.[After Poston and Poe (2008)]. . . . . . . . . . . . . . . . . . 122
3.37 Production decline curves for an innite conductivity fractured
well, centrally located in a closed, cyllindrical reservoir [After Po-
stone and Poe (2008)]. . . . . . . . . . . . . . . . . . . . . . . . . 123
3.38 Production decline curves for a nite-conductivity vertically frac-
tured well centrally located in a closed [after Poston and Poe
(2008) ]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
4.1 A fractured-horizontal well of length, L, with three transversal
fractures of half-lengths, L
)
. The reservoir is non-bounded or
innite in the x and y directions (Top view). . . . . . . . . . . . . 132
4.2 A fractured-horizontal well of length 1 , with three transversal
fractures of half-lengt 1
)
. The reservoir is non-bounded or innite
in the A
c
and the 1
c
directions (Cross section view). . . . . . . . 137
4.3 An innite conductivity vertical fracture fully penetrating the x
direction of a reservoir and the formation in the vertical z direc-
tion. The no-ow outer boundary condition denes the closed
rectangular reservoir (Areal cross section). . . . . . . . . . . . . . 143
4.4 A fractured-horizontal well of length L with three transversal frac-
tures of half-length 1
)
. The reservoir is bounded and no-ow
boundaries are X
c
and Y
c
(Areal cross section). . . . . . . . . . . 145
4.5 Avertical-fractured, and a horizontal- fractured well (with transver-
sal and longitudinal single fractures). . . . . . . . . . . . . . . . . 149
4.6 The model for a vertical-fractured well rate,
1
. versus s. The
fracture is longitudinal and of innite conductivity (Laplace"
space solutions). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
4.7 The model for a fractured-horizontal (with a transversal fracture
of uniform ux) well rate,
1
. versus s (Laplace" space solutions).150
4.8 The two model solutions for the rate,
1
. versus s (Laplace"
space solutions). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
4.9 Models for a fractured-horizontal well, the eective wellbore ra-
dius of a vertical well, and the eective half-length of a horizontal
well with a single transversal fracture. . . . . . . . . . . . . . . . 158
5.1 The dimensionless pressure, j
1
, as a function of the dimensionless
time, t
1
and radial distance, :
1
( with 1 = 1, and speed r = 10). 172
5.2 The dimensionless pressure, j
1
. as a function of the dimensionless
time, t
1
. for the dimensionless distance, :
1
= 1 ( with 1 = 1,
and r = 10). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
xii LIST OF FIGURES
5.3 The dimensionless pressure, j
1
. as a function of the dimensionless
time, t
1
. for the dimensionless distance, :
1
= 10 ( with 1 = 1
and r = 10). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
5.4 The dimensionless distance, :
1
. of a no-ow moving boundary as
a function of a constant, i, and the dimensionless time t
1
. . . . . 176
5.5 The dimensionless position, :
1
. versus the dimensionless time, t
1
.
for a coecient of no-ow moving boundary i = 1,3,5,7 and 9. . 177
5.6 The velocity of a no-ow moving boundary,
1
=
ov
T
ot
T
. as a func-
tion of the dimensionless time, t
1
. for various constant values of ,
i. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
5.7 The dimensionless distance, :
1
. of a no-ow moving boundary as
a function of the dimensionless time, t
1
. for various constants of
the no-ow moving boundary, i = 1. 3. 5. 7.and 9. . . . . . . . . . 178
5.8 A 3D plot of Arps equation rate, . versus time, t. and initial
decline, 1
i
( for a decline exponent / = 0.33, / = 0.5, / = 1, and
/ = 2). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
5.9 A 2D plot of Arps equation rate, q, versus time, t. and initial
decline, 1
i
(for a decline exponent / = 0.33, / = 0.5 . / = 1, and
/ = 2). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
5.10 The rate, . versus time, t, for / = 0.33 plotted in circles, and
various decline exponents (/ = 0.5. 1. and 2) all plotted as solid
line. The rate versus time is calculated for a specic initial decline,

i
= 5000, and an initial decline, 1
i
= 0.01). . . . . . . . . . . . . 180
5.11 The constant A as a function of the constant i of a no-ow moving
boundary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
6.1 Individual fracture rates,
)vi
(i = 1. ...5) and the rate of a well
(with fractures), (//|,d), versus time, t, in days. . . . . . . . . . 196
6.2 Individual cumulative fracture production, Q
)vi
(i = 1. .... 5), and
the cumulative production of a well with fractures, Q (bbl), versus
time, t, in days. . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
6.3 Rate and cumulative production proles for various values of poros-
ity, c, (i.e., 0.24, 0.34, 0.44). . . . . . . . . . . . . . . . . . . . . . 198
6.4 Rate and cumulative production proles for various eective thick-
nesses, /, (i.e., 95 , 75 and 55 ft). . . . . . . . . . . . . . . . . . . 198
6.5 The individual-fracture rate and individual-cumulative fracture
production versus time for various values of initial pressure, 1
i
,
(i.e., 6425, 5700, and 4975 psi). . . . . . . . . . . . . . . . . . . . 199
6.6 The rates and cumulative productions vs. time for varying values
of the initial wellbore pressure, 1
&)
, (i.e., 2176, 2900 and 3626 psi).199
6.7 Individual fracture rates and cumulative productions (
)v2
and
Q
)v2
) vs. time for various values of wellbore pressure, 1
&)
, (i.e.,
2176, 2900 and 3626 psi). . . . . . . . . . . . . . . . . . . . . . . 200
LIST OF FIGURES xiii
6.8 Individual fracture rates and cumulative productions (
)v2
and
Q
)v2
) vs. time for various values of wellbore pressure, 1
&)
, (i.e.,
2176, 2900 and 3626 psi). . . . . . . . . . . . . . . . . . . . . . . 200
6.9 Individual fracture rates and cumulative productions (
)v3
and
Q
)v3
) vs. time for various values of wellbore pressure, 1
&)
, (i.e.,
2176, 2900 and 3626 psi). . . . . . . . . . . . . . . . . . . . . . . 201
6.10 Individual fracture rates, and cumulative productions, Q vs.
time for various values of permeability, 1
I
= 1

(i.e., 40, 4, 0.4


mD). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
6.11 The individual fracture rate, . and cumulative production, Q. vs.
time for various vertical permeabilities, 1

(i.e., 0.4, 2, and 4 mD). 202


6.12 The individual fracture rate, . and cumulative production, Q.
vs. time for various total compressibility values, c
T
(i.e., 7.25
e-05, 6.26 e-05, 1.86 e-06). . . . . . . . . . . . . . . . . . . . . . . 203
6.13 The individual fracture rate, , and the cumulative production,
Q, vs. time for various oil viscosities, j (i.e., 3, 4 and 5 cp). . . . 203
6.14 The individual fracture rate, , and the cumulative production,
Q, vs. time for various oil viscosities, 1o (i.e., 1.35, 1.55 and 1.75
rb/stb). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
6.15 The individual fracture rate, , and the cumulative production, Q,
vs. time for various fractures (longitudinal fracture vs. transversal
fractures). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
6.16 Productivity index, PI for a horizontal well with 5 transversal
fractures and cumulative rate, Q with time for various fracture
partial penetrations with height, h (of 75, 55 and 35 ft). . . . . . 206
6.17 Productivity index, PI for a horizontal well with a longitudinal
fracture positioned along the 2100 ft horizontal well, and cumu-
lative rate, Q with time for various fracture partial penetrations
with height, h (of 75, 55 and 35 ft). . . . . . . . . . . . . . . . . 206
6.18 Productivity index, PI for a horizontal well with a longitudinal
fracture and transversal fractures, and cumulative rate, Q with
time for the partial penetrated fracture the height, h = 55 ft. . . 207
6.19 Productivity index, PI for a horizontal well with transversal frac-
tures, and cumulative rate, Qwith time for the various half-length,
1
)
(of 170, 85 and 42.5 ft). . . . . . . . . . . . . . . . . . . . . . 207
6.20 Productivity index, PI for a horizontal well with transversal frac-
tures, and cumulative rate, Q with time for the unequal fracture
(case 2 and 3) compared to the equally sized fractures (case 1). . 209
6.21 Productivity index, PI for a horizontal well with transversal frac-
tures, and cumulative rate, Q with time for the well length, L (of
2520, 2100 and 1680 ft). . . . . . . . . . . . . . . . . . . . . . . . 209
xiv LIST OF FIGURES
6.22 Productivity index, 11 for a horizontal well with transversal frac-
tures, and cumulative rate, Q with time for the various number
of fractures, on a xed well-length, 1=2100 ft for various number
of fractures, : (of 7, 5 and 3). . . . . . . . . . . . . . . . . . . . . 210
6.23 The productivity index, 11. for a horizontal well with transversal
fractures, and the cumulative production, Q. vs. time for the var-
ious fracture characters (uniform ux, innite conductivity, and
nite conductivity (with 1
C
= 1000 mDft). . . . . . . . . . . . . . 211
6.24 The productivity index, 11. for a horizontal well with transversal
fractures, and the cumulative rate, Q. vs. time for the innite
conductivity fracture and nite conductivity fractures (with 1
C
values of 1000 and 50 mDft). . . . . . . . . . . . . . . . . . . . . . 212
6.25 The rate, . for a horizontal well with transversal fractures, and
the cumulative rate, Q. vs. time for an innite conductivity frac-
ture, and nite conductivity fractures with 1
C
values of 1000 and
50 mDft. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
6.26 The rate, . for a horizontal well with transversal fractures, and
individual fracture rates
)vi
, where i = 1. .... 5 versus the cu-
mulative rate, Q, for an innite conductivity fracture and nite
conductivity fractures with 1
C
values of 1000 and 50 mDft. . . . . 213
6.27 The rate, q, versus the cumulative production, Q, for a wellbore
with 5 transversal fractures. Each fracture conductivity, 1
C
. is 50
(mDft). The wellbore friction reduces both the well rate produc-
tion and the cumulative production. . . . . . . . . . . . . . . . . 214
6.28 Responses of rate, q, versus time, t. The wellbore inner boundary
conditions, IBC, correspond to a variable pressure for the innite
and nite conductivity fractures of 50 mDft. . . . . . . . . . . . . 215
6.29 The rate, q, versus cumulative production ,Q, responses. The well-
bore inner boundary conditions, IBC are variable pressure for the
innite conductivity and nite conductivity, 50 (mDft), fractures.
Each individual fracture rate vs.cumulative rate is graphically pre-
sented. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
6.30 Inuence of fracture choking eect (for variable rate IBC) on pres-
sure dierence and well with fractures PI. Fractures are nite con-
ductivity (2500 mDft). . . . . . . . . . . . . . . . . . . . . . . . . 217
6.31 Inuence of fracture choking eect (for variable rate IBC) on cu-
mulative indifvidual fracture production Q
)vi
(i=1,5 and 3) and
well with fractures PI. Fractures are nite conductivity (2500
mDft). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
LIST OF FIGURES xv
6.32 Wellbore contant rate to constant pressure IBCs. The value of
the employed owing pressure following the constant rate period
corresponds to the pressure determined by the model at the end
of this period. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
6.33 Longitudinal versus transversal fracture rates and the cumulative
production (for a horizontal well with ve equally spaced innite
conductive, and equal half-length fractures). . . . . . . . . . . . . 219
6.34 Longitudinal vs.transversal fracture rates and the cumulative pro-
duction (for selective fractures: 1, 5 and 3). . . . . . . . . . . . . 219
6.35 The productivity index, PI, versus time, t, for a horizontal well
with longutudinal fractures of n = 7, 5, and 3. . . . . . . . . . . 220
6.36 The rate, q, versus time, t, for the BOX model (5000 ft by 5000
ft). IBCs of variable pressure of 4900, 3900, and 2900 psi. Innite
conductive fractures with varying fracture-half lengths, L
)
, of 120,
85 and 50 ft. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
6.37 The pressure dierence, P, versus time, t, for the BOX model
(5000 ft by 5000 ft). IBCs of variable rate of 150, 110, and 70
bbl/d. Innite conductive fractures with varying fracture half-
lengths, L
)
, of 120, 85 and 50 ft. . . . . . . . . . . . . . . . . . . 221
6.38 The rate, , versus time, t, for b=0.33 plotted in circles, and for
other values of the decline exponents (b=0.5, 1, and 2) all plotted
as solid line. The rate versus time is calculated for a specic initial
decline
i
= 5000 and a initial decline, 1
i
= 0.01 ). . . . . . . . . 223
6.39 The constant as a function of the constant , i. of the no-ow
moving boundary. . . . . . . . . . . . . . . . . . . . . . . . . . . 224
6.40 The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and the dimensionless distance, :
1
(for i = 1 and / =
1
3
). . . . . . 225
6.41 The dimensionless pressure, j
1
. versus the dimensionless time, t
1
.
and with a distance :
1
= 1 (for i = 1, and the decline exponent
/ =
1
3
). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
6.42 The dimensionless pressure, j
1
. versus the dimensionless time, t
1
.
and with a distance :
1
= 4 (for i = 1, and the decline exponent
/ =
1
3
). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
6.43 The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and with a distance :
1
= 10 (for i = 1, and the decline exponent,
/ =
1
3
). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
6.44 The dimensionless pressure, j
1
, versus a distance, :
1
, at a di-
mensionless time t
1
= 16 (for i = 1, and the decline exponent
/ =
1
3
). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
6.45 The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 3). . . . . . . . 228
xvi LIST OF FIGURES
6.46 The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 5). . . . . . . . 228
6.47 The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 7). . . . . . . . 229
6.48 The dimensionless pressure, j
1
. versus the dimensionless time, t
1
.
and versus the dimensionless distance, :
1
(for i = 9). . . . . . . . 229
6.49 The rate, , versus time, t, for / = 0.5 (plotted in circles). The
other parameters were considered constant, i.e., the initial decline
decline 1
i
= 0.01, and the initial decline rate
i
= 5000). . . . . . 230
6.50 The constant (i) as a function of the constant , i. of no-ow
moving boundary. . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
6.51 The dimensionless pressure, j
1
. versus the dimensionless time,
t
1
. calculated at a dimensionless radius :
1
= 1 (i = 1, and a
decline exponent / = 0.5). . . . . . . . . . . . . . . . . . . . . . . 232
6.52 The pressure dierence, j
1
versus the dimensionless time, t
1
.
calculated at a dimensionless distance, :
1
= 4 (coecient: i = 1,
and decline exponent: b = 0.5). . . . . . . . . . . . . . . . . . . . 233
6.53 The pressure dierence, j
1
. versus the dimensionless time, t
1
,
calculated at the dimensionless distance r
1
= 10 (for i = 1, and
decline exponent: / = 0.5). . . . . . . . . . . . . . . . . . . . . . . 233
6.54 The dimensionless pressure, j
1
. versus the distance, :
1
. at a di-
mensionless time t
1
= 16 (for i = 3, and decline exponen: / = 0.5).234
6.55 The dimensionless pressure, j
1
, versus the dimensionless time,
t
1
. and versus the dimensionless distance, :
1
(for i = 1, / = 0.5). 234
6.56 The dimensionless pressure, j
1
, versus the dimensionless time,
t
1
, and versus the dimensionless distance, :
1
(for i = 3, / = 0.5). 235
6.57 The dimensionless pressure, j
1
, versus the dimensionless time,
t
1
, and versus the dimensionless distance, :
1
(for i = 5, / = 0.5). 235
6.58 The dimensionless pressure, j
1
, versus the dimensionless time, t
1
.
and versus the dimensionless distance, :
1
(for i = 7, / = 0.5). . . 236
6.59 The dimensionless pressure, j
1
, versus the dimensionless time,
t
1
, and versus the dimensionless distance, :
1
(for i = 9, / = 0.5). 236
6.60 The rate, , versus time, t, for / = 1.0 (plotted in circles). With
a specic initial decline
i
= 5000 and an initial decline rate 1
i
=
0.01). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
6.61 The constant as a function of the coecient i. (i) is cal-
culated for an inner boundary condition of variable rate. Arps
decline exponent / = 1. . . . . . . . . . . . . . . . . . . . . . . . . 238
6.62 The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
calculated at a dimensionless radius :
1
= 1 (for: i = 1 and / = 1). 240
LIST OF FIGURES xvii
6.63 The pressure dierence, j
1
, versus the dimensionless time, t
1
,
calculated at a dimensionless distance :
1
= 4 (for: i = 1, and
/ = 1). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
6.64 The pressure dierence, j
1
, versus the dimensionless time, t
1
,
calculated at a dimensionless distance :
1
= 10 (for: i = 1, and
/ = 1). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
6.65 The dimensionless pressure, j
1
, versus the dimensionless distance,
:
1
, at a dimensionless time t
1
= 16 (for: i = 1, and / = 1). . . . 241
6.66 The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 1, / = 1). . . . 242
6.67 The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 3, / = 1). . . . 243
6.68 The dimensionless pressure, j
1
versus the dimensionless time, t
1
and versus the dimensionless distance, :
1
(for i = 5, / = 1). . . . 243
6.69 The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 7, / = 1). . . . 244
6.70 The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 9, / = 1). . . . 244
6.71 The rate, , versus time, t, for decline exponent / = 2.0 (plotted
in circles). The initial decline
i
= 5000, and the initial decline
rate, 1
i
= 0.01. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
6.72 The constant, (i), as a function of the coecient, i, of the
no-ow moving boundary, . . . . . . . . . . . . . . . . . . . . . . . 246
6.73 The dimensionless pressure, j
1
, versus the dimensionless time,
t
1
,calculated at the dimensionless radius, :
1
= 1 (for, i = 1 and
for, b = 2). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
6.74 The pressure dierence, j
1
. versus the dimensionless time, t
1
,calculated
at a dimensionless distance, :
1
= 4 (for, i = 1, and for, / = 2). . 247
6.75 The pressure dierence, j
1
. versus the dimensionless time, t
1
,
calculated at a dimensionless distance, :
1
= 10 (for i = 1, and
decline exponent, / = 2). . . . . . . . . . . . . . . . . . . . . . . . 248
6.76 The dimensionless pressure, j
1
, versus the distance, :
1
, at the
dimensionless time, t
1
= 16 (for i = 1, and and decline exponent,
/ = 2). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
6.77 The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and the dimensionless distance, :
1
(for i = 1, / = 2). . . . . . . . 249
6.78 The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 3, / = 2). . . . 250
6.79 The dimensionless pressure, j
1
, versus dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 5, / = 2). . . . 250
6.80 The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus dimensionless distance, r
1
(for i = 7, / = 2). . . . . . 251
xviii LIST OF FIGURES
6.81 The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus dimensionless distance, :
1
(for i = 9, / = 2). . . . . . 251
7.1 The pressure-dierence versus time for three models (multi-fractured
horizontal well (MFW-open outer boundary-SLAB), vertical-fractured
well (VFW-open outer boundary-SLAB) and partially perforated
horizontal well (PPHOW-closed, BOX ) [After Cvetkovic et al.
(2000)]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
7.2 The semi-analytical model cumulative production compared to
the other models (2 numerical and a semi-analytical models). . . 258
7.3 Comparisons of model-calculated cumulative productions (two nu-
merical models and a semi-analytical single-phase model). The
production history comprises 1200 days. . . . . . . . . . . . . . . 259
7.4 A comparison of observed (oil rate) data with that calculated by
the model at an IBC of variable pressure (from 7 selected time
intervals). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
7.5 A comparison of observed (oil rate) data with that calculated
by the model at IBCs of variable pressure. The daily measured
pressures at the wellbore are devided into 3 pressure intervals. . . 261
7.6 The matching of observed and calculated pressure dierences ver-
sus time (for variable-rate IBCs and changes in fracture nite
conductivities 1
C
). . . . . . . . . . . . . . . . . . . . . . . . . . . 262
7.7 The matching of observed and calculated pressure-dierences ver-
sus time (for variable-rate IBCs and changes in the fracture half-
length, 1
)
, from 50 ft and 25 ft, and assuming a maintained frac-
ture conductivity, 1
C
, of 20 mDft). . . . . . . . . . . . . . . . . . 263
7.8 The matching of observed and calculated pressure-dierences ver-
sus time (for variable-rate IBCs and changes in the fracture half-
length, 1
)
, from 50 ft and 25 ft, and assuming a maintained frac-
ture conductivity, 1
C
, of 20 mDft). The initial pressure, 1
i
, is
reduced from 4400 psi to 3850 psi. . . . . . . . . . . . . . . . . . . 264
7.9 Observed and calculated rates as functions of time. The IBC of
the model are of variable pressure, and the fracture conductivities,
1
C
, change from 70, 40 down to 20 mDft. . . . . . . . . . . . . . . 265
7.10 Matching of well observed cumulative oil data with a model cal-
culated. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
7.11 Observed and model pressure dierences vs. time, in addition to
calculated and observed rates vs. time. The IBC of the model are
of variable rate. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
7.12 The fractured horizontal well, SA-P1 penetrating 14 transversal-
fractures in an oil reservoir (cross section). . . . . . . . . . . . . . 267
7.13 The measured wellbore pressure data, j
&)
(psi), versus time, t (d). 268
LIST OF FIGURES xix
7.14 The measured oil rate,
c
(bbl/d), the equivalent oil rate,
cc
(bbl/d), and the gas rate,
j
(Scf3/d), versus time, t (d), for a
horizontal well SA-P1 with 14 transversal-fractures. . . . . . . . . 268
7.15 The measured oil rate,
c
(bbl/d), the equivalent oil rate,
cc
(bbl/d), and the gas rate,
j
(Scf3/d), versus time, t (d), for a
horizontal well SA-P1 with 14 transversal-fractures on a log-lin
scale. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
7.16 The well and fracture input data. . . . . . . . . . . . . . . . . . . 269
7.17 The reservoir input data. . . . . . . . . . . . . . . . . . . . . . . . 270
7.18 A comparison of the calculated and measured well data (model
data obtained with an IBC of constant pressure). . . . . . . . . . 270
7.19 The well cumulative production, Q (bbl), and the fracture pro-
duction, Q
)vi
(i = 1. ...14) (bbl), versus time (d). The IBC of the
model is of constant pressure. . . . . . . . 271
7.20 The model well rate, (bbl/d), and the fracture rates,
)vi
(i =
1. ...14) (bbl/d), versus time (d). The IBC of the model is of
constant pressure. . . . . . . . . . . . . . . 272
7.21 The pressure dierence, 1
i
1
&)
(psi), versus time (d) for an IBC
of variable rate. (Well production rates for the rst 800 days are
considered as 1 rate-interval). . . . . . . . . . . . . . . . . . . . . 272
7.22 The pressure dierence, 1
i
1
&)
(psi), versus time (d) for an IBC
of variable rate (The well production rates for the rst 800 days
of are devided into 3 rate intervals). . . . . . . . . . . . . . . . . 273
7.23 The step function match obtained with an IBC of constant-rate to
constant-pressure processed in a single run. Both pressures and
rates are matched within the single run. . . . . . . . . . . . . . . 273
7.24 A fractured horizontal well, o 12, penetrating 14 transversal
fractures in an oil reservoir. Cross section view. . . . . . . . . . . 274
7.25 The measured wellbore pressure, j
&)
(psi), versus time, t (d). . . 275
7.26 The measured oil rate,
c
(bbl/d), the equivalent oil rate,
cc
(bbl/d), and the gas rate,
j
(Scf3/d), versus time, t (d), for a
horizontal well SA-P2 with 14 transversal fractures. . . . . . . . . 275
7.27 The measured oil rate,
c
(bbl/d), the equivalent oil rate,
cc
(bbl/d), and the gas rate,
j
(Scf3/d), versus time, t (d), for a
horizontal well SA-P2.with 14 transversal fractures on a log-lin
scale. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
7.28 The well and fracture input data. . . . . . . . . . . . . . . . . . . 276
7.29 The reservoir input data. . . . . . . . . . . . . . . . . . . . . . . 277
7.30 A comparison of the calculated and measured well data (model
data obtained with an IBC of constant pressure). . . . . . . . . . 278
xx LIST OF FIGURES
7.31 The measured versus calculated data for the cumulative rate, Q
(bbl), versus time (d). The calculated data are dened with the
half-length, 1
)
, (of 34 and 50 ft) and the fracture penetration
height, /
)
, (of 40 and 50 ft). . . . . . . . . . . . . . . . . . . . . 278
7.32 The measured versus calculated data for the cumulative rate, Q
(bbl) versus time (d). The calculated data are dened with the
half-length, 1
)
(of 20, 34 and 50 ft) and the fracture perforation,/
)
(of 40 and 50 ft). . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
7.33 The well cumulative production, Q (bbl), and the individual frac-
ture cumulative production, Q,:i (i = 1. .... 14), versus time, t
(d). (Each fracture half-length 1
)
=34 ft and the fracture partial
penetration height, /
)
=40 ft). . . . . . . . . . . . . . . . . . . . . 280
7.34 The well rate production, q (bbl/d), and the individual fracture
rate production, q
)vi
(i = 1. .... 14), versus time, t(d) (Each frac-
ture half-length 1
)
= 34 ft, and the fracture partial penetration
height /
)
= 40 ft). . . . . . . . . . . . . . . . . . . . . . . . . . . 280
7.35 Fracture rate for individual fractures (1, 6, 7, and 14) for varying
fracture half-lengths, 1
)
(34, 50 ft) and partial penetration heights
(40, 50 ft). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
7.36 Valhall eld with several multi-fractured horizontal wells [After
Norris et al. (2001)]. . . . . . . . . . . . . . . . . . . . . . . . . . 282
7.37 Rate and PI data versus measured values of the cumulative well
production. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
7.38 The wellbore pressure and GOR data versus time. . . . . . . . . 284
7.39 Productivity Index versus Time (IBC = Constant Rate). [Model
and Well Data: PI - MATCH]. . . . . . . . . . . . . . . . . . . . . 286
7.40 Model and well data PI - match (for an IBC of costant rate - the
fracture permeability and fracture width are constant). . . . . . . 286
7.41 Rate versus time (for an IBC of variable pressure). . . . . . . . . 287
7.42 Calculated wellbore pressure matches model observed data for
variable rate IBCs. . . . . . . . . . . . . . . . . . . . . . . . . . . 287
7.43 The step-function procedure calculates dimensionless pressure for
the IBCof constant-rate and rates for the IBCof constant-pressure.
Within the same run the IBCs are changing from constant-rate to
constant-pressure. . . . . . . . . . . . . . . . . . . . . . . . . . . 288
7.44 The match of the models (SLAB & BOX) with the well rates. . . 288
7.45 The water injection rate and the cumulative injection rate versus
time for a horizontal well with 16 transversal fractures.Well: SA-
WI1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
7.46 The wellbore pressure versus time for the SA-WI1 well. . . . . . . 290
7.47 The well and fracture input data. . . . . . . . . . . . . . . . . . . 291
7.48 The reservoir input data. . . . . . . . . . . . . . . . . . . . . . . 291
LIST OF FIGURES xxi
7.49 The model water injection rate,
i
(bbl/d), and the water injection
cumulative production, Q (bbl), versus time, t (d), for the SA-WI1
water injection well. . . . . . . . . . . . . . . . . . . . . . . . . . . 292
7.50 The fracture water injection rate, q (bbl/d), for a horizontal well
with 16 fractures, and the individual fracture injection rates, q
)vi
(i = 1. .... 16). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
7.51 The cumulative fracture water injection, Q (bbl), for a horizontal
well with 16 fractures and the individual fracture water injection,
Q
)vi
(i = 1. .... 16). . . . . . . . . . . . . . . . . . . . . . . . . . . 293
7.52 The productivity index, PI (bbl/d psi), versus tme, t (d), for the
water injection horizontal well penetrating 16 fractures. . . . . . . 293
7.53 A horizontal well with 14 transversal-fractures producing from a
synthetic gas reservoir. The cumulative oil production is con-
verted into its cumulative gas equaivalent. The IBCs are either
constant or of variable pesudo-pressures. . . . . . . . . . . . . . . 295
7.54 The dimensionless pressure, j
1
, versus the dimensionless time, t,
for an inverse decline exponernt : = 1 and a dimensionless radius
:
1
= 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
7.55 The dimensionless pressure, j
1
, versus dimensionless time, t, for
an inverse decline exponernt : = 1 and a dimensionless radius, :
1
= 10. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
7.56 The dimensionless pressure, j
1
, versus the dimensionless time,
t, for an inverse decline exponernt : = 1 and a dimensionless
radius, :
1
= 20. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
7.57 The dimensionless pressure, j
1
, versus dimensionless time, t, for
an inverse decline exponernt : = 1 and dimensionless radius, :
1
=
50. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
7.58 The dimensionless pressure, j
1
, versus dimensionless time, t. for
an inverse decline exponernt : = 1 and dimensionless radius, :
1
=
100). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
7.59 The dimensionless pressure, j
1
, versus dimensionless time, t, for
an inverse decline exponernt : = 1 and dimensionless radius, :
1
=
200. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
7.60 The dimensionless pressure,j
1
, versus dimensionless time, t, for
an inverse decline exponernt : = 1 and a dimensionless radius,
:
1
= 500. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
7.61 The dimensionless pressure, j
1
, versus the dimensionless time,
t, for an inverse decline exponent, : = 10 and a dimensionless
radius, :
1
= 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
7.62 The dimensionless pressure, j
1
, versus the dimensionless time, t,
for an inverse decline exponent : = 10 and a dimensionless radius
:
1
= 10. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
xxii LIST OF FIGURES
7.63 The dimensionless pressure, j
1
, versus the dimensionless time, t,
for an inverse decline exponent : = 10 and a dimensionless radius
:
1
= 20. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
7.64 The dimensionless pressure, j
1
, versus the dimensionless time, t,
for an inverse decline exponent : = 10 and a dimensionless radius
:
1
= 50. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
7.65 The dimensionless pressure, p
1
,versus the dimensionless time, t,
for an inverse decline exponent : = 10 and a dimensionless radius
:
1
= 100. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
7.66 The dimensionless pressure, j
1
. versus the dimensionless time, t,
for an inverse decline exponent : = 10 and a dimensionless radius
:
1
= 200. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
7.67 The dimensionless pressure, j
1
, versus the dimensionless time, t,
for an inverse decline exponent : = 10 and a dimensionless radius
:
1
= 500. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
7.68 The dimensionless pressure, j
1
, versus the dimensionless time,
t, for an inverse decline exponent : = 100 and a dimensionless
radius :
1
= 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
7.69 The dimensionless pressure, j
1
, versus dimensionless time, t, for
an inverse decline exponent : = 100 and a dimensionless radius
:
1
= 10. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
7.70 The dimensionless pressure, p
1
. versus the dimensionless time,
t, for an inverse decline exponent : = 100 and a dimensionless
radius :
1
= 20. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
7.71 The dimensionless pressure, j
1
. versus the dimensionless time,
t, for an inverse decline exponent : = 100 and a dimensionless
radius :
1
= 50. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
7.72 The dimensionless pressure, p
1
, versus the dimensionless time,
t, for an inverse decline exponent : = 100 and a dimensionless
radius :
1
= 100. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
7.73 The dimensionless pressure, j
1
, versus the dimensionless time,
t, for an inverse decline exponernt : = 100 and a dimensionless
radius :
1
= 200. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
7.74 The dimensionless pressure, j
1
, versus the dimensionless time,
t, for an inverse decline exponent : = 100 and a dimensionless
radius r
1
=500. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
7.75 The dimensionless pressure, 1
1
, versus the dimensionless time, t
for an inverse decline exponent : = 1000 and a dimensionless
radius r
1
=10. The singularity case. . . . . . . . . . . . . . . . . . 310
Acknowledgements
I would like to take this opportunity to express my sincere gratitude to my
advisor Professor Jon Steinar Gudmundsson for his encouragement, guidance,
and patience throughout this work.
I owe a special thanks to Dr. Gotskalk Halvorsen for his mathematical sup-
port and enthusiasm in designing the model and for all our useful discussions.
Thanks to Jan Sagen for his contribution in programming the model options.
I was privileged to work and study at NTNU, the Department of Petroleum
Engineering and Applied Geophysics, where I got the opportunity to learn fun-
damental and advanced issues of various disciplines in petroleum engineering,
thus making years spent there enjoyable and unforgettable.
I also extend my sincere thank to Professor Jon Kleppe and to other profes-
sors for the provided support throughout my studies. Appreciation is extended
to Mrs. Marit Valle Raaness for all administrative assistance continuous encour-
agements.
Financial aid during the course of study at NTNU was provided by the
Phillips Petroleum Company Norway, and is gratefully acknowledged. I would
also like to thank to IFE (Institute for Energy Technology at Kjeller) for sup-
porting the development of the model and enabling me to nalise the thesis. In
particular, I would like to thank Arne Westeng and Jan Egil Arneberg in Bay-
erngas Norge AS, for the provided support. I wish to express my gratitude to
all individuals who aided in the completion of the thesis.
xxiii
xxiv 0. Acknowledgements
Preface
The work presented in this thesis has developed from the Dr.ing. study period
at NTNU (NTH) between 1991-1994 at the Institute for Petroleum Engineering
and Applied Geophysics where I was given the task to study the decline curve
analysis in order to derive how the drive mechanism is related to the decline
exponent, b. Without monitoring the full production history data, but merely
investigating the connection of the curvature of decline curves, dened by the
decline exponent, b, it was dicult to establish a relation between the decline ex-
ponent, b, and the drive mechanism. The study ended with 3 reports, all related
to decline curve analysis. Due to rate time relations being empirical and derived
for a vertical well operating under conditions of constant pressure during the well
depletion time, it was extremely challenging to derive a single relation. This was
due to the fact that there were numerous possible solutions (as a result of the
problem being inverse). Until now, decline curve analysis has been considered
as a convenient empirical procedure for analysing well performance. However,
only limited signicance has been put in relation to the values of exponent b.
Fetkovich (1980) related the empirical solutions of Arps (1945) to single-phase
ow solutions, thereby providing the theoretical framework for Arps solutions.
Fetkovich also related exponent b to the exponent of the deliverability curve, n.
The drive mechanism and its relation to the rate decline have yet to be the-
oretically determined. The work of providing explanations and unique solutions
has been challenging, particularly for a well operating in the North Sea. Such
wells generally operate under conditions of restricted well pressure, changing
from transient to depletion mode. This is caused by the production plateau, and
the pipeline transportation constraints. In addition, the well is mostly deviated
or even horizontal with fractures.
My interest in well-rate decline continued while working at IFE Kjeller, par-
ticularly when Wiggo Holm from Phillips Petroleum Company, Norway, asked
if it was possible to provide a well response for a fractured-horizontal well pen-
etrating up to 50 fractures. The theoretical model approach for coupling a frac-
tured well to a reservoir was presented under a poster session of the conference
Mathematical Modelling of Flow through Porous Media which was held in St-
Etienne, France in (1995). The tentative model was presented to Martin Rylance
from BP, who decided to nance the project. The model development was fur-
xxv
xxvi 0. Acknowledgements
ther nanced by industry (BP, CONOCO, and PHILLIPS), NFR, and IFE, and
ended in 2001. The nal model was successfully presented at the Seminar with
workshop organised by BP-AMOCO, Norway, in 2001. Obtained project results
were summarised in IFE internal publications, two published SPE papers (2000
and 2001) and two posters (presented at the Schlumberger GeoQest FORUM
in London, England, in 2001, and at the International Petroleum Engineering
Conference in Zadar, Croatia, in 2001).
The fractured-horizontal well data helped me to evaluate and validate the
overall model design and its solutions. Model features were further evaluated
with fractured-horizontal well data from North Africa provided by ENI (2007).
North Sea eld test data were obtained from Vallhal, Ekosk and Syd-Arne elds
(last obtained in 2008).
I conducted the unconventional stimulation study and investigated the rate
decline (while working at Schlumberger). The physics of such decline is unknown,
and was not considered by Fetkovich (1980). Unconventional stimulations cause
an increased production rate and there is no physical explanation why a stimu-
lated well increases the oil production of nearby wells. The overall eld project
was proposed by Schlumberger and ENI, with the task to physically explain
the unconventional drive mechanism. The project idea was presented in Kuala
Lumpur, Malaysia, in 2007, and further inspired my interest in developing the
model for examining the nature of pressure in the vicinity of a producing well
with an Arps rate decline. By solving a diusion equation with specied bound-
ary conditions, it was possible to obtain analytical solutions presenting values of
pressure within the drainage area of a producing vertical well.
The present thesis comprises analyses of transient and depletion well rate
decline. I have carried out their design and contributed in their development
with implementations of the fractured-horizontal well transient-rate-pressure so-
lutions. I also created the model for a vertical well with a variable rate decline
of Arps type. The following statements summarise the novelty and contribution
in the thesis:
1. The design of a fractured horizontal well model, involving the integra-
tion of numerical and program routines into a screening tool for quick transient
test analyses of a fractured horizontal well. The study on transient decline pro-
vides semi-analytical derived solutions related to the fractured horizontal well
production as:
a. Screening options that include the wellbore condition or inner boundary
conditions of constant and variable rate or pressure, with the specic constant-
rate-to-constant-pressure feature.
b. Late time approximations including equivalent wellbore radii, and equiv-
alent fracture half-lengths as measures of the eciency of a horizontal well with
fracture production.
c. Individual fracture production quantities with productivity indices.
0. Acknowledgements xxvii
d. A validation of mathematical and software model features, with fractured
well case studies from the North Sea.
e. The proposition of novel solutions for more sophisticated reservoirs with
extended heterogeneity options, new fracture features (e.g., fracture skin) and
well features (e.g., well skin) to be adapted in the future. Since the existing
model is robust and stable, it can be extended to the multilateral options.
2. The investigation of the nature of a vertical well with a rate decline
production of Arps type. The creation of a physical model and the provision
of analytical pressure responses to a variable rate wellbore condition of Arps
type. In order to solve diusion equations with variable rate wellbore conditions
(approximating Arps rate decline for large times), the approach introduces a no-
ow specied outer moving boundary. The inclusion of the velocity of the no-ow
moving boundary proportional to the square root of time renders it possible to
analytically derive wellbore pressure responses. Further modelling includes:
a. That of pressure responses for declining rates, each dened with the se-
lected decline exponent, b, in turn represented by the drive mechanism following
the concept introduced by Fetkovich (1980).
b. The contribution of the overall work to a transient and depletion rate
decline that is relevant to wells producing oil and gas.
c. A study on the depletion decline; investigating the nature of pressure
responses for variable-rate wellbore conditions of Arps decline. These pressure
responses are solutions to the diusion equation with inner boundary conditions
of variable-rate (i.e., a known decline exponent, b) and no-ow specied outer
boundary conditions (moving outward from a vertical well axis).
d. Solutions that can be extended to a horizontal well, and further calibrated
with measured pressure and rate data from a wellbore.
I have contributed to the mathematical, numerical, and programming work,
and a particular contribution consists in the novel late-time approximations and
step-function features. I worked together with Gotskalk Halvorsen regarding the
mathematical and numerical modelling of the designed model features. Also,
work was equally shared with Jan Sagen, Frederik Martin and Aage Stangeland
concerning the programming of the designed model options.
xxviii 0. Acknowledgements
Chapter 1
INTRODUCTION
1.1 Background
Well testing and rate testing have been the subjects of frequent in depth studies
over the last few decades (from the 1950s). Unlike reservoir simulation that
deals with multidimensional and multiphase uid ow in porous media with so-
lutions obtained by nite dierence methods, both well testing and rate testing
transient responses are mostly solutions based on diusion equation. These so-
lutions are predominantly semi-analytical thus yielding results quickly, which is
why single well studies are helpful in dening basic reservoir parameters. Due to
the methodology being thoroughly tested, well responses are able signicantly
to improve reservoir descriptions. Well testing or pressure-time analysis has
been used less frequently in the North Sea elds during the last decades due to
increased operational costs during testing (as a consequence of the augmented
daily drilling-rig costs). Pressure testing and rate testing analyses are based
on an identical modelling theory and respective solutions. Pressure testing re-
quires experimental data (various test data), further data modelling and the
interpretation of the measured data. Rate testing data, on the other hand, are
continuously monitored and further modelled and tested.
The method for analysing "rate response with time" while keeping the well
in production has improved continuously since the 1980s and made a huge leap
forward during the last twenty years. Although rate-time analysis predates the
pressure-time analysis of well testing the break-through came rst in the 1970s
when Fetkovich combined empirical and analytical equations and as a result
provided type curves to be matched with real observation of well-production
data. This historical milestone introduced type curves and interpretation tech-
niques thus questioning the physical understanding of the inow performance of a
well. A drive mechanism causing the rate curvature was introduced by Fetkovich
(1980) who also made the rst attempt to dene the initial rate decline,
i
, the
decline exponent, /, and the initial decline, 1
i
, by physical means.
1
2 1. INTRODUCTION
Well testing without closing of a well is now possible due to new measurement
techniques and novel modelling solutions. Made possible by new near wellbore
measuring instrumentation developed during the last decade, current research in
well testing is based on modelling the de-convolving variable pressure with time
to a rate condition.
In parallel, rate-time analysis continues to be developed throght the combina-
tion of new modelling techniques and new type curves, in addition to improving
the interpretation of rate-time responses. Anderson et al. (2006) dened pressure
transient data as "high-frequency high resolution data" and rate testing data as
"low-frequency low-resolution data". Compared to well-testing data analysis,
rate-testing analysis is characterised by poor quality data with reduced quan-
tity. Methodology and data interpretation of rate transient data depends on the
frequency and accuracy of the recorded information. Also, pressure transient
data are acquired as part of a controlled experiment, performed as a specic
event (pressure build-up, BU or pressure draw-down, DD). The production data
represent long term monitoring data, usually followed by considerable variance
occurring during its acquisition. Poston and Poe (2008) have provided recent
advances in decline-curve analysis that have aided in improving well production
performance analysis in addition explaining production-forecasting techniques
currently in use in the industry.
1.2 Scope of the Work
Decline curves and particularly decline curvature dening a decline exponent, /,
have been discussed in literature; by Fetkovich (1973) and Raghavan (1993). The
question is how to relate drive mechanism (generally multiphase ow) analytical
solutions to the empirical models dened by Arps (1945) and later improved
by Fetkovich (1973). One possible approach involves combining multiphase ow
solutions with the empirical model solutions as done and discussed by Raghavan
(1993). He evaluated the solution of a diusion equation solved for multiphase
ow and postulated conditions for which decline curvature empirical values could
be matched. This Thesis describes an attempt of modelling the empirically
derived depletion rate-time decline stems. A vertical well was produced from a
circular reservoir with various decline stems. Rate-time stem curvature is created
by the drive mechanism and the reservoir heterogeneity. Developed diusion
models use variable wellbore rates and no-owouter boundaries moving outwards
from a wellbore axis as inner and outer boundary conditions. The speed of an
outwards no-owmoving boundary is predened. The model solutions are unique
in the sense that, for a predened rate and other model input parameters, it is
possible to model the system pressure responses at various points. The present
Thesis provides a novel pressure solution to the empirical decline curvatures.
Moreover it relates rate-time stems to the pressure responses within various
1. INTRODUCTION 3
points in the circular system.
The aim of this work was to develop a novel approach in creating model
transient rate-time responses of a well with fractures to be matched to real
observed data, and thus characterising well performance and improving well
intervention in time. An analytical model was developed in order to provide a
rapid assessment of the productivity of various fracture congurations along a
horizontal well. New modelling techniques for a well with fractures were applied
and model solutions were veried in several case studies. New eective values
were tabulated and listed for a well with fracture models. This work has been
concentrated on a fractured horizontal well, its modelling and interpretation.
Within the same tool, various IBC of pressure and rate are combined. The step-
function features enable a change in wellbore conditions from a constant pressure
to a constant rate within the same run. The transient rate decline features are
covered by a model introducing new late time solutions for a well with fractures
coupled to a reservoir.
For selected rate-time stems each dened with the decline exponent, /.by
solving diusion equation it is possible to obtain pressure solution in time in
various points within a drainage area of a vertical well. So, at the same point
within drainage area it is possible to generate various pressure-time proles for
selected Arps stems dened with the decline exponent, b. The wellbore variable-
rate conditions (of Arps type) are combined with an outer-boundary that moves
outwards from a well with a certain speed. As drive mechanism and exponent,
b, known relations are empirically derived this modelling approach provide basis
for analysis of decline exponent, b, analytically. This can be achieved once both
rate and pressure data are continuously monitored. The speed of the no-ow
boundary should be related to decline stems by using this analytical approach
as its basis. This forms the tentative proposal for future studies. The extensive
research on multiphase ow modelling continues as the physical understanding
of the curvature decline dened with the decline exponent, b, is still gaining
attention in petroleum literature.
The main objective of chapters 2 and 3 is to review well rate-time proles
based on transient-analytical and depletion-empirical models. These discussed
models include the selection of the following parameters:
- Model geology (homogeneous or heterogeneous such as varying permeability,
layering, composite, naturally fractured),
- Types of uid in a reservoir (incompressible, slightly-compressible and com-
pressible uids),
- Flow regime within a reservoir (steady state, unsteady state, pseudo-steady-
state),
- Reservoir geometry (radial, linear, spherical and hemispherical),
- Number of owing uids (single or two-phase),
- Various well positioning (vertical or horizontal wells),
4 1. INTRODUCTION
- Well-fracture coupling to a reservoir (vertical fractured well, horizontal frac-
tured well),
- Well operating conditions (constant or variable pressure),
- Special features, e.g., a well with laterals, and
- Multiple well realizations.
This thesis mainly consider solutions of the three fundamental combined
equations, i.e., the Darcy equation, the continuity equation and the equation of
state given as:

n =
/
j
(\1 2). (1.1)
\(j

n ) + +
J (cj)
Jt
= 0 (1.2)
j = j
0
c
[c(jj
0
]
(1.3)
All three equations, solved with initial and boundary conditions describe uid
ow through porous media. The initial condition is most often one of pressure
boundary conditions are given in terms of pressure, 1. and velocity eld term,

n .
Further simplication predominately related to the uid and the rock proper-
ties include: a constant porosity, , a constant compressibility, c, a and constant
permeability, /. that leads to solutions for pressure or rate distribution with time.
For the slightly compressible uid, van Everdinger and Hurst (1949) derived a
diusion equation that is similar to that concerning the conduction of heat ow
as presented by Carslaw and Jaeger (1947). Physical diusion signies that the
rate of pressure change at a given point is a function of the number of parameters
describing porous media and the curvature of the pressure around the selected
point. In a case of linear ow, the rate of pressure change,
01
0t
. is proportional to
the permeability, /. as well as to a local curvature of the pressure prole,
0
2
1
0a
2
.
It is also inversely proportional to the uid viscosity, to the reservoir porosity,
and to the total compressibility.
More complex semi-analytical or approximative solutions of diusion equa-
tion are derived by introducing:
- Anisotropy and permeability, / variations with radius, :.
- Porosity as a function of time,
- Density and viscosity as functions of pressure,
- Gas ow,
- Multiphase ow,
- Inclusion of fractures coupled to a well within a reservoir, and
- Multiple wells.
1. INTRODUCTION 5
This thesis comprises a review of various formulations and presents a number
of analytical and semi-analytical solutions for a heterogeneous reservoir produc-
ing at a constant pressure from a single fully-perforated well.
One should be aware of potential limitations in using the semi-analytical ap-
proach. Particularly when simulating nature and describing a multidimensional
ow with a simplied single phase ow, ideal preconditions for the modelling set-
up should be presumed. As a result the matching procedure, although fast, is
not accurate. Nevertheless these decline analyses, or DCA testing techniques are,
due to their limited time of interpretation, still considered useful methodologies
that are powerful for screening single-well response analyses.
A single well that penetrates a reservoir with dened parameters and that
is lled in with a liquid is considered to be at equilibrium. We would like to
determine the rate response of the well in time by imposing a perturbation, e.g.,
starting the production of uid through a well under specic conditions. The
solution of such a physical problem determining the rate-time well response is
unique and this is mathematically referred to a direct or forward problem.
It is also possible to determine reservoir parameters if one knows rate-time
response to a given well side perturbation. This is perceived as an inverse prob-
lem that is usually not unique. There are several realisations that can provide
identical responses to a given perturbation. Both pressure testing and rate test-
ing disciplines have several common and complementary features that have been
historically considered as presented in the Table (1.3). Here, we refer to two
comprehensive review papers by Gringarten (2006), and Anderson et al. (2006).
1.3 Organisation
Chapter 2 reviews the transient rate decline caused by uid expansion with a
continuously increasing drainage area, it considering a model with a well pro-
ducing under constant BHP. A well positioned in a circular drainage area, for
which there is no-ow at the drainage boundaries, is the basic model for gener-
ating transient rate-time proles. Various models provide the latest rate testing
"state of the art technology" review of theory with selected solutions, all given
in transient mode.
Chapter 3 provides a review of the depletion decline. This decline begins after
the drainage radius reaches the outer boundaries that dene the drainage area.
We introduce decline curve analyses of the production data from a depletion
period only. An extensive type curve summary comprises the latest theoretical
solutions to ow equations, all based on Fetkovichs (1980) type curve approach.
We mainly consider a special case of the transient solution i.e., the depletion
solution. This chapter also provides a "state of the art technology" review of
production data analysis. It gives a comprehensive review of reservoir mod-
elling tools that are helpful in diagnosing a reservoir model and characterising a
6 1. INTRODUCTION
reservoir.
The rst two chapters consider a number of analytical and semi-analytical
answers to the forward and inverse solutions of a homogeneous or heterogeneous
reservoir producing at constant pressure from a single well. Both areally and
radially heterogeneous reservoirs have been taken into account. For each con-
sidered case, the adequate system of units is stated. The available literature on
rate testing is far too extensive to be summarised within the scope of these chap-
ters. We thus refer only to selected available publications and present modelling
methodologies of rate testing analyses for certain complex well geometries.
Chapter 4 focuses on new solutions for a well with fracture full-time responses
and also present a late-time approximation of equivalent wellbore parameters
(radius and fracture half-length). It summarises multi-fractured horizontal well
model features (both transient and depletion rate time solutions). As the model
is designed for wellbore inner-boundary-conditions of both, pressure and rate,
it can be used in order to handle rate-to-pressure changes with regards to well-
bore conditions. This unique feature is presented through step modelling. All
solutions are solved in dimensionless form and converted into a rate-pressure-
time for the predened units. Individual fracture rate and cumulative rate are
novel model features. These fracture features are available in well-producing and
well-injection mode.
Chapter 5 comprises transient pressure-time solutions of a vertical well po-
sitioned in an innite reservoir. New model features involve the moving no-ow
outer boundary conditions with a selected speed. As the boundary moves out-
wards, the drainage volume changes, giving rice to a transient model. At the
same time the rate decline is of Arps type. The inner-boundary conditions of
the model are of variable-rate, and with such a wellbore condition it is possible
to generate pressure proles over time within a selected spatial distance from
a wellbore axis. This model provides pressure proles for the wellbore variable
rate conditions of Arps type. The model drainage volume changes with the
outward-moving boundary, and thus the model investigates the transient ow
behaviour for both oil and gas.
Chapter 6 describes the verication of modelling features and the generation
of various rate and pressure versus time curves (in a predened unit system).
Most model features presented graphically are evaluated with an input from a
case study. A basic model with the no-ow moving boundary generates pressure-
time proles and provides type curves related to the inner boundary condition
of constant rate. The newly developed transient model is valid for the wellbore
variable rate conditions. Even in a transient model, a rate-time steam decline
is presumed to occur at the wellbore. The purpose is to derive the pressure
prole that matches the decline stems. With our use of no-ow and moving
outer-boundaries, we further contribute the Raghavan (1993) observation to the
inclusion of transient data while matching the decline in depletion. Several model
1. INTRODUCTION 7
Table 1-1: Pressure testing and rate testing comparisons of events, interpretation
methods and historical emphasis (Following the 2006 review by Gringarten and
Anderson et al.]
8 1. INTRODUCTION
solutions are provided in a dimensionless forms of pressure versus time. Once
pseudo-steady-state time is reached it would be possible to derive solutions to
the case when a no-ow boundary moves inwards. The transient-decline ow
analysis should also include the pressure normalization procedure and should
relate the speed of a moving boundary to the physics, and potentially to a drive
mechanism that causes the curvature of the decline stems.
Chapter 7 extends rate-time solutions for a horizontal well with transversal
and longitudinal fractures. New model solutions cover both the transient and
depletion rate-time solutions. At intermediate time, the model also includes
the pressure-time solutions. The new step function feature combines wellbore
conditions of contant-pressure to those of constant-rate, thus allowing a quick
and easy screening and matching of well data. Late time approximations are
recent solutions related to the eective wellbore parameters (radius and fracture
half-length). Several case studies have provided details to overall fractured-
horizontal well model features. Investigations in an oil reservoir present the
screening potential of the available tool features. As the model is dened for
oil and water ow, it can also be used for water injection studies. Since it is
monophase it only provides individual fracture rates, and such individual water
injection quantities should be further considered by reservoir simulation. The
model is furthermore dened as a basic gas screening tool under the assump-
tion that the pseudo-pressure, :(j), is used instead of the reservoir pressure, 1.
Overall, common features of the model is in creating individual fracture rates
and cumulative rates thus leading to a well with fracture productivity indices.
These fracture injection-production features are complementary to the commer-
cial numerical simulation. Further in Chapter 7, model solutions of a vertical
well with a no-ow moving boundary are represented by the dimensionless pres-
sure in surroundings of a wellbore that changes with time. Since the pressure is
calculated by the rate decline of an exponent b almost equal zero, it should be
veried with the measured pressure data at the site. The dimensionless pressure
versus the dimensionless time is derived with no-ow moving boundary solutions
in Appendix A.
Chapter 2
TRANSIENT RATE DECLINE
REVIEW
This chapter review the transient rate-time performance of a well positioned in
an oil and gas reservoir. Rate-time analysis or rate-testing investigates reservoir
responses measured as a rates in the producing well. During transient rate-time,
a pressure wave created by the well has not reached the boundary of the reser-
voir. A produced rate is a response to a specic pressure history at a wellbore.
Generally it is possible to use rate-time transient analysis for the purpose of de-
scribing reservoirs. We here distinguish rate-time analysis from its pressure-time
counter part. Pressure transient tests are early reservoir responses to wellbore
conditions representing a constant ow rate. Usually the constant ow rate is
easier to control in a short transient test interval of a few hours or days. Pressure
transient tests (as drawdown, DD; buildup, BU; and interference test, IT) diag-
nose near-wellbore conditions such as the reservoir conductivity dened by the
permeability-thickness product, //, the wellbore storage and skin, :. There is no
fundamental dierence between pressure-transient and rate-transient analyses.
They describe the same process and are governed by identical reservoir charac-
teristics. In both analyses, as discussed by Horne (1995), pressure transmission
is an inherently diusive process and is largely governed by average conditions
rather than by local heterogeneities. The diusion equation solutions can be
interpreted to estimate bulk reservoir properties due to them being insensitive
to most local scale heterogeneities.
In general the equation of a well performance relates well rates and pressures
to the properties of reservoir formation and within a uid. Such well performance
relations are solutions to the diusion equation for selected initial and boundary
conditions. This review comprises a number of selected published solutions for
inner boundary conditions at a well with either constant or varying pressure.
The outer boundary condition, denoted innite acting, denes unsteady-state
ow within the reservoir. In such unsteady-state ow, uid ow is due entirely
to rock or uid expansion. We refer to the type of reservoir heterogeneity that
9
10 2. TRANSIENT RATE DECLINE REVIEW
is penetrated by various well types.
Several methods can be used for obtaining the solutions to the diusion equa-
tion and these methods may be grouped into either analytical, semi-analytical
or numerical methods (with nite-dierence grids or exible grids).
The solution to the diusion equation was rst derived for a heat conduction
problem (Kelvin line source solution), and applied by Theis (1935) to ground
water hydrology problems. The fundamental theoretical work of Carslaw and
Jaeger (1947) is used as a basis for engineering studies. A further development
for the petroleum industry followed by Muskat (1937), and van Everdingen and
Hurst (1949).
The following three principals dene the ow equation for unsteady-state
ow: The Law of conservation of mass; Darcys Law, and Equation of state. A
mathematical expression of the Law of conservation of mass is the continuity
equation. For a liquid this continuity equation combined with Darcys Law and
the equation of state derives a radial diusion equation for a uid of constant
compressibility. It applies to ows of oil or water.
In the Darcy equation the permeability, /, relates the driving force, macro-
scopic phase uid velocity, n, to the phase pressure gradient, \j.
The velocity eld as dened by Darcys law is expressed as:

n =
/
j
(\1 2). (2.1)
Liquid ow is described with permeability of medium, /, phase viscosity, j,
phase pressure, 1, pressure gradient, , depth, 2. For the compressible gas ow
and turbulence non-Darcy ow the velocity eld can be modied.
For a small compressibility, c
t
, and small pressure gradients equation is lin-
earised, to the known diusion equation:
1
:
J
J:
(:
Jj
J:
) =
cjc
t
/
Jj
Jt
(2.2)
where, hydraulic diusion, j, is constant and equal to
I
jc
I
. Hydraulic diusion
is a measure of the expendability of the system. The following:assumption are
made: the ow take place along a radial path towards a wellbore; the physical
dimensions of the rock are not time dependent; the porosity and permeability
are constant in space and time; the uid saturations are constants and the uid
ow is single phase; the uid viscosity is constant, and both the compressibility
of the uid and the pressure gradients are small.
It is possible to solve the above equation provided that the following condi-
tions are met:
- the outer boundary conditions, OBCs, are innite, no ow and constant
pressure, and
- the inner boundary conditions, IBCs are of constant or variable rate, ,
and constant or variable pressure, j
&)
. The boundary conditions are usually
2. TRANSIENT RATE DECLINE REVIEW 11
Table 2-1: Pressure gradients and dimensionless pressure functions for radial
reservoir ow at the well - After Valko and Economides (1995)
Radial ow
Liquid ow
Pressure gradient, j
Gas ow Pressure
gradient, :(j)
Dimensionless
pressure, j
1
Transient
(OBC - Innite acting)
j
i
j
&)
:(j
i
) :(j
&)
) j
1
=
1
2
Ei(
1
4t
T
)
Semilogarithmic approximation
at t
1
100 where t
1
=
I
jc
I
v
2
u
t
Steady state
(OBC-Constant pressure)
j
c
j
&)
:(j
c
) :(j
&)
) j
1
= ln
v
c
v
u
Pseudosteady state
(OBC-No ow)
j
i
j
&)
:(j
i
) :(j
&)
) j
1
=
ln 0.472v
c
v
u
expressed in terms of a Darcy velocity eld,

n , or pressure, j. The pressure
solution obtained at IBCs of a constant rate, , can be converted to a rate,
, value according to by Valko and Economides (1995). The rates, , for the
constant rate solutions are:
=
2://j
jj
1
(2.3)
In order to reduce the number of unknowns and obtain solutions that are in-
dependent of any unit system the dimensionless pressure, j
1
and dimensionless
rate,
1
, are introduced. Earlougher (1977) demonstrated that the dimenionless
pressure, j
1
, for the constant rate production is almost equal to the
1
q
T
for a
well producing at constant pressure. Thus the transient rate, , obtained at
IBCs with constant pressure, j
&)
. IBC has the following form:
=
2://(j
i
j
&)
)
j
1
q
T
(2.4)
The ow pressure gradients and dimensionless pressure functions for a radial
reservoir according to Economides and Ehlig-Economides (1994) and Valko and
Economides (1995) are presented in Table 2-1.
Table 2-1 can also be used for the compressible gas. Instead of j, Dake
(1978) and Economides et al. (1994) suggested an expression with :(j). The
innite acting reservoir conditions are met under conditions with a signicant
pressure drop at any outer boundary. This may be dened with a small dimen-
sionless time at the outer boundary:
t
1c
=
/
cjc:
2
c
t _ 0.1 (2.5)
A gas compressible ow to be reviewed later in a the chapter is based on a real
12 2. TRANSIENT RATE DECLINE REVIEW
gas pseudopressure, m(p) provided by Al-Hussainy and Ramey (1966) in the
following form:
:(j) =
j
_
j
0
2j
j2
dj (2.6)
This yields to a diusion equation
1
r
0
0r
(r
0:(j)
0r
) =
cjc
t
/
0:(j)
0t
The present chapter reviews selected solutions dened with constant or vari-
able IBCs and innite acting OBCs for various fractured wells positioned within
an oil and gas reservoir. Dake formulated a general solution of the radial diusion
equation describing transient ow in a reservoir as:
1
:
J
J:
(:
J,
J:
) =
cjc
t
/
J,
Jt
where parameter , is dened for: undersaturated oil as , = j; real gas as
, = :(j); gas-oil (two-phase) , = :(j)
0
, where :(j)
0
=
1
_
1
0
I
ro
(S
0
)
j
0
1
0
dj is pseudo-
pressure as stated by Raghavan (1993).
Ikoku (1984) published general solutions by solving the dimensionless form
of the diusion equation for the inner boundary of constant rate.
1
:
1
J
J:
1
(:
1
J,
1
J:
1
) =
cjc
t
/
J,
1
Jt
The general solution for the constant rate inner boundary conditions is
_
c

_
,(j) = ,
1
(t
1
) + o
For undersaturated oil, real gas and two-phase ow, the denition of the above
solution terms includes:
o
q
being the function of
II
j
; ,(j) being the pressure
dierence; and o being the skin. For real gas, skin o is the function of o =
o(o. 1) where 1 stands for a non-Darcy ow.
2.1 Oil Flow
To perform a conventional well test analysis on a well, one common procedure is
to ow the well at a constant rate for several days before carrying out the test.
This procedure is not always eective, and often the delay can be avoided by
performing transient rate tests instead. The most important test is the analysis
2. TRANSIENT RATE DECLINE REVIEW 13
of the rate response to a step change in the producing pressure. This test allows
a type-curve analysis of the transient rate response without the complication of
wellbore storage eects.
The ow of a single oil phase through such porous media is generally accepted
to be described by a linear diusion equation. This is strictly valid only for
slightly compressible uids, such as undersaturated oil or water for which the
properties are little aected by changes in pressure. This transient rate analysis
review is focused on innite and circular reservoirs with concentric wells, i.e.,
wells that are arbitrarily located in regularly or irregularly shaped reservoirs.
2.1.1 Vertical Well
Most reservoirs can be produced by the release of pressure and the consequent
expansion of underground uid. During part of the production history of a
reservoir, uid compressibility can be considered as small and constant. Solu-
tion assumptions include a constant owing pressure at the wellbore, which fully
penetrates the reservoir. The reservoir contains a slightly compressible uid of
single phase and constant viscosity, and the uid ow is horizontal in a homoge-
neous and isotropic porous medium of uniform thickness with constant perme-
ability and porosity. Single-phase liquid solutions based on these assumptions
are widely used in hydrology and petroleum engineering.
Homogeneous Reservoir
When considering homogeneous porous media with a constant porosity, the per-
meability is also isotropic provided that /

= /
I
. No saturation gradients occurs
within such a system. For an oil ow, the oil saturation is equal wherever it is
constant i.e., o
c
= (1 o
i&
). The pressure is considered to be above the bubble
point of the uid. The eect of gravity is ignored, leading to uid properties be-
ing uniform over the constant thickness of a non-dipping formation. The inow
to a well is horizontal-radial since a well is perforated over the entire reservoir
thickness. The pressure in time and space in the porous media for a single phase
uid is described with the general diusion equation as
1
:
J
J:
_
/
j
j:
Jj
J:
_
= jcc
t
Jj
Jt
(2.7)
This diusion equation is non-linear because of the rock and uid properties
(c
t
. c. /. j. j) being pressure dependent. The above equation is linearised by
assuming
0j
0t
as being small to a most known form of
1
:
J
J:
_
:
Jj
J:
_
=
1
j
Jj
Jt
(2.8)
14 2. TRANSIENT RATE DECLINE REVIEW
Here, the hydraulic diusion, j =
I
jc
I
, for small changes in pressure is assumed to
be constant. By imposing small changes in pressure, or with a pressure increase,
c
t
decreases and the parameters /, c, and j increase. For oil ow, the total
compressibility (uid and pore volume), c
t
, is assumed to be c
t
= c
0
o
0
+c
&
o
&
+c

.
The above linearisation is however valid only for
c
t
j 1 (2.9)
as shown by Dranchuk and Quon (1967). The limitation of the linearisation
approach presented above is that, close to a wellbore, we may expect errors
in the local estimation of pressure, j(:. t), while the second derivative of the
pressure on radius,
_
0j
0v
_
2
, is neglected.
The linearised diusion equation can be solved analytically. The above equa-
tion is similar to the thermal diusion equation
1
:
J
J:
_
:
J1
J:
_
=
1
1
J1
Jt
(2.10)
where the thermal diusion, 1(:
2
,:), corresponds to its hydraulic counter part,
j, and the temperature, 1(1), to the pressure, j. Solutions of the thermal
diusion equation for various initial and boundary conditions were published
by Carslaw and Jaeger (1947) and can be applied in rate-testing analysis as an
advantage of similarity of the two diusion equations.
The transient rate decline can be employed as a tool for identifying the char-
acteristics and predicting the behaviour of reservoir systems. The reservoir per-
meability, porosity, and wellbore skin factor can be determined by matching
type-curves. Conditions under which a constant-pressure ow is maintained at
a well include production into a constant-pressure separator or pipeline, or open
ow to the atmosphere. The dimensionless form of the diusion equation of a
well producing at the constant pressure IBCs, or Dirichlet conditions and innite
OBCs is:
1
:
1
J
J:
1
_
:
1
Jj
1
J:
1
_
=
Jj
1
Jt
1
(2.11)
Here the dimensionless radius, :
1
=
v
v
u
, the dimensionless time, t
1
=
j
v
2
u
t, and
the dimensionless pressure, j
1
(:
1
. t
1
) =
2II
qj
(j
i
j
(v,t)
). For an ideal well we
dene the innite acting or transient rate decline with the initial and boundary
conditions. The initial condition are: j = j
i
at t = 0 for all :. The outer
boundary conditions are innite acting or: j = j
i
for all t, at : = . The
inner boundary conditions include: a constant rate where is constant for all
t and a constant pressure where j is constant for all t. For a constant rate,
=
2IIv
u
j
_
0j
0v
_
v+
. By assuming that :
&
is negligibly small, |i:
v!0
(:
0j
0v
) =
qI
2II
=
constant.
2. TRANSIENT RATE DECLINE REVIEW 15
The pressure-time solution of the radial diusion equation is the line source
solution for the constant rate production at a wellbore. These solutions have been
developed by well testing. Moreover, the rate-time solution based on constant
pressure inner boundary conditions has been the subject of rate-decline analysis.
Analytical Methods (Integral Transform Methods) The problem of con-
stant pressure production from a well located at the centre of a homogeneous
and isotropic cylindrical reservoir was rst studied by Moore et al. (1933) and by
Hurst (1934). The results for an innite, unbounded reservoir were presented in
graphical form in terms of a dimensionless ow rate decline with dimensionless
time. Results were put forward for innite slightly compressible, single phase
radial systems. Furthermore, Jaeger and Clarke (1942) studied the heat con-
duction problem. An integral solution for the temperature drop during a ow
period in an innite cylinder was derived and numerical values of the integral
were tabulated. van Everdingen and Hurst (1949) presented a series of solutions
for the ow rate decline with time. Their work included the study of an innite
reservoir. Jacob and Lohman (1952) described the dimensionless ow rate be-
haviour for a well producing at constant pressure. Both the dimensionless ow
rate and the dimensionless time applied only to the innite reservoir system.
The relation for dimensionless ow rate versus the dimensionless time was given
as:

1
=
2
ln t
1
+ 0.80907
(2.12)
with the
1
and the t
1
dened as:

1
(:
1
. t
1
) =
1j
2://(j
i
j
&)
)
(2.13)
t
1
=
j
:
2
&
t (2.14)
Ferris et. al. (1962) tabulated rate-time values for the innite reservoir case.
Furthermore, Earlougher (1977) provided constant pressure equations to obtain
the permeability and skin factor from rate-time production data.
Semi Analytical (Inverse "Laplace" Transform) Techniques The use of
the integral "Laplace" transform technique to obtain a solution for the pressure
behaviour was attempted by Clegg (1967). In order to obtain the "Laplace"
inverse function, Clegg (1967) employed an approximation dened by Schapery
(1962). Ehlig-Economides (1979) and Ehlig-Economides and Ramey (1981) pre-
sented the model set-up according to the given assumptions for a saturated liquid
owing as a single phase, in an isothermal reservoir of the constant uid viscos-
ity, j, and for a constant and small total compressibility of the uid and porous
16 2. TRANSIENT RATE DECLINE REVIEW
medium, c
I
. In the radial geometry, the radial ow is described with negligible
gravity eects according to the diusion equation. The ow through a porous
medium exists from nite wellbore radius, :
&
, to the innite or nite external
reservoir radius, r
c
. Here, the porous medium is homogeneous and isotropic
with a constant diusion (permeability, /, porosity, c, and thickness, /). Thus,
the idealised ow through porous media can be described by the fundamental
partial dierential equation, also known as the diusion equation. The partial
dierential equation is reduced to the ordinary dierential equation with the
spacial variable, :
1
, :
1
=
:
:
&
and the "Laplace" variable, :, as an unknown. It
is then solved analytically for the "Laplace" transform of the pressure j
1
by the
"Laplace" integral transformation.
A common method for solving the radial-diusion dimensionless Equation
(2.11) under dened inner and outer boundary conditions is to use the "Laplace"
transformation. The advantages of this method consist in transforming partial
dierential equations into ordinary dierential equations that can be solved an-
alytically for the "Laplace" variable, :, and the space variable, :
1
, as described
by van Everdingen and Hurst (1949). The "Laplace" transformation is dened
by:
1(:) =
1
_
0
c
ct
T
1(t
1
)dt
1
(2.15)
Moreover, the "Laplace" transform applied to the dimensionless partial dieren-
tial Equation (2.11) and can be expressed as:
d
2
j
1
d:
2
1
+
1
:
1
d j
1
d:
1
= : j
1
(2.16)
The "Laplace" space solution for a production from the centre of a circular
reservoir under constant pressure inner boundary conditions and innite outer
boundary conditions is given as:

1
(:) =
1
1
(:)
_
:(1
0
_
: + o
_
:1
1
_
:)
(2.17)
For the cumulative rate, Q
1
we obtain
Q
1
(:) =
1
1
(:)
:
_
:(
_
:(1
0
_
: + o
_
:1
1
_
:))
(2.18)
Since the cumulative rate, Q
1
, is dened as Q
1
=
t
_
0
(t)dt, the dimensionless
ow rate,
1
, in eld units becomes:
2. TRANSIENT RATE DECLINE REVIEW 17

1
=
141.2j1
//(1
i
1
&)
)
(t) (2.19)
and the dimensionless time, t
1
in eld units is:
t
1
=
0.0063//
cjc
t
:
2
&
t (2.20)
These equations are exact integral transform solutions in "Laplace" space as pub-
lished by Ehlig-Economides (1979). The inverse integral Laplace transformation
can only be obtained through the Mellin inversion integral transformation. The
Stehfest (1970), Crump (1976) or other approximate numerical inversion proce-
dure can be applied to the given solutions. van Everdingen and Hurst (1949)
presented the relationship between the Laplace transformed solutions for the
constant pressure and the constant rate in the form of:

1
j
1
=
1
:
2
(2.21)
where
1
. is dened under constant pressure inner boundary conditions and j
1
.
is dened under constant rate conditions. Expression (2.21) reveals that any
solution for j
1
. or constant rate production has an analog solution,
1
.for pro-
duction under constant pressure. Both variables are "Laplace" space variables.
Equation (2.21) can be derived from the principle of superposition. A numerical
inversion technique, consists in inverting "Laplace" space variables to a dimen-
sionless rate,
1
, and dimensionless time, t
1
. now in the space-time domain.
Numerical Methods (Finite Dierence) Juan-Camus (1977) presented an
alternative numerical method for obtaining solutions for a constant pressure ow.
He derived the constant pressure solutions from the constant rate solutions by
way of superposition. The vertical direction of the reservoir was also considered
when solving the diusion equation. The derivation did not require the integral
transformation of "Laplace".
Uraiet (1979) and Uraiet and Raghavan (1980) presented a study of the
transient rate behaviour through numerical methods. They considered the classic
problem of the ow of a slightly compressible uid in a cylindrical, homogeneous,
isotropic reservoir of constant thickness. The well was located at the centre
of the cylinder, and uid was produced at a constant pressure. Initially, the
pressure was uniform throughout the reservoir. The skin region in this model
was assumed to be an annular region that was concentric with the wellbore
displaying a permeability dierent from the formation permeability. Wellbore
storage eects were not considered since the well owed at a constant pressure.
The authors determined that the semi-log analysis method for constant rate
testing could be applied to the transient rate if the reciprocal rate was used.
18 2. TRANSIENT RATE DECLINE REVIEW
Kleppe and Cekirge (1980) presented approximate solutions to the diusion
equation for an innitive radial reservoir for the constant well pressure under
inner boundary conditions. They used a numerical simulation of well tests to
verify the obtained expressions and to present examples of their applications.
Raghavan (1993) observed that the late time solution for the nite wellbore
radius case was identical to the late-time solution for a line source well. This
result justies the use of the line source solution for practical problems.
The "Laplace" time domain expansions published by van Everdingen and
Hurst (1949) applied to homogeneous reservoir only. Technique comprises inte-
gral computation along contour with a search for poles of a function over the
complex plane.
Heterogeneous Reservoir
Specic types of heterogeneous reservoir systems have received much attention
in the oil and gas industry in recent years. Several analytical and numerical
models exist in order to consider the permeability, /, changes with distance,
:, naturally fractured, layered and composite reservoir systems. Heterogeneous
reservoirs have been well documented in research papers. In reality, reservoir
systems usually combine all the eects of these types of heterogeneities.
In a radially heterogeneous reservoir, the ow is supposed to be purely radial
with a permeability as a function of only the radial distance from the well. The
porosity is here constant. The diusion equation is solved for systems where
the permeability varies continuously from one well to another. Loucks (1961)
presented transient pressure solutions for a case with a radial heterogeneous
reservoir. The reservoir permeability varies with the power of the radial distance
from the wellbore, i.e., as /(:) = /(:
a
), for values of : = 0
1
3
1. c:d 2.
Oliver (1990) presented an approximative solution to the problem of a well
producing at inner boundary conditions of constant rate from an areally hetero-
geneous reservoir. Here, the permeability is considered as an arbitrary function
of position, i.e., / = / (:. ). By using perturbation theory and the "Laplace"
transform he derived approximate solutions for the transient wellbore pressure.
A two-dimensional permeability distribution, / = /(:. ), is identical to the solu-
tions for an equivalent radial permeability, / = /(:). The permeability /(:)were
taken as the harmonic average of / = /(:. ) over 2. This was valid for only
small variation of permeability in direction of . A study by Feitosa (1993)
considered a well producing a heterogeneous reservoir characterised by a contin-
uously variable radial permeability. It was also possible to consider the eect of
porosity variations in an areally heterogeneous reservoir.
Areally and Radially Heterogeneous Reservoir A reservoir with proper-
ties, as permeability, porosity and thickness being arbitrary function of position
2. TRANSIENT RATE DECLINE REVIEW 19
dened by radius, :, and azimuth, . is called the areally heterogenous reservoir.
The radially heterogeneous reservoir refers to a reservoir in which one or more
of the basic parameters (permeability, porosity, and thickness) varies only with
the radial distance, :. in the (:. ) coordinate system.
The Oliver (1990) solution is based on a perturbation theory and the "Laplace"
transforms. By introducing a small variation in permeability about a mean
value the dimensionless permeability, /
1
. as a function of a small number and
,(:
1
. ) that is on order of 1 becomes
/
1
(:
1
. ) =
1
1 ,(:
1
. )
(2.22)
A perturbation series approach combined with "Laplace" transforms leads to the
approximate pressure solution in "Laplace" space
j
1
(:
1
= 1. . :) = j
10
(:
1
= 1. :) + j
11
(:
1
= 1. . :) (2.23)
Here, j
10
refers to a constant-permeability solution given as
j
10
(:
1
= 1. :) =
1
0
(
_
:)
:
_
:1
1
(
_
:)
(2.24)
and j
11
refers to the rst order perturbation of constant permeability with a
time-correspondent "Laplace" variable, :
j
11
(:
1
= 1. . :) =
1
:1
2
1
(
_
:)
1
_
1

_
_
_
1
2
1
(
_
:)
_
_
1
2:
+
_

_
1
1
/
1
(:. )
_
d
_
_
_
_
_
.d
(2.25)
These equations were converted to dimensionless rate-dimensionless time solu-
tions through

1
=
1
:
2
[j
10
(:
1
= 1. :) + j
11
(:
1
= 1. . :)]
(2.26)
Both dimensionless-pressure and dimensionless-rate solutions can be numerically
inverted by Stehfest to obtain the real time solutions of pressure j(:
1
= 1. t)
= j
10
+ cj
11
and accordingly of rate (:
1
= 1. t). Oliver (1990) derived a
late time approximation to the dimensionless pressure solutions by letting the
"Laplace" variable, :, approaching zero:
j
10
(:
1
= 1. :) =
1
0
(
_
:)
:
(2.27)
j
11
(:
1
= 1. . :) =
1
_
1

_
_
_
1
2
1
(
_
:)
_
_
1
2:
+
_

_
1
1
/
1
(:. )
_
d
_
_
_
_
_
d (2.28)
20 2. TRANSIENT RATE DECLINE REVIEW
An analytical inversion of the "Laplace" space solutions comprises a part con-
sisting of a homogeneous solution (or constant-permeability solution)
j
10
(:
1
= 1. t
1
) =
1
2
Ei
_

1
4t
1
_
for t
1
> 25, and when using a logarithmic approximation and Eulers constant
= 0.57721566
j
10
(:
1
= 1. t
1
) =
1
2
ln
_
4t
1
c

_
The rst order perturbation solution to the costant-permeability solution is
j
11
(:
1
= 1. . t
1
) =
1
_
1
_
_
_
G(. t
1
)
_
_
1
2:
+
_

_
1
1
/
1
(. )
_
d
_
_
_
_
_
d
where G is the kernel function in terms of Whittakers function.
Feitosa (1993) developed a numerical solution to the above stated areally
heterogeneous reservoir with the nite-dierence approximations. For an arbi-
trary heterogeneous reservoir it is possible to determine the pressure response in
a reservoir with known radial permeability distribution, / = /(:). Feitosa (1993)
calculated the pressure response with time in an areally homogeneous reservoir
producing a slightly compressible uid from a single well. The physical model
that was considered comprised: an inner boundary condition of constant rate,
an outer boundary condition of closed upper and lower boundaries with a later-
ally innite reservoir, a constant porosity, thickness and rock compressibility, a
single phase uid with constant viscosity and compressibility; negligible gravity
and capillary eects; fully penetrating well, rock and uid properties indepen-
dent of pressure, a uniform initial pressure throughout the reservoir; as well as
negligible wellbore storage and skin eects.
In recent research Berard (2007) considered the use of a generalised Weber
expansion in the radius domain in well-testing analysis. A generalized Weber
function is also used in the rate-time analysis in Appendix A.
Layered Reservoirs
Reservoir rocks are usually not uniform, whether in horizontal or vertical direc-
tions. The reservoir is thus not homogeneous or isotropic. Heterogeneities can
exist in rock and uid properties from deposition, folding and faulting, post-
depositional changes in reservoir lithology and changes in uid-type properties.
The small scale heterogeneities may result in carbonate reservoir rocks matrix
and fractures, vugs or solution cavities. On the large scale, heterogeneities result
in physical barriers. Layering is the most common form of heterogeneity.
2. TRANSIENT RATE DECLINE REVIEW 21
Layered reservoirs can be divided into reservoirs with crossow, commin-
gled reservoirs and composite reservoirs. Layered reservoirs with crossow are
hydrodynamically communicating at the contact planes, whereas for commin-
gled systems, or layered reservoirs without crossow, the layers communicate
through the wellbore. The ow between adjacent and connected layers resulting
from capillary, gravitational, or viscous forces, has been studied for many years.
A composite reservoir is made up of two or more regions, and each region has
its own rock and uid properties. A composite system can either be created
articially or be naturally occurring.
Natural formations are vertically heterogeneous because of stratication and
the various depositions. The vertical sequence of deposits alternates between lay-
ers with good and poor permeability. The diusivities, dened by the physical
properties of stratied deposits, dier from layer to layer, and the layers re-
spond to production at various rates. This causes dierential pressure depletion
between the layers, thereby altering the radial ow pattern of a well.
Pressure transient analysis of layered systems has been described by numer-
ous papers. However, only few studies have addressed rate transient analysis.
Most of the reviews discuss testing methods, well testing, and rate testing in-
terpretation techniques. A comparison of the layered model solution methods
and a precise study of the available numerical techniques applied in the in-
verse "Laplace" transform solution procedure has been treated. Although, the
"Laplace" transform has been more widely used in pressure and rate transient
modelling, other integral transforms can be very useful in the solution of the
boundary value problems of layered systems as presented by Cvetkovic (1992).
Layered reservoir models are developed assuming that the reservoir pressure
can be directly measured. The pressure recorder is located in the wellbore, and
the wellbore constitutes a link between the reservoir and the recorders. The
recorded pressures are representative of the pressure in the reservoir and can be
aected by the number of wellbore-related phenomena. These wellbore phenom-
ena must be recognised in order for the diagnosis of the reservoir characteristic
to be achieved. The wellbore dynamics, such as the eects of temperature on
a wellbore uid, gas-oil solution-liberation or retrograde condensation, liquid
inux-e-ux, phase redistribution, wellbore and near-wellbore clean-up, dier-
ences between drawdown and build-up, recorder and others, not precisely de-
ned, eects all transient analysis.
Ehlig-Economides and Joseph (1987) reviewed various models of layered
reservoirs, of which the layered reservoir model treats the heterogeneous reser-
voir system as consisting of separate homogeneous strata. These layers may
communicate in the reservoir through cross-ow. Sabet (1991) presented dier-
ent physical and mathematical models according to the various solution methods,
number of layers, interlayer ow, well specication, inner and outer boundary
conditions. The report published by Cvetkovic (1992) is an attempt to incorpo-
22 2. TRANSIENT RATE DECLINE REVIEW
rate model understanding and numerical model analysis of the published layered
reservoir solution. Ehlig-Economides (1993) presented an investigation on model
diagnosis for layered reservoirs. The question is when interpretation models for
layered reservoirs can be used eectively, and how to identify a model for each
reservoir layer.
Tariq and Ramey (1978) introduced the solutions of a bounded (circular)
multilayer reservoir system producing at constant rate in order for skin eects
to be included in each layer. The total wellbore storage model was also included
into solutions. The dimensionless "Laplace" space solutions were transformed
to the real domain with the numerical inversion of Stehfest (1970). Spath et
al. (1994) proposed a stable and robust algorithm to compute pressure and
rate responses from a well producing the commingled reservoir. This approach
determined the well responses for constant or variable-rate productions. The
following dimensionless pressure for a commingled reservoir in the "Laplace"
domain is based on the constant-rate dimensionless pressure solution for each
layer:
j
&1

(:) =
1
a

)=1
I)I)
II
1
j
uT

(c)
Further, Spath et al. (1990) formulated the rate of each layer:

c1)
(:) =
/,/,
//
j
&c1
(:)
:j
&1

(:)
Blasingame et al. (1991) computed the eect of the wellbore storage and
the wellbore phase redistribution. The obtained solutions were accurate when
compared to the results obtained from the numerical inversion of the "Laplace"
domain solutions.
Lolon et al., (2008) formulated the new solution for the multilayer reservoir
and developed approximate semi-analytical solutions for the wellbore pressure
and fractional owrate responses for commingled layered reservoirs for which a
crossow was permitted in the wellbore and not in the reservoir. He referred
to the study of Lefkovits et al., (1961) concerning the pressure behaviour for
layered reservoir systems assumed to be homogeneous, isotropic and saturated
with a uid of small and constant compressibility. The reservoir comprised :
layers and each layer was dened by: permeability, /, thickness, /, porosity, c,
viscosity , j, uid compressibility, c
t
, wellbore and outer radii, :
&
and :
c
, and
skin, :. The obtained solutions provided accurate approximations only for late
times.
Lolon et al., (2008) validated the developed explicit and approximate time
solutions in the real domain for the modelling of the performance of a multilayer
2. TRANSIENT RATE DECLINE REVIEW 23
reservoir system. This was done after considering certain approximate behav-
iours of the individual layer solutions for the purpose of creating algebraically
convenient results in the "Laplace" domain for the total system behaviour. The
chosen basic functions were able to yield forms that were inverted analytically to
the real domain. The process provided solutions for multilayer reservoir system.
Solutions for a single well in a multilayer reservoir system can be used to model
well test or production data.
Crossow Models A crossow reservoir is considered to consist of continu-
ous layers of permeabilities /
i
(for i = 1. 2. ...). For the 2 layers system, when
permeability /
1
is greater than permeability /
2
. the pressure gradient in the rst
upper layer is of substantial size. The vertical pressure gradient is created in
the lower layer of permeabilities /
2
according to the pressure drop in the layer of
permeability /
1
. The direction of ow is determined by the diusivity contrast
and is upwardly oriented towards the high permeability layer.
Russell and Prats (1962) reviewed the practical features of the interlayer
crossow systems. They concluded that a well producing from a layered reservoir
with crossow behaves similarly to one in a homogeneous, single-layer reservoir
with the same pore volume, and a ow capacity equal to the total ow capacity
of the stratied system.
The eect of contrasting permeabilities for two perforated layers with the
same storage and thickness was considered by Prijambodo at el. (1985). On
the basis of several numerical simulation runs, it was noticed that whenever the
vertical-to-horizontal permeability ratio of the rst layer was greater than or
equal to 0.01 and the permeability ratio between two layers was bigger than
or equal to 5, the interlayer crossow was a signicant factor in establishing the
drawdown behaviour. In the case of the lager contrasting permeabilities between
two layers, the crossow equalises the pressure so rapidly that the well behaves
as if it were producing from a single layer.
These results are similar to those derived by Javandel and Witherspoon
(1969) on the basis of a nite element method. Moreover, Javandel and Wither-
spoon (1969) concluded that the permeability contrast had a considerable eect
on the transitional pressure behaviour. The smaller the contrast in permeability,
the more rapid is the convergence to a single-layer behaviour. In all cases, at
large values of time, the direction of ow becomes almost radial regardless of the
permeability contrast.
Commingled Reservoirs If the layers are separated by a completely imper-
meable boundary, interlayer crossows will be absent within the reservoir. Cross-
ow can take place only at the wellbore itself and only if the well is completed
in more than one layer. The pressure pattern for this case diers signicantly
from that for interlayer crossow.
24 2. TRANSIENT RATE DECLINE REVIEW
During production, crossow can occur at the wellbore provided that the
layers have an unequal initial pressure. Whenever the owing pressure is higher
than that of the layer with the lower pressure, only a portion of the total ow
from the layer with the higher pressure is produced at the surface. The remainder
of the production enters into the lower pressure layer through crossow at the
wellbore. When the owing well pressure is lower than that of the lower pressure
layer, both layers contribute to the surface production.
The solution of the fully penetrated well in a circular, bounded and commin-
gled reservoir with homogeneous and isotropic layers lled with uid of small
and constant compressibility and constant viscosity was given by Letkovits et al.
(1961). Under their assumption, if the reservoir is initially at a uniform pressure
j
i
at all times t
i
; and the production rate q, measured at initial reservoir condi-
tions, is held constant, the pressure must satisfy at any point the known partial
dierential equation for ow of a uid. The solution to this equation for the
proper boundary conditions is obtained with the "Laplace" transform. Letkovits
et al. (1961) presented rigorous equations describing the pressure behaviour at
a constant terminal rate well producing from a bounded, noncommunicating,
layered system with contrasting properties. Letkovits et al. (1961) also provided
the report describing the pressure behaviour of a well producing at a constant
terminal rate from an innitely large, multi-layer non communicating reservoir.
Composite Reservoir Often, the region surrounding the wellbore is either
more or less permeable than the reservoir because of the various drilling and com-
pletion practices. The drilling-uid invasion reduces the permeability whereas
the operation of fracturing or acidising increases it. Composite reservoir mod-
els consider reservoir systems made up of two concentric zones of varying rock
and uid properties separated by a discontinuity. Examples include reservoirs
damaged by drilling or completion uid invasion, acid-stimulated wells. Such
composite systems were studied in connection with heat ow by Jaeger (1941).
He presented a solution for temperature distribution in a radial system with an
innitely large outer radius. Penner and Sherman (1947) studied similar heat
ow problems. Closman and Ratli (1967) presented a solution for a well pro-
ducing at a constant pressure from a closed, radial composite reservoir. Turki
(1986), and Olarewaju and Lee (1987) presented solutions in "Laplace" space
for a well producing at a constant pressure from a radial, innite composite
reservoir. Turki et al. (1989) studied a constant-pressure well in the centre of a
two-composite reservoir with wellbore skin. For innite composite reservoirs, the
eect of the mobility ratio, storativity ratio, wellbore skin and the discontinuity
distance on rate decline are described.
The two-regions assumptions include: a single phase ow with only one uid,
a formation that is horizontal and of uniform thickness, a sandphase wellbore
pressure that is maintained constant, as well as a constant distance to the radial
2. TRANSIENT RATE DECLINE REVIEW 25
discontinuity. Turki et al. (1989) described each region with the following partial
dierential diusion equation
J
2
j
11
J:
2
1
+
1
:
1
Jj
11
J:
1
=
Jj
11
J:
1
for 1 < :
1
< 1
1
(2.29)
J
2
j
12
J:
2
1
+
1
:
1
Jj
12
J:
1
=
Jj
12
J:
1
for 1
1
< :
1
< (2.30)
With initial conditions and boundary conditions dened for each composite re-
gion, and after applying the "Laplace" transformation to the equations above,
Turki and al.(1989) published the "Laplace" space well rates
11
and
12
(where
j is the region diusion constant):

T1
= :
1
_
:
_
C
11
1
1
(:
1
_
:) C
12
1
1
(:
1
_
:)

. ,o:1 _ :
1
_ 1
1
(2.31)

T2
= :
1
_
j:
`
[C
21
1
1
(:
1
_
j:) C
22
1
1
(:
1
_
j:)] . ,o:1
1
_ :
1
< (2.32)
The well rate,
1
, in "Laplace" space is

1
=
_
:
_
C
11
1
1
(
_
:) C
12
11(
_
:)

(2.33)
and the cumulative well production in "Laplace" space, Q
1
is
Q
1
=
1
_
:
_
C
11
1
1
(
_
:) C
12
1
1
(
_
:)

(2.34)
To solve the transient rate decline in real space of composite reservoir inverse
transform Stehfest algorithm can be applied. Constants C
11
, C
12
, C
21
and C
22
for
the OBC innite active are provided by Turki et al. (1989). Furthermore, Issaka
and Ambastha (1998) investigated innite reservoir responses from a composite
reservoir with regards to a two-region mobility
(/,j)
1
(/,j)
2
and storativity
(c
1
)
1
(c
2
)
2
.
Naturally Fractured Reservoir
Laminar Flow The naturally fractured reservoir models, also referred to as
dual-porosity reservoir models, consider a heterogeneous system made up of two
distinct porous media. The primary porosity medium contains a majority of
uid stored in the reservoir but possessing a low conductivity. The secondary
porosity medium acts as the conductive medium but has a low storativity. The
study of transient rate decline in a double-porosity system was started by Mavor
and Cinco-Ley (1979) and continued by Da Prat et al. (1981), Raghavan and
26 2. TRANSIENT RATE DECLINE REVIEW
Ohaeri (1981), and Ozkan et al. (1987). Sageev et al. (1985) published a type
curve method for analysing rate-time data in a double-porosity system with a
well skin. Moreover, Grasman and Grader (1990) have presented an analytical
method to determine double-porosity reservoir properties.
The model of Warren and Root (1963) was extended by Mavor and Cinco-Ley
(1979). Da Prat (1981) formulated the statement of the problem for naturally
fractured reservoirs with the following partial dierential equation
J
2
1
)1
J:
2
1
+
1
:
T
J1
)1
J:
1
= (1 .)
Jj
&1
Jt
1
+ .
J1
)1
J:
1
(2.35)
(1 .)
Jj
n1
Jt
1
= `(1
)1
1
n1
) (2.36)
where parameters . and ` are associated with reservoir and uid properties.
The storage of secondary porosity to a total storage (both matrix and fracture),
. is dened as . =
(\
c
)
]
(\
c
)
]
+(\
c
)
r
. An interporosity ow controls `, and the
interporosity ow shape factor, c, is equal to ` = c
1
r
1
]
:
2
&
. Further, the dimen-
sionless fracture pressure, 1
)1
, and the dimensionless time, t
1
, are expressed as
1
)1
=
1
]
I(1
.
1)
141.2qj1
and t
1
=
2.637a10
7
1t
[
(C)
]
+(C)
r
]
jv
2
u
, respectively.
Appropriate initial and boundary conditions include: 1
)1
(:
1
. 0) = 0 and
IBCs of a constant producing pressure 1
)1
o(
01
]T
0t
T
) = 1, where o is the
skin factor. The innite OBC gives |i:
v
T
!1
1
)1
(:
1
. t
1
) = 0. The dimensionless
ow rate,
1
, into the wellbore is
1
(t
1
) =
_
01
T
0t
T
_
v
T
=1
, and the cumulative
production, Q
1
, related to the ow rate is Q
1
=
t
T
_
0

1
dt
1
.
The "Laplace" transformations convert partial dierential equations into a
system of ordinary dierential equations that can be solved analytically. Solu-
tions in "Laplace" space are functions of the complex variable, :, and the space
variable, :
1
. Mavor and Cinco-Ley (1979) published innite or unbounded reser-
voir transient rate solutions with approximations for early and late time. The
dimensionless ow rate in "Laplace" space given by Da Prat (1981) is:

1
(:) =
_
:,(:)1
1
_
_
:,(:)
_
:
_
1
c
_
_
:,(:)
_
+ :
_
:,(:)
1
_
_
:,(:)
__ ,(:) =
.(1 .): + `
(1 .): + `
(2.37)
Moreover, with an Inverse "Laplace" transform it is possible to invert the above
solutions to a real time and space. For early times, the dimensionless ow rate
is
1
=
p

(
t
T
.
)

1
2
and the cumulative dimensionless rate is Q
1
=
2
p

(.t
1
)
1
2
.
2. TRANSIENT RATE DECLINE REVIEW 27
A late time approximation of equation (2.37) becomes equal to a homogeneous
reservoir solution, as the one provided by Jacob and Lohman (1952). Da Prat
et al. (1980) derived the late time approximation for a dimensionless rate
1
=
2
ln t
T
+0.80907
.
Non-Laminar Flow Rodriguez-Roman and Camacho-Velazquez (2002) pre-
sented analytical expressions for non-Darcy liquid ow in dual porosity systems.
The dierential equation describing ow of a matrix-fracture systems is given as:
1
:
1
J
J:
1
_
_
_
_
:
1
2
1 +
_
1 + 4,
1
_
0j
]T
0v
T
_
_
_
_
_
= (1 .)
Jj
n1
Jt
1
+ .
Jj
)1
Jt
1
(2.38)
(1 .)
Jj
n1
Jt
1
= `(j
)1
j
n1
) (2.39)
where, ., `, ,
1
, :
1
, j
)1
, t
1
are dimensionless variables. In other words, . =
(\
c
)
]
(\
c
)
]
+(\
c
)
r
, ` = c
1
r
1
]
:
2
&
, ,
1
=
Ijq
2v
V
Ij
, 1
)1
=
1
.
1
1
.
1
u]
, :
1
=
v
v
u
, with a unit
conversion constant
|
, and the dimensionless time t
1
=
|
I
jc
I
v
2
u
t . With the
Forchheimer equation, it is possible to include non-Darcy ow:
Jj
)
J:
=
j
/
i
)
+ ,ji
2
)
(2.40)
there, the coecient of inertial ow resistance, ,, is equal to , =
48511

5.5
]
I
0.5
]
. The
approximative transient rate solution for early times thus becomes:

1
=
_
.
:t
1

,j/
2
)
j
_
(j
i
j
&
)
:
&
j
_
.
:
_
:t
1
_
2

t
1
(2.41)
valid for a range of

t
1
, 2 _
1
_

t
T
_ 0.5. It was dicult to assign a value to

t
1
when comparing this analytical solutions to simulation results. The lam-
inar ow solution was obtained when neglecting non-Darcy ow eects. The
approximative transient rate solution for late times is equal to:

1
=
_
_
2
1
_
0
_
4
c
2

t
1
(r + 1)
dr
_
_
a
_
_
1
4
ojI
2
]
j
_
(j
i
j
&)
)
:
&
j
_
1
_
0
_
4
c
2
t

a
r
(r + 1)
dr
_
_
(2.42)
28 2. TRANSIENT RATE DECLINE REVIEW
Combined Reservoir Olarewaju and Lee (1991) presented a model that pre-
dicts production performance from a naturally fractured reservoir with a ra-
dial discontinuity around the wellbore. This type of reservoir (composite dual-
porosity reservoir) conguration exists when a well is damaged, acidised or
gravel-packed. Zhang et al. (1993) presented solution techniques for rate decline
behaviours in a complex system comprising a well that is arbitrarily located in
a regularly or an irregularly shaped reservoir or in a composite reservoir. For
the composite model, the analytical "Laplace" solution is based upon placing
a constant pressure well with an arbitrary location in a two-composite radially
concentric domain. The model can then provide the characteristic responses
of such composite systems by varying the properties and the geometries of the
domains.
2.1.2 Horizontal Well
Horizontal wells have received considerable attention in the literature. The hor-
izontal well technology enables the exploitation of numerous reserves that may
not have been economically viable with conventional drilling methods. Naturally
fractured reservoirs, discontinuous reservoirs, water and gas conning reservoirs,
tight reservoirs, heavy oil reservoirs and EOR applications are major applica-
tions for horizontal wells. Theses wells have the potential of producing at higher
production rates than their vertical counterparts but also have other advantages
such as enhancing reservoir management and accessing reserves that cannot be
exploited by other means. Advantages of horizontal wells consists is in an in-
creased well productivity, enhanced reservoir management, and the access to
incremental reserves. However, the parameters aecting horizontal well perfor-
mance involve a higher level of uncertainty as compared to vertical wells.
Nevertheless, horizontal wells have a potential advantage over vertical or
deviated wells based on the following main reasons: an increased exposure to
the reservoir giving higher productivities (PIs); the ability to connect laterally
discontinuous features; e.g. fractures, fault blocks; and the ability to change the
geometry of drainage, e.g., reduce drawdown in oil rims. A horizontal well has
a higher productivity in laterally extensive reservoirs, as the productivity index
is a function of the length of the reservoir drained by a well. The productivity
improvement factor (PIF) compares the productivity of a horizontal well that of
a vertical in a same reservoir. It is estimated to
111 =
1
p
I
I
1
p
I
r
=
1
/
_
/

_
I
I
where L is the length of the horizontal well section, h the height of the reser-
voir, /
I
and /

horizontal and vertical permeabilities, respectively. The vertical


2. TRANSIENT RATE DECLINE REVIEW 29
permeability, /

reduces the production of a horizontal well, rendering the pro-


duction rate lower than in a vertical well.
Predicting the performance of a horizontal well for a wide range of reservoir
applications has constituted a continuous research topic as of 1977, when the rst
horizontal well was drilled in the oshore Raspo Mare Field in the Adriatic area.
Special opportunities for a horizontal well technology appeared with the decrease
of drilling costs to about 1.3 times those for vertical wells. Joshi (1987) and Norris
et al. (1991) have reviewed the horizontal well technology. A large amount of
research has been focused on the topics of reservoir engineering. Many analytical
solutions for productivity of transient pressure responses for various boundary
conditions have been derived. Besides boundary conditions, near wellbore eects,
such as formation damage, non-Darcy ow and arbitrary completions have been
studied. Several investigations have predicted inow performance relationships
(IPRs) for horizontal wells. Transient pressure analysis studies further extend
to various boundary conditions and derive pressure and rate responses. Brekke
(1996) studied how wells are aected by geological variations. A long horizontal
well increases the potential both for success and failure. A better understanding
of the total reservoir and wellbore interaction and ow behaviour raises the
potential to success before the well is completed. Moreover, horizontal wells
improve the recovery factor in the oil and gas elds.
Solutions for a horizontal well in a form of a vertical-stripe being a part of
a vertical-fractured well, are presented by Joshi (1991). Vicente et al. (2000)
presented the fully implicit, three-dimensional simulator with local renement
around the wellbore, developed to simultaneously solve reservoir and horizontal
well ow equations, for single-phase liquid as well as gas cases. The model
involves the conservation of mass and Darcys law in the reservoir, in addition
to mass and momentum conservation in the wellbore for isothermal conditions.
The coupling requirements are satised by preserving the continuity of pressure
and mass balance at the sandface. The proposed simulator is tested against and
veried with the results obtained from a commercial "Black Oil" code, available
public domain simulators and semi-analytical models. The model can be used
to simulate the transient pressure and ow rate behaviour of both the reservoir
and the horizontal wellbore. The eects of permeability, formation thickness,
well length, and uid compressibility were also studied.
Cheng (2003) investigated the productivity evaluation and well test analysis
of horizontal wells. The major components of this work consist of a 3D coupled
reservoir /wellbore model , a productivity evaluation, a deconvolution technique,
and a nonlinear regression method improving horizontal well test interpretation
.The 3D-coupled reservoir and wellbore model was developed using the bound-
ary element method for realistic description of the performance behaviour of
horizontal wells. The model is able to exibly handle multiple types of inner
and outer boundary conditions, and can accurately simulate transient tests and
30 2. TRANSIENT RATE DECLINE REVIEW
long-term production of horizontal wells. The work comprises a comprehensive
literature review of the modelling of a horizontal well.
Thompson et al. (1991) used a line source solution, while Besson (1990)
and Economides et al. (1996) employed a point source solution to develop their
semi-analytical models for performance evaluation. Spivey et al. (1992) system-
atically presented an eective method for obtaining the new solution for pressure
transient responses of a horizontal well at an arbitrary azimuth in an anisotropic
reservoir. This method transforms the relevant parameters (permeability, reser-
voir thickness, wellbore length and radius, and vertical position of wellbore) to an
equivalent isotropic system. Ding (1999) used the boundary integral equation
method to obtain a coupled reservoir/wellbore model. Dings model not only
considered wellbore hydraulics, but also considered the wellbore as a cylindrical
surface source instead of a line source.
As mentioned earlier, the semi-analytical models described above are applica-
ble only to reservoirs with the same geometrical shape and boundary conditions
as those of the source functions used to develop the model. In reality, the bound-
ary situation is problem-dependent; however, these models cannot exibly deal
with changing conditions.
A signicant breakthrough in Greens function method involves the use of the
Green function in the free space and boundary element method, BEM in order
to develop the solutions. Based on the BEM, Koh and Tiab (1993) developed
a reservoir model that can prescribe arbitrary boundary shapes and conditions.
They discretised the reservoir boundaries and wellbore surface with triangular
elements. Their solution was developed in the "Laplace" domain and the so-
lution in the timedomain was numerically inverted using Stehfests algorithm.
However, the frictional pressure loss in a horizontal wellbore was not considered
in their model.
2.1.3 Vertical-Fractured Well
Hydraulic fracturing is widely used to increase the productivity of damaged wells
or wells producing from low to moderate permeability formations. A fracture has
a much greater permeability than the formation it penetrates; hence, it inuences
the pressure or rate response of a well. This fracture can be either vertical or
horizontal and the relation between the overburden or vertical stress and the
horizontal stress denes the fracture type. If the overburden stress is larger than
the horizontal stresses, the fracture is vertical whereas for the horizontal stresses
greater than the overburden stress, the fracture is horizontal.
Hydraulically induced fractures are vertical for reservoir depths greater than
1000 m. Bellow such depths, in shallow formations, hydraulic fractures tend
to be horizontal. The hydraulic fracture length and width vary according to
the formation permeability. In moderate or high-permeability formations, the
hydraulic fracture should be short and wide, as opposed to long and narrow in
2. TRANSIENT RATE DECLINE REVIEW 31
low permeability formations.
Thus, much research has been carried out determine the eect of hydraulic
fractures on pressure-rate transient behaviour and well performance. Besides
analytical modelling for a single-phase ow, numerical modelling of ow has
been investigated. Based on fracture ow, analytical models fall under three
categories: nite conductivity, innite conductivity and uniform ux. These
three concepts have been used to describe the uid ow behaviour within the
fracture. In the nite conductivity fracture, pressure drops due to uid ow
within the fracture has been found to be signicant. In innite conductivity
fractures on the other hand pressure drops are negligible and close to zero. The
fracture conductivity is very large in comparison to the formation permeability.
This ideal condition renders the production near fracture tips higher than that at
the well centre. A uniform ux fracture corresponds to a fracture for which the
ux per unit length entering the fracture is constant along the fracture length.
Moreover, the production along the fracture length is constant, causing the well
owing pressure at the fracture centre to be smaller than that at the fracture
tips.
A vertical fracture is generally described as fully penetrating the formation
symmetrically across the well and as having a uniform width, n. The fracture
half-length, 1
)
, is dened as the distance from the axis of the well to the fracture
tip. The fracture conductivity, C
)1
, is the product of dimensionless fracture
conductivity, /
)1
, and the dimensionless fracture width, n
)1
. The dimensionless
terms, /
)1
and n
)1
, are equal to /
)
,/ and n,1
)
, respectively. Here /
)
is the
fracture permeability and / is the formation permeability.
The well productivity is known to increase with hydraulic fracturing and a
consequence of such fracturing is a single crack or a fracture. Fractures are
hydraulically induced and do not resemble the natural fractures known as s-
sures. In general, fractures can be horizontal for depths less than 1000 m, but
are mostly vertical. Rock mechanics describe stress forces and physical fracture
modelling processes that cause fracture propagation. The physics of appearance
of fractures will however not be discussed in the review. Rather, we present
properties related to each fracture to be further considered in fracture ow mod-
elling, e.g., the fracture half-length, 1
)
; the dimensionless radius, :
c1
=
v
c
a
]
,
the fracture height, /
)
, usually assumed to be equal to a formation thickness;
the fracture permeability, /
)
; the fracture width, n
)
, the fracture conductivity,
1
C
= /
)
n
)
, the dimensionless conductivity (conductivity group) 1
C1
=
I
]
I
&
]
a
]
,
the dimensionless radius (fracture group) :
c1
=
v
c
1
]
.
The identication of well-reservoir variables that impact future well perfor-
mances have been carried out both through well testing and rate testing. Con-
ductive fracture properties 1
)
, n
)
and /
)
. are mostly unknown and as well are
fracture geometric features. Since the fracture permeability is much greater than
the formation permeability inuence rate response signicantly. When analysing
32 2. TRANSIENT RATE DECLINE REVIEW
rate tests data from a vertical-fractured well it is possible to consider three rate-
transient models. Each model denes adequate fracture character.
Uniform-ux fractures The ow rate from a formation to a fracture is uni-
form along the entire fracture length. Due to a variable pressure along the frac-
ture, the transient pressure behaviour includes two ow periods, a linear ow,
and an innnite-acting pseudo-radial ow. Acid treatment is the most common
and creates uniform fractures.
Innite-conductivity fractures The owinto the wellbore occurs only through
the fracture. The fracture is highly conductive and considered as innite. There
is no pressure drop from the tip of the fracture to the wellbore and, accord-
ingly no pressure is lost in the fracture. Since the ow in the wellbore occurs
only through the fracture, the transient pressure behaviour includes three ow
periods: a fracture linear ow, a formation linear ow, and an innite-acting
pseudo-radial ow.
Finite-conductivity fractures The ow and pressure is characterised by
measurable pressure drops in the fracture. The transient pressure behaviour
includes four ow periods: linear ow within the fracture, bilinear ow, linear
ow in the formation, and innite-acting pseudo-radial ow. The conventional
hydraulic fracturing with a large quantity of propping agent maintain the fracture
open. The fracture permeability, /
)
, is lower than that of innite-conductivity
fractures.
The literature on vertical fractures goes back to Muskat (1937), and maybe
even further. Prats (1961) presented a solution to a cylindrical, homogeneous,
isotropic reservoir with a vertical well intercepted by a vertical fracture with a
nite fracture conductivity. His work was based on the assumption of an incom-
pressible uid. This study introduced for the rst time the idea of modelling a
fractured well with an unfractured well having a larger equivalent wellbore ra-
dius. The eective wellbore radius has been presented as a function of fracture
length and relative fracture capacity.
Prats et al. (1962) extended Prats work to a compressible uid depleted
through a constant-pressure or constant-rate production. This work assumed an
innite conductivity vertical fracture that fully penetrated the formation in the
vertical direction. It was found here that both the terminal rate and terminal
pressure cases could be modelled by an elliptical reservoir with a larger eective
wellbore radius. Prats et al. (1962) solved the problem of an innite conductivity
fracture producing from a reservoir that had an elliptical outer boundary. The
"Laplace" space solution was expressed as a series of Mathieu functions. These
functions always arise when solving unsteady problems in elliptical geometries.
2. TRANSIENT RATE DECLINE REVIEW 33
Scott (1963) studied the transient behaviour of a single vertical fracture inter-
secting a vertical well by means of a heat ow analogue. The results lie between
the cylindrical well and a line source for a cylindrical well with a wellbore radius
of one forth of the total fracture length, thus conrming previous ndings. Morse
and von Gonten (1972) investigated the behaviour of innite conductivity frac-
tures prior to pseudo steady state. Their work involved the productivity index
ratio between fractured cases as well as unfractured pseudo steady state cases.
Gringarten and Ramey (1973) presented the use of Green and source func-
tions to solve variety of fractured well problems, and their conclusions were
represented in a library of functions that facilitate the use of the technique. The
rst work on pressure transients in fractured wells using the Greens function
technique is that of Gringarten et al. (1974). These authors posed the problem
of an innite conductivity fracture producing at a constant rate as an integral
equation. In this integral equation, the ux distribution was the unknown and
the free space Greens function was the kernel. The solution procedure consisted
in discretising the equation in time and space, thus numerically obtaining the
ux distribution. Such a procedure has been used in numerous elds and has be-
come known as the boundary integral equation method. Gringarten et al. (1974)
presented the ux distribution for various times, but apparently did not use it
to calculate pressures. Instead, they investigated the discretised equations and
demonstrated that a uniform ux fracture evaluated at a certain point along the
fracture (they gave r = 0.732 1
)
) equals innite conductivity fracture whether
at early or late times. Kuchuk et al. (1991) indicated that this pressure point
method will not capture the character of the innite conductivity fracture at
intermediate times.
Cinco-Ley et al. (1978) provided a semi-analytical solution to a homoge-
neous, isotropic, slab reservoir with a vertical well crossed by a vertical, nite
conductivity fracture. The results showed that, for times of interest, the wellbore
pressure solutions can be correlated to a single parameter (i.e., the dimensionless
fracture conductivity). The ux within the fracture was found to depend on the
fracture conductivity. For large fracture conductivities, approximately 67 % of
the total ow originates from the far end of the fracture while for low fracture
conductivities, about 70 % of the total ow comes from the half of the frac-
ture nearest to the wellbore. The pressure distributions within the fracture were
found to depend on the fracture conductivity; i.e., the larger the conductivity,
the smaller are pressure drops. Also for the fracture permeabilities close to the
reservoir permeability, the fracture pressure drop corresponds to that of radial
ow. Furthermore, it was reported that the uniform ux solution behave like
an innite conductivity fracture at early times, then like a variable conductivity
fracture at medium times and nally approached a nite, constant dimension-
less fracture conductivity at late times. Cinco-Ley et al. (1978) used a version
of the boundary integral method to evaluate the pressure response of a nite
34 2. TRANSIENT RATE DECLINE REVIEW
conductivity fracture of rectangular cross-section. The nite conductivity model
accounts for pressure drops along the fracture. The procedure in question em-
ployed the integral of Gringarten et al. (1974) to represent reservoir pressure and
a second integral to account for fracture pressure. Equating these two integrals
at the fracture face resulted in an integral equation that could be discretised and
solved numerically. The use of this technique provides more accurate pressure
results than the nite dierence method used by Agarwal et al. (1979) or the
nite element treatment of Barker and Ramey (1978).
Kucuk and Brigham (1979) used the approach given by Tranter (1951) to ex-
press the solution of an elliptical wellbore producing at a constant rate from an
innite system. They investigated the innite conductivity fracture responses for
the fracture producing at the constant pressure and the constant rate. Agarwal
et al. (1979) addressed the problem of a nite conductivity fractures intersect-
ing a vertical well being produced at constant rate or constant pressure. They
were able to solve the diusion equation numerically. The authors used a two-
dimensional, quarter of a square model. The reservoir simulation model was
rened at the wellbore, fracture tip and parallel to the fracture face.
Cinco-Ley and Samaniego (1981) have reviewed the concepts of bilinear, lin-
ear and pseudo-radial ow nite conductivity, fractured wells. Their work also
addressed the eects of wellbore storage. Dierent type curves are presented
to facilitate the analysis of fractured wells and several scenarios are given for
limited available well test data. The estimation of parameters, the uniqueness
of the solution and the limitations encountered in each of these scenarios were
discussed. This work re-evaluated the eective wellbore radius as a function of
the dimensionless fracture conductivity.
Uniform ux ow model is similar to the innite-conductivity ow model.
The dierence only occurs at the boundary of the fracture. Among the various
models, that of Sheng-Tai and Brockenbrough (1986) provided an approximate
analytical solution for nite-conductivity vertical fractures. The solution for this
model account for the eects of skin, wellbore storage and fracture storage in
early time.
The use of the "Laplace" transformation eliminates the time integral, thereby
leaving only an integral in space. The "Laplace" transform has been used in
vertical fracture problems from 1987; in fact, the rst evaluation of the uniform
ux solution in "Laplace" space was given by Kuchuk (1987).
Papatzacos (1987) presented the solution for reservoir pressure for an innite
conductivity fracture producing at constant rate. He approached the problem
through the integral formulation of Gringarten et al. (1974), and the exact
solution was derived by means of a Mathieu function expansion of the kernel of
the integral. Papatzacos (1987) stated that the dierences between the exact
solution and the uniform ux approximation were as large as four percent.
Cinco-Ley et al. (1987) focused on analyses of wells with low fracture con-
2. TRANSIENT RATE DECLINE REVIEW 35
ductivities (i.e., 1
C1
< 0.1). Their work presented three important concepts: (1)
the equivalent wellbore radius :
&
for 1
C1
< 0.1 is not a function of the fracture
length, (2) given a value of /
)
&
I
, there is no increase in well productivity, and
(3) for a low fracture conductivity, there are only three ow regimes: bilinear,
transition and pseudoradial. The paper presents a type curve and equations for
the estimation of fracture conductivity, the formation ow capacity (//), the
equivalent wellbore radius and the skin factor. Finally, the paper put forward
the idea, that in very low dimensionless fracture conductivities, up to 99 % of
the wellbore ow comes from the 33 % fracture length closest to the wellbore.
The boundary integral method solution of the nite conductivity problem
in "Laplace" space, for the case of double-porosity reservoirs, was obtained by
van Kruijsdijk (1988) and Cinco-Ley and Meng (1988). The most accurate ap-
proximate model to date is that of Wilkinson (1980). This model neglects all of
the ow in the reservoir that is not adjacent to the fracture; the reservoir thus
becomes an innitely long, strip of nite width for which a two-dimensional ow
is allowed but the sides are closed to ow. The solution was presented in terms
of a "Fourier" cosine series. Wilkinson then combined this solution with the
innite conductivity solution to obtain an approximate well solution. This was
found to work well for high conductivity fractures, whereas a correction term
was required for low conductivity fractures.
Riley et al. (1991) investigated the pressure solutions for a nite conductivity
fracture with an elliptical cross-section. The main conclusion of this work was
that the behaviour of an elliptical fracture was essentially the same as that of a
rectangular one.
It is not our intent to give an exhaustive account of the literature, but rather
to highlight the studies that consider fracture ow modelling. Cvetkovic (1992)
reviewed innite conductivity and uniform ux solutions and a review of great
details is given by Villegas (1997).
2.1.4 Horizontal-Fractured Well
Hydraulically fractured horizontal wells represents a proven technology for pro-
ducing oil and gas from tight formations. Thus undeveloped low permeability
reservoirs can be produced. Induced hydraulic fractures reduce well drawdown,
and increase the productivity of horizontal wells by increasing the surface-area
in contact with a formation making fracture as a high conductivity path to a
formation. Hydraulic fractures, depending on in-situ stress orientations, can be
either parallel or perpendicular to the horizontal, longitudinal or transversal well
axis. The question is how to drain a reservoir, what is, the number of hydraulic
fractures and how to design an ecient well spacing.
A literature survey shows numerous analytical solutions for multi-fractured
horizontal well systems considering single-phase ow. Although exact, solutions
obtained with analytical tools are limited and restricted to the scope of assump-
36 2. TRANSIENT RATE DECLINE REVIEW
tions and simplications imposed to describe the system. Few attempts have
been made to employ numerical solutions. The diculties are related to nu-
merical instabilities, especially during the transient period, due to the use of
excessively ne grid blocks when representing the fractures. Several authors
have contributed with solutions of coupling a well with fractures to a reservoir.
For instance Soliman et al. (1990) investigated pressure-transient analysis of a
well with fractures.
In 1991, Mukherjee and Economides presented a parametric comparison be-
tween fractured vertical wells and horizontal fractured and unfractured vertical
wells. This work demonstrated how to calculate a minimum number of transverse
fractures in a horizontal well, and work uses simple relationships to treat trans-
verse fractures as vertical fractures in vertical wells with an additional pseudo-
skin. However, the study does not account for the interference between transverse
fractures and is therefore only valid for very short times. In 1991, Economides
et. al., presented results from a numerical simulation study re-evaluating the
productivity expression for an unfractured horizontal well and extending it to
anisotropic formations. Longitudinally fractured horizontal wells are simulated
showing productivity indices for isotropic formations and for several values of
fracture conductivity. Transverse fractures are simulated and the numerical re-
sults conrm the analytical expression previously presented.
Hareland and Rampersad (1995) have put forward the fractured well perfor-
mance model, being steady state and in which the variations in time are handled
by using the production, e.g., a single time interval (as one day) to re-estimate a
new initial pressure. This was followed by calculating the production of the next
time period, etc. The include it being model limitations a 2-dimensional model,
made quasi-3D (by using a technique originally justied through an appeal to
laboratory experimental results in electrostatics). Interferences between frac-
tures were absent. On the contrary, the given solutions are based on (articial)
prolongations of the fractures into no-ow walls connecting to the outer no-
ow boundary. Only a very specialised geometry (reservoir/fractures/wellbore)
is treated, and moreover, uid ow to the wellbore, both from the formation
and from a fracture is forced to be linear/radial/uniform at xed distances. Its
ecient numerical realisation has been acknowledged through the innovative use
of Gauss-Jacobi quadrature methods.
Hegre and Larsen (1995) have reported on results with a theoretical basis,
mainly presented in other papers. Their investigations concerned pressure tran-
sient analysis of multifractured horizontal wells. The approach is a classical
semianalytic one, where "Laplace" transformed pressure equations are solved
analytically and then numerically inverted to real time by use of the Stehfest
algorithm. Although this was a very solid piece of work, its limitations, through
the simplifying assumptions made in order to obtain a manageable system, need
to be pointed out here. We would specically like to comment on the discreti-
2. TRANSIENT RATE DECLINE REVIEW 37
sation of nite conductivity fractures used, in the manner of solving the inte-
gral equations by the Cinco-Ley and Samaniego (1981) method. Here, no-ow
edges (tips) of fractures were assumed which has become common in this kind
of modelling. Specic assumptions are also made of simplied ow regimes in
the fractures. Flow directly to the wellbore was generally not treated.
Raghavan, et al. (1994) presented a study explicitly treating a simple in-
nite slab reservoir with longitudinal and fully penetrating fractures, for which
also multiple perforations were allowed along the well. With proper alterations,
transverse fractures may be treated as well. All fractures are produced at a
common wellbore pressure. Referring to one of the authors, the solution method
has been found to be too slow for practical purposes. Thus, either a change of
method or a change of model, or both, would seem necessary. The basic method
requiring excessive computer time in this model is used for treating discretised
nite conductivity fractures.
A further contribution is made by Horne and Temeng (1995). In 1996, Valk
and Economides presented the rst semi-analytical solution for longitudinally
fractured wells in isotropic formations. This work shows that the productivity
of a fractured horizontal well can be three to ve-fold that of a fractured vertical
well. An economic analysis reveals that longitudinally fractured horizontal wells
are competitive in isotropic formations. Moreover, the publications by Raghavan
et al. (1997), Cvetkovic et al. (1999), Wan and Aziz (1999), Cvetkovic et al.
(2000, 2001), Al-Kobaisi et al. (2006), and Medeiros et al. (2007) contributes
with new solutions to a coupling of a horizontal well with fractures to a reservoir.
2.2 Gas Flow
2.2.1 Vertical Well
The very rst solutions of a gas diusion equation were derived assuming small
pressure gradients and constant gas properties. The variation of gas and rock
properties with pressure was ignored due to analytical diculties. The gas ow
behaviour is most accurately described using the real gas pseudo-pressure (or
real gas potential), since it takes into account the variation of gas viscosity and
gas deviation factor as a function of the pressure. If one assumes the
j
jZ
product
to be constant, one obtains a solution in terms of pressure. This solution is
appropriate only at higher pressures i.e., over a limited pressure range, of 3000
psi (or 210 bar). At high pressure, gases presents a behaviour that is similar to
that of oil. The pseudo-pressure varies linearly with pressure, and the pressure-
squared solution assumes the j2 product to be constant. It is only appropriate
to use at lower pressures, such low pressure of less than 2000 psi (140 bar).
To overcome the limitations of these solutions, Al-Hussainy et al. (1965)
proposed the use of the real gas pseudo-pressure, a term they dened as:
38 2. TRANSIENT RATE DECLINE REVIEW
Figure 2.1: The domain in which the pseudo-pressure, . varies linearly with j
and j
2
[After Bourdarot (1998)].
(j) = 2
j
_
j
l
j
j(j)2(j)
dj (2.43)
It has been demonstrated that the gas ow behaviour can be most accurately
described using the pseudo-pressure function, (j), which takes into account the
variability of gas viscosity and the gas deviation factor as a function of pressure.
The application of the real gas potential reduced a rigorous partial dieren-
tial equation for the transient ow of real gases to a quasi-linear ow equation
without assumptions of small pressure gradients or a slow variation of the gas
viscosity and gas deviation factor with pressure. Although the use of real gas
potential did not fully linearise the diusion equation, and the fact that the
hydraulic diusion had to be assumed constant in order to arrive at solutions,
the authors justied its use for small ow rates and small production times with
comparisons to numerically simulated data. According to Chien (1993), after
the pressure had declined more than 10 percent from its initial value, however,
the solutions started to deviate signicantly. In order to adjust the solutions
after a reduction of the pressure by more than 10 percent, Kacir (1990) applied
an approach to reset bounding rates and times periodically for changes in the
term of the gas viscosity multiplied by the compressibility term. The results for
calculations of the owing bottomhole pressure were satisfactory for a pressure
decline down to 70 percent of the initial reservoir pressure. Such adjustments or
resets were however not based on any mathematical formulation. Furthermore,
the pressure solution started to fall o at lower pressure values. In 1990, Prats
introduced a new form of the real gas potential to reduce the nonlinear gas ow
equation to a quasi-linear diusion equation. Considering the hydraulic diu-
2. TRANSIENT RATE DECLINE REVIEW 39
sivity, as a function of pressure the author successfully transformed the real gas
diusivity equation into a solvable form. Another formulation of the diusion
equation for general uid ow was proposed in terms of a porosity-density prod-
uct with the intent to account for pressure-dependent uid and rock properties
by Fair (1992).
Due to the highly nonlinear variation of gas density and viscosity with re-
spect to pressure, no analytical solution to the real gas diusion equation has
ever been presented in the literature. Analytical solutions used in gas well test-
ing and pressure analysis are based on idealised assumptions, such as small and
constant gas compressibilities and constant hydraulic diusion. These solutions
though widely used and easily applied, are inaccurate. They are neither ap-
plicable to a broad range of pressure changes nor to dierent ow periods. As
discussed in the literature, by Chien (1993) these solutions start to deviate sig-
nicantly after the pressure has declined more than 10 percent. Moreover, a
variety of approximate analytical solutions are used for various ow periods, due
to a wide range of time and boundary conditions. All of these limitations are
caused by the inability to analytically solve the more general nonlinear gas ow
equation. In 1993, Chien presented a new real gas potential that was rigorously
implemented. Moreover, a general solution with pressure-dependent uid and
rock properties were analytically derived from the nonlinear gas ow equation.
The obtained solution was more accurate than those available in the literature
and also applicable to a wide pressure range.
Gas Diusion Equation The principle of conservation of mass for isothermal
ow of a single uid through a porous medium, assuming negligible gravity eects
is expressed by the continuity equation as:
\(j
!
n) =
J(jc)
Jt
(2.44)
The uid ux is given by Darcys law for laminar ow as:
!
n =
/(j)
j(j)
\j (2.45)
Consider j(j), j(j), /(j) and c(j) to be functions solely of pressure, and sub-
stitute Equation (2.45) in Equation (2.44). This yields:
\(
j(j)/(j)
j(j)
\j) =
J(j(j)c(j))
Jt
(2.46)
Equation (2.46) is the most general form of the diusion equation describing an
unsteady state ow of uid in porous media. A classical solution to the diusion
equation is for single-phase, one-dimensional radial ow to a well in an innite
40 2. TRANSIENT RATE DECLINE REVIEW
reservoir. The one-dimensional radial form of Equation (2.46) in a cylindrical
coordinate system is:
1
:
J
J:
_
:
j(j)/(j)
j(j)
Jj
J:
_
=
J(c(j)j(j))
Jt
(2.47)
The initial conditions set an initial reservoir pressure everywhere in the system:
1C j(:. t = 0) = j
i
while the rst boundary condition makes sure that the system remains in an
unsteady-state ow condition:
1.C.1 j(: . t) = j
i
The second boundary condition states that the ow must approach a steady-state
condition as the uid approaches the innitely small wellbore:
1.C.2
_
:
Jj
J:
_
v=v
u
=

&
j(j
&
)
2://(j
&
)
In order to transform Equation (2.47) into a diusion equation and to preserve
j(j), j(j), /(j) and c(j) as functions of pressure during the entire derivations,
Prats (1990) dened a new real gas potential as:
m(j) =
j(j
b
)
j(j
b
)/(j
b
)
j
_
j
l
j(j)/(j)
j(j)
dj
where j
b
is a low base pressure (Chien (1993) used j
b
= 14.7 psia).
To formulate a mathematical model for real gas ow, the following assump-
tions were made: isothermal ow; negligible gravity eects; laminar Darcy ow
through porous media; owing gas of constant composition; single phase ow;
homogeneous porous media; constant reservoir thickness, isotropic porous me-
dia. After taking derivatives on pressure and time and introducing the isothermal
compressibility of gas c
j
(j) =
1
(j)j(j)
o((j)j(j))
oj
, the diusion equation in terms of
:(j) is equal to:
1
:
J
J:
_
:
Jm(j)
J:
_
=
c(j)j(j)c
j
(j)
/(j)
Jm(j)
Jt
(2.48)
The initial and boundary conditions are:
1C :(j)(r, t = 0) = :(j
i
)
1.C.1 : (j)(r , t) = :(j
i
)
2. TRANSIENT RATE DECLINE REVIEW 41
1.C.2
_
:
Jm(j)
J:
_
v=v
u
=

&
j
&
2:/
j(j
b
)
j(j
b
)/(j
b
)
It is possible to extend the above solutions for an anisotropic porous media with
the variable transformation developed by Caudle (1967). An analytical solution
of Equation (2.48) with a special case of constant
jc

I
, has been presented by
Al-Hussainy and Ramey (1966). Equation (2.48) cannot be solved directly since
:(j) is a function of pressure, which in turn is a function of both time and
position. In 1993, Chien transformed the partial dierential Equation (2.48)
into an ordinary dierential equation by substituting variables according to a
procedure known as the Boltzmann transformation.
2.2.2 Vertical-Fractured Well
Finite-dierence reservoir simulation models can be used to history-match the
production performance of fractured wells. These models are often are cum-
bersome, requiring enormous amounts of data preparation and analysis time to
obtain reasonable matches of the production performance of a fractured well.
So, and alternative lies in modelling a fractured well with the semi-analytical
methods.
Prats, in 1961, presented a solution to a cylindrical, homogeneous, isotropic
reservoir with a vertical well intercepted by a vertical fracture with nite frac-
ture conductivity. This work was based on the assumption of an incompressible
uid, and showed the pressure distribution inside the fracture and in the reser-
voir; thereby providing tools to calculate the productivity improvement due to
a fracturing job and to analyse the eects of fracture face damage. The eec-
tive wellbore radius was presented as a function of fracture length and relative
fracture capacity; a parameter that is proportional to the inverse of the dimen-
sionless fracture conductivity (i.e., c = :,2,1C1). It is demonstrated that
the eective wellbore radius decreases along with the fracture conductivity. The
eective wellbore radius was shown to vary from a maximum of 0.51
)
for an in-
nite conductivity fracture to a minimum of r
&
for a fracture conductivity equal
to n,1
)
.
In 1962, Prats et al., extended the earlier work of Prats to a compressible
uid depleted through constant pressure or constant rate production. This work
assumed an innite conductivity vertical fracture that fully penetrated the for-
mation in the vertical direction. Moreover, it considers a maximum fracture
penetration or partial length of 50 % of the radius of investigation. It was found
here that both the terminal rate and terminal pressure cases can be modelled by
an elliptical reservoir with a larger eective wellbore radius.
In 1964, Russell and Truitt obtained a numerical solution to the case of a
fractured vertical well with an innite conductivity fracture in a square reservoir.
42 2. TRANSIENT RATE DECLINE REVIEW
The reservoir was closed and was depleted at constant production rate. Their
work shows that a transient ow regime is characterised by a region near the
fracture where the ow is linear, and a region away from the fracture where the
ow is pseudo-radial (elliptical). The wellbore pressure behaviour during the
pseudo-radial regime was shown to depend greatly on the partial fracture length
or penetration. The larger the fracture penetration, the closer the performance
approaches that for a pure linear ow. The eect of the innite conductivity frac-
ture during pseudo-steady state can be represented by an equivalent reservoir
that has a wellbore radius of r,2. Additionally, this work numerically corrobo-
rates previous observations on analogues and analytical solutions for linear ow.
Moreover, it demonstrates that innite conductivity fractured wells with a small
fracture penetration (i.e.< 0.1) can be analysed with radial unfractured models
within a 10 % error.
In 1969, Wattenbarger and Ramey extended the theory of fractured wells to
fractured gas well testing including wellbore storage and turbulence. The work
uses numerical methods similarly to those of Russell and Truitt, and conformal
mapping similarly to that of Prats. Here, innite conductivity fractures in an
innite reservoir were considered. The study gives a good explanation on treating
turbulence, fracture face damage and fracture length as pseudo-skins.
In 1972, Morse and von Gonten investigated the behaviour of innite con-
ductivity fractures prior to pseudosteady-state. Their investigation revisits the
work by Russell and Truitt and presents it in terms of productivity indices.
The productivity index ratio between fractured cases and cases of unfractured
pseudosteady-state cases a decrease with time until stabilisation. The productiv-
ity index ratio increases very rapidly as the partial fracture length,
1
]
1
c
increases.
A two-dimensional numerical simulation is run to constant pressure depletion,
and the results reveal again that, the larger the
1
1c
, the larger the increase in the
productivity index ratio.
2.2.3 Horizontal-Well
The development of low permeability gas reservoirs, conventional or unconven-
tional, is one of the solutions to the energy supply and demanding problems of
today. In low-permeability gas reservoirs, the creation of a ow path is critical,
and horizontal wells have been extensively used to increase the reservoir contact
area. Hydraulic fracturing can further expand the contact between wellbores and
formations. For horizontal wells, both with and without hydraulic fracturing,
the well performance becomes very sensitive to permeability and the anisotropic
ratio when the reservoir permeability is low. If the vertical permeability in the
formation is extremely low (high anisotropic ratio), then the benet of horizontal
wells starts to diminishing. In such a case, hydraulic fracturing provides another
option to increase well productivity. When hydraulically fracturing a horizontal
well, created fractures can be single longitudinal, multiple longitudinal, single
2. TRANSIENT RATE DECLINE REVIEW 43
transverse, or multiple transverse. The orientation and placement of fractures
along a horizontal well greatly aect its performance. Depending on the for-
mation condition and fracturing design, a fracture may in some cases result in
unavoured productivity, which has been evidenced in the eld. Predicting well
performance for fractured and non-fractured horizontal wells can help to obtain
the best estimates of production from low permeability gas formations.
In the early 1980s, major production successes through horizontal wells were
reported at the Prudhoe Bay eld and the Rospo Mare eld, oshore Italy. The
reported increase in production was on the order of at least two to three times
the equivalent production of vertical wells. The Rospo Mare eld happens to
be the ideal application of horizontal wells because of its producing formation
type. Giger et al. reported that the Rospo Mare pay consists of karsts made
up of very-low-permeability, compact carbonates. The oil resides mainly in the
fractures and vugs of the karstic matrix system. A horizontal well is more apt
to intersect many of these discrete natural fractures or vugular systems in such
formations.
Recently, with the improvement in horizontal well drilling and completion
technology, the feasibility of horizontal wells is seriously considered for such
dierent reservoirs as the naturally fractured Austin chalk formations, the low-
permeability Spraberry formations in west Texas, the Hugoton formations in
the Kansas/Oklahoma region, and the naturally fractured Bakken formation in
the Williston basin. Improvements in technology and operating procedures have
also resulted in a substantial cost reduction. Wilkinson et al. (1980) reported
a reduction in cost per foot of horizontal wells on the order of 40% based on
the average cost of the original three horizontal wells drilled at the Prudhoe Bay
eld. Drilling costs, however, are still reported to be 1.3 to 2 times higher than
for comparable vertical wells. Abdat (2000) reviewed selected papers of transient
behaviour of a horizontal well.
2.2.4 Horizontal-Fractured Well
The evaluation of multifractured horizontal well performance, or the selection of
an optimum perforation/stimulation design for such wells, may be approached
through ne grid reservoir simulations. However, while reservoir simulation is
the most advanced method of predicting well performance, it is often too time
consuming to be used for a parametric screening studies. The data required
is often unavailable and the eort may be unwarranted. As an alternative to
simulation, the application of semi-analytical models can readily yield wellbore
responses to various boundary conditions. Frequently, this is sucient to provide
an understanding of the factors with the most inuence on well performance. If
simulation work is warranted, it can then proceed with the insight obtained from
the analytical models.
Tight gas reservoirs are becoming increasingly popular candidates for mul-
44 2. TRANSIENT RATE DECLINE REVIEW
tifractured horizontal wells. An accelerated production accompanied by more
moderate recovery increases, pays for the initial capital expense, especially as
horizontal drilling and stimulation costs continue to decline. The multi-fractured
horizontal well technology has recently been applied to enhance productivity
from a tight gas eld located oshore from the Netherlands. A horizontal well
with two hydraulic fractures was completed in the tight (permeability = 0.2-
1.0 md) Ameland East reservoir, which constitutes a classic example of how a
poor candidate for horizontal wells can yield a substantially improved produc-
tion when produced from a multi-fractured horizontal wellbore. The reservoir
exhibits a low ratio of vertical to horizontal permeability rendering the non-
stimulated horizontal well uneconomic. Simulations of vertical inll wells and
various combinations of multi-fractured horizontal wells, combined with eco-
nomic evaluations, have demonstrated that the case of a horizontal well with
two hydraulic fractures provided the best economic return. The actual produc-
tivity improvement of this well, over the horizontal well with no fractures, is
estimated to be a factor of four.
For wells with hydraulic fractures, Prats (1961) started working on analytic
inow performance correlations for a single fracture. Van Kruijsdijk (1988) used
a combination of "Laplace" transformation and a Boundary Element formula-
tion to model the transient response in fractured reservoirs. This model was later
extended to include tight gas reservoirs, where non-Darcy ow in the fracture
must be taken into account. Kuppe and Settari (1996) have performed a num-
ber of reservoir simulations to cover multi-fractured reservoirs, and to provide
engineering correlations for variety of scenarios.
Methods for Predicting Productivity
With the increasing use of fractured and multi-fractured horizontal wells, it
seems appropriate to expect a more accurate method for determining the po-
tential productivity index enhancement of these wells. The manner in which a
productivity index is calculated, with consideration taken to a range of variables
(i.e., fracture height and length, well length, reservoir and fracture permeability,
etc.), could inuence the size of hydraulic fractures, the number of fractures in
the horizontal wellbore or whether or not a horizontal well should be drilled. As
mentioned previously, the use of the available analytical solutions, for fractured
or unfractured horizontal wells, could lead to errors if the limiting assumptions
are not taken into consideration or overlooked. There is a growing trend to
marry numerical simulation technology with analytical solutions. Improvements
can be made to existing analytical solutions by comparing their predictions to
numerical simulation results. Economides et al. (1991)
.
used a simulator with
a exible grid scheme (i.e., not following standard Cartesian orthogonality)
to modify Joshis solution in anisotropic permeability conditions. The original
form of one version of Joshis equation was:
2. TRANSIENT RATE DECLINE REVIEW 45

1
=
2:/
I
/1
j1
_
ln
_
o+
_
o
2
(12)
2
12
__
+
oI
1
ln
_
oI
2v
u
_
where: L = well length, c = large half-axis of elliptical drainage area, , =
_
/
I
,/

.
Tight Gas Reservoirs
In 1998 El-Banbi presented a collection of models and solutions useful for analysing
pressure and production data of tight gas reservoirs. An investigation was carried
out of the linear ow since many tight gas wells produce predominantly under
linear ow conditions for long times. The causes behind linear ow in tight
gas reservoirs are numerous. Among these can be mentioned: linear reservoirs;
high permeability streaks; wells between two no-ow boundaries; transient dual
porosity behaviour for radial reservoirs; wells intercepted by vertical, horizontal,
or diagonal fractures; horizontal wells; and horizontal wells with fractures. Lin-
ear reservoirs are those that show predominantly linear ow because of the shape
of the reservoir. The reservoir would impose one-dimensional linear ow. This
situation may occur in vertically fractured vertical wells whose fractures extend
laterally to the reservoir boundaries. It may also occur in horizontal natural
fractures and high permeability streaks. In this case, the linear ow develops
in the vertical direction. Such reservoir congurations may give rice to a linear
ow from the start of production. Linear ow may also persist for long times
before any boundary eects are felt.
2.3 Multiphase Flow
It is of common interest to describe multiphase ow in a reservoir. It is how-
ever dicult to obtain a simple solutions of ow within a reservoir as equations
describing multiphase ow are highly nonlinear. Muskat and Meres (1936) for-
mulated the fundamental equation that governs the multiphase ow in porous
media. The major contribution is the extension of Darcys law from single phase
to multiphase ow problems. This was possible due to the key concept of eec-
tive (or relative) permeability. Kato and Serra (1991) commented on diculties
in characterising the reservoir parameters under multiphase ow, according to
which there do not exist very many publications on multiphase welltest analysis.
Perrine (1955) was able to modify modify single phase solution with a pressure
approach. Based on empirical observations, the single phase properties (mobility,
compressibility) could thus be replaced by total system properties. Furthermore,
it was possible to estimate eective phase permeabilities (not absolute perme-
ability) and wellbore skin. Martin (1959) showed that Perrines approach was
46 2. TRANSIENT RATE DECLINE REVIEW
based on the pressure diusion equation derived when assuming negligible pres-
sure and saturation gradients. Perrines approach was further investigated by
Weller (1966), Chu et al. (1986) and Ayan and Lee (1988) and it has remained
the most commonly applied approach. Fetkovich (1973) intuitively contributed
in understanding the multiphase ow, and Raghavan (1976) suggested pseudo-
pressure for solution gas drive reservoirs. This approach is analogous to the one
proposed by Al-Hussainy and Ramey (1966). The pseudo-pressure approach was
further elaborated by Aanonsen (1985), who studied non-linear eects of solution
gas-drive reservoirs and noticed that small inaccuracies in relative permeability
data greatly inuence the exactness of the pseudo-pressure approach. Be et
al. (1989) used a similarity transform (the Boltzman transform) to solve radial
problems in well test analysis under multiphase ow. Al-Kalifah et al. (1987)
derived a diusion equation for multiphase ow with pressure squared, j
2
, as the
dependent variable.
2.4 Flow Under Variable Rate and Pressure
Ilk et al. (2006) sorted the references involving methods for variable-rate reser-
voir performance into the following categories:
Superposition and Convolution
Rate Normalization and Material Balance Deconvolution
Deconvolution
The transient ow rates are generated by a constant owing pressure of the
well bottom hole and analysed by decline curve analysis. In reality, due to
changes in operating procedures, the downhole owing pressure seldom remains
at a constant level over a long period of time. Recently, the deconvolution
technique, has become employed for well testing converts measured transient
pressure due to variable sand-face rate into the transient pressure response as a
result of equivalent constant owing rate. This technique can also be applied to
transient owing rate analysis. Since the wellbore pressure usually varies when
deconvolved, it appears to be equivalent to a constant pressure. Therefore,
deconvolved constant pressure at a wellbore is able to generate rate-time well
responses.
The basic assumption of all deconvolution techniques resides in the consis-
tency of the measured pressure and rate data with the linear Duhamel model,
which is based on the principle of superposition. The approach is limited due to
linearity of the system in which only one well disturbance appears. Moreover,
the technique cannot be applied if nearby wells cause interference in pressure
in a system. Further deconvolution cannot apply if well pressure behaviour is
inuenced by aquifer or gas cap. Additional requirements for linearity of the
system include the single-phase ow, which signies that deconvolution applies
to pressures above bubble point pressure in oil reservoirs. Finally the initial
2. TRANSIENT RATE DECLINE REVIEW 47
Table 2-2: Variable rate publications the history [After Gringarten (2006)]
48 2. TRANSIENT RATE DECLINE REVIEW
uniformity of pressure, within the whole investigated part of the reservoir and
well rate from the entire production by this well, must be satised all the way
from the initial equilibrium state.
The existing deconvolution methods can be classied into spectral and time
domain techniques. The spectral methods are based on the convolution theo-
rem of spectral analysis, and the convolution product is obtained by applying a
spectral transform such as the "Laplace" or "Fourier" transform. Kuchuk and
Ayestaran (1985), Roumboutsos and Stewart (1988), Cheng et al. (2003), and
Ilk et al. (2006) have all applied the spectral method. The time domain meth-
ods discretise the convolution integral using an interpolation scheme and then
proceed to solve the linear system. A set of smoothing constraints were imposed
on the solution when reducing the solution oscillation. The time domain method
deconvolution algorithm was recently published by von Schroeter et al. (2004).
His method works in a time-domain when a reasonable level of noise is present
in both pressure and rate data. In 2005 Levitan improved the algorithm.
Zheng and Fei (2008) created a deconvolution algorithm and code, which were
only tested with single phase oil data. They considered both pressure-rate and
rate-pressure deconvolution. Generally deconvolution methods applied to the
reservoir system are given by Duhamels integral or principle of superposition,
as a function of time. The pressure drop across the reservoir corresponds to the
convolution product of rate and reservoir response as given below:
j(t) = j
i
j(t) =
t
_
0
(t)q(t t)dt (2.49)
there, (t) is the measured ow rate, j(t) the pressure at the wellbore, and j
i
the initial pressure. Equation (2.49) is referred to as the impulse response of
the reservoir system. With the deconvolution, or an inversion of the convolution
integral, it is possible to estimate the reservoir system response. Two types of
deconvolution are related to pressure-time and rate-time analyses.
Well testing deals with pressure-rate deconvolution. The reservoir system
responds with transient pressure due to wellbore constant-rate conditions as
described with the following convolution integral:
j(t) = j
i
j(t) =
t
_
0
(t)
dj
&v
(t t)
dt
dt (2.50)
where (t) is the measured ow rate, j(t) is the measured bottomhole pressure,
j
i
is the initial reservoir pressure and j
&v
is the unit-rate pressure response. We
are interested in rate testing that deals with rate-pressure deconvolution in which
the unit constant pressure transient rate response of the reservoir system is given
by the following convolution integral:
2. TRANSIENT RATE DECLINE REVIEW 49
(t) =
t
_
0

&j
(t t)
dj(t)
dt
dt (2.51)
there
&j
is the transient rate response of the reservoir system obtained when
a well produces under unit constant pressure conditions. Multi-phase ow and
multi-well interferences need further investigations and new algorithms.
2.5 Other Transient Models
2.5.1 Multilateral Model
When obtaining solution for both vertical and horizontal wells the line source
method has been commonly used. The study of well behaviour by solving diu-
sion equations is dicult due to a dierence in scale between the wellbore diam-
eter and the size of a reservoir size (scale of 10
5
). The well behaviour was solved
with the a semi-analytical source function method, rst presented by Carlslaw
and Jaeger (1959) and Gringarten and Ramey (1973). The following authors
have since then studied pressure transient response for a horizontal well with the
source function method: Clonts and Ramey (1986), Goode and Thambynayagam
(1987), Daviau et al. (1988), Ozkan et al. (1998), Rosa and de Carvalho (1989),
and Odeha and Babu (1990). A further extension of the method to advanced
well studies has been performed by Besson (1990), Economides et al. (1994),
Azar-Nejad et al. (1996), Jasti et al. (1997). In 1998 Ouayang et al. applied
the line source approach to model nite conductivity wells.
Multi-lateral wells are increasingly used in reservoir engineering to improve
the oil recovery. To optimise the eciency of these wells, it has been necessary to
develop semi-analytical methods which are simple, fast and accurate for study-
ing the transient pressure behaviour. The pressure drop in the wellbore has a
strong impact on the pressure transient behaviour. Ding (1999) presented the
boundary integral equation, based on a single layer heat potential, to describe
the transient phenomena This approach included the pressure drop along the
well length. A mathematical method of Galerkin-type was employed to solve
the boundary integral equation leading to a quick and accurate evaluation of
the pressure drop along the length. The method was used to study the pressure
solution for multilateral wells and also for selectively perforated wells. Boundary
integral methods (BIEs) represent an alternative for determining the advanced
well behaviour. Scale problems between the wellbore diameter and the reservoir
size could be solved by representing the equation at the wellbore boundary and
the reservoir boundary. Among the authors that have used BIEs to study the
ow behaviour by applying the integral equation to the reservoir boundary can
be mentioned Kikani and Horne (1993), Sato and Horne (1993), Pechera and
50 2. TRANSIENT RATE DECLINE REVIEW
Stanislav (1997), Oguztorelli and Wong (1998) and Jongkittinarukorn and Tiab
(1998). In 1998 and 1999 publications Ding applied the BIE to the wellbore
boundary. This approach permitted an accurate modelling of pressure and ow
in the near well region. Furthermore, in 1999, Ding published more accurate
numerical techniques in order to solve the BIE for advanced well modelling. The
high-order Galerkin approach was used for space discretisation, and, compared
to the line source method, the linear Galerkin approach. The study included
calculations of the pressure drop along the wellbore. For each well in a clustered
system, both inner boundary conditions of pressure and rate were imposed. It
was possible to obey the coupled modelling of the reservoir and wellbore ow
for the single phase owing in a homogeneous anisotropic media. The transient
pressure behaviour of an advanced multilateral well included calculations of the
pressure drop along the well length. The transient pressure behaviour and inow
distribution can be calculated with any well congurations by for each well im-
posing both the rate and bottom hole pressure. The advantage of the modelling
approach was to validate the well modelling features in a reservoir simulator.
Calculations of the numerical PI and transmissibility in the vicinity of a well
were performed to improve the well modelling in a reservoir simulator. Further
model extensions were considered to a multi-layer reservoir with an arbitrary
reservoir geometry by applying the boundary integral equation to the reservoir
boundary and to the interfaces between layers.
Ozkan et al. (1998) investigated the transient pressure behaviour of dual-
lateral wells. The inuences of the length, phase angle and vertical and horizontal
separations of the laterals were discussed, and the inuence of anisotropy in the
horizontal plane on dual lateral well responses were determined. The eect of the
horizontal anisotropy may reduce the eective total length, and results indicated
that the best dual-lateral conguration was obtained when opposing laterals are
drilled along the minimum permeability direction.
Umnauayponwiwat et al. (2000) studied the transient pressure responses
and inow performance of a multiple well. The analytical model developed by
Umnauayponwiwat et al. (2000) evaluated the inow performance of multiple
horizontal wells in a closed system. Moreover, the transient model considered
the ow of a single phase liquid in a simple homogeneous and isotropic porous
medium. It was possible to simulate a mixture of vertical, fractured and hori-
zontal wells, arbitrarily located in a reservoir, with varying production and shut
in sequences. The model corresponds to a "Laplace" transformation domain and
the results are inverted into a time domain by the Stehfest numerical inversion
algorithm. The approach investigated the eects of transient ow periods on
the estimation of inow performances and the analysis of build-up responses of
horizontal wells. A simple homogeneous and isotropic porous medium can be
extended to a naturally fractured reservoir and anisotropy can be incorporated
into the model.
2. TRANSIENT RATE DECLINE REVIEW 51
2.5.2 Multiple Wells Model
Numerous solutions of the diusion equation were reviewed under various bound-
ary conditions, providing rate responses of a producing single well. The con-
ventional theories of transient pressure or rate analysis and inow performance
consider a single well with xed drainage boundaries, that can be constituted
of either the physical boundaries of the reservoir or the xed boundaries result-
ing from a stabilised ow conditions (pseudo-steady state or steady ow). In a
multi-well production system, wells interfere with each other due to the tran-
sient ow conditions resulting from changing wellbore conditions. A multi-welll
production system may be a mixture of vertical, horizontal, and fractured wells
creating a system with complex interactions among the wells.
In a closed system with multiple wells, Onur et al. (1991) and Valko et
al. (2000) studied the most relevant items related to the pressure transient
behaviour and well inow performances. In 2000, Umnuayponwiwat et al. in-
vestigated transient pressure behaviour of multiple wells in closed rectangular
systems. The main objective was to provide an understanding of the complex
interaction among wells in a multi-well system. The study involved the eect
of pressure transients due to changes in the production rates and estimations of
the wells drainage areas.
Fokker et al. (2005) presented a new semi-analytical method for calculating
the productivity of vertical, horizontal or multilateral wells draining either gas
or oil reservoirs. They considered well interference eects and the presence of
natural or induced fractures. By introducing moving pressure boundaries they
calculated the pressure eld in a three-dimensional reservoir containing multiple
wells and nite-conductivity fractures. Anyhow the moving-boundary approxi-
mation of the transient pressure response had limitations in accuracy for small
times at large distances from the well. Thus, this approach was considered valu-
able for screening purposes and for quick analysis. In their approach the solution
of a fully penetrating vertical well was further used for the complex well geome-
try or fracture geometry. The moving pressure boundary was developed for the
vertical well. Hence, for short times there is no inuence of the boundaries of
the reservoir, thus the reservoir can be treated as if it were innite. In 1978
Dake provided the solution of the pressure behaviour within a reservoir with a
producing fully penetrating vertical well as
j(:. t) = j
i


4:`/
1
_
a=
r
2
4TI
c
c
d:
:
(2.52)
further, simplifying they got
j(:. t) - j
i


4:`/
ln(c

:
2
41t
) = j
i


4:`/
ln(
_
41t,c

:
) (2.53)
52 2. TRANSIENT RATE DECLINE REVIEW
with ` =
I
j
, and 1 =
I
jc
and Eulers constant = 0.5772. This pressure
approximative solution is based on the simplication valid for small value of r,
i.e., r < 0.01, where the integral in Equation (2.52) is equal to
1
_
a=
r
2
4TI
c
c
d:
:
- ln(r) (2.54)
They calculated the time-dependent external boundary
r
e
(t) =
_
41t
c

by comparing the simplied Equation (2.53) to the steady-state solution of


a vertical well with a drainage radius, :
c
. Thus, by using a moving external
boundary it was possible to approximate the transient solution as to solution to
the steady-state problem.
Busswell et al. (2006) presented novel analytical solutions to a single layer
model for a cuboid shaped reservoir by employing a method of integral trans-
forms. The model solutions of a diusion equation apply to variety of boundary
and initial conditions. The well model comprises partially penetrating vertical,
horizontal and fractured wells and provides solutions in multi-well and multi-
rate scenarios. All three fracture conditions are implemented (uniform ux,
nite and innite conductivity). The gas model also comprises non-Darcy ow,
wellbore storage, and a naturally fractured reservoir. Wellbore conditions in-
clude both pressure and rate. Model solutions are constituted of "Laplace"
space solutions and are inverted to real space. This method may solve problems
where any permutation of the Neuman (ux) and Dirichlet (pressure) and Rubin
(ux+pressure) are specied over the multi-well closed box model boundaries.
The key reference on integral transform methods is that of Thambynayagem
(2006), unfortunately unavailable. Several other authors have recently presented
a multi-well concept that appears to be a research topic of great interest. De-
spite the existence of many advanced reservoir simulation technology models,
there is a need for an alternative analytical tool that is quick and at the same
time honours the physics of uid ow providing a broad understanding of the
reservoir dynamics.
Jordan et al. (2008) contributed by creating a simple method for predict-
ing the performance of multiple gas wells in complex reservoir shapes. Using
an approximation of the traditional "image well" method, pressure and produc-
tion proles could be generated for arbitrarily shaped reservoirs. The inclusion
of pseudo-time, which handles variable viscosity and compressibility (jc
t
), im-
proved the quality of late-time forecasting of gas productions.
Chapter 3
DEPLETION RATE DECLINE
REVIEW
Depletion rate decline also known as decline curve analysis, is a method for
matching the observed production rates of an individual well, group wells or
reservoirs by a mathematical function for reserve estimation and production fore-
cast. In the 1950s, Arps presented traditional decline curve analysis including
exponential, hyperbolic and harmonic methods. In the 1980s, Fetkovich cre-
ated type curves. He used constant-pressure analytical mono-phase solutions for
transient production analysis with empirical multiphase depletion decline curves
taken from Arps. As a result, the interpretation of wellbore rate responses was
more rigorous due to the possibility of dividing rates into transient and the de-
pletion parts. It is generally dicult to precisely measure transient rates. Also,
the depletion rates can be determined with a certain accuracy, assuming that
the down-hole owing pressure is constant. However, in reality, due to condition
constraints or changes in operating procedures, the down-hole owing pressure
is seldom kept constant over long periods of time. In other words, the method
cannot be directly applied. Further improvements in rate time interpretations
were made in the 1990s with methods that include various types of superposi-
tion and normalisation among which the most important are the investigations
conducted by Palacio and Blasingame (1993) and Agarwal et al. (1999). Kuchuk
et al. (2005) proposed a deconvolution-based method for diagnostics in decline-
curve analysis.
3.1 Empirical Models (Arps)
The earliest reported eort to study the production drop over time was per-
formed by Arnold and Anderson in 1908. They proposed that the production
rates during equal time intervals formed a geometric series and stated that the
production drop expressed as a fraction was approximately constant. Arnold and
53
54 3. DEPLETION RATE DECLINE REVIEW
Anderson called this fraction the decline. These authors were also the rst to
notice that the rate-versus-time curve exhibited a straight line on semilog paper.
Between 1908 and 1944, extensive research was carried out in this area and
although the studies are too numerous for all of them to be cited, the work
performed by Cutler in 1924 warrants highlighting. Cutler noticed that the
geometric or exponential type of decline curve gives conservative estimates of
volume as well as a conservative production forecast. He also stated that a
hyperbolic relationship on log-log paper would better describe the production
decline.
Attempts to theoretically model production rate decline and cumulative pro-
duction curves of gas and oil wells, date as far back as the early part of the 20
tI
century. In 1921, a detailed summary of the most important ndings of the early
research activities in this area was documented in the Manual for the Oil and
Gas Industry. This treatise, which is mainly a compilation of the research work
of the U.S. Bureau of Mines personnel, rst noted the exponential decline model
for oil wells, as well as the use of graphical techniques in the form of percentage
decline curves (i.e., hyperbolic declines) for the analysis of production rate
data from gas wells.
In 1927, Johnson and Bollens introduced the so-called loss ratio method,
which was dened as the ratio between production rates and production drops
at equal time intervals. This ratio was found to remain approximately constant,
thus providing an easy method for extrapolation.
Arps Model
In 1945, Arps published a comprehensive review of the previous eorts regard-
ing decline curve analysis. Based on these results, he was able to empirically
verify the equations for the exponential and hyperbolic decline behaviour. The
continuous use of these equations, even presently, is basically due to the ease
of application and their acceptance in industry. Arps also showed how to ex-
trapolate rate-time data following an exponential or hyperbolic decline. While
the exponential decline represents the simplest model to use, it also yields con-
servative estimates and remains the most popular method within the petroleum
industry.
Several eorts were made during the years that followed, and the most signif-
icant contribution towards the development of the modern decline curve analysis
concept is probably the classic paper by Arps presented in 1945. In this paper,
Arps described a set of exponential and hyperbolic equations for production rate
analysis. Although the basis of Arps development was purely statistical, and
therefore empirical in nature, these historic results have found widespread ap-
peal in the oil and gas industry. The continuous use of these so-called Arps
equations to date is basically due to the explicit nature of the relations and the
ease of application of these equations to eld data.
3. DEPLETION RATE DECLINE REVIEW 55
Arps also introduced the concepts of a decline exponent, b, and a decline
rate constant, 1
i
, both of which have become the cornerstones of many sub-
sequent research eorts. By estimating these important parameters through
history matching, Arps demonstrated the technique of extrapolating rate-time
data following exponential and hyperbolic declines using a semilog plot. Al-
though the exponential decline is the simplest model to use, especially in decline
curve analysis of oil wells, this model also yields the most conservative estimates
of in-place uids and production rates. On the other hand, the harmonic decline
model, provides the most optimistic estimates when used for predicting future
production rates.
Research following Arps publication concentrated on improving the fore-
casting of production data based on a general hyperbolic model. In 1968, Slider
presented a new curve matching technique for obtaining a more accurate ex-
trapolation of production rate data following the hyperbolic decline model. This
approach was also based on semilog analysis. In addition, the author demon-
strated a practical curve-tting method using preconstructed theoretical decline
curves based on Arps equations. The technique was presented as simple and
more eective than other decline curve analysis methods although a signicant
amount of work was required in data preparation.
Arps (1945) also carried out a fundamental study of the mathematical basis
of decline curves. He showed that equations for the semilog and log-log graphs
could be derived from the basic dierential equations. Moreover, he introduced
exponential, hyperbolic and harmonic decline curves, which were determined by
the decline exponent, /. The curves were dened according to the drop in pro-
duction rate per unit time, represented by the fraction of production rate directly
proportional to the production rate and fractional power of the production rate.
The fractional power is dened by the decline exponent, /, which has a value
between zero and one. Expressed in the form of a dierential equation, the above
statement becomes:
1(t) = 1(t)
b
=
oq(t)
ot
(t)
(3.1)
This equation can be solved and presented in the form of a general empirical
hyperbolic equation:
(t) =
i
(1 +/1
i
t)

1
l
(3.2)
With the integration of the rate time relationship, the cumulative production
can be expressed as:
`j =
_
(t)dt =

i
/
(1 /)1
i
(
(1b)
i
(t)
(1b)
) (3.3)
56 3. DEPLETION RATE DECLINE REVIEW
The Arps equation (3.2) is the most commonly used. It is empirical and selected
for depletion behaviour under producing conditions where the compressibility
is modied. An example of compressibility changes involves solution gas drive
mechanisms. It was noticed that the decline exponent, /, can be inuenced by the
reservoir ow conditions and that value of / determines the degree of curvature
from the straight line or exponential semi-log decline (b= 0.0) to the harmonic
decline (b = 1.0). Exponential and harmonic decline equation are particular
solutions to the general hyperbolic solution.
An exponential decline can also be referred to as a constant percentage decline
since the decline rate, 1, remains constant with production. Ahyperbolic decline
equation contains two unknowns: the hyperbolic exponent, /, and the initial
decline rate, 1
i
. Once these two unknowns are established the determination of
remaining reserves and future production can be done.
The extrapolation of production decline curves provides us with the remain-
ing quantity of oil and gas reserves as well as the time of abandonment of a well
or lease. Arps assumed that the extrapolation procedure was strictly empirical
and that everything causing the decline curve trend in the past uniformly con-
tinues to maintain it. The decline exponent, /, must be between zero and one.
Empirical extrapolation signies a wide range of interpretations. Experience,
integrity and objectiveness of the evaluator are related to the interpretation.
Available data also can be controlled by the interpretation. Production data
can be controlled by the nature of reservoir rocks, uid characteristics and drive
mechanisms or by the production strategy, works on wells, producing equipment
limitations or personnel policies.
3.2 Analytical-Numerical Models
3.2.1 Oil Flow
The purpose of this subsection is to discuss, based on examples, the general
model that generates the rate responses of a vertical perforated well in a stratied
homogeneous reservoir. A single-phase oil production takes place only through
vertical well perforations. Boundaries are circular and either no-ow or bounded.
Generally, the model can incorporate inner boundary conditions of variable rate
or variable pressure at the wellbore. This study considers constant pressure
IBCs and does not include a wellbore skin analysis. Outer Boundary Conditions,
OBCs, are no-ow and xed, or located from a wellbore axis at a distance :
c
.
The model yields a reservoir rate that varies with radial distance and time.
Volumetric Bounded Reservoir In 1981, Ehlig-Economides and Ramey
established an overview of the transient rate decline analysis for well produc-
tion at constant pressure. From 1981 until today, the area of decline rate solu-
3. DEPLETION RATE DECLINE REVIEW 57
tions were partly presented in various publications and literature. Transient rate
analysis is currently an alternative to the well test analysis.
The conditions considered diered fromthe previous model in outer boundary
condition. The closed or bounded reservoir model was characterised by a no-uid
ow across an outer boundary (t) = 0 and in dimensionless form, expressed for
:
1
= :
1c
as:
0j
T
0v
T
= 0.
For a well producing at constant pressure from the limited drainage volume,
the resultant behaviour is an exponential decline in the rate. This case was
denoted exponential depletion. A well producing at a constant rate from the
limited drainage volume represented a pseudosteady state behaviour. The expo-
nential depletion for the closed boundary system derived by Ehlig-Economides
and Ramey (1981) was given as:

1
(t
1
) =
1
ln
4
C
/
v
2
u
_
4:t
1
ln
4
C
/
v
2
u
_
(3.4)
This was an exact equation for the exponential depletion for t
1
(tj::)1,
where t
1
was the dimensionless time based on the drainage area, , and (tj::)
was the dimensionless time at the beginning of the pseudosteady state ow, or
in other words the time required for the development of the true pseudosteady
state at the constant rate inner boundary condition. It was dependent on the
reservoir shape as published by Earlougher and Ramey (1968). The eect of skin
was included by using an apparent, eective wellbore radius :
&o
, instead of the
radius :
&
in (3.4). This ctitious radius, :
&o
, was dened as :
&o
= :
&
c
(c)
.
Tsarevich and Kuranov (1956) were the rst to publish that the exponential
decline was the nal form of ow rate decline for constant pressure inner bound-
ary production from a circular reservoir. They provided a theoretical basis for
decline curve analysis and presented tabulated solutions for the cumulative pro-
duction for the closed boundary reservoir.
A special case derived for a well located in the centre of a circular reservoir
was proposed by Fetkovich (1980). An analytical solution applied to a circular
reservoir case conrmed that the exponential rate time decline was a late time
solution of the volumetric reservoir under constant pressure inner boundary con-
ditions:

1
(t1) =
1
ln(0.472:
1c
)
c

2I
T
r
2
Tc
ln(0.472r
Tc
)
(3.5)
The unsteady ow rate declined to the point where the cumulative production
was constant depending on the reservoir size, :
1c
, as published by Uraiet and
Raghavan (1980):
Q
j1
=
1
2
(:
2
1c
1) (3.6)
58 3. DEPLETION RATE DECLINE REVIEW
Uraiet and Raghavan solved partial dierential equations with dened bound-
ary conditions by the nite dierences method. They analysed build-up be-
haviour of a well producing at a constant wellbore pressure. Pressure buildup
equations can be obtained by the principle of superposition, and a nite dier-
ences model was developed due to the diculty in obtaining a simple analytical
expression that can describe the bottomhole pressure buildup behaviour.
Ehlig-Economides and Ramey presented solutions for three existing outer
boundaries by implementing the numerical "Laplace" transform inversion algo-
rithm according to Stehfest.
A method for converting constant rate solutions to constant pressure so-
lutions developed by Cox (1979) was also applicable to bounded reservoirs.
Bounded reservoirs producing under pseudosteady state ow conditions against
well drawdown displayed exponential declines in the production rate. The pres-
sure response for a well producing under pseudosteady state ow conditions was
expressed by Ramey and Cobb (1971):
j
1
= 2:1
1
+
1
2
ln(

:
&
2
) +
1
2
ln(
2.2458
C

) + : (3.7)
Equation (3.7) was transformed by Cox (1979) in the form of an exponential
decline for the production rate,
1
:

1
(t
1
) =
1
1
1O
c
(2t
T/
1
TC
)
(3.8)
where 1
1O
is the intercept in equation (3.8).
The constant pressure outer boundary associated with a gas cap or bottom
water did not change the pressure distribution with time. A steady state condi-
tion was described with : = :
c
and the pressure j = j
i
. The outer boundary
condition in dimensionless form involved, for :
1
= :
1c
dimensionless pressure
j
1
= 0. The exact solution, including the skin factor, was derived by Ehlig-
Economides and Ramey (1981) was:

1
=
1
ln :
1c
+ o
(3.9)
The solution was valid for a dimensionless time, t
1
of:
t
1
=
1
2.2458:
(3.10)
3.2.2 Gas Flow
The decline analysis of gas wells has been reported by Stewart (1970) and Gurley
(1963). The gas well rate equation can be expressed as:

j
(t) = C
j
(j
2
1
1
2
&)
)
n
(3.11)
3. DEPLETION RATE DECLINE REVIEW 59
If j
&)
= 0, and assuming that the gas compressibility 2 = 1, we get:
1
1
=
_
1
1
G
1i
_
G
j
+ 1
1i
(3.12)
Alternatively, the cumulative gas production as a function of the initial gas can
be expressed as:
Gj = G
1i
_
1
_
1
1
1
1i
_
I
_
(3.13)
For G
j
= G
j
( P
1
,t), Equation (3.13) can be written as:
dG
1
dt
= /G
1i
1
1
I1
1
1
I
d1
1
dt
(3.14)
Moreover, the rate equation for gas wells was dened as:

j
(t) = C
j
(1
v
2
1
2
&)
)
n
(3.15)
In comparison to the rate oil equation, this results in:
C
j
J

0
1
1
1
1i
(3.16)
For the known initial rate and pressure, the gas well backpressure curve coe-
cient, C
j
, can be calculated by:
C
j
=

ji
_
1
1i
2
1
2
&)
_
n
(3.17)
which yields:

ji
=
_
1
1
2
1
2
&)
1
1i
2
1
2
&)
_
n
(3.18)
By assuming that P
&)
is very small, this equation can be simplied to:

ji
=
_
1
1
2
1
1i
2
_
2n
(3.19)
A combination of Equations (3.14)) and (3.19) results in:

j
=
dG
1
dt
=
ji
_
1
1
1
1i
_
2n
= /G
1i
1
1
I1
1
1
I
i
d1
1
dt
(3.20)
By separating the variables and integrating, the equation can take the form:
60 3. DEPLETION RATE DECLINE REVIEW
1
T
_
1
T.
1
1
(2n+I1)
d1
1
=

ji
1
1i
(2n+I)
/G
1i
t
_
0
dt (3.21)

j
(t) =
ji
1
_
2nI
I
q
.
G
T.
t 1
_ 2r
2rI
(3.22)
The rate time equation for a gas well in the case where (-2m+k),=0 and k,=2m
can be expressed as:
1
T
_
1
T.
1
1
1
d1
1
=

ji
/G
1i
t
_
0
dt (3.23)
The general rate time equation for gas wells:

j
(t) =
ji
1
c
_
2r
I
q
.
C
T.
t
_ =
1
c
_

q
.
C
T.
t
_ (3.24)
The unit solution of Equation (3.24) was plotted as a log-log type curve. For
the lower limit of the backpressure curve slope i.e., : = 0.5 decline exponent
was / = 0.0, which was recognised as an exponential decline. The upper limit of
: = 1.0 resulted in the decline exponent / = 0.5 The eect of the backpressure
on the gas well was thus found to alter the type of decline. This situation diered
from the liquid case solution. The backpressure was expressed as a j
)
,j
i
ratio
and for the j
&)
j
i
(i.e., j 0), the type curve approached the exponential
decline with / = 0. Cumulative rate time log-log type curves could be prepared
by integration of the rate time Equation (3.24).
3.2.3 Multiphase Flow
A multiphase ow approach based on the assumptions employed for Arps equa-
tions was done by Camacho and Raghavan (1989). They examined a well per-
formance in solution-gas-drive reservoirs with a closed boundary ow. The Arps
decline exponent, /, and the initial decline rate, 1
i
, expressed in terms of physical
properties. The conditions for decline analysis can be described by a homoge-
neous closed cylindrical model with a fully penetrating well located in its centre.
The inner boundary condition was dened for a well producing at constant well-
bore pressure. The eect of a skin region was included through an annular re-
gion with a permeability diering from that of the formation. Gravity, capillary
pressure and non-Darcy ow eects were not considered. Cammancho (1987)
developed the dimensionless pseudopressure:
3. DEPLETION RATE DECLINE REVIEW 61
j
T
(:. t) =
//
141.2
0
(t)
_

_
v(t)
_
v
_
c(j. o
0
)
Jj
J:
_
t
d: +
t
_
0
_
c(j. o
0
)
Jj
Jt
0
_
v
dt
0
_

_
(3.25)
where c is a function of pressure and saturation, c(j. o
0
) = /
vc
(o
0
), [(j
0
(j)1
0
(j)],
and : is radius corresponding to the position in the reservoir at which pressure,
j(:), is equal to average pressure, j. During the boundary-dominated ow pe-
riod, : ~ 0.54928 :
c
. The dimensionless pseudopressure was calculated for inter-
val close to a wellbore, where 1 6 :
1
6 :
c1
, where :
1
= :,:
&
is dimensionless
radius, , and :
c1
is the dimensionless radius of the skin zone. For an outer
interval:
j
T
(:. t) = j
j1
(t) +
_
ln
:
c1
:
1

3
4
_
+
_
/
/
c
1
_
_
1
4
v
4
sT
:
4
c1

v
2
sT
:
2
c1
+
1
2
(:
2
c1
1)
:
2
c1
_
+
1
2
(:
2
c1
1)
:
2
c1
(3.26)
for :
c1
_ :
1
_ :
c1
. The volumetric average of the pseudopressure is calculated
by using Muskat (1945) material balance equation:
j
T
(t) =
//
141.2
0
(t)
j
.
_
j(t)
c(j
0
. :)dj
0
= 2:

t
1
(3.27)
This dimensionless pseudopressure may also be considered a generalization of the
material-balance equation for production at a variable rate in solution-gas-drive
reservoirs. Here,

t
1
=

t
1
:
2
W

(3.28)
and the dimensionless time,

t
1
dened as

t
1
=
0.006328/
:
2

0
(t)
t
_
0

0
(t
0
)`
t
(t
0
)
c
t
(t
0
)
dt
0
(3.29)
The system compressibility, c
t
, and mobility, `
t
,corresponding to j (and o
0
) are
c
t
=
o
0
1
0
(j)
_
d1
0
dj
_
j

o
j
1
j
(j)
_
d1
j
dj
_
j
+ o
0
1
j
(j)
1
0
(j)
_
d1
c
dj
_
j
(3.30)
and
62 3. DEPLETION RATE DECLINE REVIEW
`
t
=
_
/
vc
j
0
+
/
vj
j
j
_
(
j,S
0)
(3.31)
For the constant oil-rate dimensionless pressure in termof time, t
1
=
0.006328I
v
2
t
_
0
A
I
(t
0
)
c
I
(t
0
)
dt
0
is:
j
T
(t) = 2:t
1
:
2
&

(3.32)
Equation (3.32) is an extension of the material-balance equation for single-phase
liquid ow.
3.3 Type-Curves
3.3.1 Vertical Well
Authors such as Tsarevich and Kuranov (1956), Ehlig-Economides and Ramey
(1981), Uraiet and Raghavan (1980) have considered production rate decline
analysis given a constant wellbore pressure. They assumed a constant diusiv-
ity, j for various dimensionless radii :
1
. The analytical solutions presented by
these authors form the transient portion of Fetkovichs (1980) type-curves where
constant pressure innite (early transient period) solutions were combined with
the empirical decline curve equation developed by Arps (1945).
Fetkovich Type-Curve
Decline curve analysis is founded on the same basic uid ow principles that are
used in pressure transient analysis. Rate time curve analysis is based on constant
wellbore pressure solutions for various physical models. This concept included
the depletion period and pressure transient period of time. Constant wellbore
pressure solutions and their corresponding log-log type curve plots represented
the inverse of the constant rate solution.
Single dimensionless unit type curves were composed of an analytical constant
wellbore pressure solution and the Arps exponential, hyperbolic and harmonic
decline curve solution. Depletion steam values range between an exponential,
/ = 0.0, and a harmonic, / = 1.0, decline accepted as the lower and upper limits.
The exponential depletion was taken as common to the Arps equation depletion
part and to the transient part of the analytic solution on the dimensionless plot.
Dimensionless
1o
and t
1o
values were dened by Fetkovich (1980). He based
this denition on the Arps exponential equation:

1o
(t
1
) =
(t)
i
=
1
c
1
.
t
=
1
c
t
Tu
(3.33)
3. DEPLETION RATE DECLINE REVIEW 63
From the hyperbolic equation, the dimensionless variables
1o
and t
1o
can be
formulated as:

1o
(t
1o
) =
(t)

i
=
1
(1 +/1
i
t)
1
l
=
1
(1 +/t
1o
)
1
l
(3.34)
Moreover, for a dimensionless harmonic decline, / = 1, was dened as:

1o
(t
1o
) =
1
(1 +1
i
t)
1
l
=
1
(1 +t
1o
)
(3.35)
The unit solutions, for an initial decline exponent 1
i
= 1, were plotted as a
set of log-log type curves. For each decline curve the decline exponent, /, was
between 0 and 1, increasing with increments value of 0.1. On the dimensionless
graph, data previous t
o1
= 0.3 will be on the exponential decline regardless of
true value of /. All curves were dened in the depletion area on the plot, and
the dimensionless time between 0.2 and 0.3 separated the depletion from the
transient period.
The dimensionless time and rate of the decline curve were dened in terms
of reservoir variables for the transient period with the following expressions,
t
1o
=
t
1
1
2
[ln(
vc
v&
)
2
1][ln(
vc
v&
) 1]
(3.36)
t
1
=
0.00634/
cjc
t
:
2
&o
t

1
=
1
[ln(
:
c
:
&

1
2
)] and
1
=
141.3j1
//(j
i
j
&)
)
(t).
All analytically derived depletion stems become exponential solutions and col-
lapse into a single curve with the above denition of
1o
and t
1o
.
The published values of t
1
and q
1
for the innite solution data were ob-
tained from Ferris et al.(1962). For the nite constant pressure solutions, data
were obtained from Tsarevich and Kuranov (1966). The t
1
and q
1
values were
transformed into a dened decline dimensionless time, t
1o
, and rate,
1o
, for
various values of :
1c
=
v
c
v
u
.
Late-time production analysis is based on the Arps decline curves (by match-
ing real data to empirical Arps expressions). Fetkovich (1973) combined Arps
decline curves with semi-analytical rate-time well responses, which were then
transformed and plotted in dimensionless form of associated rate and time. Semi-
analytical expressions are solutions of a diusion equation solved for inner bound-
ary conditions of constant pressure and outer boundary conditions of no-ow at
a xed distance :
c
. Again, the distance :
c
from a well situated in a middle of a
cylinder was kept constant and the drainage area does remain unchanged with
time. Under single-phase ow conditions in a homogeneous reservoir of height
/, the well rate response with time declines exponentially after a time t
1SS
.
64 3. DEPLETION RATE DECLINE REVIEW
Figure 3.1: Dimensionless Arps curves (Decline / = 0.0; 0.5, and 1.0) [After
Cvetkovic (1992)].
Figure 3.2: Semi-analytical dimensionless rate-time type curves (for various di-
mensionless radii, :
1
) [After Cvetkovic (1992)].
3. DEPLETION RATE DECLINE REVIEW 65
Figure 3.3: Combined transient-depletion dimensionless Fetkovich (1973) rate-
time type curves [After Cvetkovic (1992)].
Pseudo-Steady-State Time Single-phase ow solutions for various drainage
areas are plotted in a transient part of the type curves transformed by Fetkovich
(1973). Each drainage radius is represented with a transient decline curve as
demonstrated in Figure (3.2). After a time t
1SS
, this single-phase response is
exponential, i.e., all drainage areas end with the same exponential decline as
in Figure (3.5). Contrary to transient decline, depletion or Arps decline may
also be relevant to a multiphase ow. The well produces the same drainage
area. After the time t
1SS
(the time in which the reservoir boundary is reached),
the semi-analytical expression for the rate becomes exponential, overlaying the
Arps exponential decline curve. Other Arps curves have rate-time curvatures
expressed with a decline exponent, /, ranging from 0.1 to 1.0. In other words,
the rate decline changes with time from exponential to hyperbolic and nally to
harmonic. Arps expressions for rate-time decline are empirical, based on analyse
of data collected through numerous years of well production involving certain
dened drainage areas. The approach of Fetkovich (1973, 1980) introduced more
physics into the decline parameters (initial rate,
i
, initial decline, 1
i
, and decline
exponent, /). The time t
1SS
, on a unique combined curve, actually divides rate-
time well responses into the transient decline and the depletion decline. After
time t
1SS
, the well is producing a constant drainage volume and the radius of
drainage reaches the outer boundaries of no-ow. In the depletion decline of
a well, several drive mechanisms may be considered (solution gas drive, gravity
drainage or partial water drive). The well may also be producing a single layered,
multilayered or heterogeneous reservoir.
Transient decline curves have been derived for a vertical well with one ow
regime during transient ow. Golan and Whitson (1986) derived :
1
expressions
for early times for a conventional vertical well situated in the centre of a radial
reservoir. The extension of Fetkovich type curves to a nonconventional well as
66 3. DEPLETION RATE DECLINE REVIEW
Figure 3.4: Transient dimensionless rate-time curves (for two values of :
1
) [After
Cvetkovic (1992)].
Figure 3.5: Transformed depletion dimensionless rate-time curves (for two di-
mensionless r
1
) [After Cvetkovic (1992)].
3. DEPLETION RATE DECLINE REVIEW 67
Figure 3.6: Transient dimensionless rate-time curves (for various dimensionless
:
1
values) [After Cvetkovic (1992)].
Figure 3.7: Arps dimensionless rate-time curves [After Cvetkovic (1992)].
68 3. DEPLETION RATE DECLINE REVIEW
a well with fractures should include several ow regimes in the transient decline
part.
Blasingame et al. (1991) implemented method for analysing rate-time data
when the bottom hole pressure is variable. He introduced a material balance time
function, t
cj
, (that was calculated by dividing the cumulative oil by the oil rate
for each time period) and applying it to convert the constant pressure solutions
for liquid and gas to an equivalent constant rate liquid solution for a single layer
system. During transient ow conditions the constant pressure and constant rate
methods are the same, while they are quite dierent during depletion, when the
constant rate system solutions are harmonic. However, the constant pressure
solutions are exponential. The method required that the drawdown normalised
rate be plotted vs. the material balance time, t
cj
. The method smooth the data
and may thus improve the type-curve matching. Nevertheless, it was noticed
that, with this method, the depletion data was "forced" to match the harmonic
depletion stem and not the value of exponent / < 1. So, it is evident that by
applying the method someone loose information on drive mechanism, recovery
eciency, and layered no cross-ow behaviour.
Callard et al. (1995) presented type curves in plots of the pressure-normalised-
rate versus the pressure-normalised cumulative production. Both curves for con-
stant rate and constant pressure overlaid the same rate-cumulative curve during
transient and pseudosteady-state ow conditions. The equation for the dimen-
sionless cumulative production, Q
1
, was given by
Q
1
=
0.8936Q(t)1
0
c/:
2
&o
(j
i
j
&)
)
(3.37)
Pressure-Dependent Fluid and Rock Properties Samaniego and Cinco-
Ley (1980) performed a numerical investigation of the inuence of pressure-
dependent uid and rock properties, during a single-phase ow, on well produc-
tion decline caused by constant wellbore pressure conditions. The variable rock
properties included the permeability, porosity, pore compressibility and forma-
tion thickness and the variable uid properties included the density, compress-
ibility and viscosity. A ow equation considering the pressure dependence of rock
and uid properties when expressed as a function of a pseudo-pressure, :(j),
resembled the diusion equation. Samaniego et al. (1976 and 1977) evaluated
this variable property problem for various ow conditions.
The mathematical model was based on: horizontal ow, with no gravity, a
fully penetrating well, an isothermal single phase uid obeying Darcys law, and
an isotropic homogeneous formation. Samaniego at al., (1976 and 1977) and
Samaniago (1974) noted that the assumption of horizontal ow was not quite
valid.
With uid and rock properties held constant, the computer model obtained
data that was veried against published solutions of van Everdingen and Hurst
3. DEPLETION RATE DECLINE REVIEW 69
Figure 3.8: A type-curve match for a constant- pressure drawdown test with
variable property solutions [After Samaniego and Cinco (1980)].
(1949) and Fetkovich (1973). A good match with dimensionless solutions of
van Everdingen and Hurst was obtained. Numerical results obtained from the
computer model and the analytical technique by Fetkovich (1973) were created
for 3 reservoirs of varying :
c1
and a good agreement was found. It was noticed
that, for all ratios of j
i
,j
&)
during transient ow conditions, the production
rate decline expressed in terms of a dimensionless rate,
1
, was the same as
the production rate decline for constant property liquid ow as given in the
transient part of Figure (3.8). Nevertheless, solutions for a bounded reservoir
deviated from the classic
1
solutions once the ow was aected by the outer
boundaries, as in Figure (3.9). It was concluded that the production rate in
pressure sensitive-systems declined faster than in constant-property systems.
Stratied No-Crossow Reservoirs Most reservoir are heterogeneous and
consist of several layers without crossow, each with its reservoir properties.
In a reservoir with crossow the adjacent layers can be combined into a single
equivalent layer that can be described as homogeneous through an averaging of
the reservoir properties of the crossowing layers. The decline curve exponent,
/, for a single homogeneous layer ranges from 0, to 0.5, whereas for layered no-
crossow systems, values of / range from 0.5 to 1. Thus, those / values greater
the 0.5 can identify the reservoir stratication.
Fetkovich et al. (1996) suggested that the decline exponent, /, and the decline
rate, 1
i
, can be expressed in terms of the back-pressure curve exponent, :. Both
expressions were derived from the back-pressure equation,
70 3. DEPLETION RATE DECLINE REVIEW
Figure 3.9: The dimensionless ow rate compared to the Arps decline rates
[After Samaniego and Cinco (1980)].

j
= C(j
v
j
&)
)
a
(3.38)
where : is the back-pressure curve exponent; C is the performance coecient;and
j
v
is the reservoir pressure. The Arps decline exponent, /, and the decline rate,
1
i
, (with an initial in place gas in place, G) were respectively dened as:
/ =
1
2:
_
(2: 1)
_
j
&)
j
i
_
2
_
(3.39)
1
i
= 2:(

i
G
) (3.40)
Equation (3.39) shows that, as the j
i
approaches j
&)
, the hyperbolic decline
shifts to the exponential decline, thus changing the / exponent from a value
not equal to zero, to zero. Further, for those wells producing at a very low
bottom-hole owing pressure with j
&)
= 0, the decline exponent is reduced to
/ = 1
1
2:
Fetkovich (1980) derived the expressions in Table (3-1), by combining Arps
hyperbolic equation with the material balance equation that relates j,2 with
G
j
, and the back-pressure equation. He expressed the rate-time equation for a
gas well in terms of the back-pressure exponent, :, with a constant j
&)
of 0 that
also implies that
i
=
inoa
, as given by
3. DEPLETION RATE DECLINE REVIEW 71
Table 3-1: The rate-time equation for a gas well in terms of the back pressure
exponent, n, with constant "p
&)
" of 0 as dened by Fetkovich (1980)
:
t
G
j
(t)
0.5 < : < 1.0
0 < / < 0.5 qi
[
1+(2a1)
(
q
.
C
)
t
]
2n
2n1
G
_
1
_
1 + (2: 1)
_
q
.
G
_
t
1
12n
_
: = 0.5
/ = 0

i
c

(
q.
C
)
t
G
_
1 c

(
q.
C
)
t
_
: = 1
/ = 0.5 qi
[
1+
(
q
.
C
)
t
]
2
G
_
1
1
1+
(
q
.
I
C
)
_

i
=
i max
=
//j
2
i
14221(j
j
2)
oj
_
ln(
v
c
v
u
)
3
4
+ :
_ (3.41)
Here,
inoa
(in `:c,,d) is a stabilized absolute open-ow potential, i.e., at
j
&)
= 0; G (in Mscf) is the initial gas in place;
i
(in `:c,,d) is gas ow rate at
time t; and G
j
(t) (in `:c,) is the cumulative gas production at time t. Ahmed
(2006) presented a case of a commingled well producing from two layers at a
constant j
&)
. The total ow rate, (
t
)
tcto|
is the sum of the ow rates of the
two layers according to (
t
)
tcto|
= (
t
)
1
+ (
t
)
2
. For a hyperbolic decline with an
exponent, / = 0.5, using the expression from Table (3-1), we get
(
max
)
tcto|
_
1 +
_
q
max
G
_
tcto|

2
=
(
max
)1
_
1 +
_
q
max
G
_
1

2
+
(
max
)2
_
1 +
_
q
max
G
_
2

2
(3.42)
Evidently, the composite rate-time value of / = 0.5 can be achieved only if
_
q
max
G
_
1
=
_
q
max
G
_
2
, assuming that decline exponent, /, of each layer is equal to
0.5.
Carter Type Curves
Fetkovich (1980) type-curves were developed for a well producing under constant
pressure in an oil and gas reservoir. For a pressure drawdown that is moderate to
large, Fetkovich (1980) liquid ow curves are however not recommended for gas
production type-curve analysis. Carter (1985) developed type-curves for a gas
well production from a boundary reservoir. These type curves are theoretical,
and provide understanding and implicit guidelines to eld data analysis. Carter
also noticed that the changes in uid properties with pressure aect the reser-
voir performance predictions, especially the gas viscosity-compressibility prod-
uct, j
j
c
j
. In order to represent changes in j
j
c
j
during depletion and to measure
72 3. DEPLETION RATE DECLINE REVIEW
the magnitude of pressure drawdown on j
j
c
j
, he introduced the "dimensionless
drawdown", `, according to
` =
_
j
j
c
j
_
i
_
j
j
c
j
_
oj
(3.43)
` =
_
j
j
c
j
_
i
2
_
:(j
i
) :(j
&)
)
ji
Z
.

j
u]
Z
u]
_
(3.44)
By introducing the magnitude of the pressure drawdown in gas wells, `, Carter
(1985) presented gas type-curves. These curves are based on dimensionless para-
meters: the dimensionless time, t
1
; the dimensionless rate,
1
; the dimensionless
geometry parameter, j; the dimensionless radius, :
c1
, and the ow geometry;
dimensionless drawdown correlating parameter, `. The-type curves were gener-
ated with the radial gas simulation model. For the exponential decline, / = 0,
corresponding to ` = 1.0 and indicating a negligible drawdown eect, also the
gas decline was dened as exponential. An increasing magnitude of pressure
drawdown was dened with ` = 0.75 and ` = 0.55. Gas reserves are better
estimated with Carter type curves as those presented in Figure (3.10). This set
of curves is similar that of Fetkovich in the aspect of plotting scales, but are not
as straightforward and general as those of Fetkovich.
Chen and Teufel Type Curves
In 2000, Chen and Teufel extended the "Fetkovich" type-curves to linear/near-
linear-ow features that are important in tight-gas production data analysis.
They used Fetkovichs transient decline and Arps depletion decline by simul-
taneously considering Carters linear and radial single phase ow. The derived
solutions in "Laplace space" for a vertical well producing at constant pressure in
a closed drainage area were provided for linear ow and derived from the tem-
perature solution published by Carslaw and Jaeger (1959, p. 309) in the form
of

1
=
1
_
:
tanh(
_
:) (3.45)
The radial ow in "Laplace" form was presented by van Everdingen and Hurst
(1949, Eq. VII-4) as

1
=
1
_
:
1
1
(
_
:) 11(
_
:)
1
1
(v
cT
p
c)
1
1
(v
cT
p
c)
1
0
(
_
:) + 1
0
(
_
:)
1
1
(v
cT
p
c)
1
1
(v
cT
p
c)
(3.46)
Linear and radial ow are presented in Figures (3.11). Dimensionless log-log
type curves with linear and radial ow of a reservoir with :
c1
< 10 are plotted
3. DEPLETION RATE DECLINE REVIEW 73
Figure 3.10: Radial-linear gas reservoir type curves [After Carter (1985)].
74 3. DEPLETION RATE DECLINE REVIEW
Figure 3.11: The linear and the radial ow geometry [After Chen and Teufel
(2000)].
in Figure (3.13) with
1
and Q
1
there functions of o. (o is also dened as
the fraction of 2: radians that is open to ow). Chen and Teufel referred the
main diculty when constructing "Fetkovich" curves to dening dimensionless
plotting variables. Finding a proper set of dimensionless variables should give
rise to a unique curve during theoretical boundary-dominated ow period for
both linear and radial ow, and thus also for all types of ow in a closed system.
The simplied dimensionless set of Fetkovich curves (1980 and 1996) was found
to be inadequate for cases of small values of :
c1
. Carters dimensionless set
was used for a smooth transition from linear to radial ow and for a decent
convergence of boundary dominated ow. Further, Chen and Teufel dened the
dimensionless set, the ow rate,
o1
, the cumulative rate, Q
o1
, and the time,
t
o1
, according to Figure (3.12). Moreover the authors provided detailed analyses
of dimensionless parameters with an explanation of the coupling procedure.
Palacio and Blasingame Type Curves
For a varying bottomowing pressure, Blasingame et al. (1991) created an equiv-
alent rate liquid solution from liquid and gas constant pressure solutions. These
methods smooth data, thus improving the type-curve match. The depletion data
was forced to match the harmonic depletion stem instead to the Arps decline
exponent / < 1. This approach exclude the Fetkovich (1980) concept of the drive
3. DEPLETION RATE DECLINE REVIEW 75
Figure 3.12: The dimensionless rate,
1
, and the cumulative production, Q
1
,
versus the dimensionless time, t
1
[After Chen and Teufel (2000)].
mechanism, recovery eciency, and layered no-cross ow behaviour. For each
time period, the cumulative oil can be divided by an oil rate. Consequently,
in a single-layer system, it is possible to nd a material balance time function,
t
cj
. Both constant pressure and constant rate solutions during transient ow are
equivalent. Contrarily during depletion decline, constant rate solutions follow a
harmonic decline and constant pressure solutions follow an exponential decline.
This method requires that the drawdown-normalised rate be plotted versus the
material balance time function, t
cj
. As constant pressure data are replotted with
the time, t
cj
, and compared to the constant rate solution, they become equiva-
lent and overlay the harmonic stem. Palacio and Blasingame (1993) represented
Carters curves in terms of Fetkovich plotting variables, including the issues of
changing uid properties and operating conditions.
Aqarwal et al. (1999) commented that a constant rate solution takes ad-
vantage of pressure transient analysis techniques for plotting decline curve data.
Moreover type curves utilise plots of pressure, rate, cumulative rates, time and
derivatives for result verication.
Mattar and Anderson Type Curves
Mattar and Anderson (2003) applied the concept of a normalised rate and a
material balance pseudo-time to create a simple linear plot, which can be ex-
trapolated to the uid in place. The method is similar to that of Palacio and
76 3. DEPLETION RATE DECLINE REVIEW
Figure 3.13: The composite type-curves: (A) The ow rate vs. time; (B) The
cumulative production vs. time; (C) The ow rate vs. the cumulative production
[After Chen and Teufel (2000)].
3. DEPLETION RATE DECLINE REVIEW 77
Blasingame regarding the use of available production data. The ow system of a
depletion drive gas reservoir under peeudosteady-state conditions was described
by

:(ji) :(jn,)
=

:(j)
= (
1
G/
0
jcc
)Q` +
1
/
jcc
0
(3.47)
where, Q
.
, is the normalised cumulative production
Q
.
=
2
i
j
i
t
o
(ctj
i
2
i
) :(j)
and t
o
is the Blasingame normalised material balance pseudo-time
t
o
=
_
j
q
c
q
_
i

i
2iG
2j
i
[:(ji) :(j)]
Furthermore, /
jcc
0 was dened as the inverse productivity index (j:i
2
,cj
``:c,) according to
/
jcc
0 =
1.41710
6
1
//
_
ln(
:
c
:
&o
)
3
4
_
(3.48)
Ansah-Knowles-Blasingame Type Curves
Ansah et al. (2000) noticed that, during depletion, a signicant change in gas
properties aect the reservoir characteristics. This change is due to a variation
in the gas viscosity-compressibility product, j
j
c
j
, with pressure.
The gas property changes were described by material balance equation in
dimensionless form
j
1
= (1 G
j1
) (3.49)
which was derived from
j
2
=
j
2
(1
Gj
G
)
where, j
1
=
jZ
j
.
Z
.
and G
j1
= G
j
,G
The authors proposed that j
j
c
t
be expressed in a the form of dimensionless
ratio
(
j

c
I)
.
(
j

c
I)
. Here, c
ti
is the total system compressibility at j
i
(j:i
1
), and j
i
is the gas viscosity (cj) at j
i
. Further, they stated the dimensionless ratio as a
function of dimensionless pressure according to Table (3-2).
The type-curves by Ansah et al., are given as a set of dimensionless variables,

1o
, t
1o
, :
c1
, and the correlation parameter is a function of the dimensionless
pressure, ,, as presented in Figures (3.15, 3.16, and 3.17).
78 3. DEPLETION RATE DECLINE REVIEW
Figure 3.14: The distribution of the viscosity-compressibility function [After
Ansah et al. 2000].
Table 3-2: The dimensionless ratio as a function of dimensionless pressure as
dened by Anash et al. (2000)
dimensionless ration
vs. dimensionless pressure
Pressure ranges
First order polynomial
j
.
c
I.
jc
I
= j
1
ji < 5000
Exponential model
j
.
c
I.
jc
I
= ,
0
c
(o
1
j
T
)
ji 8000
General polynomial model
j
.
c
I.
jc
I
= c
0
+ c
1
j
1
+c
2
j
2
1
+ c
3
j
3
1
+ c
4
j
4
1
c:
3. DEPLETION RATE DECLINE REVIEW 79
Figure 3.15: The "rst-order" polynomial solution for real-gas ow under
boundary-dominated ow conditions. A viscosity-permeability, jc
t
, is linear
with dimensionless pressure, j
1
[After Ansah 2000].
Figure 3.16: The "exponential" solutions for real-gas ow under boundary-
dominated ow conditions [After Ansah (2000)].
80 3. DEPLETION RATE DECLINE REVIEW
Figure 3.17: "General polynomial" solution for real-gas ow under boundary-
dominated boundary conditions [after Ansah 2000)].
Analysis of Production Data The paper by Mattar and Anderson (2003)
provides a comprehensive presentation of the methods developed by Arps, Fetkovich,
Blasingame, and Agarwal-Gardner, as well as a new method named the Flow-
ing Material Balance. Some methods yield recoverable reserves, while others
give hydrocarbons in place. Traditional (Arps) decline analysis (exponential or
hyperbolic) can underestimate or overestimate the reserves as it ignores the ow-
ing pressure data. Nonetheless, it gives reasonable answers in many situations.
With new developed electronic data measurements both owing pressure and
ow rate are readily available and thus more sophisticated techniques of data
interpretation are needed in data interpretation. Each method has its strengths
and limitations.
Arps methodology is simple and does not require knowledge of reservoir or
well parameters. It uses an empirical curve match to predict the future per-
formance of the well. It applies to production through any type of reservoir or
reservoir drive mechanism. A practical guidelines has been assembled through
extensive eld analysis which suggests what curves belong to which type curves
through Fetkovich (1980). Arps decline analysis is not able to include changes in
operating constraints as it inherently assume that historical operating conditions
remain constant in the future. It is not applicable to transient ow conditions.
Thus, predicting ultimate recovery with Arps curves has very limited application.
Fetkovich (1980) was the rst to use type curves to analysis of production
data. Type curves combine depletion stems describing boundary dominated ow
with constant pressure type curves that originated by Van Everdingen and Hurst
for transient type of ow. This type curves valuable feature lies in diagnostic
and less in analysis of production data. Matching production data with type
3. DEPLETION RATE DECLINE REVIEW 81
curves it is possible to dene transient and depletion decline. Method calculates
expected ultimate recovery and is constrained to existing operating conditions.
The transient part of type curves assumes constant bottomhole owing pressure
what is a limitation. Usually when well is rate restricted approach does not apply.
This technique can quantify hydrocarbon-in-place by using recovery factor only.
Blasingame et al. (1989) and Agarwal et al. (1999) methods are similar to
Fetkovich, as being used for production data analysis. The main dierence is
that the these new methods incorporate the owing pressure data along with
production rates. In addition these methods use analytical solutions to calculate
hydrocarbon-in-place meaning that expected recoverable reserves can be quan-
tied independently of production constraints The main two features are in
normalizing of rates using owing pressure drop and handling changing com-
pressibility of gas with pressure Mattar and Anderson (2003) explained how
pseudo time works.
Moreover, instead of type curves there are modern analytical methods as
Flowing Material Balance. For a reservoir under volumetric depletion by using
production rate and owing pressure data it is possible to calculate hydrocarbon-
in-place. As these methods are analytical it comprises simplications about the
reservoir and production data. Mostly method assume single phase ow and
account for interference eects and a volumetric reservoir. The non volumetric
eects such as water-drive and interference among multiple wells can be handled
eectively using inuence functions. As an example, Blasingame type curves
have a multiple-well feature that can accommodate and account for interfer-
ence eects. Single phase ow is considered valid especially for gas wells. For
multiphase ow pressure loss from surface to bottomhole conditions should be
considered.
3.3.2 Vertical Fractured Well
Special type-curves designed for hydraulically fractured wells have also been
proposed for various degree of complexity, e.g., planar/elliptical fracture with
innite/nite conductivity partially/fully penetrated in an innite/closed reser-
voir. Thus many type curves have been developed. In 1975, Locke and Sawyer
generated a constant pressure type curve numerically that included boundary
eects and an innite conductivity fracture. The constant pressure solution for
a nite reservoir exhibits an exponential decline for large value of time. To rep-
resent the change in owrate they used a stepwise method. Further, an integral
was developed for converting the constant rate solution to constant pressure.
In 1979, Agarwal et al., determined dimensionless rates for constant terminal
pressure from wells with nite-conductivity fractures. Type-curves were intro-
duced at early times by To simulate the eect of a nite conductivity vertical
fracture on ow behaviour they developed a nite-dierence model. Vertical
fractures were assumed to have high storage capacity. In 1985, Fetkovich et al.
82 3. DEPLETION RATE DECLINE REVIEW
Table 3-3: The vertical well, the vertical and the horizontal fracture
Vertical Fractured Well
Vertical Horizontal
Plane is perpendicular
to earths surface
due to overburden stress
being too great to overcome.
Plane is parallel to the
earth surface, and is usually
associated with shallow wells
of less than 3000 ft (914 m) depth.
Fracture gradient < 0.8 j:i,,t Fracture gradient 1.0 j:i,,t
demonstrated that lowpermeability hydraulically fractured wells can be analysed
using the transient radial ow solution with very little dierence in results from
the same data matched to an innite conductivity fracture type curve. In 1986
Fraim developed semi-analytical type curves for history matching hydraulically
fractured wells. In 1996 Cox et al. presented new hydraulically fractured type
curves that comprised the eect of reservoir geometry, skin eect, and varying
compressibility.
Agarwal et al. (1999) used the constant rate solution in combination with
transformation Blasingame applied before to create rate-time, and cumulative-
time plots along with their derivative plots. Type curves were generated for
radial systems and vertically fractured nite and innite conductivity system.
Fetkovich et al. (2006) found that in same case the derivative may be noisy.
Production type curves that assume a constant bottom owing pressure have
been developed and used to predict production.
Vertically fractured well can exhibit ve distinct ow regimes: fracture linear,
bilinear, formation linear, pseudo-radial and boundary dominated ow. Flow
regimes are separated by the transition periods. During linear ow uid in the
fracture expands and enters the wellbore with the following characteristics: it
ends quickly; not usually seen in production data; fracture uid cleanup occur-
ring; an remark is that it is not useful for analysis. During the bilinear ow the
following occurs: uid ows linearly from the formation to fracture and from
fracture to wellbore; the lower C
)1
the longer is bilinear ow; the formation ow
is compressible and fracture ow is incompressible;and the fracture tip eects
could not yet be seen at the well. During formation linear ow the pressure drop
in fracture is negligible compared to other factors driving the system by means
that the higher is C
)1
the longer is formation linear ow. Further, during the
pseudo radial ow, after bilinear or formation linear ow, it is independent of
C
)1
;the higher the C
)1
the larger t
1
before the pseudo radial ow is reached; it
can be approximated by the well large :
&o
; this ow may not be exhibited due
to the boundary eects.
3. DEPLETION RATE DECLINE REVIEW 83
Figure 3.18: The vertical well, the vertical and the horizontal fracture.
Figure 3.19: The vertical fractured well in a rectangular drainage area [After
Chen et al. (1991)].
84 3. DEPLETION RATE DECLINE REVIEW
Figure 3.20: Type of ow for a vertcal fractured well
3.3.3 Horizontal Well
Several theoretical studies of inow performance of a horizontal well are made
long time ago by Slichter (1897) in 2 D space, and further Kozeny (1933) and
Muskat (1937) studies in 3 D space.
Decline curves for horizontal wells under various boundary conditions have
been developed using the Greens and source functions. These decline curves
can be used to estimate the production forecast and the ultimate recovery of
a horizontal well. Poon (1991) studied inuence of wellbore stimulation, well
spacing and length on eectiveness of a horizontal well. Most of equations for
predicting the production rates of a horizontal well are based on the assumption
of stady-state ow conditions. These equations have been developed using the
Greens and source functions.
The most widely used methods are the steady-state-eqations published by
Merkulov (1958), Borisov (1964), Giger (1983), Karcher et al. (1986), Renard
and Dupuy (1990) and Joshi (1986 and 1991). These equations require an es-
timation of the horizontal well drainage radius which is not known until the
well has been on production and well tests were conducted. Also, steady state
production rate cannot be used to estimate the ultimate recovery of the well.
Recently Michelevichius and Zolotukhin (2002) provided an alternative approach
to the productivity evaluation of a horizontal well. The approach was based on
the idea of representing a well as a chain of spheres and on averaging technique
derived by Muskat.
However, several analytical horizontal well analytical model have been devel-
3. DEPLETION RATE DECLINE REVIEW 85
Figure 3.21: The dimensionless rate, qD versus the dimensionless time, tDXf for
the horizontal well [After Cox et al. (1996)].
86 3. DEPLETION RATE DECLINE REVIEW
oped which do not require the assumption of the drainage radius meaning that
there are not based on the steady-state Darcy equation. Babu and Odeh (1988)
presented an equation for calculating the productivity of a horizontal well in a
bounded reservoir during pseudosteady-state ow. The rst is the geometric fac-
tor which accounts for the eect of anisortopic permeability, the well location and
the drainage area. The second is the skin factor which accounted for the eect of
well length. Goode and Kuchuk (1991) developed an equation for predicting the
inow performance of a horizontal well in a bounded rectangular reservoir being
under constant pressure outer boundary. Duda and Aminian (1989) and Chang
et al. (1989) developed type curves for predicting the cumulative production of
a horizontal well with numerical simulation. Poon (1991) presented the decline
curve development. He used the analytical model solutions for a single well de-
veloped by Clonts and Ramey.(1986) and implemented the Duhamels theorem
introduced by van Everdingen and Hurst (1949). For a single horizontal well
in a bounded rectangular reservoir the Greens function can be integrated with
respect to time and "Laplace" transform was taken directly, thus providing an
exact solution. However, approximate solution method should be considered
to avoid diculties in applying "Laplace" transform. Thus, de Carvalho and
Rosa (1988) suggested the use of the short and long time approximations to the
Greens function to evaluate the "Laplace" transform of the pressure function.
Further, Cox (1979) introduced another approximate solution method that t
a simple correlation equation through the transient pressure data. He provided
dimensionless production rate versus time for a wellbore producing under the
constant pressure conditions. Poon plotted the dimensionless rate,
1
, versus
dimensionless cumulative production, `
j1
, for a single horizontal well located
in a bounded reservoir
Locke and Sawyer (1975) developed a constant pressure type curve for a in-
nite conductivity fracture with a numerical simulator that included boundary
eect. Agarwal et al. (1979) considered well with nite-conductivity vertical
fracture. Further, Fetkovich et al. (1985) used the transient radial ow solution
to analyse rates from low permeability hydraulically fractured well. Fraim et
al. (1986) developed semi-analytical type curves to history match hydraulically
fractured wells. Cox et al. (1996) presented new hydraulically fractured type
curves that included the eect of reservoir geometry, skin eect and varying com-
pressibility. Chen and Teufel (2000) extended Fetkovich type curves by including
the linear ow regime.
3.3.4 Horizontal Fractured Well
It is known that horizontal wells have been successful in naturally fractured
reservoirs and in reservoirs with gas and water coning problems. However, there
are situations where a fractured horizontal well is preferable. Thus, the fractur-
ing of a horizontal well must be considered before the well is completed. Many
3. DEPLETION RATE DECLINE REVIEW 87
Figure 3.22: Decline curve for a horizontal well ina bounded reservoir [After
Poon (1991)].
Figure 3.23: The eect of the aspect ratio on horizontal well productivity (the
ratio of the length to the width of a rectangular well pattern) [After Poon (1991)].
88 3. DEPLETION RATE DECLINE REVIEW
authors studied the productivity aspects of fractured horizontal wells. One im-
portant aspect of a positioning of a horizontal well is the determination of the
stress eld about the proposed well. Once, principal stress is known it is possible
to create transversal fractures or longitudinal fractures as presented in Figure
(3.24). In 2006 Demarchos et al., discussed the operational challenges of a frac-
turing project and provided recommendations for the successful treatment of a
transversally fractured horizontal well. Wei and Economides (2005) had studied
the performance of horizontal wells with multiple transversal fractures. Further,
it was found that at depth where exists producing formation, the stress elds
leads to a hydraulic fracture that is vertical and normal to the minimum horizon-
tal stress. Thus, the fracture direction and azimuth aect the well orientation.
Study by Villegas et al. (1996) concluded that with equal fracture length and
conductivity, the performances of a fractured vertical well and a longitudinally
fractured horizontal well are almost identical. It was found that almost all of the
reported applications of fracturing of horizontal wells are transversal fractures.
The fracturing of horizontal wells has been mostly considered in the United
States and the North Sea. Fractured horizontal gas wells were also considered
in Germany and Australia.
The increased productivity of multiple transversal fractured horizontal well
has been studied by several authors. as Giger (1985), Karcher et al. (1986),
Mukherje and Economides (1988). Soliman (1990) was among rst to study
the behaviour of a horizontal well with a single fracture. By disregarding the
presence of the well they created the simple model that was a solution for a nite
connectivity disk in a one dimensional innite slab. In 1989 van Kruijsdijk and
Dullart provided a boundary element solutions of the transient pressure responses
of multiply fractured horizontal well. Larsen and Hegre (1991) examined well
performance during the linear-radial ow regime. Further, Roberts et al., (1991)
investigated a well with fractures pressure responses and major ow regimes. In
1993 Guo and Evans provided a two-dimensional solution for a horizontal well
with multiple fractures. Raghavan et al. (1994) provided model solutions with
a comprehensive report of understanding of the performance of multi-fractured
horizontal wells. Horizontal wells with multiple fractures were further studied
by Guo and Evans (1994), Kuchuk and Kadar (1994).
An extensive analytical work has been done to investigate pressure-transient
analysis and short and long term productivity of horizontal well with single
or multiple hydraulic fractures by Soliman et al. (1990), Larsen and Hegre
(1994), and Kuchuk and Hubusky (1994). Moreover, Horne and Temang (1995),
Raghavan et al. (1997), Soliman (1998), further investigated the eect of number
of transversal fractures on a well productivity.
In 1997 Guo and Schechter presented a simple mathematical model for esti-
mating productivity of a vertical and horizontal wells intersecting long fractures.
They observed that the rapid decline in wellbore productivity was mainly at-
3. DEPLETION RATE DECLINE REVIEW 89
tributed to stress-sensitive fracture conductivity. Their study provided overview
on several analytical solutions for transient ow in fractured reservoirs and se-
lected numerical models developed for simulating uid ow in fractured reser-
voirs. The review comprised Economodes et al. (1991) general simulation scheme
that can handle horizontal wells. Hydraulic fractures that penetrate the well in
both the transversal and longitudinal directions were eectively simulated, and
predicted performance was in excellent agreement with analytical solutions.
In 1999 Wan and Aziz derived analytical solution for the well pressure that
can be combined with a numerically computed gridblock pressure to obtain the
well pressure(WI). They presented an overview of existing methods for hydraulic
fractures modelling. They used in the study the modifying the eective wellbore
radius method. The rening the fracture grid method; and modifying the frac-
ture transmissibilities method were only reviewed.
Al-Kobaisi et al. (2006) provided a hybrid, numerical-analytical model for
the pressure-transient response of a nite-conductivity fracture intercepted by a
horizontal well.
In addition Mederios (2006 and 2007) presented semi-analytical solutions of
fractured horizontal wells with transverse and longitudinal fractures in hetero-
geneous, tight gas formations.
3.3.5 Multilateral Well
Ozkan et al.(1998) presented computational methods with solutions for dual
lateral wells in homogeneous formations. In 1998 Larsen derived solutions for
pressure-transient behaviour of multibranched wells in layered reservoirs. The
computational methods was based on Laplace transforms and numerical inver-
sion to generate type curves for use in direct analyses of pressure-transient data.
The approach can handle any number of branches with arbitrary direction and
deviation. Earlier in 1997 Larsen provided solutions for deviated wells in layered
reservoirs. The results applied for any deviation, and hence, also for horizon-
tal segments within dierent layers. The approach was restricted, however, to
cover at most one segment within each layer with no overlap vertically. The
approach cannot handle the boundary condition at the wellbore for nonvertical
segments, thus, each perforated layer segment has to be replaced by a uniform-
ux fracture. This approach is illustrated in Figure (3.25) for a deviated well in
a three-layered reservoir. Further, in 1998 Larsen investigated productivity of
fractured and non-fractured deviated well in commingled layered reservoir Direct
analytical methods are introduced to determine productivity indices of fractured
and nonfractured deviated wells in commingled layered reservoirs, including cases
with horizontal wells. For non-fractured wells, the total productivity index (PI)
can be obtained by adding the individual layer PIs. For a fractured wells, a
more direct approach covering all well and fracture elements was needed if at
least one of the fractures penetrates more than one additional layer above or
90 3. DEPLETION RATE DECLINE REVIEW
Figure 3.24: The fracture orientation along a horizontal well.
3. DEPLETION RATE DECLINE REVIEW 91
Figure 3.25: A well in a three layered reservoir with perforated segments replaced
by uniform-ux fractures.
Figure 3.26: The multibranch and multiple-fracture congurations for horizontal
wells [After Economides at al. (2001)].
below the wellbore layer.
In 2001 Economides et al., presented advances of the complex well-fracture
congurations. A rather sophisticated conceptual conguration involved the
combination of multiple-fractured vertical branches from a horizontal "mother"
well drilled above the producing formation in a non-producing interval. The
simplify perforation strategy of a vertical section over a highly deviated or hori-
zontal section makes multibranch well with vertical branches more advantageous
compared to a horizontal well with multiple transverse fractures as presented in
Figure (3.26). However, a complex well design procedures need advance mod-
elling tool for better understanding of a complex fractured well performances.
Cvetkovic et al. (2007) numerically investigated a complex well with a com-
plex lateral geometry. Numerical simulation were performed with synthetic reser-
voir and uid data. A producing well was positioned vertically and horizontally
92 3. DEPLETION RATE DECLINE REVIEW
Figure 3.27: The multilateral well types [After Louis J. Durlofsky TAML, 1999
presentation].
3. DEPLETION RATE DECLINE REVIEW 93
Figure 3.28: A vertical and horizontal well with laterals positioning within an
oil reservoir [After Cvetkovic et al., (2007)].
with over 100 laterals. Liu (2009) presented an overview of multilateral wells
that become a standard IOR practice to enhance hydrocarbon recovery in both
oil and gas reservoirs. Accurately forecasting the performance of multilateral
wells is challenging particularly in a complex reservoir such as highly faulted or
naturally fractured reservoirs. Reservoir simulation is considered as a reliable
and economic method to asses the benet of multilateral wells in terms of in-
creased oil production and improved sweep eciency. Technological advances in
measurement and geological modelling provide a detailed description of the reser-
voir, especially in the vicinity of the wells, thus accurate well models are essential
for reservoir and production engineering applications. Recently Karimi-Fard et
al. (2009) presented an overview of dierent numerical techniques developed to
study the well productivity in complex situations including fractured reservoir.
3.3.6 Multi Wells
Rodriquez and Cinco-Ley (1993) developed a model for production decline in a
bounded multi-well system. The primary assumptions in their model are that
the pseudosteady-state ow conditions exists at all points in the reservoir, and
that all wells produce at a constant bottomhole pressure. They concluded that
the production performance of the reservoir was shown to be exponential in all
cases , as long as the bottomhole pressures in individual wells are maintained
94 3. DEPLETION RATE DECLINE REVIEW
constant. Later in 1996, Camacho and Galindo-Nava, improved the Rodriquez
and Cinco-Ley model by allowing individual wells to produce at dierent times.
Moreover, Camacho et al., also assumed the existence of the pseudosteady-state
condition and that all wells produce at constant bottomhole pressures.
Valko at al. (2000) presented a concept for an arbitrary number of wells in a
bounded reservoir system named as "muti-well productivity index". These au-
thors also assumed the existence of pseudosteady-state ow, but proved that the
concept was valid for constant rate, constant pressure, or variable rate/variable
pressure production. Marhaendrajana and Blasingame (2001) developed a gen-
eral multi well solution that was accurate and provided mechanism for the analy-
sis of production data from a single well in a muti-well reservoir system. The
methodology was applicable for both oil and gas reservoirs. Their approach used
the single well decline type curve (i.e., Fetkovich-McCray type curve) coupled
with the appropriate data transforms for the multi-well reservoir system. Fur-
ther, method includes a "total material balance time" plotting function that
comprised the performance from all of the producing wells in the multi well
reservoir system. Method was applicable to estimate ow capacity (i.e., per-
meability), the original uid in place. Method applied for homogeneous and
heterogeneous reservoir systems.
In 2007 Gilchrist et al., presented novel semi-analytical solutions to the lay-
ered reservoir produced with multiple wells. Applying a method of integral
transform they derived an analytical solution within each layer. Solutions were
applicable to partially penetrating vertical, horizontal, deviated and fractured
well taking into account superposition eects in multi-well and multi-rate sce-
narios. Further, they derived solutions for innite conductivity fracture and
a nite conductivity fracture with non-Darcy ow. Inner boundary conditions
were both, constant pressure and rate for the overall multi-well scenario. All
published solutions were related to the interpretation of generalized multiple
layer, multiple well problems in single phase hydrocarbon reservoirs as presented
in Figure (3.29). Selected derived expressions were taken from the Thamby-
nayagams work that is internal and unfortunately not available. According to
authors, Thambynayagam provided practical and elegant solutions to a variety
of congurations by the use of successive integral transforms. In an earlier paper
Busswell et al., (2006) presented the "Laplace space" solutions derived from a
generalised single layer analytical model that handle multi vertical and horizon-
tal wells. In order to avoid problems in converting "Laplace space solutions" to a
real domain, mainly caused by the Stehfest inversion algorithm and its handling
a discontinuous nature of rate history, they chose the Chen and Raghavan (1996)
approach. They implemented Chen and Raghavan approach in order to handle
discontinuities of the Stehfest algorithm.
3. DEPLETION RATE DECLINE REVIEW 95
Figure 3.29: The multiple vertical, horizontal and deviated completioned wells
in the layered reservoir [After Gilchrist et al. (2007)].
96 3. DEPLETION RATE DECLINE REVIEW
3.4 Decline Curve Analysis Physics
The following constitutes a review of Arps gas decline analysis. Arps equation
continues to apply as an empirical relation in both history match and production
forecast. The initial rate,
i
, initial decline rate, 1
i
, and decline exponent, /, are
constants that dene the rate-vesus-time relation. An exponential decline gives
rise to a decline exponent / = 0 , and a decline rate 1 that is constant with
time, or 1 =
1
q
(
oq
ot
). The plot of such a rate logarithm versus cumulative gas
production corresponds to a straight line. A hyperbolic decline is obtained with
/ between 0 and 1, and a decline rate 1 that decreases constantly with time.
A plot of the rate logarithm versus the cumulative gas production (and time)
appear upward concave. Moreover, the reserves calculated for the same initial
decline rate vary. Consequently, the exponentially calculated reserves are more
conservative as compared to those calculated hyperbolically.
The choice of the decline exponent, /, inuences the estimates of reserves,
and economic evaluations of well production (duration of a well production and
well rate). We presume that the empirical hyperbolic equation (3.50) is valid
only for the boundary-dominated ow and when the wells owing pressure is
constant.
=

i
(1 +/1
i
t)
1
l
(3.50)
Fetkovich type-curves combine transient mono-phase solutions with the em-
pirical boundary-dominated stems of the Arps equation. The Fetkovich type-
curves allow the entire transient-decline data set analysis with limitations in-
volving the fact that a well produces under a constant owing pressure. Recent
work introduced the variable owing pressure analysis. The theoretical exponen-
tial (mono-phase) depletion decline and the empirical depletion (multi-phase)
decline with an exponent / = 0 can be superimposed. Fetkovich et al. (1996)
investigated the exponential decline / and the drive mechanism relation, and
found the volumetric depletion driving force to be inuenced by the total system
compressibility. In a gas system (contrarily to a single-phase liquid system), the
compressibility varies approximately as the inverse of the average reservoir pres-
sure and is not constant. As a result, the / value of the gas system is larger than
0. Among factors that aect gas rate-decline can be mentioned: wellbore ow-
ing pressure, turbulence and multiple no crossow layers. Okuszko et al. (2008)
investigated the eect of various parameters on gas decline by means of reservoir
simulation. It was found that the / value depended on the magnitude of the
owing pressure. For a low drawdown (j
i
j
&)
) or higher backpressure, j
&)
, the
exponent / approached exponential decline. As the backpressure, j
&)
,decreased
the decline exponent, /, increased from the exponential to the hyperbolic type.
In a near-wellbore region the wellbore turbulence, :, was found to also aect
the decline exponent, /. In the backpressure equation (3.51), the turbulence
3. DEPLETION RATE DECLINE REVIEW 97
factor, :, relates to the degree of turbulence. As a consequence, the laminar
ow represents a value of : = 1, and the turbulence ow represents a value of
: = 0.5.
= C(j
2
1
j
2
&)
)
a
(3.51)
The second relationship is between the decline exponent, /, and the tur-
bulence factor, :, was made by Fetkovich et. al. (1996). They coupled the
backpressure equation (3.51) with the material balance equation to obtain the
expression (3.52), valuable for conditions of a very low owing pressure. With
an increase in turbulence, or for a decreasing :, / approaches zero.
/ =
2: 1
2:
(3.52)
In a single layer gas reservoir the decline exponent, /, is between 0 and 0.5. If
uid properties are constant during depletion (a liquid-like uid behaviour), / is
equal to zero. If uid properties changes signicantly during depletion, decline
exponent, /, approaches a value of 0.5. The change in uid properties is more
signicant at lower reservoir pressures. Moreover, that exists a proportionality
between decline exponent / and reservoir depletion. Both drawdown and tur-
bulence aect the pressure depletion and the decline exponent, /. The reservoir
pressure is reduced with a high draw-down, which is followed by higher uid
property changes and a higher value of the decline exponent, /.
Fetkovich et. al. (1996) declared that the decline exponent can be as high
as one provided that the gas reservoir is layered. In a tight gas reservoir, it
is possible to obtain exponent / higher than one. This seems to signify that
the well produces partly in transient mode instead of exclusively in a boundary
dominated mode.
The decline exponent, /, may change at the end of production. Okuszko et
al. (2008) found that deviation from a constant decline exponent, /,occurs after
production of 90% of the expected reserves. For a constant value of the decline
exponent, /, they noticed only a marginal overestimation of reserves of ca. 5%.
As the decline exponent decreases, the late-time eect on gas reserve estimates
is below 5%. For this reason, a constant decline exponent, /, represents a good
approximation.
Drive Mechanisms Decline curve analysis is used for interpreting rate-time
data and to predict the future performance of a well or reservoir. The production
trend in a well or eld is a reection of the characteristics of the formation, the
uid in the formation, the well, the mechanism by which the uid is driven
into the well and the mechanism by which it is lifted to the surface. If these
characteristics remain unchanged, the past trend will continue into the future.
98 3. DEPLETION RATE DECLINE REVIEW
The following presents certain comments on reservoirs with various drive
mechanisms. The oil in an oil reservoir is not ordinarily produced by only one
or two of the principal oil-recovery mechanisms. Gas-cap and edge-water drives
may function simultaneously. This can take place in the early oil-productive
reservoir. Gravity drainage might predominate as the oil-recovery mechanism
in the later oil-productive life of the reservoir. Lefkovits and Matthews (1958)
studied gravity-drainage reservoirs and found a decline exponent / of 0.5. A
high-pressure oil-reservoir can, in its early life, have unlimited edge-water drive,
assuming that it is combined with gravity drainage as the eld is developed,
especially if the oil productive strata are thick and dip steeply.
Gravity drainage is an ineective oil-recovery mechanism in a thin and at-
lying reservoir. If such reservoir-rock is lled with a high-pressure gas-saturated
oil as well as immobile interstitial water, and if no natural gas-cap energy or
natural water-drive energy is available, the solution-gas-drive mechanism would
be the primary agent for recovering the oil. Some of the gas that is originally
in solution in the oil is released when the uid pressure is reduced below that
of the gas-saturated oil. This gas expands, thus displacing liquid oil into the
well. As the process continues, the uid pressure becomes reduced at increasing
distances from the well. Moreover, the fraction of pore space of the oil reservoir
rock occupied by gas increases and the oil fraction decreases. The volume ratio
of gas to oil experiences a rise.
The solution gas-drive has received the most attention in theoretical studies.
An exponential decline of / = 0 is obtained for : = 0.5. The harmonic decline is
not possible. The highest value of : given by Fetkovich was : = 1.0 which leads
to / = 1,3 or 0.333. According to Fetkovich, j
&)
= 0 is a realistic assumption for
a well on wide-open decline. Fetkovichs inow-performance curve equation or
IPR curve equation for : = 1.24 are approximately identical to Vogels reference
curve, and provide / = 0.43. This ts Arps nding that, for the majority
of decline curves, the range of / is 0 < / < 0.4. However, the result is in
disagreement with that of Ramsey and Guerrero (1969).
3.4.1 Solution Gas Drive Decline
Raghavan (1993) provided the rate,
0
(t), as

0
(t) =
d`
j
jj
=
c
t
c
`
t
c/
5.614
(3.53)
The decline rate, 1
i
,was expressed as
1
i
=
d ln
0
dt
t
2:0.006328/
1
2
_
ln
4
c

C
/
v
2
u
+ 2:
_
c
`
t
c
t
(3.54)
Moreover, the decline exponent, /, can be written
3. DEPLETION RATE DECLINE REVIEW 99
/ =
d
dt
_
1
o ln q
0
ot
_
t
1
2
_
ln
4
c

C
/
v
2
u
+ 2:
_
c
2:0.006328/
d
dt
_
c
t
`
t
_
(3.55)
which states that as long as the ratio of the average total compressibility and
mobility varies linearly with time, the decline exponent, /. is constant Both
the initial decline rate, 1
i
, and the decline exponent, /, depend on the relative
permeability and uid properties, and thus material-balance can be used to
study the variation of the mobility and compressibility. It is also indicated that
predictions of the future performance are strong functions of the uid properties,
the well conditions and well spacing. Fetkovich (1973, 1987) suggested that /
should be in the range 0.333 _ / _ 0.667, while Camacho and Raghavan (1989)
published that / was in the range 0.4 _ / _ 0.8. They noticed that responses
cut across several curves as a result of the total compressibility-mobility ratio
not being a linear function of time. They also introduced the idea of a variable
skin factor causing non-Darcy ow to yield a constant value of /.
Decline Exponent / Raghavan (1993) studied the character of exponent /,
and clearly stated conditions under which the decline exponent can be constant.
By assuming a xed drainage area, he stated that exponent / should vary with
time, i.e., it can not be constant for a well producing under IBCs of constant
pressure. He also mentioned that it would be possible for a well to produce with
the xed decline exponent, by assuming the existence of well skin incorporated
into solution. Moreover, he pointed out that this assumption was relevant to
the well producing from a xed drainage area where the distance to the no-ow
boundaries remained unchanged with time.
The following summarises the discussion of decline exponent /, based on a
modelling approach and analytically obtained expressions. The decline exponent,
/, is the measure of the change in loss ratio
/ =
d
dt
_

oq
ot
_
(3.56)
According to the above expression, the rst dierence of the loss ratio is constant,
signifying that the loss ratio is a linear function of time. The decline exponent,
1
i
, is, in turn, related to the loss ratio by
1
1i
=
_

oq
ot
_
i
(3.57)
By integrating and combining these two equations, it is possible to obtain the
dimensionless rate,
1o
, known as the hyperbolic decline, which describes the
implicit boundary dominated ow
100 3. DEPLETION RATE DECLINE REVIEW

o1
=

i
=
1
(1 +/1
i
t)
1
l
(3.58)
For an exponential decline

o1
=

i
= c
1
.
t
(3.59)
Decline curve analysis was considered as a convenient empirical procedure
until Fetkovich (1973) tried to attribute signicance to / and 1
i
. Fetkovich type
curves combined empirical Arps (1945) solutions to the analytical single phase
ow solutions. Following the Raghavan (1993) nomenclature, the dimensionless
ow rate
1
is

D
(t
AD
) =
1
a
c

2t
AD
a
c =
1
2
ln
_
4
c

:
2
&
_
+ o
Introducing the time, t
1,i
, and the corresponding rate,
1i
, that relates to
1
,
we obtain

o1
=

1

1i
=

i
= c

2r(I
/T
I
/T,.
)
a
(3.60)
1
i
=
2:c
2
/
ccc
t
j
(3.61)
t
o1
= 1
i
t =
2:c
2
/
ccc
t
j
t (3.62)
With an initial decline
i

i
=
2://
c
1
1j
(j
i
j
&)
)
_
1
2
ln
_
4
c

C
/
v
2
u
_
+ o
_ (3.63)

o1
(t
o1
) =
(t)

i
=
c
1
1j
2://(j
i
j
&)
)
_
1
2
ln
_
4
c

:
2
&
_
+ o
_
(t) (3.64)
For two owing phases, Raghavan (1993) provided an expression for the pore
volume, \
j
,
\
j
=
c
1
c
2
1j
(j
i
j
&)
)jc
t
_
t
t
o1
_
A
_

o1
_
A
_

o1

_
A
=
c
1
1jc
2://(j
i
j
&)
)
and
_
t
t
o1
_
A
=
ccc
t
j
2:c
2
/
3. DEPLETION RATE DECLINE REVIEW 101
Fetkovich (1980) related decline exponent, /, to the production mechanism
and to the exponent of the deliverability curve. The deliverability equation for
multiphase ow as derived by Fetkovich (1973), with : ranging from 0.5 _ : _ 1
can be written

0
= J
0
0
(j
2
j
2
&)
)
a
(3.65)
The starting production at initial conditions, i, and producing cumulative oil,
`
j
, that can be expressed with the material balance for the single phase ow
(under boundary dominated conditions) as
j j
i
=
j
i
`

i
`
j
Further, assuming j
&)
j and J
0
0
_ ,, it is possible to obtain the following
expression where exponent : varies between
1
2
_ : _ 1

0
= J
0
0i
j
j
i
j
2a
=

ci
_
1 + 2:
_
q
o.
.
.
_
t
_2n+1
2n2
(3.66)
The decline exponent / =
1
2
for : =
1
2
, and / =
2
3
for : = 1. For the multiphase
ow condition Fetkovich (1980) derived the following expression
j
2
j
2
i
=
j
2
i
`

i
`
j
(3.67)
and the derived rate can be expressed according to

0
=

ci
_
1 +
2a1
2
_
q
o.
.
.
_
t
_2n+1
2n2
Here the decline exponent / =
1
3
for : = 1 and / = 0 for : =
1
2
. Presuming
that reservoir boundaries inuence the rate response Fetkovich (1980) related /
to the drive mechanism as:
/ = 0 for undersaturated oil or gravity drainage without a free surface
/ =
1
2
for gravity drainage with a free surface
/ =
2
3
for a solution gas drive
During the transient rate, the decline exponent / 1. The dimensionless
rate and time according to Fetkovich (1973) can be expressed as

o1
(t
o1
) =
c
1
1j
2://(j
i
j
&)
)
_
ln
_
:
c
:
&
_

1
2
_
(t)
t
o1
=
2:c
2
/
cc
t
j
_
ln
_
v
c
v
u
_

1
2
_t
102 3. DEPLETION RATE DECLINE REVIEW
Figure 3.30: The dimensionless rate vs. the dimensionless time Fetkovich type
curves [After Fetkovich (1980)].
Fetkovich et al. (1986) explained that the term
_
ln
_
v
c
v
u
_

1
2
_
, as opposed to
_
ln
_
v
c
v
u
_

3
4
_
, better matched the depletion exponential decline stem of / = 0 on
the right-hand side of the combined type curves. The term also better correlated
the transient
v
c
v
u
stems on the left side of the type-curves in Figure (3.30). So,
the term
1
2
better match analytical transient
v
c
v
u
stems and at the same time
provided a better t to the exponential empirical decline stemof / = 0. Raghavan
commented in 1993 that it is important to realise the dierence and consider the
error of this matching approach by using type-curves.
According to Fetkovich (1973), a unique match of transient stems
v
c
v
u
is only
possible if the same set of data matches one of the depletion stems dened by
/. If a match is obtained only for the transient
v
c
v
u
stems, the decline exponent /
should be greater than 1.
As presented earlier, Camacho and Raghavan (1989b) provided a semi-analytical
explanation for matching / values in transient rate decline. Material balance
equations derived by Muskat (1945) with terms `
t
and c
t
are
do
0
dj
=
o
0
(j)
1
0
(j)
d1
0
dj
+
`
c
`
t
c
t
(3.68)
`
t
= `
c
+ `
j
=
_
/
vc
j
c
_
+
_
/
vj
j
j
_
(3.69)
3. DEPLETION RATE DECLINE REVIEW 103
c
t
=
o
0
1
0
(j)
d1
0
dj

o
j
1
j
(j)
d1
j
dj
+
o
0
1
j
(j)
1
0
(j)
d1
c
dj
(3.70)
where
_
I
r.
j
.
_
is the volume average of
_
I
r.
j
.
_
with i = o, q. Combining the above
equations with
d,
dt
=
c
3

0
(t)
c/
Raghavan (1993) expressed
_
c
t
`
t
_
/
v0
j
0
1
0
dj
dt
=
c
3

0
(t)
c/
(3.71)
and dened pseudo-pressure related to average pressure as an integral with a
reference pressure independent of time j
v
as
:(j) =
j(t)
_
j
r
/
v0
j
0
1
0
dj (3.72)
The above equation then becomes
_
c
t
`
t
_
d:(j)
dt
=
c
3

0
(t)
c/
(3.73)
The Constant Pressure Production Conditions After integration and
assuming a rate-normalised pseudopressure, :
1
(
w
t
1
), corresponding to the av-
erage pressure :(j), the following expression is obtained
:
1
(
w
t
1
) = 2:
w
t
1
:
1
(
w
t
1
) =
2://
c
1

0
(t)
[:
j
(j
i
) :
j
(j)] =
2://
c
1

0
(t)
j
.
_
j(t)
_
/
v0
j
0
1
0
_
dj
104 3. DEPLETION RATE DECLINE REVIEW
w
t
1
=
c
2
/
c
0
(t)
t
_
c

0
(t
0
)`
t
(t
0
)
c
t
(t
0
)
dt
0
The Constant Rate Production Conditions
:
1
(
=
t
1
) = 2:
=
t
1
:
1
(
=
t
1
) =
2://
c
1

0
(t)
[:
j
(j
i
) :
j
(j)] =
2://
c
1

0
(t)
j
.
_
j(t)
_
/
v0
j
0
1
0
_
dj
=
t
1
=
c
2
/
c
t
_
c
`
t
c
t
dt
Both inner boundary obtained expressions can be used under conditions changing
from a constant rate to a constant production. The gas drive is given by the
expression:
:
1
(
w
t
1
) = 2:
w
t
1
. (3.74)
During late-time and boundary-dominated ow, the obtained results dier for
liquid ow. For a well producing under constant pressure, a gas drive value for
:
1
(
w
t
1
) is obtained from
:
1
(
w
t
1
) = c
_
1 c
2r
=
I
/T
a
_
(3.75)
c =
1
2
ln
_
4
c

:
2
&
_
+ o
A single liquid dimensionless pressure drop j
1
=
20II
c
1
q(t)1j
(j
i
j) satises the
above expression.
2::
1
(
w
t
1
) = c
_
c
2r
=
I
/T
a
1
_
Expanding the exponential function c
2r
=
I
/T
a
leads to inequality
w
t
1

=
t
1
(3.76)
The derivative of the cumulative oil, `
j
, over the average pressure, j, is obtained
by dierentiating above equations with time
3. DEPLETION RATE DECLINE REVIEW 105
d`j
dj
= c
3
c
t
c
0
`
t
o/
This expression corresponds to the prediction performance as a function of rela-
tive permeabilities, uid properties and pore volume. Further, Raghavan (1993)
derived the following relation:
d ln
0
dt
-
2:c
2
/
cc
`
t
c
t
(3.77)
Here, if
A
I
c
I
is constant then log
0
versus time t can be presented as a straight
line. This suggests that exponential decline / = 0 solutions can be used for the
performance prediction if
=
t
1
is employed. Raghavan (1993) introduced decline
curve parameters as a function of reservoir properties. Based on the transient-
ow analytically derived equations, he obtained the following expressions for 1
i
and /:
1
i
=
_
2:c
2
/
cc
`
t
c
t
_
/ =
d
dt
_
1
o ln q
0
ot
_
=
cc
2:c
2
/
d
dt
_
c
t
`
t
_
With theses two expressions, derived from Arps (1945) exponential decline re-
lation, it is possible to comment on both rate data analyses and future rate
performances.
The Variation of
c
I
A
I
and Decline Exponent / For a solution gas drive,
Fetkovich (1980) dened / as
1
3
_ / _
2
3
whereas Raghavan (1993) dened it as
2
5
_ / _
4
5
. Since, / = co::tc:t for
c
I
A
I
corresponds to a linear function of time, the
decline exponent / is constant when uid properties, relative permeabilities and
the in place volume all combine into a linear function of time. Since, / ,= co::tc:t
for
c
I
A
I
is not a linear function of time, `
0
(j) `
0
(j
&)
) as time increases and
calculated rates should cross over several stems on type-curves or over several
values of the decline exponent /. This was also stated by Carter (1985) and
Fraim and Wattenbarger (1987).
Camacho and Raghavan (1989) considered the following: the possibility to
derive analytically the constant decline exponent, /; the wellbore pressure be-
haviour supporting the constant decline exponent, /.
They published that, for the single layer system; the decline exponent, /, was
not constant for a constant pressure production case. Thus, / can be obtained
under the following conditions: when a well produces under a variable pressure-
variable rate; if the wellbore pressure increases with time (while skin and drainage
106 3. DEPLETION RATE DECLINE REVIEW
area remain constant); if the expression c
o
_
c
I
A
I
_
ot
is constant during the constant
pressure production; and when the variable skin factor o exists in term c. For
a layered reservoir (both commingled or with interlayer communication), it is
possible to achieve the conditions yielding a constant decline exponent /.
The Use of Pseudo Time The following expression relates
0
(t) to the
pseudotime
=
t
1
and is derived from Equation (3.77)

0
(t) =
0i
c
_
2
=
I
/T
a
_
The expression is based on the following simplication
d ln
0
d
=
t
1
-
2:
c
Plots of
0
(t) versus pseudotime
=
t gives rise to a straight line. We now need
to compute the pseudotime
=
t. All above calculations are based on a simple
material balance equation of Muskat (1945). Raghavan (1993) suggested the use
of other simple material balance equations for studying the inuence of
_
c
I
A
I
_
on
parameters 1
i
and /.
Decline Curves and Transient ow If only transient rates are available and
no depletion rates are measured, the value of decline exponent, /, is greater then
unity. Camacho and Raghavan (1991) derived the following expression during
radial ow for skin o
o =
1
2
_
1
1
o ln q
0
o ln t
ln
4t
1
c

_
and then combined it with decline expression for / thus giving
/ = 2o + ln
_
4t
1
c

_
(3.78)
Raghavan (1993) stated that in Equation (3.78) during transient ow the decline
parameter / is a function of time t. For a decline exponent / greater then unity,
the following inequality is valid for large and small reservoirs as well as for large
and small values of the dimensionless time t
1
o
1
2
_
1 ln
4t
1
c

_
Thus, parameter / is in most cases greater than 1 provided that transient re-
sponses are used to predict the performance.
3. DEPLETION RATE DECLINE REVIEW 107
Figure 3.31: The rst decline on Fetkovichs type curve, for / 0 [After Padilla
and Camacho (2004)].
3.4.2 Solution Gas Drive and Gravity Drainage Decline
Padilla and Camacho (2004) examined the well and reservoir performance under
the combined eects of solution gas-drive and gravity drainage in homogeneous
system. The inuence of various parameters like oil rate, position of the pro-
ducing interval, wellbore pressure level, skin factor, and vertical permeability
were investigated by numerical simulation. Moreover, they found that during
the boundary dominated ow period, when gravitational forces are important,
the production decline presents two decline periods with a stabilisation period
in between. This stabilisation period depends on wellbore pressure, geometrical,
petrophysical and uid properties. Camacho and Raghavan (1989) and Cama-
cho (1987) showed that the production decline during the boundary dominated
ow under solution-gas-drive does not follow a xed decline curve Figure (3.31)
shows a match of several rate responses
corresponding to dierent values of wellbore pressure and three skin values
of Fetkovichs type curve. It is evident that the data points, that also include
gravity segregation forces, do not follow a xed decline curve. As a consequence
108 3. DEPLETION RATE DECLINE REVIEW
Figure 3.32: The rst decline on Fetkovichs type curve, for / = 0 [After Padilla
and Camacho (2004)].
of variation of
A
I
c
I
with time production decline under gravity segregation does
not follow the type curves of Fetkovich. It was also observed that the data
points do not follow the / = 0.5 curve as pointed out by Mathews and Lefkovits
(1956). During the rst decline period, when gravitational forces are important
the
A
I
c
I
function is approximately constant with time and the production decline
is exponential as presented in Figure (3.32). However, when a second decline
period is present the production decline is does not follow an exponential form
as given in Figure (3.33). This was in agreement with results of Gentry and
McCray (1978).
3. DEPLETION RATE DECLINE REVIEW 109
Figure 3.33: The second decline on Fetkovichs type curve, for / < 0. The decline
exponent is negative and constant [After Padilla and Camacho (2004)].
110 3. DEPLETION RATE DECLINE REVIEW
3.5 Analysis of Well Production data
3.5.1 Type Curves and Decline Curve Analysis
A Transient Radial Flow Regime Transient ow conditions start as a well
becomes to ow. Transient rate and pressure data are used to calculate the
permeability, thickness and skin. Once production from a well aects the entire
drainage area the well ows under pseudo-steady-state conditions also known
as boundary-dominated ow conditions. Pseudo-steady state data are used to
calculate the decline exponent / and to determine the corresponding original
oil in place. Transient and pseudo-steady state or boundary-dominated decline
behaviour are dierent. It is found that a value of the decline exponent greater
than one matches the transient rate-time data. Fetkovich et al. (2006) stated
that a / value greater than one is physically impossible.
It is possible to determine unknown parameters from dened reservoir eects
and uid characteristics. The future production can thus be calculated with-
out a prior production history. Such a method was implemented in a reservoir
with known physical quantities and composition. A numerical simulation of the
dened geology and reservoir showed that a uid system can change its initial
decline rate 1
i
. A deviation in the relative permeability has a greater eect
on decline exponent / than changing uid properties The initial production rate

i
(t) depends on the permeability of the formation and initial water saturation
and its magnitude depends on the uid characteristics. Reservoir heterogeneities
thus have a predictable eect on the production history.
Production history data can be plotted on a logarithmic scale versus time on
a linear scale and further extrapolated as a straight line into the future. This
extrapolation is denoted as a constant percentage decline or exponential decline
(/ = 0) under the estimated production. A hyperbolic decline with a better
reliability is used to describe future production trends. On the same plots of
semilog scale rate versus time, hyperbolic declines have been found to exhibit
concave upward behaviour. A technique used to t the production data to a
hyperbolic curve involves repetitive plotting of data points by trial and error
to obtain a straight line. Decline curve determination of the future production
can be used by a trial and error procedure, graphical methods, and dened
mathematical expressions.
Data displaying a concave upwards trend indicate transient ow, while data
presenting a concave downward bend indicate a pesudosteady-state ow. From-
early time data, it is possible to determine a dimensionless external radius and
to calculate permeability and skin,S. The two dimensionless plots of
1o
and t
1o
and the real data plot of q versus t are during matching shifted by the coecients
of rate, q in q
1o
, and time, t in t
1o
. Once matching
1o
, t
1o
, :
c1
and / has been
done it becomes possible to calculate the following reservoir variables: //, :
c
,:
&
,
:
c
and skin, o. Rate-time data is history matched on an appropriate log-log type
3. DEPLETION RATE DECLINE REVIEW 111
curve and then extrapolated to make a forecast.
OOIP was calculated from type curves by:
CC11 =
_
j
0
1
0
j
i
c
ti
(j
i
j
&)
)
__
t
t
1o
(t)

1o
_
notcI
The initial declining rate period can be considered as an extended drawdown
test. Early-time data on a rate-time type curve can be matched to obtain the
permeability, /. Depletion data exhibit a value of the decline exponent, / 0,
indicating changing values of
_
I
ro
j
0
1
0
_
and (j)
j
(c
t
)
j
, thus representing a reec-
tion of an increasing total compressibility with an increasing gas saturation, as
stated by Fetkovich et al. (2006). The permeability-thickness, //, is calcu-
lated with rate-time transient data using the dimensionless rate,
1
, versus. the
dimensionless time, t
1
, as in Figure (3.21), also known as Cox type curves:
// =
_
141.2j
0
1
0
(j
i
j
&)
)
__
(t)

1
_
notcI
The skin, o, is calculated from o = ln
v
ua
v
c
.
A Transient Linear Flow Regime Due to a low mobility (heavy oil), the
orientation of a horizontal well and the well length, the ow regime can last for
a long time until either a transient pseudo-radial or pseudo-steady state ow
begins. Pseudo-radial ow is not typically observed in low mobility horizontal
wells. Their ow usually undergoes a transition from being linear transient to
becoming boundary-dominated. The estimated time needed for a horizontal well
positioned in a centre of a square to reach pseudo-steady-state can be determined
with the time, t
jcc
, according to
t
jcc
=
379cj
0
c
t

/
Joshi compared a horizontal well to a controlled innite conductivity vertical
fracture of limited height. Well productivity of a horizontal well is a strong
function of the reservoir thickness, /, and the permeability ratio,
I
r
I
I
.
Pseudo-steady-state productivity indices, 11, were estimated by Babu and
Odeh (1989). They assumed a uniformux along a horizontal wellbore, described
by a set of simplied equations. Goode and Kuchuk (1991) took for granted a
uniform pressure that was represented by a more complicated innite series.
Rate Testing and Well Testing
Transient and Boundary-Dominated Flow Stems The coupling of the
transient and boundary-dominated ow stems may be accomplished in an em-
pirical manner, such as that used by Fetkovich et al. (1980, 1987), or with a
112 3. DEPLETION RATE DECLINE REVIEW
theoretical basis as that used by Doublet and Blasingame (l995a, 1995b) and
Shih and Blasingame (1995). The Fetkovich empirical approach was previously
addressed in this text. The present section provides, a theoretical basis for cou-
pling the transient and boundary-dominated ow production decline behaviours
into a composite production decline curve set. The production decline behaviour
of a well is governed by a number of variables, among which can be mentioned
the time level of interest and the specic properties of the reservoir and well
completion.
The transient behaviour of the well is governed by the intrinsic properties of
the reservoir and the well completion eciency. The eects of a nite drainage
areal extent of the reservoir do not aect the transient production decline be-
haviour of the well. Examples of the properties of the reservoir which govern the
early transient behaviour of the well include its eective permeability,.porosity,
uid saturations, and uid and rock properties. The eects of a dual perme-
ability or dual-porosity system are also factors in the early transient behaviour
of the well. Some of the well-completion eciency properties that aect the
early transient behaviour of the well are the systems characteristic length (L),
the fraction of the productive formation height that is open to ow to the well,
near-well stimulation or damage, and the specic completion design eciencies,
such as those caused by perforations and gravel-pack completions. Other factors
may also aect the early transient behaviour of a well, such as inertial and/or
multiphase ow in porous media that are ow-rate- or time-dependent.
The late-time boundary-dominated ow behaviour of a well is also gov-
erned by the previously addressed reservoir and completion properties controlling
the transient behaviour. However, the boundary-dominated ow behaviour of
the well is more predominantly governed by the extent and shape of reservoir
drainage area, the location of the well within that drainage area, and the types
of boundaries that exist along the perimeter of the drainage area of the well,
just as suggested by the name of the ow regime. During the fully developed
boundary-dominated ow regime, the eects of all boundaries of the reservoir
are exhibited in the production decline behaviour of the well. A unique pro-
duction decline-curve analysis of the historical production performance of a well
can actually only be obtained when at least some of the production performance
history spans at least a portion of both the transient and boundary-dominated
ow regimes.
Theoretical Basis for Coupling Stems The development of a set of compos-
ite production decline curves for evaluating the reservoir and completion prop-
erties using solutions of the rate-transient behaviour of a well in a nite closed
reservoir is actually quite simple and straightforward. Doublet and Blasingame
(1995) presented both pressure and rate-transient approaches for establishing the
necessary ordinate and abscissa scaling parameters required to obtain conjuga-
3. DEPLETION RATE DECLINE REVIEW 113
tion of the transient and boundary-dominated ow regime production decline be-
haviour stems. The rate-transient approach is more applicable and directly pro-
vides the required scale shift parameter values. The dimensionless rate-transient
behaviour of any well type (e.g., unfractured vertical, vertically fractured, or
horizontal) located in a closed nite reservoir, during the late-time fully devel-
oped boundary-dominated ow regime, can be generalised in the form given by
Equation (3.79).

&1
(t
1
) =
1

exp(
2:t
1

) (3.79)
The dimensionless time referenced to drainage area, t
1
, is dened in terms of
the dimensionless time, t
1
, and the dimensionless drainage area,
1
. The di-
mensionless superposition time is determined in a manner similar to that used
for the dimensionless material balance time, except that it relates the superposi-
tion time function of a variable ow rate history to its equivalent for a constant
drawdown pressure (i.e., inner-boundary condition) history.
t
1
(t) =
/
cjc
t
1
2
c
t
c
(t) (3.80)
The dimensionless time referenced to the drainage area, t
1
, is thus dened as
t
1
=
t
1
(t)

1
The dimensionless drainage area is simply the drainage area of the reservoir
divided by the square of the system characteristic length, 1
c
, or

D
=

1
2
c
The system imaging function, , that appears in Equation (3.79), is specic
for a given set of well completion and reservoir properties at a certain well lo-
cation. The mathematical denitions of the system imaging function for the
more commonly considered well completion and reservoir types were presented
by Postone and Poe (2008) in Chapter 8. They observed from such a construc-
tion of a dimensionless decline analysis reference decline rate variable, and a
dimensionless decline time function results in a collapse of the whole family of
boundary-dominated ow production decline stems to a single decline stem for
the late-time ow regime. The dimensionless decline analysis reference decline
rate variable is the product of the dimensionless well ow rate and the sys-
tem imaging function, . The dimensionless decline time function is shown to
conveniently incorporate the elements of the argument of the exponential func-
tion. The resulting dimensionless decline ow rate relationship obtained with
this variable substitution, applicable to the boundary-dominated ow, is given
by
114 3. DEPLETION RATE DECLINE REVIEW

1o
(t
1o
) = exp(t
1o
) (3.81)
Here, the dimensionless decline ow rate,
1o
, and the the dimensionless decline
time, t
1o
, reference functions are respectively given by:

1o
(t
1
d) =
&1
(3.82)
t
1o
=
2:

t
1
(3.83)
The dimensionless cumulative production during the boundary-dominated ow
regime can also be generalised for the rate-transient production decline behaviour
of a well. The dimensionless cumulative production of a well (e.g., unfractured
vertical, vertically fractured, or horizontal) in a nite closed reservoir is described
for the boundary-dominated ow by the expression:
Q
j1
(t
1
) =

1
2:
_
1 exp
_

2:

t
1
__
(3.84)
The denition of the dimensionless decline cumulativeproduction function, Q
j1o
,
is therefore an integration of the dimensionless decline ow rate with respect to
the dimensionless decline time (including all values of t
1o
):
Q
j1o
(t
1o
) =
t
Tu
_
0

1o
(t)dt =
2:Q
j1

1
(3.85)
For the boundary-dominated ow we have:
Q
j1o
(t
1o
) = 1 exp(t
1o
) = 1
1o
(t
1o
)
Imaging Function The non dimensional image function, , applicable for an
unfractured vertical well that is centrally located in a closed, circular reservoir,
is given by:
= ln(:
c1
)
3
4
there, the dimensionless drainage radius, :
c1
, is dened in a conventional manner
by:
:
c1
=
:
c
1
c
=
:
c
:
&
.
Similarly, the image function appropriate for a fully penetrating, unfractured
vertical well, located at a reservoir spatial position given by the coordinates
(A
&1
. 1
&1
) in a closed, rectangular reservoir of dimensions (A
c1
. 1
c1
) . whose
3. DEPLETION RATE DECLINE REVIEW 115
reference origin according to Poe (2003) is located at the lower left corner of the
rectangle, is given by:
= 2:
1
c1
A
c1
_
1
3

1
1
1
c1
+
1
2
1
+ 1
2
&1
21
2
c1
_
+ 2
1

n=1
1
:
cos(::
A
&1
A
c1
) cos(::
A
1
Ac1
)
.
.
cosh
_
::
Y c1jY
T
Y
uT
j
A
cT
_
+ cosh
_
::
Y c1jY
T
+Y
uT
j
A
cT
_
sinh(::
Y
cT
A
cT
)
. (3.86)
Here, the dimensionless spatial parameters (A
1
. 1
1
. A
&1
. 1
&1
. A
c1
and 1
c1
)
are dened as the ratio of the corresponding dimensional spatial dimensions
A. 1. An. 1 n. Ac.and 1 c) to the characteristic system length (1
c
= :
&
).
During the early transient behaviour of a vertical well, the wellbore solution is
commonly evaluated using a line source well solution at a dimensionless reservoir
spatial position away from the centre of the well, equal to the dimensionless
wellbore radius, :
1
= 1. However, under boundary-dominated ow conditions,
it is generally sucient to simply evaluate the solution at the reservoir spatial
position, equal to the midpoint of the wellbore (A
1
= A
&1
. 1
1
= 1
&1
).
It follows that the above solution can be simplied into a more readily com-
putable form given as:
= 2:
1
c1
A
c1
_
1
3

1
&1
1
c1
+
1
2
&1
1
2
c1
_
+ 2
1

n=1
1
:
cos(::
A
&1
A
c1
) cos(::
A
1
Ac1
)
.
_
1 + exp(2::
1
c1
A
c1
) + exp(2::
1
&1
A
c1
) + exp(2::
1
c1
1
&1
A
c1
)
_
.
_
1 +
1

a=1
exp(2:::
1
c1
A
c1
_
(3.87)
The dimensionless fracture conductivity, 1
C1
, is dened as a relative mea-
sure of the ratio of the fracture conductivity (/
)
/
)
) to the formation eective
permeability, /, and the characteristic system length (1
c
= A
)
) . or convention-
ally dened as:
1
C1
=
/
)
/
)
/1
c
=
/
)
/
)
/A
)
Poe (2005) evaluated the equivalent dimensionless fracture-spatial position A

1
as:
116 3. DEPLETION RATE DECLINE REVIEW
A

1
= 0.73551.5609(
1
1
C1
)+1.5313(
1
1
C1
)
2
179.4346(
1
1
C1
)
3
+3928.97.(
1
1
C1
)
4
40211.24(
1
1
C1
)
5
+ 183267.48(
1
1
C1
)
6
305367.26(
1
1
C1
)
7
(3.88)
Moreover, A

1
was used to accurately reproduce the wellbore rate or pressure-
transient behaviour of a nite-conductivity fracture (for 1
C1
_ 4.1635) during
the pseudoradial and boundary-dominated ow regime. The equivalent fracture
spatial position away from the well location at which to evaluate the uniform
ux fracture solution to obtain the equivalent wellbore response as that of an
innite-conductivity vertical fracture, A
1
, obtained in above equations is equal
to 0.7355. This value is only slightly larger than that commonly reported in the
literature (i.e., 0.732) for evaluating an innite-conductivity fracture response
from the uniform ux solution, originally presented by Gringarten et al. (1974).
The nite conductivity fracture responses and the innite-conductivity fractured
well responses are evaluated with the uniform ux solution by the A

1
correla-
tion. This A

1
correlation was used by Ozkan (1988) for dening the pseudoskin
function caused by the bounded nature of the reservoir (o), determined as:
o(A

1
. :
c1
) =
_
(r

1
+ 1)
3
(r

1
1)
3

12:
2
c1
The image function that applies for a nite-conductivity vertically fractured
well located in a closed, rectangular reservoir of dimensions A
c1
by 1
c1
with the
midpoint of the fracture located at (A
&1
. 1
&1
) has been given by Poe (2002) in
the form of:
= 2:
1
c1
A
c1
_
1
3

1
1
1
c1
+
1
2
1
+ 1
2
&1
21
2
c1
_
+
2A
c1
:
1

n=1
1
:
2
sin(::
1
A
c1
) cos(::
A
&1
A
c1
) cos(::
A
1
Ac1
)
.
cosh
_
::
Y c1jY
T
Y
uT
j
A
cT
_
+ cosh
_
::
Y c1jY
T
+Y
uT
j
A
cT
_
sinh(::
Y
cT
A
cT
)
(3.89)
The solution for the image function of a vertically fractured well in a closed rec-
tangle can also be simplied into a more readily computable form for a wellbore
spatial position of (A
&1
. 1
&1
) with the nite-conductivity fracture evaluation
spatial location of (A
1
= A
&1
+ A

1
. 1
1
= 1
&1
), as given by:
3. DEPLETION RATE DECLINE REVIEW 117
= 2:
1
c1
A
c1
_
1
3

1
1
1
c1
+
1
2
1
+ 1
2
&1
21
2
c1
_
+
2A
c1
:
1

n=1
1
:
2
sin(::
1
A
c1
) cos(::
A
&1
A
c1
) cos(::
A
1
Ac1
)
.
_
1 + exp(2::
1
c1
A
c1
) + exp(2::
1
&1
A
c1
) + exp(2::
1
c1
1
&1
A
c1
)
_
.
_
1 +
1

a=1
exp(2:::
1
c1
A
c1
_
(3.90)
Similar expressions of the appropriate image functions can also be derived
for a horizontal well located in a closed circular or rectangular reservoirs. For a
uniform ux horizontal well with the midpoint of the eective wellbore length ex-
posed to the reservoir centered in a closed, cylindrical reservoir, the appropriate
image function is presented in a readily computable form by Ozkan (1988):
= ln(:
c1
) +
1
4
+o(A1. 0) +o(A
1
. 0. :
c1
) +
1
:1
1
1

a=1j
cos(::2
1
) + cos(::2
&1
)
:
_
_
_
a1
T
(1+A
T
)
_
0
1
0
(n)dn +
a1
T
(1A
T
)
_
0
1
0
(n)dn +
1
1
(::1
1
:
c1
)
1
1
(::1
1
:
c1
)
_
_
a1
T
(1+A
T
)
_
0
1
0
(n)dn +
a1
T
(1A
T
)
_
0
1
0
(n)dn
_
_
_
_
_
The above solution is expressed in terms of modied Bessel functions of the rst
kind of orders zero and one (1
c
and 1
1
), modied Bessel functions of the second
kind, of orders zero and one (1
c
and 1
1
), as well as integrals of the modied
Bessel functions of the rst and second kind, of order zero.
For a horizontal well, centrally located in a cylindrical, bounded reservoir,
the dimensionless drainage radius is dened as the ratio of the eective drainage
radius of the circular reservoir divided by the characteristic system length, which
in this case is equal to half the eective horizontal wellbore length in the pay
:
c1
=
:
c
1
c
=
2:
c
1
I
The imaging function for a uniform-ux, horizontal well, located in a closed,
rectangular reservoir, can also be derived using the late-time solutions reported
by Ozkan (1988). The image function for an innite-conductivity, horizontal
118 3. DEPLETION RATE DECLINE REVIEW
well located in a closed rectangular reservoir can be evaluated as the sum of the
corresponding image function of an innite-conductivity, vertical fracture in a
closed, rectangular reservoir,
)
, given by Equation (3.90).
Composite production decline curves for innite-conductivity horizontal wells
and its uniform ux solution are evaluated at A
c
= A
&1
+0.732 after Gringarten
et a1. (1974) in closed, rectangular reservoirs have also been reported by Shih and
Blasingame (1955). The appropriate wellbore solution is evaluated at the reser-
voir spatial position (A
1
= A
&1
+ A

1
. 1
1
= 1
&1
.and 2
1
= 2
&1
+ :
&1
).with
A

1
= 0.732. The late-time imaging function, , for a horizontal well (with all
its components) can be written as:
=
)
+ 1
1b
+ 1
1b1
+ 1
1b2
+ 1
1b3
Values of 1
1b
, 1
1b1
, 1
1b2
and 1
1b3
when expressed in a readily computable
form is quite lengthy and has not been here included. The interested reader can
nd it in the book of Postone and Poe (2008) on page 131-132.
The use of the dimensionless decline variables given by Equations (??, 3.83
and 3.84) and the corresponding scaling associated with each variable for both
the transient and boundary-dominated ow behaviours of the production decline
of a well results in a composite reference decline curve set with only a single late-
time stem for the sake of ease with regards to graphical matching purposes.
Integral Decline Curve Analysis Functions The dimensionless decline
ow rate-integral and rate-integralderivative functions are introduced to im-
prove the uniqueness of the graphical matching procedure. The dimensionless
decline ow rate-integral function is equivalent to the dimensionless decline cu-
mulativeproduction function (3.85), normalised by the dimensionless decline
time (3.83). Moreover, the dimensionless ow rate integralderivative is equal to
the derivative of the dimensionless ow rate integral with respect to the natural
logarithm of the dimensionless decline time function. These graphical analysis
relationships tend to display the same general trend as the dimensionless decline
ow rate function with respect to the dimensionless decline time. Moreover, they
can provide a clearer demarcation of the ow regimes exhibited in the decline
behaviour of production performance of a well.
The dimensionless owrate integral function introduced as an aid in graphical
production decline-curve analysis matching procedures is equal to the dimension-
less decline time normalised dimensionless decline cumulative-production func-
tion. The function,
1oi
, is applicable for all values of dimensionless time, t
1o
,
and is expressed as:

q1oi
(t
1o
) =
1
t
1o
t
Tu
_
c

1o
(t)dt =
Q
j1o
(t
1o
)
t
1o
3. DEPLETION RATE DECLINE REVIEW 119
The function,
1oi
, only applicable for boundary-dominated ow, is expressed
with

q1oi
(t
1o
) =
1
t
1o
[1
q1
(t
1o
)]
This function is signicantly smoother than the dimensionless decline ow rate
function (3.81), yet, despite this, it does not suer any appreciable loss in char-
acter as a result of the production decline trend for a particular ow regime.
The derivative of the dimensionless ow rate integral function with respect
to the natural logarithm of the dimensionless decline time function has also
been used to provide a more distinctive character of the transient production
decline behaviour than either the dimensionless decline ow rate or the ow
rate integral functions. The enhanced signature that is characteristic of the
dimensionless decline ow rate integral-derivative function renders it extremely
useful for identifying the start and end of a particular ow regime, as well as
for improving the uniqueness of the decline-curve analysis graphical matching
procedure. The derivative function applicable for all values of t
1o
is expressed
as:

1io
(t
1o
) =
d
1oi
(t
1o
)
d ln(t
1o
)
= t
1o
d
1oi
(t
1o
)
t
1o
=
1oi
(t
1o
)
1o
(t
1o
) (3.91)
For the boundary-dominated ow,
1io
, is represented by

1io
(t
1o
) =
1
t
1o
[1
1o
(t
1o
) [1 + t
1o
]] (3.92)
A graphical representation of the reference production decline curves for an
unfractured vertical well centrally located in a closed, cylindrical reservoir, pre-
sented in the more conventional manner (
1o
versus t
1o
) is given in Figure (3.34).
The dimensionless decline ow rate response is displayed as black curves,
the dimensionless ow rate integral behaviour is given as red curves, and the
derivative response is presented in blue.
Two types of single-phase ow boundary ux models were proposed by Dou-
blet and Blasingame (1995) for an unfractured vertical well centered in a cylin-
drical bounded reservoir. These are the step-rate and ramp-rate boundary ux
models. The inux at the outer boundary is initially equal to zero (no-ow
outer-boundary condition), which permits the use of the closed-boundary rate-
transient decline curve analysis development procedure previously discussed, at
least for the limiting case of a no-inux outer-boundary condition.
The no-inux case results in the exponential production decline given by
equation comprising terms of the modied Bessel function of the rst and sec-
ond kinds, of orders of zero and one. The Laplace space transform parameter
120 3. DEPLETION RATE DECLINE REVIEW
Figure 3.34: Production decline curves for a nite-conductivity, vertcally frac-
tured well positioned in a closed rectangular reservoir.[after Poston and Poe
(2008)].
(s) corresponds to the dimensionless time values at which solution should be
evaluated.
v
j
&1
(:) =
1
0
(
_
:)1
1
(
_
::
c1
) + 1
0
(
_
:)1
1
(
_
::
c1
)
o
_
:
_
1
1
(
_
:)1
1
(
_
::
c1
) 1
1
(
_
:)1
1
(
_
::
c1
)

+
v

1cat
(:) [1
0
(
_
:)1
1
(
_
:) + 1
0
(
_
:)1
1
(
_
:)]
_
::
c1
_
1
1
(
_
:)1
1
(
_
::
c1
) 1
1
(
_
:)1
1
(
_
::
c1
)
(3.93)
a specied ux condition,
1cat
to several inux models is given as:
v

1cat
=
_
_
_
0 :o ,|on

1
c

1cat
exp(t
1ctovt
:) :tcj :ctc
q
TciI
c(1+t
TsIarI
c)
:c:j :ctc
_
_
_
At a specic point in time, t
1ctovt
, in the production history, the outer-boundary
condition is switched from the initial no-ow condition to a specied ux condi-
tion,
1cat
.
The real space rate-transient solution of the boundary ux model can be
readily obtained by evaluation of the Laplace space dimensionless wellbore ow
rate solution given by application of Duhamels theorem followed by a numerical
inversion of the result into the real space domain by Stehfest (1970).
The ramp-rate boundary ux model assumes that the outer-boundary condi-
tion (initially at zero inux) smoothly and slowly increases from time-zero to a
3. DEPLETION RATE DECLINE REVIEW 121
Figure 3.35: The production decline curves for a vertical well positioned in
a cylindrical reservoir with a step-rate ux outer-boundary condition .[After
Poston and Poe (2008)].
xed larger value at a later time. This decline analysis model was developed to
approximate the production decline behaviours of reservoirs with natural water
inux or slowly responding waterood systems. The production decline behav-
iour obtained with the ramp-rate boundary ux model is often found to be virtu-
ally indistinguishable from the production decline behaviour of a dual-porosity
reservoir.
The step-rate boundary ux model assumes that the outer-boundary condi-
tion is abruptly switched to a specied ux condition at a prescribed point in
time, after which the inux rate is held constant. This decline analysis model
was developed to address issues related to waterood operations. The produc-
tion decline behaviour obtained with the model exhibits production "humps"
similar to those commonly observed in producing wells in waterooded elds.
Multiphase numerical reservoir simulation runs were carried out in the inves-
tigation by Doublet and Blasingame (1995) to validate the use of the single-phase
analysis step-rate and ramp-rate boundary ux models for the production decline
analysis of two-phase systems, in which water displaced oil. It was found that
the single-phase model developed in Doublet and Blasingame (1995) should be
applicable for oil/water two-phase systems with mobility ratios close to unity,
potentially relevant to an injection-production system analogous to that of a ve
or nine-spot pattern.
The application of the recent developments in decline-curve model construc-
tion was also employed for developing production/injection decline curves for
122 3. DEPLETION RATE DECLINE REVIEW
Figure 3.36: The production decline curves for a vertical well positioned in
a cylindrical reservoir with a ramprate ux outer-boundary condition.[After
Poston and Poe (2008)].
innite-conductivity vertically fractured wells located in closed, cylindrical reser-
voirs (Doublet and Blasingame 1995). The uniform ux fracture solution of
Ozkan (1988) was in that study used to rapidly (and with reasonable accu-
racy) estimate the innite-conductivity fractured well response at a dimension-
less fracture spatial position from the wellbore equal to 0.732 after Gringarten
et al. (1974). The generated set of reference decline-curves based on this solu-
tion is presented in Figure (3.37) for a set range of dimensionless drainage radii,
:
c1
=
v
c
A
]
, ranging from 1 to 1000.
A generally more appropriate production reference decline-curve set can be
constructed for an innite conductivity vertical fracture in a closed, rectangu-
lar reservoir using the uniform ux solution developed by Ozkan (1988). It has
been reported in the literature by a number of investigators that the eectively
draining the reservoir area (over which the pressure distribution was inuenced)
by a vertically fractured well in a low permeability reservoir is a direct function
of the eective fracture half-length. This results in a drained area in the reser-
voir that is typically more elliptical in shape during the transient ow regimes.
An elongated rectangular drainage area is therefore generally considered to be
more appropriate than a circular one during modelling of transient behaviour
of vertically fractured wells in nite reservoirs. Since the innite-conductivity
fracture is simply a special case of the nite-conductivity (1
C1
500) fractured
well in a closed, rectangular reservoir, this limiting case can be readily included
in the reference decline-curve sets for a nite conductivity vertical fracture in a
3. DEPLETION RATE DECLINE REVIEW 123
Figure 3.37: Production decline curves for an innite conductivity fractured
well, centrally located in a closed, cyllindrical reservoir [After Postone and Poe
(2008)].
rectangular reservoir.
Vertical Wells Intersected by FiniteConductivity Fractures An ex-
ample of a dimensionless production decline-curve set for a nite-conductivity
fracture is given by Figure (3.38). This decline-curve set was developed for a frac-
tured well, centrally located in a rectangular drainage area. A complete family
of reference production decline curves for a nite-conductivity fractured well in-
cludes other decline-curve sets for a range of dimensionless fracture conductivity
and drainage area and/or aspect ratio.
lnnite-Conductivity Horizontal Wellbore Dimensionless production decline
curves for an innite-conductivity horizontal wellbore also have three or more
independent parameters that govern their transient dimensionless production
decline behaviour. For an innite conductivity horizontal wellbore centrally lo-
cated in a closed, circular reservoir, the required image function for coupling the
transient and boundary-dominated ow production decline behaviours is given
by Postone and Poe (2008). The independent variables determining the tran-
sient behaviour of the well in this case include the dimensionless wellbore length,
1
1
=
1
I
2I
, the dimensionless drainage radius, :
c1
=
v
c
1
c
=
2v
c
1
I
, the dimensionless
wellbore radius, :
&1
=
v
u
I
, the dimensionless wellbore vertical spatial position
(or stando from the bottom) in the reservoir, and the dimensionless decline
time.
124 3. DEPLETION RATE DECLINE REVIEW
Figure 3.38: Production decline curves for a nite-conductivity vertically frac-
tured well centrally located in a closed [after Poston and Poe (2008) ].
The production decline behaviour of an innite-conductivity horizontal well
located in a closed, rectangular reservoir is even more complex. The transient
production decline behaviour of an innite-conductivity horizontal well located
in a closed rectangle is a function of the dimensionless wellbore length, the di-
mensionless wellbore radius, the stando from the bottom of the reservoir, the
reservoir drainage area, the wellbore midpoint location in the drainage area,
and the dimensionless decline time. This list of parameters represents the mini-
mum that must be considered. Additionally, the eects of reservoir permeability
anisotropy and dual-porosity behaviour may also be included in the family of
reference production decline curves, thus expanding the required number of ref-
erence curve sets even further.
Calculation Procedure for Plotting Functions The analysis of the histor-
ical production decline data of a well using the dimensionless production decline
solutions that are expressed in terms of the dimensionless decline variables of ow
rate, ow rate integral, and ow rate integral-derivative requires that the appro-
priate dimensional production decline functions be directly computed from the
historical production data. The dimensional graphical analysis variables derived
from the historical production data used are the pressure drawdown normalised
ow rate, the pressure drawdown normalised ow rate integral, and the pressure
drawdown normalised ow rate integral-derivative functions.
Production Decline-Curve Analysis With Partial or Absent Pressure
Record A problem commonly encountered in the evaluation of the production
3. DEPLETION RATE DECLINE REVIEW 125
performance of a well not all o the required production data being available for
analysis. This problem arises quite frequently and may result from a number
of reasons. Whether performing a production decline-curve analysis using the
material balance time approach, or a history match of the well performance data
based on a numerical model employing superposition of the varying ow rate and
owing pressure history, the ow rates of each of the uid phases as well as the
bottomhole-owing pressure are required at each time level in the production
history to correctly perform the analysis.
Assumptions and Limitations The eective convolution analysis technique
previously described in this chapter is subject to limitations and assumptions.
One such limitation, related to the use of the Horner approximation of the
pseudoproducing time (e.g., material balance time), concerns the assumption
that the ow rate is only permitted to be smoothly varying. Therefore, an er-
ratic well-ow history shortly before a production data time level of interest can
result in a signicant error in the estimation of the equivalent superposition time
function value for that time level. This limitation is also true for other produc-
tion analysis methods that use the material balance time function as a substitute
for the superposition time function.
126 3. DEPLETION RATE DECLINE REVIEW
Chapter 4
RATE DECLINE OF A
FRACTURED WELL
Remaining world-wide recoverable hydrocarbon resources exist in reservoirs pos-
sessing poor permeability. At present, low production rates accompanying such
poor permeability imply that some form of permeability enhancement or stimu-
lation must be carried out within these reservoirs in order for hydrocarbons to
be economically exploited. Even where initial permeabilities are relatively high,
stimulation may still be required to overcome problems associated with localised
permeability damage due to, for example, drilling mud invasion.
Hydraulic fracturing is widely used to increase the productivity of damaged
wells or wells producing from low permeability formations, and consists in in-
creasing well eective areal contact within a reservoir. Fractures can be posi-
tioned either parallel or transversally to well length. Moreover the stimulation
of a horizontal well in a low-permeability reservoir may further increase its pro-
ductivity. Unlike a vertical well, a horizontal well may be fractured at more
than one point along the well length. A fracture has a much greater permeabil-
ity than the formation it penetrates; hence, it inuences the pressure and rate
response of a well. Thus, much research has been carried out to determine the
eect of hydraulic fractures on pressure-transient and rate-transient behaviours
in addition to well performance.
This chapter discusses and develops the fractured well model responses. The
fractured well is positioned in a non-bounded or innite reservoir, and in a
bounded or closed reservoir. Solution for a vertical-fractured well solutions are
extended from pressure-time to rate-time solutions. Furthermore solutions for
a horizontal fractured well are new in development A vertical well fracture is
longitudinal or parallel to the wellbore axis, whereas horizontal well fractures
are transversal and longitudinal. Here, we present model solutions that couple
a well with fractures to an oil reservoir. Within the model, both fracture ow
and wellbore ow may be simulated, and for each model, initial and boundary
conditions for a diusion equation are set up. The diusion equation is then
127
128 4. RATE DECLINE OF A FRACTURED WELL
solved for the inner boundary condition of constant pressure, giving rise to well-
rate responses in time. The inner boundary condition of the constant pressure
can be extended to a variable pressure condition. The new inner boundary
condition feature is the restart option. This feature combines the constant rate
and constant pressure inner boundary condition within the same time interval,
where the IBCs consist in a constant rate for a selected time. At the end of that
time, the model nds the wellbore pressure and changing IBC to the constant
wellbore pressure that is maintained unaltered until the end of the time interval.
This restart feature is developed, implemented and tested, and the solution is
described in the third section.
In the section fourth, we present the late time approximation in an innite
reservoir for a horizontal well that is transversally fractured. The developed
approximation relates rates of a multi-fractured horizontal well to those of a
vertical well.
The approximations are useful and may convert complicated rate-time solu-
tions to simple expressions to be used in a screening analysis of a single well
production. The rate-time approximation developed for a horizontal-fractured
well can be compared to a vertical well, a vertical-fractured well or a horizon-
tal well. Eective wellbore radii and half-length expressions are developed to
measure the eectiveness of a multi-fractured horizontal well.
The models presented in this chapter do not consider formation damage.
This parameter, a common problem associated with eld operations which is
of great importance for the ecient exploration and production of hydrocarbon
resources, should be considered in future studies, in which fracture skin and non-
Darcy ow may also be incorporated. The model is applicable to compressible
gas ow, i.e., pressure, j, should be replaced by the pseudo-pressure, :(j), with
a remark that inclusion of non-Darcy ow should be further implemented.
4.1 Transient Oil Flow
The choice was made to develop rate-time solutions for two models; a vertical-
fractured well model, and a horizontal fractured well. The two model solutions
were then summarised and well rates were plotted versus time solutions. Since
the reservoir was an oil non-bounded, the ow regime was transient or innite
acting.
4.1.1 Fractured-Vertical Well Model
Hydraulic fracturing is a widely used method for increasing well productivity.
Such productivity increases occur due to fracturing eectively increases the well
surface area, thus rendering ow to the well much more ecient. The increase
in a surface area is achieved by injecting uids into the formation at pressures
4. RATE DECLINE OF A FRACTURED WELL 129
above the formation parting pressure. Injection of a uid at high pressure ini-
tiates the fracture and causes it to propagate. The subsequent injection of a
proppant allows the fracture to remain open. The ow of uids to a propped
fracture is much more ecient than ow to a wellbore. This is due to the wellbore
having a small surface area whereas that of a fracture may be very large. Frac-
turing thus drastically increases well production, often rendering unprotable
wells highly protable. Although hydraulic fracturing is usually very eective,
it is also very costly. It is thus essential that methods be available to evaluate
the eectiveness of the fracturing process. The most widely used evaluation tool
for hydraulic fractures is pressure transient testing or recently, equivalently, rate
decline analysis.
Rate-time transient analysis consists of two phases, of which the rst concerns
posing and solving the equations that govern ow in an idealised model. This
model is assumed to reect the mechanisms at work in the reservoir. The second
phase consists in matching the rate and pressure measurements, taken in the
eld, to those from the assumed model. The present work mainly investigates
the rst aspect of pressure transient testing of fractured wells posing and solving
the equations that govern ow in the model. Nevertheless the prediction of
the pressure response of fractured wells is not a new topic. Numerous models
have been investigated which consider various aspects of the problem. However,
these models either consider only part of the problem, or only allow approximate
solutions of the governing equations.
The most comprehensive model to have been investigated is the nite con-
ductivity fracture model developed by Cinco and co-authors and presented in
several papers. The two most important are Cinco-Ley et al. (1978), where
the model is proposed and the governing equations solved, and Cinco-Ley and
Samaniego (1981), where the behaviour of the solution is investigated.
The solution procedure used in the rst of these papers is a numerical solution
of an integral equation. This technique has become known as the boundary
integral equation method. The computation method is intensive, and since it is
numerical, yields only approximate results. The pressures computed using this
method appear to be accurate, although it is dicult to state just how accurate
they are.
In Cinco-Ley and Samaniego (1981), the behaviour of the well pressure is
investigated. The intent of the present work is to put forward a model that
depicts the ow of uids into and through a nite conductivity vertical fracture.
We seek an exact solution that can be used to determine pressures anywhere in
the fracture and reservoir system. The reason for pursuing an analytic solution is
that it is expected to be highly accurate and but time consuming. The ultimate
goal is to provide a solution to which the computation is suciently rapid to be
of use in computer-aided rate-time interpretations.
Here, we nd a rate-time solution from the pressure solution of Cinco-Leys
130 4. RATE DECLINE OF A FRACTURED WELL
model. The solution, was determined by altering the inner boundary condition
of constant rate production to a constant pressure production. The following
expressions were implemented into the existing code and the output was tested
versus cases found in the literature. We further compare the vertical fractured
well model, the model by Cinco-Ley at al. (1978) to that of a horizontal well
with one transversal fracture within an oil innite reservoir.
By presenting rate solutions for a vertical well with a nite conductivity
fracture and a horizontal well with a transversal nite conductivity fracture,
we rst compared solutions for rate versus time. Moreover model dierences
and model similarities were discussed. The model for a horizontal well with a
transversal fracture has more options that may be included into the vertical well
model. The model comparison helps to understand the dierences between a
horizontal fractured versus a vertical fractured model.
The work of Cinco-Ley and Samaniego (1981) describes modelling features
of a vertical fractured well in an oil reservoir. Based on this, a pressure solution
was extended to a rate solution for a vertical fractured well.
The diusivity equation describes unsteady-state ow in the system (of nite
conductivity fractured vertical well in an innite reservoir of SLAB geometry).
The properties of both the reservoir and the fracture are independent of pressure
and the ow in the entire system obeys Darcys law. The pressure gradients are
small, gravity eects are negligible, and the ow into the wellbore is believed to
come through the fracture.
The dimensionless wellbore pressure (drop) is given in "Laplace" space as:
j
&1
=
C
: (: + d
_
:)
1
2
(4.1)
where C and d are constants, given by
C =
: j
1
2
)1
(/
)
/
)
)
1
. d =
2j
)1
(/
)
/
)
)
(4.2)
where again
j
)1
=
/
)
cc
t
/ c
)
c
)t
(4.3)
(/
)
/
)
)
1
=
/
)
/
)
/ r
)
(4.4)
with /
)
, r
)
representing the fracture width and half-length.
The corresponding dimensionless time is given by
t
1a
]
=
j/t
cjC
t
r
2
)
(4.5)
4. RATE DECLINE OF A FRACTURED WELL 131
where parameter j = 3.6 10
9
Also,
j
&1
=
//(j
i
j
&)
)
c
c
q 1
c
j
(4.6)
with c
c
= 1.842 (eld unit solutions).
The model by Cinco-Ley (1982) denes the dimensionless pressure, j
&1
, and
we calculate the dimensionless rate,
1
. with the known transformations in Equa-
tion (4.7). Subsequently, all transformations and solutions are carried out in
Laplace space.
Hence,
j
&1
=
1
:
2
(4.7)
which gives
=
1
:
2
_
1
j
&1
_
(4.8)
=
1
:
2
: (: + d
_
:)
1
2
C
(4.9)
=
1
:
(: + d
_
:)
1
2
C
(4.10)
there, . is the rate ("Laplace" space) solution for a vertical fractured well based
on the work of Cinco-Ley and Samaniego (1981). A real rate solution can be
obtained by the known Stehfest (1970) inverse "Laplace" technique (presented
in Appendix B).
4.1.2 Horizontal Well with Transversal Fractures
The general assumptions and limitations for a model of a multi-fractured hori-
zontal well include: an innite (or geometry SLAB) reservoir that is isotropic or
anisotropic, with no-ow boundaries above and below. The well passes through
the centre of transversal, rectangular, and fully penetrating fractures. Flow to
a well occurs only through fractures, that are either of uniform ux or nite
conductivity type, the latter being introduced through the "equivalent pres-
sure point" method from references Gringarten and Ramey (1972), Chen et al.
(1991), of Raghavan and Joshi (1993), Blasingame and Poe (1993). For the gen-
eral case, fractures are uniformly spaced and sized. However, a limited number
of variations of sizing and spacing may be treated additionally. Inner boundary
conditions include: a constant rate or constant pressure, nevertheless, variable
132 4. RATE DECLINE OF A FRACTURED WELL
Figure 4.1: A fractured-horizontal well of length, L, with three transversal frac-
tures of half-lengths, L
)
. The reservoir is non-bounded or innite in the x and
y directions (Top view).
4. RATE DECLINE OF A FRACTURED WELL 133
rate inner boundary condition may be employed. A top view of the model is
given in Figure 4.1.
The work presented in the next two sections are selected topics related to the
fractured horizontal well published by Cvetkovic et al. (1999). The transversal
and longitudinal model solutions are utilised in section 3 (for the derivation of the
restart option) and section 4 (for the late-time approximations, the equivalent
wellbore radius and the equivalent half-length).
The basic pressure (diusion) equation (with sources) can be expressed as:
\
_
/
j
\j
_
cc
Jj
Jt
= (4.11)
Here, the ordinary simplifying assumptions are made i.e., one-phase Darcy ow,
Newtonian uids, isothermal conditions, gravity negligible, small pressure gra-
dients, constant compressibility, viscosity, porosity and permeability.
j, uid viscosity
/, permeability tensor
c, porosity of the medium, constant
, source volumetric production rate per unit volume.
For the principal axes of permeability (anisotropy) coinciding with the coor-
dinate the axes, we may write
\
2
j
1
j
Jj
Jt

j
/
= 0 (4.12)
where j =
I
cj
is the diusivity coecient of the porous medium. We have written
/ =
3
_
/
a
/
j
/
:
, to obtain an invariance of volume elements and the uxes, while
the new coordinate variables are r
0
=
_
I
I
i
r etc., reducing to isotropy. By
introducing the usual dimensionless variables r
1
= r
0
,|,
1
=
0
,|, .
1
= .
0
,|,
t
1
= jt,|
2
,
1
=
|
2
j
, we nally obtain the expression in dimensionless form:
\
2
1
j
Jj
1
Jt
1


1
cc
= 0 (4.13)
The application of the Laplace transform, dened by
,(:) = 1,(t
1
) =
1
_
0
c
ct
T
,(t
1
)dt
1
(4.14)
gives the simpler equation
\
2
1
j :j =
~
1
cc
j
i
(4.15)
where j
i
= j(t
1
= 0). Leaving out the details, we can assume a uniform initial
pressure and write
134 4. RATE DECLINE OF A FRACTURED WELL
j =
j
i
:
j (4.16)
Further, assuming outer boundary conditions of a normally occurring type,
Green
0
s identities can be used to obtain an integral representation of the fol-
lowing kind:
j =
j
/|
_
S
do
0
(4.17)
there, is the (Laplace-transformed) volumetric rate of extraction through the
source S, while is a fundamental solution or Green
0
s function, depending on
the geometry and boundary conditions.
Assuming now an innite reservoir geometry with impermeable boundaries
at z = 0 and z = h, one form of is:
=
1
4
+1

1
exp
_

_
:
_
(r r
0
)
2
+ (j j
0
)
2
+ (. .
0
2:/)
2
_
_
(r r
0
)
2
+ (j j
0
)
2
+ (. .
0
2:/)
2
+
exp
_

_
:
_
(r r
0
)
2
+ (
0
)
2
+ (. + .
0
2:/)
2
_
_
(r r
0
)
2
+ (
0
)
2
+ (. + .
0
2:/)
2
(4.18)
This is the direct result obtained by using the method of images on the funda-
mental (Lord Kelvin
0
s) point source solution

c
=
exp(1
_
:)
4:1
(4.19)
The integration in (4.18) is performed over the system of fractures in our case.
Having assumed uniform ux fractures, we rst obtained the simplication
j =
j
/|

_
S
do
0
(4.20)
For several fractures, we get the linear superposition
j =
j
/|
a

i
_
S
.

i
do
0
(4.21)
To make things somewhat less general, the fractures are presumed to be rectan-
gular, fully penetrating and transversal.
Another basic assumption is that there is no direct ow to the wellbore, which
is designated as being of innite conductivity and passing centrally through the
fractures. Finally, we assume an equal spacing and size, choosing our reference
length,|, as equal to the fracture half-length, 1
)
.
4. RATE DECLINE OF A FRACTURED WELL 135
Making use of Poisson
0
s summation formula enables us to express our solution
in a (formally) simple manner via modied Bessel functions 1
0
(1
_
:):
2://
j
j
i
= j
i1
=
_
/
:
/
a
/
2
a

)=1

)1
_
I
I
i
_
0
1
0
_
_
:
_
n
2
+ (i ,)
2
,
2
_
dn (4.22)
Here,
, =
_
/
a
/
j

1,1
)
: 1
(4.23)
where L is the distance between the outermost fractures. j
i1
is then the nor-
malised pressure (in "Laplace" space ) evaluated at the centre of the fractures.
When demanding that these pressures be equal, a system of equations is
obtained:

1
1
0
+
2
1
1
+ .... +
a
1
a1
= j

1
1
1
+
2
1
0
+ .... +
a
1
a2
= j
.
.
.

1
1
a1
+
2
1
a2
+ ... +
a
1
0
= j

1
+
2
+ ... +
a
=
_

_
(4.24)
Here we have again employed a simplication of / = /
a
= /
j
= /
:
, and can thus
write:
1
)
= 1
)
(:) =
1
_
0
1
0
_
_
:
_
r
2
+ ,
2
,
2
_
dr. (4.25)
Where, 1
)
(:) is the coecient in the set of Equations (4.24) , and 1
0
. is the
modied Bessel function of the second kind of zero order. From this set of
equations (4.24) it is desirable to eliminate
1
,
2
,......,
a
. This gives us a relation
of the following kind:
j = 1
a
(:) (4.26)
For one fracture it can be fairly easy determinated that : = 1 and
1
1
(:) = 1
0
(:) (4.27)
Further, if we increase the number of fractures : = 2. .... 4 we obtain
136 4. RATE DECLINE OF A FRACTURED WELL
1
2
(:) =
1
2
1
0
(:) +
1
2
1
1
(:) (4.28)
1
3
(:) =
1
2
0
(:) + 1
0
(:)1
2
(:) 21
2
1
(:)
31
0
(:) 41
1
(:) + 1
2
(:)
(4.29)
1
4
(:) =
1
2
0
(:) + 1
0
(:)1
1
(:) 1
2
1
(:) + 1
0
(:)1
3
(:) 21
1
(:)1
2
(:) + 1
1
(:)1
3
(:) 1
2
2
(:)
41
0
(:) 21
1
(:) 41
2
(:) + 21
3
(:)
(4.30)
The expressions rapidly become bigger, with 1
7
(:) being the last of these
expressions to ll less than one page. 1
10
(:). instance needs more than 8 pages!
The parameter 1(:) relates the dimensionless pressure, j to the dimensionless
rate, both in "Laplace" space. Late time solution for the equally sized and
spaced fractures details are reported by Cvetkovic et al.(1996).
4.1.3 Horizontal Well with Longitudinal Fractures
Fractures of a horizontal well are of uniform ux type, fully penetrating, evenly
spaced and have the same rectangular shape - with half-length, 1
)
. The wellbore
runs through the middle of the reservoir (and fractures); there is no direct ow to
the wellbore, and hence its length, L, is considered to extend from the beginning
of the rst fracture to the end of the last one. The top view of a model for a
horizontal well with three fractures is given in Figure (4.2).
The dimensionless pressure calculated in formula (4.22) is further used in
[4.26]. For the normalised dimensionless pressure, an evaluation was carried
out at some point along the wellbore at fracture number i with (dimensionless)
coordinate y:
j
i1
=
1
2
i
2
.

)=1
)o+2
_
)o
1
0
__
:i[
0
[

d
0
(4.31)
= 1 + i, + t
1 _ t _ 1
Here, 1
0
is the standard modied zero-order Bessel function and ` is the number
of fractures. In this case we obtain:
i =
6
_
/

/
I
. , =
1
1
]
2
` 1
4. RATE DECLINE OF A FRACTURED WELL 137
Figure 4.2: A fractured-horizontal well of length 1 , with three transversal frac-
tures of half-lengt 1
)
. The reservoir is non-bounded or innite in the A
c
and
the 1
c
directions (Cross section view).
valid for ` _ 2; and evidently when for ` = 1. , = 0. Further for 1 _ 2`1
)
implies , _ 2(` _ 2). Moreover, changing variable from
0
= 1 + ,, + t
0
to t
0
gives dimensionless pressure
j
iD
=
1
2
i
2
N

j=1

jD
1
_
1
1
0
__
:i[(i ,), + t t
0
[

dt
0
(4.32)
j
i1
=
1
2
i
2
.

)=1

)1
J
ji)j
(t)
Choosing i = 1. 2. .... `. provides us with the same form of equations as in (4.22),
(4.24), connecting fracture rates,
)1
. to the fracture (i.e., wellbore) pressure j
i1
.
In this case, however, the pressure varies along each fracture, and it is crucial to
make a choice with regard to presenting the wellbore pressure. One possibility
is to choose the midpoint (t = 0) pressure (the natural choice for transver-
sal fractures), but we will instead use the integral average, as suggested by
Raghavan and Ozkan (1995) (of page 72) to obtain a good approximation for
an innite-conductivity wellbore, although this renders the computations more
cumbersome. In other words in (4.33), each J
I
(t) is replaced by its integral
138 4. RATE DECLINE OF A FRACTURED WELL
average:
1
2
1
_
1
J
I
(t)dt =
~
J
I
Changing the variables once again, this time by a factor i
_
: , we nd:
~
J
I
=
1
4:
2i
p
c
_
0
2i
p
c
_
0
1
0
_

/,i
_
: + r

drd (4.33)
Dealing rst with the special case where coecient, / = 0 (coinciding with the
case , = 0. i.e., ` = 1) we nd after some calculations:
~
J
0
=
1
4:
_

_
4i
_
:
2i
p
c
_
0
1
0
(n)dn 2
2i
p
c
_
0
n1
0
(n)dn
_

_
Simplied, with properties of Bessel functions, we obtain:
~
J
0
=
i
_
:
_

_
2i
p
c
_
0
1
0
(n)dn + 1
1
(2i
_
:)
_

1
2:
(4.34)
Moreover, when / 0 further calculations give
~
J
I
=
1
4:
_
,((/, 2)i
_
:) 2,(/,i
_
:) + ,((/, + 2)i
_
:)

(4.35)
where we have
,(r) = r
a
_
0
1
0
(n)dn + 1
1
(r) (4.36)
For a numerical evaluation
~
J
I
given in (4.34), (4.35) and (4.36), we use Chebyshev-
polynomial expressions to compute 1
a
for large :-values. However, dierent ver-
sions are required here; furthermore, we will also cover small and moderate
:-values by the same kind of expansions. (small s corresponds to the large time
and large s to the early time).
In addition to formula (4.36) we need also the modied version for large
r-values
,(r) = r
_
_
:
2

1
_
a
1
0
(n)dn + 1
1
(r)
_
_
(4.37)
4. RATE DECLINE OF A FRACTURED WELL 139
From Luke (1969) (of pages 341-343) we may directly use tabulated values for
coecients c
a
and d
a
in
1
1
(r) =
_
:
2r
c
a
1

a=0
c
a
1

a
_
5
r
_
(r _ 5) (4.38)
1
_
a
1
0
(n)dn =
_
:
2r
c
a
1

a=0
d
a
1

a
_
5
r
_
(r _ 5) (4.39)
Once again,
1

a
(:) (4.40)
correspond to shifted Chebyshev polynomials, given by
1

a
(r) = 1
a
(2r 1) (4.41)
1
a
(r) = cos(:cos
1
r) (4.42)
with explicit formulas for polynomial expansions given by Oberhettinger and
Badii (1973) (page 15). However, there are smarter ways of carrying out these
computations, e.g., via recursion formulas given by Wimp (1962). Prior to this,
in addition to (4.38) and (4.39) we need similar expansions valid for
[r[ _ 5
From Luke, Y. (1969), ( page 452-453) we have
1
1
(cr) =
_
+ log
cr
2
_
1
1
(cr) +
1
cr
+
1

a=0
H
a
1
2a+1
(r) (4.43)
1
1
(cr) =
1

a=0
1
a
1
2a+1
(r). (0<r _ 1) (4.44)
with coecients H
a
, 1
a
tabulated for a = 5 in Wimp (1962), (page 454). Similar
coecients for expansions of
a
_
0
1
0
(n)dn have to be developed specially, which is
done below. We shall need the following coecients c
a
oa
_
0
1
0
(n)dn =
1

a=0
c
a
1
2a+1
(r). [r[ _ 1 (4.45)
In order to determine c
a
a dierentiation is carried out giving:
140 4. RATE DECLINE OF A FRACTURED WELL
c1
0
(cr) =
1

a=0
c
a
1
0
2a+1
(r) (4.46)
Furthermore, we have:
1
0
(cr) =
1

a=0
C
a
1
2a
(r) (4.47)
with C
a
calculated for c = 5 in Wimp (1962) (of page 454). Generally,
1
2a
(r) =
1
2
_
1
0
2a+1
(r)
2: + 1

1
0
j2a1j
(r)
2: 1
_
(4.48)
see e.g., Luke (1969) (page 300). Hence, we obtain directly:
c
a
=
c
2(2: + 1)
(C
a
C
a+1
) (4.49)
(: _ 1)
c
0
= cC
0
(4.50)
Similarly, starting with the assumption
oa
_
0
1
0
(n)dn =
_
+ log
cr
2
_
oa
_
0
1
0
(n)dn +
1

a=0
/
a
1
2a+1
(r) (4.51)
and then dierentiating, gives us:
c1
0
(cr) = ( + log
cr
2
)c1
0
(cr) r
1
oa
_
0
1
0
(n)dn +
1

a=0
/
a
1
0
2a+1
(r) (4.52)
This is to be compared to a formula from Wimp (1962) (of page 453):
1
0
(cr) = ( + log
cr
2
)1
0
(cr) +
1

a=0
G
a
1
2a
(r) (4.53)
Again, G
a
is tabulated for c = 5 in Wimp (1962) (page 454).
We now have a connection between /
a
, G
a
and c
a
. but in order to solve for
/
a
. according to Luke (1969), (page 298), we need to observe that:
r1
0
2a+1
(r) = (2: + 1)
_
1
2a+1
(r) + 2
a1

0
1
2I+1
(r)
_
(4.54)
4. RATE DECLINE OF A FRACTURED WELL 141
Deleting some steps of computation, this can be inverted to yield:
r
1
1
2a+1
(r) =
1
2: + 1
1
0
2a+1
(r) + 2
a1

I=0
(1)
aI
2/ + 1
1
0
2I+1
(r) (4.55)
After some additional computations, we can conclude that:
/
a
=
_
o2
2a+1
(G
a
G
a+1
)
_
+
1
2: + 1
_
c
a
+ 2
1

n=a+1
(1)
n+a
c
n
_
(4.56)
(: _ 1)
Finally, we obtain the following very ecient nesting (recursion) procedure an-
nounced above to evaluate the sums occurring in (4.38), (4.39), (4.43), (4.44),
(4.45) and (4.46), of the forms
,(r) =
.

a
1

a
(r) (4.57)
,(r) =
.

a
1

a
(r) (4.58)
q(r) =
.

0
1
a
1
2a+1
(r) (4.59)
,(r) = c
0
+ c
1
(1 2r) c
a
= (4r 2)c
a+1
c
a+2
+
a
(4.60)
q(r) = (/
0
/
1
)r /
a
= (4r
2
2)/
a+1
/
a+2
+ 1
a
(4.61)
Starting with c
.+1
= c
.+2
= /
.+1
= /
.+2
= 0 and using the backwards
recursion in (4.60), we end up with our evaluations of (4.57). The procedure is
based on the work of Clenshaw (1955).
4.2 Depletion Oil Flow
This section deals with two semi-analytical model solutions, one for a verti-
cal fractured well and another for a horizontal-fractured well. For each case
we develop and discuss the rate-time solutions by considering a close reservoir
geometry. Two fracture conductivities are considered: the innite conductivities
and a uniform ux. For each model, we set up several assumptions to facilitate
a mathematical formulation that is made both accurate and simple. In the two
model approaches, the following is assumed:
142 4. RATE DECLINE OF A FRACTURED WELL
The porous medium is isotropic, homogeneous, and bounded by upper and
lower impermeable strata.
The reservoir is bounded, corresponding to a BOX geometry, of constant thick-
ness, h. Other reservoir parameters such as porosity, c, and permeability, k, are
also assumed constant.
The reservoir contains a slightly compressible single-phase uid of constant
compressibility, c, and constant viscosity, j. The ow in the formation is as-
sumed to be of Darcy type.
The well is produced at constant pressure and an incompressible non-Darcy
ow is assumed to occur within the fracture.
4.2.1 Fractured Vertical Well Model
Suppose that a vertical well is located in a closed or bounded rectangular reser-
voir (a reservoir geometry also known as the BOX geometry). The reservoir is
positioned horizontally and extends in directions x, y and z. A fracture fully
penetrates the reservoir in its z and x directions. The above assumption for a
fracture geometry makes this model one dimensional with the variable, y, as a
space variable.
The inner boundary condition is of a constant well owing pressure, j
&)
.
Carslaw and Jaeger (1959) presented the dimensionless rate,
1
. versus the di-
mensionless time, t
1
. as a late time behaviour of the linear closed system. Vil-
legaz (1997) solved the diusion equation in one dimension for the inner bound-
ary condition of constant pressure.
The system in Figure 4.3 simulates the behaviour of a vertical fractured well.
The uid and the formation compressibility relate to pressure according to:
c
)
=
1
j
dj
dj
. c =
1
c
dc
dj
(4.62)
The one-dimensional diusion equation is:
J
2
j
J
2
=
cjc
0.00633/
Jj
Jt
(4.63)
and the initial and boundary (inner and outer) conditions are:
j(. 0) = j
i
. j(0. t) = j
&)
.
Jj
J
= 0 (4.64)
The solution is given for the pressure, j(. t). The rate is calculated by
(. t) =
Jj(. t)
Jt j=0
. (4.65)
Moreover, the variables for dimensionless time, t
11)
4. RATE DECLINE OF A FRACTURED WELL 143
Figure 4.3: An innite conductivity vertical fracture fully penetrating the x
direction of a reservoir and the formation in the vertical z direction. The no-ow
outer boundary condition denes the closed rectangular reservoir (Areal cross
section).
t
11
]
=
0.00633/
cjc1
)
t (4.66)
and the dimensionless rate,
1
. is

1
=
141.21j
(j
i
j
&)
)
(t) (4.67)
This nally denes
1
(t
11)
) as,

1
_
t
11
]
_
=
4
:
_
1
)

c
_
1

a=0
c
_
_
_
(2a+1)
2
_
:
2
_
2
_
_
1
)

c
_
_
t
T1
]
_
_
_
(4.68)
The above solution can be inverted into the Laplace space. For this, Equation
(4.68) is rst subject to the following transformation:

1
=
_
4
:
__
1
)

c
__
c
:t
1
2
o
3
_
1
2
[
c
:
2
t
_
(4.69)
144 4. RATE DECLINE OF A FRACTURED WELL
with
c =
_

c
1
)
_
2
. (4.70)
Now, by taking the "Laplace" transformation of
1
. we have
1(
1
) =
2
:
_
:
1

a=1
(1)
a
_
:
:
c
2
p
ccjaj
=
2
:
_
:
_
1 + 2
1

a=1
(1)
a
c
2
p
cca
_
2
:
_
:
_
1 2
c
2
p
cc
1 +c
2
p
cc
_
=
2
:
_
:
1 c
2
p
cc
1 +c
2
p
cc
=
2
:
_
:
tanh
_
c: (4.71)
and nally, after substituting Villegaz (1997) obtained for the "Laplace" trans-
form of
1
:

1
=
2
:
tanh
_

c
1
)
_
_
:
. (4.72)
By using the inverse "Laplace" transformation of Stehfest (1970), it is possible
to invert the rate dened in the "Laplace" space,
1
. to the real time solution,

1
:

1
= 1
1
(
1
) (4.73)
The nal comment is that owto a wellbore fromthe reservoir is facilitated by
a vertical fracture of the innite fracture conductivity. The fracture is of length
21
)
, width /
)
, and fully penetrates the wellbore. The fracture permeability,
/
)
, is assumed constant. This simplied modelling approach is one-dimensional.
The vertical fracture fully penetrates the formation and completely extends in
the r.direction of the reservoir.
4.2.2 Horizontal Well with Transversal Fractures
Suppose that a model with a horizontal well is located in the middle of a closed
and bounded (BOX type geometry reservoir). All BOX sides are no-ow bound-
aries. A well can be fractured with transversal fractures, and the fundamental
assumptions of a multi-fractured horizontal well model include:
Fractures that are vertically fully penetrating, of equal size and spacing, for
each of which is assumed a uniform ux ow.
4. RATE DECLINE OF A FRACTURED WELL 145
Figure 4.4: A fractured-horizontal well of length L with three transversal frac-
tures of half-length 1
)
. The reservoir is bounded and no-ow boundaries are X
c
and Y
c
(Areal cross section).
No direct ow to the wellbore, which passes perpendicularly through the mid-
points of the fractures.
Additionally, there is no pressure drop along the wellbore.
Omitting many details of the solution to the diusion equation with the inner
boundary condition of pressure or rate and the no-ow outer boundary condition,
we obtain an expression for the normalised dimensionless pressure (drop), j
1
,
at the midpoint of a fracture number, i, as a linear combination of individual
fracture rates all in Laplace" space.
j
i1
=
.

)=1
1
i)

)1
(i = 1. 2. .... `) (4.74)
For the coecients, 1
i)
. by solving the diusion equation with sources through
Greens function method omitting details we nd:
1
i)
=
:i1
)
r
c
c/j
_
: + c/i
_
:
_
: :/c
_
:
+
r
c
:1
)
1

I=1
sin
2I1
]
a
c
/
c/j
_
: + /
2
c + c/i
_
: + /
2
c
_
: + /
2
c :/c
_
: + /
2
c
(4.75)
146 4. RATE DECLINE OF A FRACTURED WELL
with coecient c equal to c
c =
i
c
1
)
. c =
4:
2
1
2
)
i
2
r
2
c
j = j
i)
=
i
c
1
)
[i ,[d
=
i)
=
i
c
1
)
(i + , 2)d
d =
i
1
)
1
` 1
(` 1). .......d = 0 (,o: ` = 1)
i =
6
_
/

/
I
New variables, specic to the model geometry, are r
c
and
c
giving the horizontal
extension (side lengths) of the BOX model.1
)
is the fracture half-length (in
the r-direction), and 1 is the well length (in the y-direction). The midpoint of
the well coincides with the midpoint of the reservoir.
The expression for 1
i)
is numerically unsuitable, and thus it needs to be
transformed, which is re-entered possible through the connection with elliptic
theta-functions. We nd (omitting details) that:
1
i)
=
1
)
2
c
a
0
_
0
1o
3
_
r
2:

ct
:
2
_
o
4
_
j
2c

t
c
2
_
+ o
3
_
r
2:

ct
:
2
_
o
4
_
i
2c

t
c
2
_
dr (4.76)
where 1 stands for the "Laplace" transform with respect to t (dimensionless
time), where r
0
=
2 1
]
a
c
.and where the theta-functions o
3
, o
4
are given by the
dual formulas:
o
3
_
r
2:

ct
:
2
_
=
+1

1
c
ca
2
t
cos :r (4.77)
o
4
_
i
2c

t
c
2
_
=
+1

1
(1)
a
c

r
2
a
2
n
2
I
cos
: i
c
: (4.78)
o
3
_
r
2:

ct
:
2
_
=
_
:
ct
+1

1
c

(i+2r n)
4 oI
2
(4.79)
o
4
_
i
2c

t
c
2
_
=
c
_
:t
+1

1
c

[ +(2n+1)a]
2
4I
(4.80)
4. RATE DECLINE OF A FRACTURED WELL 147
The two sets of formulas for o
3
, o
4
have extremely good convergence properties,
each one perfectly complementing the other. Using formulas (4.77, 4.78, 4.79,
and 4.80) in the expression for 1
i)
, we nally arrive at the following:
1
i)
=
: i
2 c
_
c
_
:

c/ j
_
: + c/ i
_
:
:/c
_
:

a
(1)
a

cos
: j
c
: + cos
: i
c
:
: +
:
2
c
2
:
2

:/
_
_
:
_
c
i
_
_
: +
:
2
c
2
:
2
_
:/
_
:
c
_
: +
:
2
c
2
:
2
_ (4.81)
where
c
a
=
2
1
: = 0
: 2
(4.82)
This nal formula is very suitable for numerical calculations on the basic geom-
etry BOX model.
4.2.3 Model Solutions Comparison
Model solutions of a vertical-fractured well and a horizontal-fractured well that
is coupled with a single transversal (longitudinal) fracture to an oil reservoir
are considered, and rate-time obtained results are compared. There exist model
rate-time solutions for a vertical and a horizontal single-fracture well, as sketched
in Figure (4.5), positioned in a nite reservoir geometry. Table (4-1) sumarises
"Laplace" space rate-time solutions for the vertical fractured well and the hori-
zontal fractured well (with a transversal fracture). We summarise the two model
solutions: the vertical well-fractured model and the horizontal well-fractured
model. A vertical well fracture is vertical, longitudinal and has innite con-
ductivity. A horizontal well is fractured with one transversal and uniform ux
fracture. The two model solutions are given in Laplace" space. From the
"Laplace" space solutions, the real time solutions may be obtained by the In-
verse "Laplace" transform with the procedure of Stehfest (1973). The solution
for both a vertical-fractured well, and a horizontal-fractured well (transversal
and uniform ux fracture) are developed for a closed bounded reservoir with
an inner boundary condition of constant pressure. Model solutions are given in
Table ??.
From Table (4-1), we plot the two model rates,
1
, versus, :, both in the
Laplace" space. Figure (4.6) represents a fractured vertical-well rate
1
versus
:. Figure (4.7) presents a fractured horizontal-well rate,
1
. versus, :. For
comparison both solutions are plotted in Figure (4.8). Here we can comment
that a small : means a large time in a real space, whereas large : corresponds
to early time solutions.
148 4. RATE DECLINE OF A FRACTURED WELL
Table 4-1: Solution for a fractured-vertical (longitudinal) and a fractured-
horizontal (transversal) well
4. RATE DECLINE OF A FRACTURED WELL 149
Figure 4.5: A vertical-fractured, and a horizontal- fractured well (with transver-
sal and longitudinal single fractures).
Table ( 4-1) model the vertical-fractured well (with a vertical innite con-
ductivity fracture) and the horizontal-fractured (with a transversal uniform ux
fracture) well solutions. Both solutions are plotted in Figure (4.8).
The real space solution should be obtained by the Inverse Laplace transform
method. Several Inverse Laplace transform methods are described in Appendix
B.
150 4. RATE DECLINE OF A FRACTURED WELL
Figure 4.6: The model for a vertical-fractured well rate,
1
. versus s. The
fracture is longitudinal and of innite conductivity (Laplace" space solutions).
Figure 4.7: The model for a fractured-horizontal (with a transversal fracture of
uniform ux) well rate,
1
. versus s (Laplace" space solutions).
4. RATE DECLINE OF A FRACTURED WELL 151
Figure 4.8: The two model solutions for the rate,
1
. versus s (Laplace" space
solutions).
4.3 Additional Well-Fracture Features
4.3.1 Fracture Conductivity
If we further consider, the same reservoir pressure equation as before, with a
solution for the dimensionless pressure (drop) in "Laplace" space we get:
j
v1
=
.

i=1
1
_
1

)i
(r
0
)1
0
_
_
:
_
[r r
0
[ +[
i
[
_
dr
0
(4.83)
On the other hand, now also we have an equation for the (one-dimensional) ow
in each fracture, yielding a pressure solution of the form (i=1,2,...,N)
j
)1
=
&i
,(r. :)
1
_
1

)i
(r
0
)q(r r
0
. :)dr
0
(4.84)
Identifying the pressure expressions j
v1
and j
)1
along each fracture gives a
system of integral equations for the fracture rates. By dividing each fracture
into a number of 2M gridblocks and assuming a uniform ux over each block, we
can discretise the system of integral equations into a set of M.N linear algebraic
equations for the discretised uxes,
_

na
. The centre of each gridblock is chosen
for identication of pressures.
Further, as reported by Cvetkovic, Halvorsen and Sagen (1999), the total
uxes,
_

a
,correspond to the term,
&i
. in Equation (4.84). For the total rate
_

we have the following relation:


152 4. RATE DECLINE OF A FRACTURED WELL
.

i=1
_

a
=
_
(4.85)
This gives us a total of `.` +` + 1 equations for as many unknown qualities
_

na
(: = 1. 2. .... `.and : = 1. 2. .... `),
_

a
(: = 1. 2. .... `), and either rate,
_
.
or pressure,
_
j. With a simplied notation, we obtain the following system of
equations:
.

a=1
A

n=1
c
n
0
a
0 (:. :)
_

n
0
a
0 = d(:)
_

n
0
=1
d
n
(:)
_

na
(4.86)
A

i=1
c
n
_

na
= c
0
_

_
j (4.87)
.

i=1
_

a
=
_
(4.88)
More details, including formulas are given by Cvetkovic et al. (1999).
4.3.2 Well Conductivity
To calculate the well pressure drop resulting from the nite conductivity, we
assume that the wellow is stationary and laminar (viscous or streamline ow).
We use the so-called Hagen-Poiseuille law Equation (4.89) which states that the
pressure drop is proportional to the well ow and the distance travelled. The
well ow or the volumetric ow rate, q , is a Newtonian uid in steady laminar
ow through a pipe.
dj
dr
=
8j
::
4
&
(4.89)
The Haugen-Poiseulle law, Equation (4.89), results from an averaging over the
wellbore cross section of a parabolic velocity prole. This, in turn, is obtained
from integrating an assumed radially linear shear stress relation for a Newtonian
uid. With the usual notation for dimensionless quantities we get from Equation
(4.89):
j
&1
=
16//1
)
:
4
&

&1
r
1
(4.90)
Using this relation along the wellbore, summing the contributions from fractures
1, 2, ... up to n, we obtain:
4. RATE DECLINE OF A FRACTURED WELL 153
T
a
b
l
e
4
-
2
:
F
i
n
i
t
e
a
n
d
i
n

n
i
t
e
c
o
n
d
u
c
t
i
v
i
t
y
f
r
a
c
t
u
r
e
s
-
p
a
r
a
m
e
t
e
r
s
)
F
i
n
i
t
e
C
o
n
d
u
c
t
i
v
i
t
y
I
n

n
i
t
e
C
o
n
d
u
c
t
i
v
i
t
y
c
0
2
c
p
c
c
o
t
h
_
p
c

_
2
c

c
=
2

1
]
&
:

1
c
n
4
c

c
c
o
s
h
_
L

r
+
1 2
L
_
p
s

s
i
n
h
p
s
2
L

s
i
n
h
p
s

4
c

c
1
2
A
=
2

1
]
A
&
:

1
d
(
:
)
2
c
p
c
c
o
s
h
L

r
+
1 2
p
s
L
s
i
n
h
p
s

2
c

c
=
2

1
]
&
:

1
(
=
l
i
m
c
0
)
d
n
0
(
:
)
2
c

c
c
I
p
s
2
L

c
I
p
s

_
c
/
A

j
n

n
0
j

p
c

+
c
/
A

n
0
+
1
A
p
c

_
2
c

c
1
2
A
2
=
2

1
]
A
&
:

1
(
=
l
i
m
c
n
)
d
n
4
c

c
c
I
p
s
4
L

c
I
p
s

_
c
/
p
c
4
A

_
c
/
A

2
n
+
1
A
p
c

+
c
/
4
A

1
4
A
p
c

4
c

c
2
4
A
=
2

1
)
A
&
:

1
(
=
l
i
m
c
n
0
)
c
n
0
a
0
n

n
0
+
1 2
_
r

r
0

1 2
L
1
c
_
_
:
_
r
2
+
(
:

:
0
)
2
,
2
_
d
r
+
n
+
n
0
+
1 2
_
r
+
r
0

1 2
L
1
c
_
_
:
_
r
2
+
(
:

:
0
)
2
,
2
_
d
r
c
n
0
a
0
a
s

n
i
t
e
c
o
n
d
u
c
t
i
v
i
t
y
,
=
1

1
]
.

1
(
`

1
)
&
,
=
0
(
`
=
1
)
;
c
=

1
]
b
_
I
I
]
;

=
_
I
]
I
/
)

0
&

154 4. RATE DECLINE OF A FRACTURED WELL


_
j
&1
=
16//1
&
:
4
&
(` 1)
a1

i=1
(: i)
_

i1
. : = 1. .... ` (4.91)
In Equation (4.91), fractures are numbered beginning at : = 1 at the toe of the
well and ending at : = ` at the wells heel. Incorporating this extra pressure
drop into the equations valid for an innitely conductive wellbore, gives us the
following modied set of equations, to be substituted for Equation (4.87), thus
leading to:
A

i=1
c
n
_

na
= c
0
_

_
j+
16//1
&
(` 1):
4
&
_
.1

a
(` i)
_

i
+ (` :)
a1

1
_

i
_
. : = 1. 2. .... `
(4.92)
We note here that for : = `. the last bracketed expression equals 0, correspond-
ing to the choice of a new wellbore pressure (drop), j.to be measured at the
wells heel.
4.3.3 Fracture-Well Limited Communication (Choking Ef-
fect)
Following the work of Schulte (1986), which disccused the inuence of a limited
communication interval on the transient pressure behaviour and the long-term
productivity of a fractured well, Equation ( 4.94) presents the dimensionless
pressure (drop) or the "skin factor", o
|c
. This option was only implemented
within the model for the IBC of constant rate. Derivation details was published
by Cvetkovic, Halvorsen and Sagen (1999) (pages 8-11). From the formula for
the pressure (drop)
j
1
= j
2://
`j
=
:/
`/
)
/
)
. (4.93)
and after averaging the total inow over N fractures the expression for the skin
factor, o
|c
or the dimensionless pressure (drop)
o
|c
= j
1
=
//
`/
)
/
)
log sin
:c:
&
/
(4.94)
Here, |
)
= c:
&
corresponds to an "equivalent fracture half-length". This choking
eect option is presented as a case study in Chapter 6.
4.3.4 Restart Option
Here we describe changing the Inner Boundary Condition (IBC) of constant-rate
to one of constant-pressure within time for a multi-fractured horizontal well in
4. RATE DECLINE OF A FRACTURED WELL 155
an innite reservoir SLAB model and closed reservoir BOX model. We generally
have a linear relationship in "Laplace" space as follows:
j = 1 (4.95)
For the basic restart option treated here, the input parameters include a period
of constant rate, followed by the pressure maintained constant:
=
0
,o: 0 < t < t
0
j = j
0
,o: t t
0
(4.96)
We assume that
j (0) = 0
and further that
j
0
= lim
t!t
0
j (t) (4.97)
The value of j (t) for t < t
0
is easily calculated in the usual way.
We need a new means of computing (t) for t t
0
, and the easiest way is
detailed bellow:
From (4.95) we have:
= 1
1
j (4.98)
which we rewrite
= 1
1
:
1
. :j (4.99)
This gives the convolution integral
(t) =
t
0
_
0
1
1
_
1
1
:
1
_
(t t) 1
1
(:j) (t) dt (4.100)
generally now we have
1
1
(:j) = j
0
(t) (4.101)
provided that
j (0) = 0
as was also assumed. For
156 4. RATE DECLINE OF A FRACTURED WELL
t t
0
. j
0
(t) = 0
we hence obtain
(t) =
t
0
_
0
1 (t t) j
0
(t) dt (t < t
0
) (4.102)
In Equation (4.102) 1(t) represents the rate response to a unit pressure (drop)
corresponding to
j = :
1
(4.103)
while j(t) is the pressure response of the system to the constant rate,
0
. for
t < t
0
. We may factor out the constant
0
, relating also j(t) to a unit rate
( = 1)
nally obtaining
(t) =
0
t
0
_
0
1 (t t) j
0
(t) d t
(where j(t) is now being a unit rate response.)
A simple numerical approximation can be written:
(t) -
0
.1

i=0
1 (t t
i
) [j (t
i+1
) j (t
i
)] (4.104)
with t
i
=
i
.
t
0
and a suciently large `.Equation (4.104) is valid for t t
0
and. 1(t) and j(t) are responses to the unit pressure and unit rate, respectively.
Formula (4.95) results from the elimination of (:) individual fracture rates
from a linear system of (:+1) equations. These results are obtained from solving
the basic pressure equation for the system of : fractures connected by a wellbore.
In the simplest case concerning transversal, equally spaced and fully penetrating,
uniform ux fractures in a SLAB reservoir, our basic "transfer" function 1(:)
from formula (4.95) can be rationally expressed by integrals of the following type:
1
)
= 1
)
(:) =

_
0
1
0
__
:
_
_
_
r
2
+ ,
2
,
2
_
dr (4.105)
with
4. RATE DECLINE OF A FRACTURED WELL 157
=
6
_
/

/
I
(4.106)
, =
1
(: 1) 1
)
(4.107)
Where, 1. length of wellbore, 1
)
. half length of fracture. For 4 fractures, : = 4.
we nd function 1(:)
1(:) =
1
2
0
(:) + 1
0
(:)1
1
(:) 1
2
1
(:) + 1
0
(:) 1
3
(:)
41
0
(:) 21
1
(:) 41
2
(:) + 21
3
(:)
+
1
1
(:) 1
3
(:) 21
1
(:) 1
2
(:) 1
2
2
(:)
41
0
(:) 21
1
(:) 41
2
(:) + 21
3
(:)
With an increasing number of fractures, the expression for 1(:) becomes very
long. We can nevertheless, easily handle 50 or more fractures, as we have to our
disposal fast and robust numerical methods that approximate the Bessel function
integrals by Chebychev polynomials that is combined with a exible numerical
"Laplace" inversion method.
4.3.5 Late Time Approximations
In this section, we develop approximate late time expressions for a fractured-
horizontal well. These expressions are related to a vertical well production, and a
production of a horizontal well (without fractures) production and to a horizontal
with a single transversal fracture. In order to measure the production eciency
of a fractured-horizontal well, we introduce the eective wellbore radius and the
eective single-fracture half-length. The eective wellbore radius is dened for a
vertical well without fractures, and for a horizontal well, also without fractures.
Furthermore the eective half-length is dened for a horizontal well with a single
transversal fracture. It is assumed that a transversal fracture ow signies a
uniform ux. Late time approximations are developed for a fractured-horizontal
well positioned in the middle of an innite reservoir. The developed expressions
from Section (4.1) for a multi-fractured horizontal well are utilised for dening
an eective wellbore radius and an eective fracture half-length. The production
of a horizontal-fractured well may be related to the vertical well production in
addition to the single-fracture horizontal well production. Figure (4.9) illustrates
the models mentioned above.
Vertical-Well Eective Wellbore Radius
The following relation is developed for a vertical-well eective wellbore radius,
:
&
:
158 4. RATE DECLINE OF A FRACTURED WELL
Figure 4.9: Models for a fractured-horizontal well, the eective wellbore radius
of a vertical well, and the eective half-length of a horizontal well with a single
transversal fracture.
:
&
= 21
)
c
(+
^
)
(4.108)
Here, 1
)
. as before, refers to the half-length of the fractures, is Eulers constant
( - .577), ` is the number of fractures, and
.
is given as:

.
=
0
+ 1
00
. (4.109)
for which the generalised expression 1
00
is dened in Equation (4.112) and
0
corresponds to:

0
= log 2 + 1. (4.110)
yielding
:
&
=
1
)
c
c
1
00
(4.111)
As an example, from Equation (4.111) we calculate the vertical-well eective
wellbore radius for a fractured-horizontal well with number of fractures N= 1,
2, and 3.
For ` = 1, we obtain 1
00
= 0, which gives the same value for the eective
wellbore radius, :
&
. as noted by Hegre and Larsen (1994). We can obtain a
4. RATE DECLINE OF A FRACTURED WELL 159
general, explicit value for 1
00
, and hence also for :
&
, in the case of a general
`. The generalised expression for 1
00
is:
1
00
=
_
c
.
1
1
0
c
T
.
_
1
(4.112)
where the matrix elements of 1
0
are:
(1
0
)
n,a
= : cot
1
:
1
2
log(1 + :
2
) (4.113)
and where : = [::[
11
]
.1
(` _ 2). Explicitly, we also nd, for ` = 2 and
: =
1
1
]
1
00
=
1
2
: cot
1
:
1
4
log(1 + :
2
). (4.114)
for ` = 3:
1
00
=
2c
2
4c /
. (4.115)
where
c = : cot
1
: +
1
2
log(1+:
2
). / = 2: cot
1
(2:)+
1
2
log(1+4:
2
). : =
1
21
)
(4.116)
Hence, for various `.we obtain expressions for :
&
including dimensionless ra-
dius r as : =
_
1
1
]
_
:
` = 1 : :
&
=
1
)
c
(4.117)
` = 2 : :
&
=
1
)
c

4
_
1 +:
2
c
1
2
v cot
1
v
1
)
c
c
12
:
12
(: ). (4.118)
= 3 : r
wv
=
1
f
c
exp
_

_
2
_
r cot
1
r +
1
2
log(1 + r
2
)
_
2
2r cot
1
2r
3
1 + 3r
2
+
1
2
log
(1 + r
2
)
4
1 + 4r
2
_

_
~
1
)
c
2
29
c
43
:
23
(: ). for : =
_
1
21
)
_
(4.119)
corresponding to an identication of the late time behaviour of a cylindrical,
fully penetrating well and our fractured horizontal well. When we actually com-
pare "Laplace" transforms for large time, t , (or small s-values), starting with a
160 4. RATE DECLINE OF A FRACTURED WELL
classical expression for the vertical well dimensionless pressure transform, j
1
is
given by Raghavan (1993), as
j
1
= C
1
0
(:
&1
_
:)
:
&1
_
:1
1
(:
&1
_
:)
:
1
(4.120)
where
_
C =
cq1j
2II
_
. For late time behaviour (or small s-values), we get the di-
mensionless pressure, j
1
.
j
1

1
2
C:
1
log
_
c
2
::
2
&1
4
_
(4.121)
where,
:
&1
=
:
&
1
)
(4.122)
For the fractured horizontal well:
j
1
=
C
2
:
1
(log : 2
.
+ C(: log :))
C
2
:
1
log
_
:c
2
^
_
(4.123)
By identifying the two expressions for j
1
, and dropping terms corresponding to
terms on the order of t
1
(t ) in the real time variable, we nd the result for
a vertical well eective wellbore radius, :
&
reported at the start of this section
in Equation (4.111).
Horizontal-Well Eective Wellbore Radius
An eective wellbore radius converts the productivity of a horizontal well into
an equivalent vertical well productivity.
:
0
&
= :
&
c
c
Joshi (1991) presented a steady-state solution for an eective radius. For a
horizontal and a vertical well, the drainage volumes were set to be equal, :
cI
=
:
c
, as were the productivity indices:
_

j
_

=
_

j
_
I
(4.124)
By substituting the steady-state rate solutions for a horizontal well rate,
I
and
a vertical well rate,

, we can dene the expression for an eective wellbore


radius of a horizontal well, r
&I.
_

j
_
v
=
2
/
h
/
j
o
1
o
ln
r
e
r
w
_

j
_
h
=
4. RATE DECLINE OF A FRACTURED WELL 161
=
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
2:/
I
/
j
c
1
c
ln
_

_
c +
_
c
2

_
1
2
_
2
1
2
_

_
+
_
/
1
_
ln
_
/
2:
u
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
I
(4.125)
where
c = (1,2)
_
0.5 +
_
0.25 + (2:
cI
1)
4
(4.126)
The equal productivity indices from above provide an eective horizontal well-
bore radius in an isotropic reservoir:
:
&I
=
:
cI
1
2c
_
1 +
_
1 [1, (2c)]
2
_
[/,(2:
&
)]
(I1)
(4.127)
Now, by including
1
)
c
= :
&
=
1
)
(`)
c
c
1
oo
(.)
(4.128)
we are able to express the horizontal wellbore radius, :
&
. to a single fracture half
length, 1
)
.
For an anisotropic reservoir the eective wellbore radius becomes:
:
&I
=
:
cI
1
2c
_
1 +
_
1 [1, (2c)]
2
_
[,/,(2:
&
)]
(oI1)
(4.129)
The above work can be extended to several eects such as shape factor, well
position and well skin.
A Multifractured-Horizontal Well Eective Wellbore Radius
The eective wellbore radius for a horizontal-well with N transversal fractures
being associated to a vertical well corresponds to:
:
&
(`) =
:
&c
`
=
1
)
(`)
c
c
1
oo
(.)
. (4.130)
162 4. RATE DECLINE OF A FRACTURED WELL
For one fracture, ` = 1 and 1
cc
= 0, so
:
&
(1) =
1
)
(1)
c
(4.131)
or by equalizing :
&
(1) = :
&
(`), we get,
1
)
(1)
c
=
1
)
(`)
c
c
1
oo
(.)
(4.132)
and
1
)
(1) = 1
)
(`) c
1
oo
(.)
(4.133)
or
1
)c))
= c :
&
(`) (4.134)
Further, the comparison of solutions a horizontal-fractured well (with a single
transversal fracture) and those from literature concerning a vertical-fractured
well in an innite reservoir are given in Table (4-3).
4. RATE DECLINE OF A FRACTURED WELL 163
Table 4-3: Model solutions for a fractured-vertical well as compared to those of
a fractured-horizontal (single-transversal-fracture) well
164 4. RATE DECLINE OF A FRACTURED WELL
Chapter 5
RATE DECLINE WITH A
MOVING BOUNDARY
Arps empirical depletion-rate solutions from the 1950s involving conventional
interpretation procedures were in the 1980s enlarged into combined transient-
depletion rate type curves. Together with these curves, a decline exponent,
/. and an initial decline, 1
i
. were for the rst time introduced by means of
reservoir-production parameters through the extensive study of Fetkovich (1980).
Raghavan (1993) extended the analysis of the multiphase ow in a reservoir and
presented the theoretical view on the decline exponent, /. He summarised the
most critical items required for matching the decline exponent, /. He further
stated limitations to the empirical and universal decline curve theory experienced
in the 1980s. Raghavans (1993) study evaluates in a more sophisticated manner
the possibility of matching the decline curvature dened by the decline exponent,
/.
Among others, the following issues were discussed by Raghavan (1993) (pages
527-528) concerning the production from a single-layer reservoir:
1. The possibility to obtain a constant value of exponent b for a wellbore
constant-pressure production.
2. The nature of the wellbore pressure response when the production is forced
to follow a specic value of parameter b in the Arps equations.
3. The inclusion of the variable skin factor and a constant value of exponent
b.
Camacho and Raghavan (1989) reported that exponent / in Arps equation
is not constant for an inner boundary condition of constant pressure. Moreover
they stated a wellbore condition of variable pressure for which the rate may be
forced to follow a specic stem of exponent /. They come to the conclusion that,
in some cases, even the wellbore pressure must increase with time leading to the
wellbore response following the specic value of the exponent /. This was based
on the premise that a single-layer system was produced, and was relevant to a
well without skin zone properties and with the constant drainage area. Carter
165
166 5. RATE DECLINE WITH A MOVING BOUNDARY
(1985) with Fraim and Wattenbarger (1987) supported the statement that the
theoretical solutions can not match the exponent b, and moreover that they
should cut accross several values of the decline exponent, /. Another option was
to include the wellbore skin. Then, by varying skin factor for a well producing
at a constant pressure condition, a production characterised by the constant
decline exponent, / is obtained. If the reservoir is layered (with a commingled
production or a system with interlayer communication), a case can be established
for a constant value of exponent /. Both Fetkovich and Raghavan noticed that
transient rate data matched depletion rate data, usually with an exponent /
that is grater than 1. Several case studies have demonstrated such a match as
commented by Raghavan (1993).
Further, this chapter mainly investigates the nature of the wellbore pressure
response when production is forced to follow a specic value of exponent b. We
here present a physical model and solve the diusion equation with variable-rate
inner-boundary-conditions for several selected Arps exponents, b, and for no-
ow moving outer-boundary conditions. All obtained variable pressure functions
with time were analytically derived. This way, we demonstrate the nature of the
wellbore pressure response (and also the pressure response within a drainage
area of a vertical well) when production is forced to follow a specic parameter,
/, in the Arps equations.
The present chapter deals with pressure solutions of a well positioned in an
oil and gas reservoir. In order to investigate the pressure behaviour with time,
we dene the physical model with the wellbore variable-rate condition of Arps
type, and no-ow specied moving outer boundary. The model input parameters
are rates dened by Arps and moving outer boundary, whereas output data are
pressure solutions that are analytically derived. The no-ow outer boundary
moves with one speed of outwards moving. How this speed may be related
to the reservoir condition and drive mechanism should be further investigated.
Since no-ow boundary moves with time, its speed of a no-ow boundary also
changes with time. Ideally, the speed of a no-ow boundary will stop when a
drainage area is reached. For such moving boundary production forced to follow
a specic value of the decline exponent, /. it would be possible to solve the
diusion equation for wellbore pressure responses. The model for the change in
drainage volume is transient, but at the same time the rate-time is depletion
kind of variable rate being of Arps type. Also a no-ow boundary moves and
does not have xed position at the drainage radius. Ideally, the moving no-ow
boundary should reach the drainage radius at the end of the Arps decline.
Solution to the diusion equation for initial and xed boundary conditions
have been well explained in the petroleum literature. We solve the diusion
equation for a single well situated in a circular homogeneous reservoir. To be able
to match Arps exponent, it is necessary to assume a moving no-ow boundary.
We estimate the position of a no-ow boundary, that moves outwards from a
5. RATE DECLINE WITH A MOVING BOUNDARY 167
wellbore axis with the speed proportional to square a root of time. By assuming
such speed, it is possible to solve the diusion equation for the variable-rate IBC
of Arps type. The constant of proportionality, i, may have a physical meaning.
It is supposed to be related to a driving force moving a no-ow boundary. At
each point in the drainage area, the solution consists in a variable-pressure with
time. Pressure response solutions of the diusion equation are presented in the
following sections, for both for oil and gas ow.
5.1 Vertical Well Oil Flow
This section considers the following topics related mainly to transient rate decline:
Solutions to the diusion equation for an oil well in which the wellbore
production varies with time.
A well variable rate production of Arps type, i.e., where the decline expo-
nent, b is known.
A no-ow boundary moving (not xed) outwards from a wellbore axis.
A solution corresponding to the pressure response to a variable rate decline,
dened by the exponent b and the speed of the moving boundary.
The basis for further studying the physical meaning of a coecient, i (i
relates the dimensionless position, :
1
. and speed,
ov
T
ot
T
. of the no-ow moving
boundary with the dimensionless time, t
1
).
In this chapter we consider the moving boundary solutions for the wellbore
conditions of variable rates (empirical rates dened by Arps). The function
is introduced instead of rate such function correspond to specially rates
of power with a negative exponent .The solutions are thus limited to specic
choices of rates, . In order to solve diusion equation with Arps rate decline,
it is necessary to assume a moving no-ow boundary. With time the no-ow
boundary moves outwards from a wellbore axis. In addition, we have in this
study assumed that the zone skin is zero. To our knowledge, for the diusion
equation analytical solutions for the inner boundary condition of an Arps rate
decline have as of yet not been presented. The production in a such case is forced
to follow a specic value of exponent / in the Arps equation. As the drainage
volume changes due to a moving no-ow boundary, we consider the rate to be
transient, i.e., combined with a decline dened by the Arps exponent, /.
5.1.1 Introduction to Moving Boundary Problems
Tarzia (2000) provided a comprehensive bibliography of moving and free bound-
ary problems for the heat-diusion equation. This review contained almost
6000 references in various kinds of publications; mostly western and from a
mathematical-physical-engineering literature. A very rst publication on a mov-
ing boundary is that by Lame and Clapeyron (1981). Italian literature (Fasano-
168 5. RATE DECLINE WITH A MOVING BOUNDARY
Primicerios group) has classied problems for either the heat or diusion equa-
tion, due to boundary problems being divided into: xed, moving and free.
Cryer (1978) discussed the relationship between moving boundary problems
(parabolic and time-dependent) and free boundary problems (elliptic and steady
state). Free boundary details are well presented by Tarzia (2000) in the following
citation: "The free boundary problems for the heat equation are those in which
the spatial domain of the unknown function varies with time because of a law
of movement not known a priori. The fact of not knowing boundary or part of
it, determines, of course, the mathematical need to impose new condition on the
unknown function, which will depend on the physical model studied. In general,
the new condition to be imposed on the unknown function is deduced from the
principle of conservation of energy across the boundary. Thus it follows that
this boundary is the complementary unknown of the problem, and is called free
boundary of the problem under analysis." The free moving boundary will not be
a subject of study in this thesis.
5.1.2 Fixed Boundary
Inner Boundary Conditions of (Constant) Pressure
Boundary-dominated or depletion rate decline models for a vertical well are em-
pirical, analytical and numerical. Ehlig-Economides and Ramey (1981) solved
the diusion equation analytically for an inner boundary condition of constant
pressure (dened at a wellbore), and an outer boundary condition of no-ow
(dened at the distance of drainage radius, :
c
). They provided an expression for
the rate-time decline of various dimensionless radii, :
1
. For a reservoir closed
or no-ow boundary the rate declines exponentially with time. Thus, an ana-
lytically derived exponential decline overlays the empirical Arps decline dened
by the decline exponent, / = 0. This analytically derived model solution is ideal
since it is limited to a single-phase ow within a homogeneous oil reservoir.
In the empirical modelling approach by Arps (1945), the rate declines both
exponentially and hyperbolically with time. An extensive summary of solutions
are given in Chapter 2. Cvetkovic (1992) reviewed transient and depletion lay-
ered reservoir solutions, and Cvetkovic and Gudmundsson (1993) have written
a bibliography of analytical models for constant wellbore pressure IBC. The
transient-depletion rate production was dened by OBCs that were being both
innite acting and no-ow but xed. It would be challenging to consider a tran-
sient rate decline with IBCs of constant and variable pressure and a no-ow
boundary moving outwards. We also consider rate depletion (dened as variable
rate at a wellbore) controlled by a no-ow boundary moving inwards, from the
time when the pseudo-steady-state conditions are reached.
5. RATE DECLINE WITH A MOVING BOUNDARY 169
Inner Boundary Conditions of (Variable) Pressure
Deconvolution is a diagnostic tool that provides the equivalent rate or constant
pressure response of a reservoir system aected by variable-rate or pressure pro-
duction. The method applies to both production and pressure test data analysis.
Due to the fact that production data are generally of poor quality, their analysis
may not be realistic. Deconvolution may provide signicant diagnostics values
for production data, as found in recently published references: Agarwal et al.
(1999), Araya and Ozkan (2002), von Scroeter et al. (2002), Levitan (2005),
Levitan et al. (2006), Ilk and Valko (2005), Ilk and Valko (2006), and Kuchuk
et al. (2005).
5.1.3 Moving Boundary
We consider here an isotropic, homogeneous, stratied or horizontal that is in-
nite in two horizontal directions of a moving no-ow boundary. Within the
reservoir, the thickness, permeability and porosity are uniform The formation
properties are independent of pressure. An upper and a lower impermeable layer
bind the slab reservoir, which contains a single-phase, slightly compressible uid.
The compressibility and viscosity are assumed constant.
Fluid is produced through a fully perforated vertical well, i.e., the perforations
extend over the total vertical height of the reservoir. Originally, the well is
assumed as a line source. At outer boundary conditions, of the reservoir we
incorporate no-ow boundaries, nevertheless moving outwards from the wellbore
axis. The uid ows towards a wellbore of a single-phase slightly, compressible
uid in a porous medium assuming constant reservoir parameters governed by
the linear partial dierential equation in one independent variable pressure, j.
This parabolic pressure equation is given in a known radial from by:
1
:
J
J:
_
:
Jj
J:
_
=
1
j
Jj
Jt
(5.1)
The IBC of the wellbore consists in a constant rate condition and a variable
rate condition. The pressure, j. is a function of the radial distance, :. and
the time, t. The coecient, j =

Ijc
. is known as the diusion coecient. By
introducing the dimensionless time, t
1
. as t
1
=
j
v
2
u
t, the dimensionless distance,
:
1
=
v
v
u
. and dimensionless pressure, j
1
=
j
.
j(v,t)
j
.
j
u]
. we obtain a dimensionless
form of equation (5.1):
1
:
1
J
J:
1
_
:
1
Jj
1
J:
1
_
=
Jj
1
Jt
1
(5.2)
We now introduce the time variable, t, such that t = j t. The diusion equation
(5.1) thus becomes:
170 5. RATE DECLINE WITH A MOVING BOUNDARY
1
:
J
J:
_
:
Jj
J:
_
=
Jj
J(jt)
=
Jj
Jt
(5.3)
Alternatively, in dimensionless form with t
1
= j t
1
. the Equation (5.3) can be
written:
1
:
1
J
J:
1
_
:
1
Jj
1
J:
1
_
=
Jj
1
Jt
1
(5.4)
corresponding to the diusion equation in the dimensionless form of :
1
, t
1
,
and the dimensionless pressure, j
1
. The outer boundary is no-ow and moves
outwards from the wellbore axis according to the following expression: :
1
=
i
0
_
t
1
, in which the distance, :
1
. is proportional to the square root of time, t
1
.
This expression is now changed for time t to j:
1
= ji
0
_
t
1
, or j:
1
= j
1
2
i
0
_
jt
1
,
and further :
1
= j

1
2
i
0
_
jt
1
= i
_
t
1
, where i =
i
0
p
j
. is the no-ow moving
boundary constant. The no-ow boundary moves with dimensionless velocity,
ov
T
ot
T
. equal to:
d:
1
dt
1
=
i
2
1
_
t
1
(5.5)
Inner Boundary Conditions of Constant Rate
Diusion equation solutions for j
1
as a function of the radius :
1
and time t
1
are given below:
j
1
= ,
_
:
1
t

1
2
1
_
= ,(r) (5.6)
We rst calculate j
t
T
on the left-hand side and j
v
T
on the right-hand side of
Eguation (5.4)
j
t
T
=
Jj
1
J t
1
=
J,
J t
1
_
:
1
t

1
2
1
_
=
1
2
:
1
t

3
2
1
,
0
_
:
1
t

1
2
1
_
(5.7)
j
v
T
=
Jj
1
J:
1
= t

1
2
1
,
0
_
:
1
t

1
2
1
_
(5.8)
Further derivatives of Equation (5.4) include:
0
0r
D
(r
D
j
r
D
) =
0
0r
D
_
r
D
0 j
D
0 r
D
_
=
0
0r
D
_
r
D
t

1
2
D
,
0
_
r
D
t
D

1
2
__
= t

1
2
1
,
0
_
:
1
t

1
2
1
_
+ :
1
t
1
1
,
00
_
:
1
t

1
2
_
(5.9)
Equation (5.4) now becomes:
5. RATE DECLINE WITH A MOVING BOUNDARY 171
1
: 1
_
t

1
2
1
,
0
_
:
1
t

1
2
1
_
+ :
1
t
1
1
,
00
_
:
1
t

1
2
1
__
=
1
2
:
1
t

3
2
1
,
0
_
:
1
t

1
2
1
_
(5.10)
or
_
t

1
2
1
,
0
_
:
1
t

1
2
1
_
+ :
1
t
1
1
,
00
_
:
1
t

1
2
1
__
=
1
2
:
2
1
t

3
2
1
,
0
_
:
1
t

1
2
1
_
(5.11)
and
_
:
1
t
1
1
,
00
_
:
1
t

1
2
1
_
+ t

1
2
1
,
0
_
:
1
t

1
2
1
_
+
1
2
:
2
1
t

3
2
1
,
0
_
:
1
t

1
2
1
_
_
= 0
By multiplying with t
1
2
1
we have:
_
:
1
t

1
2
1
,
00
_
:
1
t

1
2
1
_
+ (1 +
1
2
:
2
1
t
1
1
),
0
_
:
1
t

1
2
1
_
_
= 0
for,
:
1
t

1
2
1
= r (5.12)
r ,
00
(r) +
_
1 +
1
2
r
2
_
,
0
(r) = 0 (5.13)
,
00
(r) +
_
1
r
+
1
2
r
_
,
0
(r) = 0 (5.14)
,
00
(r)
,
0
(r)
+
_
1
r
+
1
2
r
_
= 0 (5.15)
Hence,
log ,
0
(r) + log r +
1
4
r
2
= C (5.16)
,
0
(r) = 1 r
1
c

1
4
a
2
(5.17)
where 1 is the integration constant that can be adapted for any additional
conditions. Further integration gives, for , (r) .
, (r) = 1
a
0
_
0
r
1
c

i
2
4
d r (5.18)
172 5. RATE DECLINE WITH A MOVING BOUNDARY
Figure 5.1: The dimensionless pressure, j
1
, as a function of the dimensionless
time, t
1
and radial distance, :
1
( with 1 = 1, and speed r = 10).
:
1
Jj
1
J:
1
= :
1
t

1
2
1
,
0
_
:
1
t

1
2
1
_
= 1 ,
0
_
:
1
t

1
2
1
_
(5.19)
:
1
Jj
1
J:
1
= 1 c

1
4
a
2
; r = :
1
t

1
2
1
(5.20)
The obtained solutions can be considered as basic due to the imposed premises.
Figure (5.1) presents the dimensionless pressure, plotted versus the dimensionless
time, t
1
, and radial distance, :
1
. The dimensionless pressure can be calculated
for a constant rate production given by the integral constant 1 = 1, and speed
of a moving boundary proportional to the constant r = 10).
A 2D plot of the dimensionless pressure versus the dimensionless time, given
in Figure (5.2) is calculated for a rate production dened by the constant 1 = 1,
a no-ow moving boundary constant, r = 10. and a constant r = :
1
t

1
2
1
.
For an IBC of constant production rate, the decline in dimensionless pressure
is more evident at times. In both Figure (5.2) and Figure (5.3), the constant
rate parameter, 1. and the moving boundary parameter, r. are constant.
Variable Rate Production
A diusion equation with an inner boundary condition of variable rate and an
outer boundary condition of no-ow, and moving outwards from the wellbore
axis, can be reduced to a system of ordinary dierential equations by the tech-
nique of separation of variables. We assume a solution that is dependent on the
dimensionless radial space, :
1
. and the dimensionless time, t
1
. As a mathe-
matical entity, the diusion equation has two very important features: linearity
5. RATE DECLINE WITH A MOVING BOUNDARY 173
Figure 5.2: The dimensionless pressure, j
1
. as a function of the dimensionless
time, t
1
. for the dimensionless distance, :
1
= 1 ( with 1 = 1, and r = 10).
Figure 5.3: The dimensionless pressure, j
1
. as a function of the dimensionless
time, t
1
. for the dimensionless distance, :
1
= 10 ( with 1 = 1 and r = 10).
174 5. RATE DECLINE WITH A MOVING BOUNDARY
and separability. The fact that the equation is separable means that we can
decompose the original equation into a set of uncoupled expressions for the eval-
uation of j
1
in a space dimension 1 and a time dimension 1. This signies that
whatever solution one nds for j
1
, it can be represented as the multiplication
of separate solutions. Thus j
1
can be written as a product of terms that each
depends on a single coordinate:
j
1
= 1(:
1
) 1 (t
1
) (5.21)
In order to solve the problem, this yields to:
:
1
11
0
= 1
_
1
0
+ :
1
1
00
_
(5.22)
After we substitute the separable form of j
1
and divide both sides by :
1
11. we
have:
1
0
1
= :
1
1
_
1
0
1
+ :
1
1
00
1
_
= / (5.23)
Dividing both sides by :
T
11 results in the left-hand side having no space de-
pendence, and the right-hand side having no time dependence. The only way
that an equation strictly in time (the left-hand side) can be equal to an equation
strictly in space (right-hand side), is if they are both equal to a constant. This is
due to the time and space variables being arbitrarily varied. By convention, the
constant was taken to be /. and this peculiar constant will make more sense
when compared to the actual solution.
The left-hand side of the equation can be integrated to nd 1(t
1
)
1(t
1
) = c
It
T
(5.24)
The right-hand side, with :
1
. is a Bessel equation:
:
1
1
00
+ 1
0
+ / :
1
1 = 0 (5.25)
Its solution is given by Bessel functions
1 (:
1
) = 2
0
_
:
1
_
/
_
(5.26)
where a general zero-order Bessel function of rst kind, 2
0
is a linear combination
of Bessel functions J
0
and 1
0
. and 1 are constants that should be chosen.
2
0
= J
0
+ 11
0
(5.27)
Being a linear equation means the equation itself has no powers or complicated
functions of the function j
1
This renders it possible to think of the equation
as a so-called linear dierential operator, 1. acting on the function j
1
, so
that 1[j
1
] = 0. According to mathematical physics, the dierential equation is
5. RATE DECLINE WITH A MOVING BOUNDARY 175
linear, and has more than one solution, a new solution can be obtained by adding
together other solutions. This means that if 1[j
1
I
] = 0 for some solution j
1
I
,
we have 1
_

I
c
I
j
1
I
_
= 0 for any arbitrary constants c
I
. From this we may
form linear combinations:
j
1
=

I
C (/) c
It
2
0
_
:
1
_
/
_
(5.28)
or
j
1
=
b
_
o
C (/) c
It
2
0
_
:
1
_
/
_
d/ (5.29)
The last expression is very useful, especially when choosing j
1
as:
j
1
=
1
_
0
C (/) c
It
2
0
_
:
1
_
/
_
d / (5.30)
We thus get:
:
T
Jj
J:
1
= :
T
1
_
0
C(/)c
It
_
/2
1
(:
1
_
/)d/
Further the derivatives of the Bessel functions, 2
0
, 1
0
and J
0
. are:
7
0
0
= 7
1;
1
0
0
= 1
1
, J
0
0
= J
1
2
1
= J
1
+ 11
1
(5.31)
Now, as an arbitrary chosen . 0 (which can also be equal to :
1
_
/), we have
for the Bessel functions, J
1
(.) and 1
1
(.):
J
1
(.) v
1
2
..and.1
1
(.) ~
2
:
.
1
Hence,
lim
v
T
!0
:
1
J j
1
J :
1
=
2
:
1
1
_
0
C (/) c
It
d / (5.32)
which is a "Laplace" integral. Putting 1 =

2
, the constant B is not dependent
on time. Meanwhile, A remains an indeterminate. Further writing:
176 5. RATE DECLINE WITH A MOVING BOUNDARY
Figure 5.4: The dimensionless distance, :
1
. of a no-ow moving boundary as a
function of a constant, i, and the dimensionless time t
1
.
lim
v
T
!0
:
1
J j
1
J :
1
= (t) (5.33)
where the rate inow from the left-hand side is expressed as any function of time,
we get:
(t) = 1 [C (/)] , or C (/) = 1
1
() (5.34)
Here, 1
1
() is an inverse "Laplace" transform of the function.
Analytical models, when applicable, have been of great importance because
of their power. The cost for this power is however a limited applicability. In
that sense, we may now in our analytical model demand that the boundary be
no-ow, moving outwards according to (5.35),
Jj
1
J:
1
= 0 for :
1
= i
_
t
1
(5.35)
where i is a constant that should be further investigated and that is possibly
related to the driving force that moves the no-ow outer boundary with the
speed dened by Equation (5.5).
Figure (5.4) represents the perspective of a position, :
1
of, the no-ow bound-
ary as a function of the dimensionless time, t
1
. and a constant, i. The 2D plot
of the dimensionless position, :
1
. versus the dimensionless time, t
1
. for a coe-
cient of no-ow moving boundary, i = 1,3,5,7 and 9 is given in Figure (5.5).The
constant i can be related to the driving force or physics of the ow, i.e., to a
driving force that causes the rate to decline by the decline exponent /. The
derivative of the distance of a no-ow boundary with time gives the speed of a
moving no-ow boundary and is presented by Figure (5.6).
5. RATE DECLINE WITH A MOVING BOUNDARY 177
Figure 5.5: The dimensionless position, :
1
. versus the dimensionless time, t
1
.
for a coecient of no-ow moving boundary i = 1,3,5,7 and 9.
Figure 5.6: The velocity of a no-ow moving boundary,
1
=
ov
T
ot
T
. as a function
of the dimensionless time, t
1
. for various constant values of , i.
178 5. RATE DECLINE WITH A MOVING BOUNDARY
Figure 5.7: The dimensionless distance, :
1
. of a no-ow moving boundary as
a function of the dimensionless time, t
1
. for various constants of the no-ow
moving boundary, i = 1. 3. 5. 7.and 9.
According to the expression for :
1
0j
T
0v
T
in Equation (5.34), we now get:
1
_
0
C (/) c
It
_
/
_

_
J
1
_
i
_
/t
_

:
2
1
1
_
i
_
/t
_
_

_
d/ = 0 (5.36)
For Equation (5.36) we need to determine when C(/) = 1
1
() is given.
The constant A values for various exponents / are provided in Chapter 6.
Overall pressure responses, j
1
. within a drainage area are derived for Arps
variable-rate wellbore conditions for selected decline exponents, /. Chapter 7
describes the derivation of such responses for a selected decline exponent, / al-
most equal to zero.
5.1.4 Variable Rate Production of Arps Type
Further, we present diusion equation solutions for a variable rate production of
Arps type. The decline exponent, /. in Arps equation varies, and in each case,
we dene an inner boundary condition of variable-rate and an outer boundary
condition of no-ow and moving boundaries. The inner boundary condition is of
Arps type, with a decline exponent, / = 0.33. 0.5. 1 and 2. The rate, . versus
time, t. and initial decline, 1
i
, for four decline exponents, b (with b raging from
0.33 for the lowest rate to 2 for the highest rate) are presented in 3D based on
the Arps equation in Figure (5.8), and in 2D in Figure (5.9).
Figure (5.9) represents the 2D plot of the rate, . versus time, t.based on Arps
equation. All rates are calculated with the same initial decline rate,
i
= 5000,
5. RATE DECLINE WITH A MOVING BOUNDARY 179
Figure 5.8: A 3D plot of Arps equation rate, . versus time, t. and initial decline,
1
i
( for a decline exponent / = 0.33, / = 0.5, / = 1, and / = 2).
Figure 5.9: A 2D plot of Arps equation rate, q, versus time, t. and initial decline,
1
i
(for a decline exponent / = 0.33, / = 0.5 . / = 1, and / = 2).
180 5. RATE DECLINE WITH A MOVING BOUNDARY
Figure 5.10: The rate, . versus time, t, for / = 0.33 plotted in circles, and
various decline exponents (/ = 0.5. 1. and 2) all plotted as solid line. The rate
versus time is calculated for a specic initial decline,
i
= 5000, and an initial
decline, 1
i
= 0.01).
and the same initial decline, 1
i
= 0.001, ranging the decline exponent, /.from
0.33 to 2.
Hyperbolic Decline (b=1/3)
In the case of hyperbolic decline the inner boundary condition of variable rate
for the model, was set by taking an Arps decline exponent, /. equal to 1,3 (This
is known as the hyperbolic decline). Figure (5.10) displays a plot of Arps-derived
curves of rate, . versus time, t. The decline curve of / = 1,3 is plotted with
circles, for a dened initial decline ,1
i
. and an initial decline rate,
i
. Here, we
refer again to work of Fetkovich (1973) which relates the decline exponent, /. of
1,3 to 2,3, for the solution gas drive mechanism. Such rates dened with the
decline exponent / = 1,3 correspond to the IBC of the variable-rate model.
In Figure (5.10), all rates, , are related to time, t, through Arps hyperbolic
relation:
(t) =
i
(1 +/1
i
t)

1
l
(5.37)
By simplifying (5.37) to
(t + t
0
)

1
l
(/1
i
)

1
l

i
= (t + t
0
)

1
l
+ Co::t. (5.38)
and by a further simplication (by taking out t
0
) of equation (5.38) we obtain
the expression:
5. RATE DECLINE WITH A MOVING BOUNDARY 181
(t) - t

1
l
+ Co::t. (5.39)
Instead of using an Arps type expression (5.37) for rate (t), we introduce an
expression (t
1
) similar to (5.39). Correspondence of (t
1
) to Arps decline is
valid for late values of time, t
1
. At early times (t
1
) . We can
now solve the diusion equation by dening the variable rate / = 1,3. The rate
function (t
1
) thus becomes:
(t
1
) = t
3
1
.
The expression is simplied and valid for late times in order to obtain inverse
"Laplace" transform of the rate function (t
1
). This inverse "Laplace" trans-
form of the rate function, (t
1
), is:
1
1
() =
/
2
2
= C(/).
Now the dimensionless pressure, j
1
, is:
j
D
=
1
_
0
1
2
/
2
_
J
0
(r
D
_
/)

2
1
0
(r
D
_
/)
_
c
k
D
d/
=
2
t
3
1
_
c

r
2
T
4:
T
_
1
1
2
:
2
1
t
1
+
1
32
:
4
1
t
2
1
__

1
2
1i
_
:
2
1
4t
1
__
+
1
16
:
2
t
1

3
4
_
(5.40)
By requiring that a no-ow boundary moves with a distance : from a wellbore,
we get:
Jj
1
J:
1
= 0. ,o: :
1
= i
_
t
1
(5.41)
The distance :, where the no-ow boundary is spaced in time, t, is related by a
constant i. Here, and i are related according to:
=
1
2
1i
_
i
2
4
_

2
i
2
1
5
8
i
2
+
1
32
i
4
3
3
4
i
2
+
1
32
i
4
c

2
4
(5.42)
We can now plot the constant A as a function of the constant, i, as shown in
Figure (5.11).
By imposing variable rate IBCs of various Arps selected decline exponent,
b model generates production proles derived from the analytically solved dif-
fusion equation. Table (5-1), summarises all developed model solutions. In the
table solutions for j
1
and (i) are given for each variable-rate inner boundary
182 5. RATE DECLINE WITH A MOVING BOUNDARY
Figure 5.11: The constant A as a function of the constant i of a no-ow moving
boundary.
condition (the Arps rate decline exponents are: / = 0.33, / = 0.5, / = 1, and
/ = 2). Further, Table (5-1) presents the solution for the dimensionless pressure,
j
1
. calculated for the early and the late dimensionless times, t
1
.
5. RATE DECLINE WITH A MOVING BOUNDARY 183
Table 5-1: The dimensionless pressure, "p
1
", calculated by the model at time
"t
1
", for a variable-rate production of Arps type, dened by exponent b (ranging
from 0.3; 0.5; 1., and 2)
184 5. RATE DECLINE WITH A MOVING BOUNDARY
5.2 Vertical Well - Gas Flow
This section presents the gas ow associated with a moving boundary model. A
linear parabolic diusion equation is derived in Equation (5.1) for an oil ow.
Under the assumption that oil has a small and constant compressibility, the
diusion equation was derived by combining the continuity equation in radial
coordinates with Darcys law and an appropriate equation of state relating the
density to the pressure. The assumption of a small and constant compressibility
is not valid for gas whose compressibility is on the order of the reciprocal of the
pressure. In addition, the assumption further states that the gas-compressibility
and pressure product, c
j
j. is certainly not much less than unity, for gas ow.
Other dierences between oil and gas include that gas viscosities are a hundred
times smaller than their lowest oil counter parts (gas viscosities are on the order
of 0.002 cp). The gas volumetric ow-ratio and hence also gas velocities, are
thus much higher. The Reynolds number near the wellbore region is such that
non-Darcy ow is occurring. Usually, this non-Darcy ow eect can be lumped
into the skin eect.
The basic continuity equation is also valid for a real gas ow:
J
J:
(jn) +
jn
:
= c(
Jj
Jt
) (5.43)
and the Darcy velocity, n is:
n =
/(j)
j(j)
Jj
J:
(5.44)
In Equation (5.44) we may neglect the dependence of permeability on the pres-
sure, /(j), or the Klinkenberg eect which can be expressed as:
/(j) = /
c))
(1 +
:
|
j
) (5.45)
there, /
c))
is the eective permeability to liquid, and :
|
is slippage in porous
media (related to the mean free path of the gas molecules, controlled by tem-
perature, pressure, and the nature of the gas). Klinkenberg and others have
observed the permeability for a gas as a straight-line function of the reciprocal
of the mean pressure of the measurement, as given by Equation (5.45) relating
permeability to gas, /(j), to the mean pressure, j.This eect is important only at
very low pressures, and for most practical purposes, the gas permeability can be
assumed constant. Further, from the equation of state for real gases, we express
density as:
j =
`
11
_
j
2(j)
_
(5.46)
5. RATE DECLINE WITH A MOVING BOUNDARY 185
Subsequently, the fundamental non-linear partial dierential equation describing
isothermal ow of real gases becomes:
1
:
J
J:
_
j
j(j)2(j)
:
Jj
J:
_
=
c
/
J
Jt
_
j
2(j)
_
(5.47)
The consequence is that the diusion equation becomes non-linear. A special
technique is necessary to linearise the basic dierential equation for the radial
ow of a real gas.
5.2.1 Pseudo-Pressure Transformation (Intermediate Pres-
sures)
Al-Hussainy et al. (1965) have introduced transformation by using the real gas
pseudo-pressure or a real gas potential, :(j):
:(j) = 2
j
_
j
0
j
j(j)2(j)
dj (5.48)
The limits of integration range from an arbitrary base pressure, j
0
, and the
pressure of interest, j. In order to calculate :(j), we must know the viscosity, j,
(dened by a correlation method) and compressibility factor 2 (dened by the
equation of state). Moreover the following assumptions need to be established.
We assume that both the permeability, /. and the porosity, c, are constant.
From the equation of state follows that the specic gravity, j. is
j =
`
11
j
2(j)
(5.49)
By neglecting the water compressibility, c
&
, and the formation compressibility,
c
)
, the total compressibility c
t
can be expressed as:
c
t
= (1 :
&c
) c
j
(5.50)
and the gas compressibility, c
j
, as
c
j
=
1
j

1
2
J2
Jj
(5.51)
Equation (5.48) then becomes
:(j) = 2
j
_
j
0
jdj
j2
=
211
`
j
_
j
o
j
j
dj (5.52)
The real gas-pseudo pressure derivatives in space:
186 5. RATE DECLINE WITH A MOVING BOUNDARY
J:(j)
J:
=
d:(j)
dj
Jj
J:
=
2j
j2
Jj
J:
(5.53)
and time:
J:(j)
Jt
=
d:(j)
dj
Jj
Jt
=
2j
j2
Jj
Jt
(5.54)
Further derivatives give
2
J
Jt
(
j
2(j)
) = cjc
J:
Jt
=
211
`
cj
Jj
Jt
(5.55)
This corresponds to
Jj
Jt
= cj
Jj
Jt
(5.56)
and
j = j
0
c
c(jj
0
)
(5.57)
The diusion equation with a pseudo pressure as a dependent variable now be-
comes:
1
:
J
J:
_
:
J:(j)
J:
_
=
c(jc
t
)
i
/
J:(j)
Jt
(5.58)
The partial dierential equation describing an unsteady state radial gas ow
is linearised by integral transformation. In drawdown, the product of viscos-
ity and compressibility (jc
t
) is assumed constant at (jc
t
)
i
the initial value of
pressure, j
i
. The viscosity, j, is proportional to the pressure and compressibil-
ity, c
t
, is inversely proportional to the pressure. In other words, the product
(jc
t
) reduces the pressure dependence. The dimension of the real gas pseudo-
pressure, :(j). is equal to the pressure squared over the viscosity or (j:ic
2
,cj)
in the eld system of units. The inner boundary condition of constant rate with
pseudo-pressure, :(j) for a nite wellbore radius are:

c)
2:/:
&
=
/
j
Jj
J:
[
v
u
(5.59)
where the sandface ow rate,
c)
.under standard conditions is,

c)
= 2Q
j
cc
j
1
1
cc
(5.60)
Further derivatives give:
Jj
J:
=
Jj
J:(j)
J:(j)
J:
=
j2
2j
J:(j)
Jj
(5.61)
5. RATE DECLINE WITH A MOVING BOUNDARY 187
We than obtain the pseudo-pressure gradient at the wellbore as a constant equal
to:
J:(j)
J:
[
v
u
=
j
cc
1
://:
&
1
cc
(5.62)
Here, all pressure-dependent variables disappear (j2. j). The inner boundary
condition for a line source well is:
lim
v!0
_
:
J:(j)
J:
_
=
j
cc
1
://1
cc
(5.63)
The outer boundary conditions concern a no-ow, but moving boundary. The
well is assumed to be located in the centre of a cylindrical reservoir of radius
: = :
c
. The no ow boundary condition implies the local pressure gradient to
be zero, i.e., for t 0.
Jj
J:
[
v=v
c.
= 0 (5.64)
Here, :
ci
. stands for a no-ow boundary that is moving with a certain velocity
as previously described in this chapter.
It is convenient to dene a dimensionless pseudo-pressure, :(j
1
), as:
:
1
(j
1
) =
:(j
i
) :(j)
j
cc
1
://1
cc
(5.65)
with the dimensionless time, t
1
, as:
t
1
=
/t
(jc
t
)
i
:
2
&
(5.66)
and the dimensionless radius, :
1
, (related to a no ow and moving boundary)
as:
:
1
=
:
ci
:
&
(5.67)
The above dimensionless equations (5.65), (5.66) and (5.67) are substituted
into the linearised gas ow equation (5.58) thus giving the dimensionless pressure
gas ow equation that is valid for 1 _ :
1
_ :
1ci
:
1
:
1
J
J:
1
_
:
1
J:
1
(j
1
)
J:
1
_
=
J:
1
(j
1
)
Jt
1
(5.68)
The initial condition species a uniform initial pressure :(j
i
)
.
It is further spec-
ied that at time t
1
= 0 and for all :
1
:
:
1
(j
1
) = 0 (5.69)
188 5. RATE DECLINE WITH A MOVING BOUNDARY
The inner boundary condition (nite wellbore radius) at :
1
= 1 and for all t
1
is:
J:
1
(j
1
)
J:
1
= 1 (5.70)
Alternatively, the line source form can be written as:
lim
v
T
!0
_
:
1
J:
1
(j
1
)
J:
1
_
= 1 (5.71)
For all t
1
0 and moving :
1
= :
1ci
, with no ow across the external boundary
the outer boundary conditions are:
J:
1
(j
1
)
J:
1
= 0 (5.72)
Thus, the linearised diusion partial dierential equation and boundary condi-
tions describing the unsteady state ow of gas have been put in exactly the same
generalised form as the dimensionless equations derived for the liquid case. This
is achieved with pseudo-pressure rather than actual pressure, along with ap-
propriate denitions of the dimensionless variables. All the analytical solutions
that have been found useful in oil ow are also applicable to gas ow. Here, we
comment that pressure analysis must be carried out in terms of pseudo-pressure,
:(j). This means that for a given recorded pressure against time we need to
convert pressure into a pseudo-pressure versus time by using the transform.
It is possible to instead use the normalised pseudo-pressure, (j), which can
be obtained by dividing m(p) with
j
i
(j2)
i
. (j2)
i
is the product evaluated at
pressure p
i
and reservoir temperature T. The normalised pseudo-pressure (j)
has all the linearising properties of :(j) as well as the units of pressure. Its
magnitude is similar to the real pressure. When normalised pressure (j) is
used rather than :(j) gas well plots look very similar to those obtained in oil
well analysis. This is typical for high pressure p 3000 psi (210 bar). To be able
to use the normalised pressure, (j
1
), we changed the dimensionless pressure
:(j
1
) to:
(j
1
) =
:(j
i
) :(j)
Qj
cc
1
://1
cc
=
(j
i
) (j)
Qj
cc
1
://1
cc
(j2
i
)
j
i
(5.73)
It is important to mention that the pseudo-pressure concept does not result in
a complete linearisation of the radial ow equation.
With the normalised dimensionless pressure, (j
1
), Equation (5.58) be-
comes:
5. RATE DECLINE WITH A MOVING BOUNDARY 189
1
:
1
J
J:
1
_
:
1
J(j
1
)
J:
1
_
=
J(j
1
)
Jt
1
(5.74)
The inclusion of the normalised pseudo-pressure, (j
1
). into the dimensionless
quantity j
1
from equation (5.73) means that the analytical solutions termed
dimensionless functions for oil, j
1
, and gas, (j
1
) ow, are identical. In high
rate oil wells with limited entry, there is some evidence of rate dependent skin
being present. This skin should also be included in gas wells. Agarwal (1978)
introduced the real gas pseudo-time pseudo-time, t
o
. by the following expression,
comparable to the pseudo-pressure:
t
o
= (jc
t
)
i
t
_
t
0
dt
jc
t
(5.75)
When implemented into the diusion, we get:
1
:
J
J:
_
:
J:(j)
J:
_
= c
(jc
t
)
i
/
J:(j)
Jt
o
(5.76)
The same pseudo-time, t
o
, can be put into the dimensionless form, i.e., in t
1
:
t
1
=
/t
o
c(jc
t
)
i
:
2
&
(5.77)
The conditions under which this pseudo-time transformation linearises the radial
ow equation have been studied by Lee and Holdich (1980). It is important to
mention that the pseudo-pressure concept does not result in a complete lineari-
sation of the radial ow equation.
5.2.2 Pressure Squared Transformation
The present section concerns the special case of the pseudo-pressure transfor-
mations valid for low pressures. This low pressure is below j < 2000 psi (140
bar), where the squared pressure j
2
, is considered instead of the pseudo-pressure
function :(j). Under the following assumptions, pressure-squared solutions can
be obtained:
1. The gas behaviour is ideal (2 = 1) and the gas viscosity, j
j
, is assumed
to be independent of pressure. Equation (5.58) becomes:
1
:
J
J:
_
:
Jj
2
J:
_
=
cj
/j
Jj
2
Jt
(5.78)
2. When the product of viscosity, j. and the gas deviation factor, 2, are
constant, the equation becomes:
190 5. RATE DECLINE WITH A MOVING BOUNDARY
1
:
J
J:
_
:
Jj
2
J:
_
=
cjc
/
Jj
2
Jt
(5.79)
3. For pressure gradients assumed to be small, i.e.,
_
Jj
J:
_
2
0 (5.80)
The dimensionless form of the diusivity equation is:
1
:
1
J
J:
1
_
:
1
Jj
1
(j
2
)
J:
1
_
=
Jj
1
(j
2
)
Jt
1
(5.81)
Here, the only new dened dimensionless variable is now j
1
(j
2
), equal to:
j
1
(j
2
) =
j
2
i
j
2
j
cc
1
://1
cc
(5.82)
5.2.3 No-Flow Moving Boundary - Gas Flow Solutions
All dimensionless pressure solutions developed for an oil ow are now also valid
for a gas ow. The only dierence is that the dimensionless form of the pressure
variable for oil ow has to be replaced by an adequate dimensionless variable for
gas. Gas dimensionless pressure variables have been previously dened based on
pseudo-pressure, normalised pseudo-pressure and squared pressure.
No-Flow Moving Boundary - Variable Rate of Arps Decline (b=0.5)
Here we present gas ow solutions with variable rate of Arps decline for a hy-
perbolic decline of / = 0.5 and a dimensionless pseudo-pressure, j
1
, is equal
to:
j
1
=
1
2t
2
1
_

_
_

_
1 +
_
1
:
2
1
4t
1
_
c

:
2
1
4t
1
1i
_
:
2
1
4t
1
_
_

_
2
_
1
:
2
1
4t
1
_
c

:
2
1
4t
1
_

_
(5.83)
The dimensionless pseudo-pressure, j
1
, is:
j
1
=
j
i
j
j
cc
1
://1
cc
(5.84)
the dimensionless time, t
1
, is:
5. RATE DECLINE WITH A MOVING BOUNDARY 191
t
1
=
/t
(jc
t
)
i
:
2
&
and the dimensionless radius, :
1
, is:
:
1
=
:
ci
:
&
Also here is the change of a no-ow moving boundary with time dened as:
:
1
= i
_
t
1
The constant as a function of the coecient of no-ow moving boundary, i,
is given as:
(i) =
1
2
1i
_
i
2
4
_

2
i
2
1
i
2
4
2
i
2
4
c
i
2
4
(5.85)
Here, are presented only the pseudo-pressure gas ow solutions. For a hyper-
bolic decline of / = 0.5 dimensionless pseudo-pressure :(j
1
) is equal to:
:
1
(j
1
) =
1
2t
2
1
_

_
_

_
1 +
_
1
:
2
1
4t
1
_
c

:
2
1
4t
1
1i
_
:
2
1
4t
1
_
_

_
2
_
1
:
2
1
4t
1
_
c

:
2
1
4t
1
_

_
(5.86)
where dimensionless pseudo-pressure, :(j
1
) is:
:
1
(j
1
) =
:(j
i
) :(j)
j
cc
1
://1
cc
(5.87)
We can further normalised the dimensionless pseudo-pressure, :(j
1
), to the
normalised pseudo-pressure, (j
1
), according to:.
(j
1
) =
1
2t
2
1
_

_
_

_
1 +
_
1
:
2
1
4t
1
_
c

:
2
1
4t
1
1i
_
:
2
1
4t
t1
_
_

_
2
_
1
:
2
1
4t
1
_
c

:
2
1
4t
1
_

_
(5.88)
where,
1
(j
1
) is:
(j
1
) =
(j
i
) (j)
j
cc
1
://1
cc
(j2
i
)
j
i
(5.89)
192 5. RATE DECLINE WITH A MOVING BOUNDARY
Pressure squared solutions, j
1
(j
2
), are equal to:
j
1
(j
2
) =
1
2t
2
1
_

_
_

_
1 +
_
1
:
2
1
4t
1
_
c

:
2
1
4t
1
1i
_
:
2
1
4t
1
_
_

_
2
_
1
:
2
1
4t
1
_
c

:
2
1
4t
1
_

_
(5.90)
We can further write all solutions obtained for oil in an adequate form for gas
ow.
5.3 Summary
Chapter 5 describes an investigation of the nature of the wellbore pressure re-
sponse when production is forced to follow the Arps decline dened by the spe-
cic value of a decline exponent, /. The physical model is dened with the inner
boundary condition of variable rate of Arps type and a no-ow boundary moving
outwards from a wellbore axis. The velocity of the no-ow moving boundary is
proportional to a coecient, i, and a square root of time. In order to solve a
diusion equation for the variable-rate IBC of Arps type and a no-ow moving
boundary, the rates are modied in such a way that Arps decline exponent, /,
is matched with the late time rates of the model. In addition, the velocity is
made proportional to the square root of time. The no-ow boundary moves and
for late time it reaches a condition where the pressure dierence is zero. The
solutions obtained for the diusion equation correspond to the pressure chang-
ing with time. Each variable pressure prole can be dened within the drainage
area of a vertical well. The obtained pressure solutions conrm Raghavans ob-
servation that matching rates dened by exponent b can only be achieved with
a pressure that changes in time. Variable pressure solutions are developed for
an oil ow but are also applicable for gas ows. Such solutions are available for
both low-pressure and high-pressure conditions.
Chapter 6
RATE DECLINE CURVES
This chapter comprises two parts, the rst deals with transient and depletion
responses for oil fractured-horizontal wells oil and the second investigates the na-
ture of the pressure responses when the rate response of a vertical well is of Arps
type for selected values of the decline exponent, b. The value of horizontal mul-
tiple fractured wells can be maximised when the parameters that most impact
the productivity of such wells are understood. As a step towards achieving max-
imum values from horizontal multi-fractured wells, a computer model, presented
in Chapter 4, has been developed, allowing the user to quickly assess the inu-
ence of various well or reservoir properties on the well performance. The model
is three-dimensional and considers full-time variability. It assumes a single-phase
oil ow and is based on semi-analytical techniques. Modelling of transversal or
longitudinal fractures of varying types (uniform ux, innite conductivity, nite
conductivity) is possible. The exibility to use various constraints and the op-
tion of a multi-run sensitivity generation can be used for obtaining prognosis of
a multi-fractured horizontal well productivity its diagnosis.
B. Cvetkovic contributed to the teamwork in designing the software tool, to
the physical and mathematical modelling with software development and test-
ing, based on the implemented model. Its main features comprehend the ability
to provide a fast screening analysis (for various inner boundary conditions and
special features of step-function and late time approximations). Most of these
features are presented in this chapter. The model solution that is developed
can be employed for a general screening and optimisation of a multi-fractured
horizontal well thanks to the implementation of fast numerical algorithms as
published by Cvetkovic et al. (1999). For any given set of reservoir, fracture and
well parameters, one can determine the oil rate cumulative production, well pres-
sure or productivity index. The evaluation of a multi-fractured horizontal well
performance, or the selection of an optimum perforation with stimulation design
for such wells, may be approached through ne grid reservoir simulation. How-
ever, while reservoir simulation is the most advanced method for predicting well
performance, it is often too time consuming for conducting parametric screening
193
194 6. RATE DECLINE CURVES
studies. Often the data required are not available and the eort may not be
warranted. As an alternative to simulation, the application of semi-analytical
models can readily yield wellbore and fracture responses to various boundary
conditions. This is often sucient to provide an understanding of factors with
the most inuence on well performance. If simulation work is warranted, it can
then proceed with the insight obtained from the analytical models.
A variable-rate and specied boundary model generates pressure proles for
the decline exponent, /, (/ = 0.33. 0.5. 1 and 2). Obtained solutions have
contributed to Raghavans (1993) discussion on wellbore pressure conditions re-
quired for the well to produce Arps-type rates with a specied decline exponent,
/. Although the producing well is vertical, the physical model describing its
solutions can be extended to a horizontal producing well.
6.1 Transient Decline of a Fractured-Horizontal
Well
We chose to use several options from the fractured-horizontal well model (for
which the general model solutions are presented in Chapter 4) to demonstrate
screening analysis capabilities with data from a North Sea well. In cases where
well data were lacking, properties that were believed to be appropriate for the
sake of this review were assumed. The model is an extremely fast mini-simulator,
modelling one-phase (slightly compressible) liquid ow into a multifractured hor-
izontal well in an open (slab) or closed (box) reservoir. The fractures are rec-
tangular and vertical, and either transversal or longitudinal relative to the well
direction. They are also alternatively of nite conductivity, innite conductivity
or uniform ux type. Further, the fractures are fully or partially penetrating,
of equal or unequal length and spacing. There is no limitation to the number
of fractures. The well is either open or perforated only at the fractures; how-
ever, one special option is a partially perforated well with no fractures. The
simulation setup is made easy for the user through a screen input interface de-
sign. Special features include multirun (several parameter values given and used
consecutively), linear or logarithmic scales on axis, and a choice of units (eld,
metric or dimensionless). To be given as input are various relevant geometric and
physical parameters for the reservoir-well-fracture system, including the alterna-
tive of well rate or pressure, each with the option of being constant or variable
or varying in time in a freely stepwise manner.
The output given is correspondingly the well pressure or rate as a function
of time, in addition to individual fracture rates. Further output parameters
comprise pressure derivatives (logarithmic and square root) and all cumulative
rates. Also, productivity indices and eective wellbore radii are given. Output
(and input) data are immediately available, both in tabular form and through a
6. RATE DECLINE CURVES 195
Table 6-1: Reservoir, Well and Fracture Data
display of curves by a specially developed application of the Excel software. Well
data required for analysis using the fractured-horizontal well model are presented
in Table (6-1).
6.1.1 Transient-Rate Response of a Well
Sensitivity analysis to reservoir, uid and well data
To start with, the ability to run sensitivities to various parameters is demon-
strated by varying the number of reservoir, uid and well data for a xed number
of fractures and a xed well length. Reservoir parameters include: the initial
pressure, 1
i
, the eective thickness, /, the horizontal and vertical permeability
/
I
, /

, the porosity, c, and the total compressibility, c


T
. Fluid parameters are:
the viscosity, j, and the formation volume factor, 1
c
. Further, well data include
the well length and the wellbore IBC of constant pressure. The initial reservoir
pressure, 1
i
. was kept constant at 5700 psi. Values of the sensitivity to initial
pressure, 1
i
, porosity, c, and eective thickness, /, are given in Table (6-2).
The case concerns production from a horizontal well in an oil reservoir with
5 transversal fractures. The fracture ow is of uniform ux. Figures 6.1 and
6.2 represent plots of rate and cumulative rate for each fracture and for a well.
196 6. RATE DECLINE CURVES
Table 6-2: The sensitivity to the initial reservoir pressure, "P
i
", the porosity and
the eective reservoir thickness, h
Figure 6.1: Individual fracture rates,
)vi
(i = 1. ...5) and the rate of a well (with
fractures), (//|,d), versus time, t, in days.
Figure 6.1 shows the production prole of fracture 1 overlaying that of fracture
5. One can also see the overlay of the fracture 2 production prole with that of
fracture 4. As expected, the prole of fracture 3 was the lowest, due to the size
of the drainage area being small as compared to those of the other producing
fractures. Production proles of cumulative rates with time are presented in
Figure (6.2). Again, fracture 1 and 5 overlay each other. The same is true
for fracture 2 and 4. The cumulative rate of fracture 3 is low as compared to
the other plotted fracture cumulative rates, again inuenced by the size of the
drainage area.
The sensitivity of the rate and cumulative rate to the variation of initial
pressure, 1
i
, is presented in Figure (6.5). Plots for three initial pressures (1
i
= 6425, 5700 and 4970 psi) demonstrate how the input pressure changes a (5-
6. RATE DECLINE CURVES 197
Figure 6.2: Individual cumulative fracture production, Q
)vi
(i = 1. .... 5), and
the cumulative production of a well with fractures, Q (bbl), versus time, t, in
days.
fractured) well rate and cumulative rate prole. The sensitivity to changes in
porosity is presented in Figure (6.3) and to the eective thickness, / (,t), in
Figure (6.4).
Figure 6.2 gives the variation in the fracture production proles and the frac-
ture proles for a well with 5 producing fractures. The gure shows changes in
the individual fracture rate,
)v
, and the individual cumulative fracture produc-
tion rate, Q
)v
, with time for each of 5 transversal fractures positioned along the
horizontal well conditioned to a well owing pressure of 2900 psi.
For a selected initial pressure, 1
i
, equal to 5700, psi we considered three
wellbore pressures (with values of 2176, 2900 and 3626 psi). Further, in Figure
(6.6), we plot the rate and cumulative rate with time at varying values of wellbore
pressure (j
&)
= 2176, 2900 and 3626 psi). The pressure, j
&)
, represents the
average pressure in the wellbore.
The contribution from fractures 1 and 5 to the well production is presented
in Figure (6.7). For each choice of the porosity production rate and cumulative
production proles, the data from fractures 1 and 5 overlay each other.
The contribution from fractures 2 and 4 to the well production is presented
in Figure (6.8). Both the production rate and cumulative production with time
correspond to overlays, and dier for each of three selected wellbore pressure.
The production proles of fracture 2 and 4 are lower as compared to those
of fracture 1 and 5.The contribution to the well production from fracture 3 is
198 6. RATE DECLINE CURVES
Figure 6.3: Rate and cumulative production proles for various values of porosity,
c, (i.e., 0.24, 0.34, 0.44).
Figure 6.4: Rate and cumulative production proles for various eective thick-
nesses, /, (i.e., 95 , 75 and 55 ft).
6. RATE DECLINE CURVES 199
Figure 6.5: The individual-fracture rate and individual-cumulative fracture pro-
duction versus time for various values of initial pressure, 1
i
, (i.e., 6425, 5700,
and 4975 psi).
Figure 6.6: The rates and cumulative productions vs. time for varying values of
the initial wellbore pressure, 1
&)
, (i.e., 2176, 2900 and 3626 psi).
200 6. RATE DECLINE CURVES
Figure 6.7: Individual fracture rates and cumulative productions (
)v2
and Q
)v2
)
vs. time for various values of wellbore pressure, 1
&)
, (i.e., 2176, 2900 and 3626
psi).
Figure 6.8: Individual fracture rates and cumulative productions (
)v2
and Q
)v2
)
vs. time for various values of wellbore pressure, 1
&)
, (i.e., 2176, 2900 and 3626
psi).
6. RATE DECLINE CURVES 201
Figure 6.9: Individual fracture rates and cumulative productions (
)v3
and Q
)v3
)
vs. time for various values of wellbore pressure, 1
&)
, (i.e., 2176, 2900 and 3626
psi).
presented in Figure (6.9). The time evolution of both the production rate and
cumulative production diers for each of the three selected wellbore pressures.
The production prole of fracture 3 is the lowest compared to other fracture
production proles. This is due to the small size of the drainage area.
For the case of an isotropic permeability (/
I
= /

), Figure (6.10) shows


the rate and cumulative production plotted against time while changing the
permeability, /
I
, from 40 mD, 4 mD and 0.4 mD. Increasing the permeability
tenfold signicantly changed production proles. On the other hand, a tenfold
decrease in permeability reduced the rate and cumulative rate proles.
For a homogeneous reservoir with a permeability of 4 mD, the vertical per-
meability, /

. was chosen as either 4, 2 or 0.4 mD. Figure (6.11) shows the


change in production prole due to a variation of the vertical permeability. An
increased heterogeneity, caused by the /

value being only 10 % of a horizontal


permeability causes a doubling of the cumulative production prole.
In Figure (6.12), the plots of the rate and cumulative production vary due to
the changes in total compressibility, c
T
.
Furthermore, changes in uid properties with viscosity, j, in Figure (6.13),
and the formation volume factor, 1
c
, in Figure (6.14), illustrate how the pro-
duction proles vary with time.
202 6. RATE DECLINE CURVES
Figure 6.10: Individual fracture rates, and cumulative productions, Q vs. time
for various values of permeability, 1
I
= 1

(i.e., 40, 4, 0.4 mD).


Figure 6.11: The individual fracture rate, . and cumulative production, Q. vs.
time for various vertical permeabilities, 1

(i.e., 0.4, 2, and 4 mD).


6. RATE DECLINE CURVES 203
Figure 6.12: The individual fracture rate, . and cumulative production, Q. vs.
time for various total compressibility values, c
T
(i.e., 7.25 e-05, 6.26 e-05, 1.86
e-06).
Figure 6.13: The individual fracture rate, , and the cumulative production, Q,
vs. time for various oil viscosities, j (i.e., 3, 4 and 5 cp).
204 6. RATE DECLINE CURVES
Figure 6.14: The individual fracture rate, , and the cumulative production, Q,
vs. time for various oil viscosities, 1o (i.e., 1.35, 1.55 and 1.75 rb/stb).
Fracture position
Ideally, fractures can be positioned perpendicular to a well axis, known as
transversal fractures, and also along a wellbore axis, known as longitudinal frac-
tures. Figure (6.15) displays production proles for longitudinal and transversal
fractures. A longitudinal fracture production rate prole oscillates. Model rates
are usually smooth functions, but for certain choices of the input parameters,
the model may occasionally generate oscillating proles. This is due to the in-
stability cased by the implemented Stehfest inversion techniques. One way to
avoid oscillation in production prole is to tune the Stehfest inversion procedure.
Another is to consider new inversion techniques (several of which are presented
in Appendix B).
Up to now, all fractures were found to fully penetrate the reservoir. If
transversal fractures partially penetrate the reservoir, the productivity index
PI and cumulative rate production proles show small variations, as plotted in
Figure (6.16). For the longitudinal transversal fracture positioned along the hor-
izontal wellbore axis which partially penetrates the reservoir, the productivity
index PI and cumulative production proles are almost identical to those plotted
in Figure (6.17). The very strong modelling feature represents fracture interfer-
ences. Opposite to the fully perforated fractures, where each fracture produces
its drainage area, in the partially penetrating fracture drainage area is distrib-
uted by another fracture production. Due to fracture interferences, the resultant
6. RATE DECLINE CURVES 205
Figure 6.15: The individual fracture rate, , and the cumulative production, Q,
vs. time for various fractures (longitudinal fracture vs. transversal fractures).
proles are varying slightly.
Figure (6.18) plots productivity indices and cumulative rates versus time for
a horizontal well with a longitudinal fracture and 5 transversal fractures (each
with a length of 85 ft). Two of the longitudinal fracture have lengths of: 2100 ft
and 425 ft. All fractures are partially penetrating (with height of 55 ft). For a
fracture area of (425 ft x 85 ft), a single longitudinal fracture displays a higher
production as compared to a production with 5 transversal fractures. Also, for
two longitudinal fractures producing oil, a single longitudinal fracture that is
twice as large will have twice the production.
Production proles change rapidly with unequal fracture half-lengths, as can
be seen in Figure (6.19). The production increases for larger half-lengths as
presented in Figure (6.19). Table (6-3) shows the variation of the fracture half-
length for 5 transversal fractures equally positioned along a horizontal well. The
production proles and PIs change for a choice of fracture half-lengths. For
equally sized fractures (in Case 1), the production prole is higher as compared
to the unequally sized fractures (in Case 2 and Case 3). Again the interference
between producing fractures (in Case 2 and Case 3) reduces the production
prole. This is demonstrated in Figure (6.20).
In a case where the fractures are unequally sized, the fracture interferes giv-
ing low productivity indexes in comparison to the equally sized fractures as
presented in Figure (6.20). The same gure comprises plots of two unequally
sized fractures. For each case, fracture half-length sizes are provided in Table
206 6. RATE DECLINE CURVES
Figure 6.16: Productivity index, PI for a horizontal well with 5 transversal frac-
tures and cumulative rate, Q with time for various fracture partial penetrations
with height, h (of 75, 55 and 35 ft).
Figure 6.17: Productivity index, PI for a horizontal well with a longitudinal
fracture positioned along the 2100 ft horizontal well, and cumulative rate, Q
with time for various fracture partial penetrations with height, h (of 75, 55 and
35 ft).
6. RATE DECLINE CURVES 207
Figure 6.18: Productivity index, PI for a horizontal well with a longitudinal
fracture and transversal fractures, and cumulative rate, Q with time for the
partial penetrated fracture the height, h = 55 ft.
Figure 6.19: Productivity index, PI for a horizontal well with transversal frac-
tures, and cumulative rate, Q with time for the various half-length, 1
)
(of 170,
85 and 42.5 ft).
208 6. RATE DECLINE CURVES
Table 6-3: Sensitivity to fracture half length sizes, Lf
Table 6-4: Sensitivity to the distance bewtween two fractures due to well length,
L changes (L= 2520, 2100, and 1680 ft
(6-3). The PIs and cumulative rates are signicantly lower as compared to the
case with the equally sized fractures. This is due to fracture interferences caused
by the unequal fracture size. Also, between unequally sized fractures (Case 2 and
Case 3), there exists a dierence between two production proles. The higher
production prole is dened by the fractures positioned at the beginning and
the end of a horizontal well. The size of a fracture half-length (Fracture 1 and 5)
denes the production prole. This screening may help in understanding to what
degree a production prole can be inuenced by an unequal fracture half-length.
Up to now, the fracture space between two fractures, has been equal, so for a
horizontal well of 2100 ft with 5 fractures, the space between two neighbouring
fractures is 520 ft. This size can vary by increasing the well length, L, as given in
Table (6-4). Productivity indexes and production proles are provided in Figure
(6.21).
The distance between two fractures changes with the number of fractures for
a well length, L (L = 2100 ft), as demonstrated in Table (6-5). Accordingly,
production indexes and production proles change according to Figure (6.22).
Table 6-5: Sensitivity to the distance bewtween two fractures due to number of
fractures changes sor the same well length, L = 2100 ft
6. RATE DECLINE CURVES 209
Figure 6.20: Productivity index, PI for a horizontal well with transversal frac-
tures, and cumulative rate, Q with time for the unequal fracture (case 2 and 3)
compared to the equally sized fractures (case 1).
Figure 6.21: Productivity index, PI for a horizontal well with transversal frac-
tures, and cumulative rate, Q with time for the well length, L (of 2520, 2100 and
1680 ft).
210 6. RATE DECLINE CURVES
Figure 6.22: Productivity index, 11 for a horizontal well with transversal frac-
tures, and cumulative rate, Q with time for the various number of fractures, on
a xed well-length, 1=2100 ft for various number of fractures, : (of 7, 5 and 3).
In all gures, the distance between two fractures was kept equal. In one
case, where the fractures were unequally distributed along the well length, L,
the production proles for each fracture varied with time, as presented in Figure
(6.22).
The fracture character changes from a uniform ux to an innite and nite
conductivity. Up to now, fractures have been considered to have a uniform ux.
In reality, a fracture is either innite conductive or nite conductive. Figure
(6.23) presents how the fracture character inuences the production. The PIs
and the cumulative rate proles are similar for the uniform ux and the innite
conductivity fracture, whereas for the nite conductivity fracture, the proles
are lower due the fracture being less conductive.
Productivity indexes and the sensitivity to cumulative production proles to
selected values of nite conductivity, 1
C
, as provided in Table (6-6), are presented
in Figure (6.24). The rate and cumulative production versus time are plotted
in Figure (6.25). Moreover, the fracture conductivity values are given in Table
(6-7), and Figure (6.26) shows plots of both the rate and individual fracture rate
versus the cumulative rate.
Figure (6.27) presents 5-transversal nite conductivity fractures, for which
each fracture conductivity, 1
C
, is 50 mDft, and compares them to the wellbore
proles with calculated wellbore frictions. The wellbore friction signicantly
reduces the cumulative production, according to the friction model feature pre-
6. RATE DECLINE CURVES 211
Figure 6.23: The productivity index, 11. for a horizontal well with transversal
fractures, and the cumulative production, Q. vs. time for the various fracture
characters (uniformux, innite conductivity, and nite conductivity (with 1
C
=
1000 mDft).
Table 6-6: Sensitivity to fracture conductivity "F
C
" (mDft) equal to 2500, 1000,
qnd 100
Table 6-7: The sensitivity to the fracture conductivity ,"F
C
" (mDft), when it is
innite, 1000, and 50 mDft
212 6. RATE DECLINE CURVES
Figure 6.24: The productivity index, 11. for a horizontal well with transversal
fractures, and the cumulative rate, Q. vs. time for the innite conductivity
fracture and nite conductivity fractures (with 1
C
values of 1000 and 50 mDft).
Figure 6.25: The rate, . for a horizontal well with transversal fractures, and
the cumulative rate, Q. vs. time for an innite conductivity fracture, and nite
conductivity fractures with 1
C
values of 1000 and 50 mDft.
6. RATE DECLINE CURVES 213
Figure 6.26: The rate, . for a horizontal well with transversal fractures, and
individual fracture rates
)vi
, where i = 1. .... 5 versus the cumulative rate, Q,
for an innite conductivity fracture and nite conductivity fractures with 1
C
values of 1000 and 50 mDft.
sented in Chapter 4.3.2.
Variable Wellbore Pressure Conditions
All plots generated up until now have been for inner boundary conditions, IBCs,
of constant pressure. For variable IBCs, as provided in Table (6-8), the rate
versus time responses for a well with 5 transversal fractures are dened as being
of innite and nite conductivity. These are plotted in Figure (6.28). Figure
(6.29) presents rates versus cumulative rates for a well with innite conductivity
and nite conductivity fractures. Individual fracture rates are plotted for each
of the ve transversal fractures versus the cumulative fracture rate.
6.1.2 Well Transient-Pressure Responses
As for the constant pressure IBC, it is also possible to repeat all cases by choosing
IBCs of constant rate instead of constant pressure. All features presented before
are available with IBC of both constant and variable rate. Also, regarding the
fracture character, it is possible to consider fracture ow as either of uniform ux,
innite conductivity or nite conductivity. The limited communication between
transverse fractures and the wellbore creates a choking eect near the well and
214 6. RATE DECLINE CURVES
Figure 6.27: The rate, q, versus the cumulative production, Q, for a wellbore
with 5 transversal fractures. Each fracture conductivity, 1
C
. is 50 (mDft). The
wellbore friction reduces both the well rate production and the cumulative pro-
duction.
Table 6-8: Wellbore inner boundary conditions, IBCs of variable pressure for the
innite conductivity and nite conductivity for fractures with "F
C
" = 50 mDft
6. RATE DECLINE CURVES 215
Figure 6.28: Responses of rate, q, versus time, t. The wellbore inner boundary
conditions, IBC, correspond to a variable pressure for the innite and nite
conductivity fractures of 50 mDft.
may cause an apparent reduction in fracture conductivity.
The limited communication of a fractured-well (choking eect) is included
in the model, as discussed in Chapter 4.3.3, and causes a decrease in the rate
production and PI, as demonstrated in Figure (6.31). The fractures are of nite
conductivity (each of 2500 mDft) and the IBCs are of variable rate. For such
variable rate IBCs, the PI and individual fracture cumulative production proles
are given in Figure (6.31).
6.1.3 Well Pressure-to-Rate Responses
Inner boundary conditions of constant and variable rates are usually applied in
pressure and rate testing analyses. Operating conditions generally change from
rate to pressure and vice versa. Most of the wells in the North Sea, for example,
operate under plateau production or constant rate wellbore conditions for a year
or longer. In Figure (6.32), we plot solutions to varying wellbore conditions
from constant rate to constant pressure. The well starts producing with the
rate of 4000 bbl/d and after 700 days it switches to the calculated constant
wellbore pressure conditions as presented in Figure (6.32). For the basic restart
option treated here, we have as input a period of constant rate followed by a
constraint involving a bottomhole owing pressure. The value of the employed
216 6. RATE DECLINE CURVES
Figure 6.29: The rate, q, versus cumulative production ,Q, responses. The well-
bore inner boundary conditions, IBC are variable pressure for the innite con-
ductivity and nite conductivity, 50 (mDft), fractures. Each individual fracture
rate vs.cumulative rate is graphically presented.
6. RATE DECLINE CURVES 217
Figure 6.30: Inuence of fracture choking eect (for variable rate IBC) on pres-
sure dierence and well with fractures PI. Fractures are nite conductivity (2500
mDft).
Figure 6.31: Inuence of fracture choking eect (for variable rate IBC) on cumu-
lative indifvidual fracture production Q
)vi
(i=1,5 and 3) and well with fractures
PI. Fractures are nite conductivity (2500 mDft).
218 6. RATE DECLINE CURVES
Figure 6.32: Wellbore contant rate to constant pressure IBCs. The value of the
employed owing pressure following the constant rate period corresponds to the
pressure determined by the model at the end of this period.
owing pressure following the constant rate period corresponds to the pressure
determined by the model at the end of this period.
6.1.4 A Well with Longitudinal Fractures
Up until now, all fractures have been transversal. The following gures comprise
rate changes for the longitudinal fracture positioned along a horizontal well.
6. RATE DECLINE CURVES 219
Figure 6.33: Longitudinal versus transversal fracture rates and the cumulative
production (for a horizontal well with ve equally spaced innite conductive,
and equal half-length fractures).
Figure 6.34: Longitudinal vs.transversal fracture rates and the cumulative pro-
duction (for selective fractures: 1, 5 and 3).
220 6. RATE DECLINE CURVES
Figure 6.35: The productivity index, PI, versus time, t, for a horizontal well
with longutudinal fractures of n = 7, 5, and 3.
6.2 Well Depletion Responses
6.2.1 Closed (BOX) Model
Rate Responses
A closed (BOX) model is very basic and thus needs further improvements. Calcu-
lated production rates are low as compared to expected values, and consequently
some modications must be applied before use. Implemented model features in-
clude constant and variable IBCs, along with a variation with regard to the
fracture half-length.
Pressure Responses
By varying the IBC of rate, we generate pressure dierence proles with time for
a horizontal well with 5 transversal fractures. As the IBC of rate is changed the
well starts producing at the rate 150 bbl/day. After 700 days, this rate reduces
to 110 bbl/day and after 1400 days it goes down to 90 bbl/day. As mentioned
before, closed or BOX model rates and pressures need to be normalised. Figures
(6.36, 6.37) plot selected BOX model features of changing fracture half-lengths
and simultaneously with changing IBCs of pressure and rate.
6. RATE DECLINE CURVES 221
Figure 6.36: The rate, q, versus time, t, for the BOX model (5000 ft by 5000
ft). IBCs of variable pressure of 4900, 3900, and 2900 psi. Innite conductive
fractures with varying fracture-half lengths, L
)
, of 120, 85 and 50 ft.
Figure 6.37: The pressure dierence, P, versus time, t, for the BOX model (5000
ft by 5000 ft). IBCs of variable rate of 150, 110, and 70 bbl/d. Innite conductive
fractures with varying fracture half-lengths, L
)
, of 120, 85 and 50 ft.
222 6. RATE DECLINE CURVES
6.3 Rate Decline with a No-owMoving Bound-
ary
The following sections further investigate the nature of the wellbore pressure
needed for the Arps rate decline production dened with a xed exponent, b.
Chapter 5 presented the physical model together with its analytical solutions.
This section aims at demonstrating the nature of pressure proles for several
b values. Each b value is selected, and can as such be related to the drive
mechanism (i.e., b = 1/3, for the solution-gas-drive, b = 0.5, for the gravity-
drainage, and b = 1, for the multilayered no-cross ow production).
6.3.1 Hyperbolic Decline (/ = 1,3)
In this case, we set the inner boundary condition of variable rate for the model
by taking the Arps decline exponent, b, equal to 1/3 (a known as hyperbolic
decline). In Figure (6.38), we plot curves of rate, q, vs. time, t. A decline curve
of / = 1,3 is plotted with circles, for a dened initial decline, 1
i
. and an initial
decline rate,
i
. Fetkovich (1980) related values of the decline exponent, /. of
1,3 to 2,3, for the solution gas drive mechanism.
In Figure (6.38) all rates, . are related to time, t, through Arps hyperbolic
relation
(t) =
i
(1 +/1
i
t)

1
l
(6.1)
By simplifying Equation (6.1) to
(t + t
0
)

1
l
(/1
i
)

1
l

i
= (t + t
0
)

1
l
+ Co::t. (6.2)
and a further simplication of Equation (6.2), we obtain the expression:
(t + t
0
)

1
l
+ Co::t. - t

1
l
+ Co::t. (6.3)
Instead of using the Arps-type expression (6.2) for the rate (t), we introduce
a an expression (t
1
) that is similar to (6.3). For large values of time the
correspondence of (t
1
) to Arps decline is t
1
. At early times (t
1
)
.Now, we solve the diusion equation by dening the variable rate / = 1,3,
i.e., the rate function (t
1
) then becomes:
(t
1
) = t
3
1
. (6.4)
Moreover, an inverse "Laplace" transform of the rate function is:
1
1
() =
i
2
2
= C(i). (6.5)
6. RATE DECLINE CURVES 223
Figure 6.38: The rate, , versus time, t, for b=0.33 plotted in circles, and for
other values of the decline exponents (b=0.5, 1, and 2) all plotted as solid line.
The rate versus time is calculated for a specic initial decline
i
= 5000 and a
initial decline, 1
i
= 0.01 ).
The dimensionless pressure, j
1
. is now:
j
1
=
1
_
0
1
2
i
2
_
\
0
(:
1
_
i)
:
2
1
0
(:
1
_
i)
_
c
it
T
di (6.6)
j
1
=
2
t
3
1
_
c

r
2
T
4I
T
_
1
1
2
:
2
1
t
1
+
1
32
:
4
1
t
2
1
__

1
2
1i
_
:
2
1
4t
1
__
+
1
16
:
2
t
1

3
4
_
(6.7)
From this, and provided that the no-ow boundary moves with a distance : from
a wellbore:
Jj
1
J:
1
= 0 . for :
1
= i
_
t
1
(6.8)
The distance :, where the no-ow boundary is spaced in time t. is related by the
constant i (i.e., the coecient of the no-ow moving boundary). The relation
between and i is given by the following equation:
=
1
2
1i
_
i
2
4
_

2
i
2
1
5
8
i
2
+
1
32
i
4
3
3
4
i
2
+
1
32
i
4
c

2
4
(6.9)
We can now plot the constant as a function of the constant, i, as shown in
Figure (6.39).
The constant i relates the dimensionless distance of a no-ow moving bound-
ary, :
1
. to a dimensionless time, t
1
.
224 6. RATE DECLINE CURVES
Figure 6.39: The constant as a function of the constant , i. of the no-ow
moving boundary.
Table 6-9: The constant A as a function of the coecient of the no-ow moving
boundary
6. RATE DECLINE CURVES 225
Figure 6.40: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and the dimensionless distance, :
1
(for i = 1 and / =
1
3
).
Further, for a coecient i = 1, we plot the dimensionless pressure, j
1
. as
a function of the dimensionless time, t
1
. and the dimensionless pressure, j
1
.
as a function of the dimensionless radius, :
1
. Subsequently, we also plot the
dimensionless pressure, j
1
. as a function of time, t
1
. and distance, :
1
, for various
coecients, i.e., for i = 3. 5. 7 and 9.
The following plots present the variation of the dimensionless pressure with
time and distance, for inner boundary conditions of variable rate (an Arps decline
of / =
1
3
) and the coecient i = 1.
In Figure (6.41) and Figure (6.43), all times between 0 and 16, correspond
to times where the dimensionless pressure, j
1
. is outside the no-ow boundary.
Consequently the calculated dimensionless pressure, j
1
. is inside the no-ow
boundary between the dimensionless time of 16 and 20. At a dimensionless
time, t
1
= 16. and a dimensionless pressure, j
1
= 0.00218, the no-ow boundary
reaches a dimensionless radius, :
1
= 4.
In Figure (6.44), we plot the dimensionless pressure, j
1
. versus the dimen-
sionless radius, :
1
. for the particular time t
1
= 16. The no-ow boundary moves
from the wellbore axis with a speed that is inversely proportional to the square
root of time. Moreover we have assumed a constant of proportionality i = 1,
and the no-ow boundary reaches the position of :
1
= 4.
For distances, of :
1
greater that 4, the calculated pressures, j
1
. are outside
the no-ow boundary.
6.3.2 Hyperbolic Decline (/ = 0.5)
In this case, the inner boundary conditions are again of variable rate, and of
Arps decline with exponent / = 0.5. This corresponds to the hyperbolic rate
226 6. RATE DECLINE CURVES
Figure 6.41: The dimensionless pressure, j
1
. versus the dimensionless time, t
1
.
and with a distance :
1
= 1 (for i = 1, and the decline exponent / =
1
3
).
Figure 6.42: The dimensionless pressure, j
1
. versus the dimensionless time, t
1
.
and with a distance :
1
= 4 (for i = 1, and the decline exponent / =
1
3
).
6. RATE DECLINE CURVES 227
Figure 6.43: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and with a distance :
1
= 10 (for i = 1, and the decline exponent, / =
1
3
).
Figure 6.44: The dimensionless pressure, j
1
, versus a distance, :
1
, at a dimen-
sionless time t
1
= 16 (for i = 1, and the decline exponent / =
1
3
).
228 6. RATE DECLINE CURVES
Figure 6.45: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 3).
Figure 6.46: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 5).
6. RATE DECLINE CURVES 229
Figure 6.47: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 7).
Figure 6.48: The dimensionless pressure, j
1
. versus the dimensionless time, t
1
.
and versus the dimensionless distance, :
1
(for i = 9).
230 6. RATE DECLINE CURVES
Figure 6.49: The rate, , versus time, t, for / = 0.5 (plotted in circles). The other
parameters were considered constant, i.e., the initial decline decline 1
i
= 0.01,
and the initial decline rate
i
= 5000).
decline, as presented in Figure (6.50).
The rate function (t
1
) is:
(t
1
) = t
2
1
(6.10)
and the inverse "Laplace" transform of the rate function is:
1
1
() = /. (6.11)
Now, the dimensionless pressure j
1
becomes:
j
1
=
1
2t
2
1
__
1 +
_
1
:
2
1
4t
1
_
c

r
2
T
4I
T
1i
_
:
2
1
4t
1
__
2
_
1
:
2
1
4t
1
_
c

r
2
T
4I
T
_
(6.12)
and, provided that,
Jj
1
J:
1
= 0 . ,o: :
1
= i
_
t
1
(6.13)
we get
(i) =
1
2
1i
_
i
2
4
_

2
i
2
1
i
2
4
2
i
2
4
c

2
4
(6.14)
We can now plot the constant as a function of the coecient of the no-ow
moving boundary, i. as shown in Figure (6.50).
As previously mentioned the coecient i from Table (6-10), relates the di-
mensionless position, :
1
. of a no-ow moving boundary to the dimensionless
6. RATE DECLINE CURVES 231
Figure 6.50: The constant (i) as a function of the constant , i. of no-ow
moving boundary.
Table 6-10: A as a function of the coecient of the no-ow moving boundary
232 6. RATE DECLINE CURVES
Figure 6.51: The dimensionless pressure, j
1
. versus the dimensionless time, t
1
.
calculated at a dimensionless radius :
1
= 1 (i = 1, and a decline exponent
/ = 0.5).
time, t
1
. By calculating the coecient (i) from Equation (6.14), we can, for
each choice of a coecient, i. compute a dimensionless pressure, j
1
. These pres-
sure conditions are those for which we are able to get the wellbore rate decline
with a decline exponent / =
1
2
. Further, for the coecient i = 1. we plot the
dimensionless pressure, j
1
. as a function of the dimensionless time, t
1
. and as
a function of the dimensionless radius, :
1
, for various coecients, i = 1. 3. 5. 7.
and 9.
From the above gures it is possible to determine how the dimensionless
pressure, j
1
, varies with the dimensionless time, t
1
. For example, in Figure
(6.52 ), we know that at the time t
1
= 16, the no-ow boundary reaches a
dimensionless radius :
1
= 4. All times t
1
< 16 are times where the dimensionless
pressure, j
1
. is outside of the no-ow boundary. Consistently, the calculated
dimensionless pressure, j
1
. is inside the no-ow boundary for dimensionless time
t
1
16.
In Figure (6.55), it is clear that, at the time t
1
= 16, the no-ow boundary
reaches a distance :
1
= 4. In order words, the pressure j
1
, calculated between
:
1
= 0 and :
1
= 4. is considered to be within the no-ow moving boundary.
All values of j
1
for :
1
4 are considered to be outside the no-ow moving
boundary.
6.3.3 Harmonic Decline (b = 1)
The third case is equivalent to the rate change at IBC of Arps, with the decline
exponent rate (the so called harmonic decline).
Here, the rate function (t
1
) is:
6. RATE DECLINE CURVES 233
Figure 6.52: The pressure dierence, j
1
versus the dimensionless time, t
1
. cal-
culated at a dimensionless distance, :
1
= 4 (coecient: i = 1, and decline
exponent: b = 0.5).
Figure 6.53: The pressure dierence, j
1
. versus the dimensionless time, t
1
, cal-
culated at the dimensionless distance r
1
= 10 (for i = 1, and decline exponent:
/ = 0.5).
234 6. RATE DECLINE CURVES
Figure 6.54: The dimensionless pressure, j
1
. versus the distance, :
1
. at a di-
mensionless time t
1
= 16 (for i = 3, and decline exponen: / = 0.5).
Figure 6.55: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
.
and versus the dimensionless distance, :
1
(for i = 1, / = 0.5).
6. RATE DECLINE CURVES 235
Figure 6.56: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 3, / = 0.5).
Figure 6.57: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 5, / = 0.5).
236 6. RATE DECLINE CURVES
Figure 6.58: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
.
and versus the dimensionless distance, :
1
(for i = 7, / = 0.5).
Figure 6.59: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 9, / = 0.5).
6. RATE DECLINE CURVES 237
Figure 6.60: The rate, , versus time, t, for / = 1.0 (plotted in circles). With a
specic initial decline
i
= 5000 and an initial decline rate 1
i
= 0.01).
(t
1
) = t
1
1
(6.15)
or
1
1
() = 1 = C (i) (6.16)
For part I, we have:
t
_
0
c
it
T
_
iJ
1
_
i
_
it
1
_
d i =
i
_
t
1
2
t
2
1
c

2
4
(6.17)
and for part II:
t
_
0
c
it
_
i1
1
_
i
_
it
1
_
d i =
i
2:
t

3
2
1
1i
_
i
2
4
_
c

2
4

2
:i
t

3
2
1
(6.18)
where
1i (.) = + log . +
1

1
.
a
: :!
. with = 0.577 (6.19)
and
1

1
.
a
: :!
=
:
_
0
c
&
1
n
dn (6.20)
238 6. RATE DECLINE CURVES
Figure 6.61: The constant as a function of the coecient i. (i) is calculated
for an inner boundary condition of variable rate. Arps decline exponent / = 1.
Hence, we get for A,
=
2
i
2
c

2
4
+
1
2
1i
_
i
2
4
_
(6.21)
We plot A from Equation (6.21), as a function of i. From Figure (6.61), we
choose several values for coecient i and calculate constant A. By substituting
A from Equation (6.22), into Equation (6.23), we nd the dimensionless pressure,
j
1
as a function of the radius, :
1
and the time, t
1.
We plot j
1
= j
1
(:
1
. t
1
) in
Figure (6.62), to Figure (6.65).
For j
1
:
j
1
=
1
_
0
c
It
_
J
0
_
:
_
/
_

:
2
1
0
_
:
_
/
__
d/ (6.22)
and
j
1
= t
1
1
c

r
2
T
4I
T
_

1
2
1i
_
:
2
1
4t
1
__
(6.23)
The formulas developed above may be found either as "Laplace" transforms, or
as "Hankel" transforms, e.g., from specic tables.
As already mentioned, the coecient, i. from Table (6-11), relates the dimen-
sionless position, r
1
. of a no-ow moving boundary to the dimensionless time,
t
1
. By calculating coecient A from Equation (6.21), we can, calculate for each
choice of the coecient, i compute the dimensionless pressure, j
1
. These pres-
sure conditions are those for which we are able to obtain the the wellbore rate
decline with a decline exponent / = 1. Further, for the coecients, i = 1 we
6. RATE DECLINE CURVES 239
Table 6-11: A as a function of the coecient of the no-ow moving boundary
can plot the dimensionless pressure, j
1
. as a function of the dimensionless time,
t
1
. as well as a function of dimensionless radius, :
1
.
From the above gures, it is visible how the dimensionless pressure, j
1
,
varies with the dimensionless time, t
1
. We can also see whether it is positioned
inside or outside the no-ow moving boundary. For all times t
1
<
_
v
i
_
2
. the
calculated pressure, j
1
. lies outside the no-ow moving boundary. In Figure
(6.61), for example, the calculated pressures, j
1
. for all times t
1
1, reside
outside the no-ow moving boundary, due to :
1
= 1 and i = 1. In Figure
(6.62), where, :
1
= 4, we see that for t
1
= 16, the no-ow boundary reaches the
dimensionless radius, :
1
= 4. In Figure (6.62), all times t
1
< 16 are such that
the dimensionless pressure, j
1
. is outside the no-ow boundary. In the same
gure, the calculated dimensionless pressure, j
1
. is inside the no-ow boundary
for dimensionless times, t
1
16. In Figure (6.63) all calculated pressure values,
j
1
. lie outside the no-ow moving boundary. In Figure (6.54), it is clear that,
t
1
= 16, the no-ow boundary reaches a distance :
1
= 4. In other words,
the pressure j
1
, calculated between :
1
= 0 and :
1
= 4, is considered to be
within the no-ow moving boundary. All values of pressure, p
1
. for :
1
4 are
therefore considered to be outside the no-ow moving boundary. In Figure (6.55)
to Figure (6.59), we plot the dimensionless pressure, j
1
. calculated according to
Equation (6.23), versus the dimensionless radius, :
1
. and versus dimensionless
time, t
1
. For each coecient of the no-ow moving boundary, i. we can for
values of :
1
and times t
1
, plot the pressure solution j
1
and perform a similar
240 6. RATE DECLINE CURVES
Figure 6.62: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
calculated at a dimensionless radius :
1
= 1 (for: i = 1 and / = 1).
Figure 6.63: The pressure dierence, j
1
, versus the dimensionless time, t
1
,
calculated at a dimensionless distance :
1
= 4 (for: i = 1, and / = 1).
6. RATE DECLINE CURVES 241
Figure 6.64: The pressure dierence, j
1
, versus the dimensionless time, t
1
,
calculated at a dimensionless distance :
1
= 10 (for: i = 1, and / = 1).
Figure 6.65: The dimensionless pressure, j
1
, versus the dimensionless distance,
:
1
, at a dimensionless time t
1
= 16 (for: i = 1, and / = 1).
242 6. RATE DECLINE CURVES
Figure 6.66: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 1, / = 1).
analysis evaluating the position in time where the calculated pressure lies.
The above calculations describe the dimensionless pressure, j
1
. as a function
of time, t
1
. and distance, r
1
, for various coecients i= 1, 3, 5, 7, and 9. These
pressure values are calculated for an inner boundary condition of variable rate.
This corresponds to a harmonic rate decline with an Arps decline exponent,
/ = 1. From Table (6-11), we can take any (i) and implement it in Equation
(6.23) in order to generate more pressure plots.
In the following subsection, we consider an inner boundary condition of vari-
able rate. We use the Arps hyperbolic expression with a decline exponent / 1.
Referring to Fetkovich (1980), such solutions are considered to be of transient
type depletion. Nevertheless, it is rather than interesting to include the solution
for / 1 in this study.
6.3.4 Decline Exponent (b = 2)
Figure (6.71) presents a plot of the rate, q, vs. time, t, for a decline exponent
/ = 2 (thick line). This decline is usually not considered as a depletion decline.
The / exponent indicates that the transient region is used instead of its depletion
counterpart from combined curves of Fetkovich (1980).
For the rate change (t) we have:
(t
1
) =
_
:
t
1
(6.24)
and with an inverse "Laplace" transform:
1
1
[ (t
1
)] = /

1
2
. (6.25)
6. RATE DECLINE CURVES 243
Figure 6.67: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 3, / = 1).
Figure 6.68: The dimensionless pressure, j
1
versus the dimensionless time, t
1
and versus the dimensionless distance, :
1
(for i = 5, / = 1).
244 6. RATE DECLINE CURVES
Figure 6.69: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 7, / = 1).
Figure 6.70: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 9, / = 1).
6. RATE DECLINE CURVES 245
Figure 6.71: The rate, , versus time, t, for decline exponent / = 2.0 (plotted in
circles). The initial decline
i
= 5000, and the initial decline rate, 1
i
= 0.01.
we get the pressure, j
1
. according to:
j
1
=
_
:
t
1
c

r
2
8:
_
1
0
_
:
2
1
8 t
1
_
+
1
2
1
0
_
:
2
1
8 t
1
__
(6.26)
which gives:
=
1
2
1
0
_
i
2
8
_
+ 1
1
_
i
2
8
_
1
0
_
i
2
8
_
1
1
_
i
2
8
_ (6.27)
With the values of i from Table (6-12), relating the dimensionless position,
r
1
. of a no-ow moving boundary to the dimensionless time, t
1
, we can calculate
the dimensionless pressure, p
1
.for which we are able to get a wellbore rate decline
with the decline exponent, / = 2. Further plots have been made for, i= 1
From the above gures, we can see how the dimensionless pressure, j
1
, varies
with the dimensionless time, t
1
. and where it is positioned, i.e., inside or outside
the no-ow moving boundary. For all times t
1
<
_
v
i
_
2
. the calculated pressure,
p
1
. lies outside the no-ow moving boundary. This is exemplied in Figure
(6.73), where j
1
is outside the no-ow moving boundary, for t
1
1. :
1
= 1 and
i = 1. Figure (6.75), shows that at the time, t
1
= 16, the no-ow boundary
reaches dimensionless radius, :
1
= 4. Further Figure (6.74), correspond to
those all times t
1
< 16, where the dimensionless pressure, j
1
is outside the
no-ow boundary. In the same gure, the calculated dimensionless pressure, j
1
.
is inside the no-ow boundary for dimensionless times t
1
16. In Figure (6.75)
all calculated pressure, j
1
. values lie outside the no-ow moving boundary.
In Figure (6.76), at the time t
1
= 16, the no-ow boundary reaches a dis-
tance :
1
= 4. In order words, the pressure p
1
, calculated between :
1
= 0 and
246 6. RATE DECLINE CURVES
Figure 6.72: The constant, (i), as a function of the coecient, i, of the no-ow
moving boundary, .
Table 6-12: The constant A as the function of the coecient of the no-ow
moving boundary
6. RATE DECLINE CURVES 247
Figure 6.73: The dimensionless pressure, j
1
, versus the dimensionless time,
t
1
,calculated at the dimensionless radius, :
1
= 1 (for, i = 1 and for, b = 2).
Figure 6.74: The pressure dierence, j
1
. versus the dimensionless time,
t
1
,calculated at a dimensionless distance, :
1
= 4 (for, i = 1, and for, / = 2).
248 6. RATE DECLINE CURVES
Figure 6.75: The pressure dierence, j
1
. versus the dimensionless time, t
1
,
calculated at a dimensionless distance, :
1
= 10 (for i = 1, and decline exponent,
/ = 2).
Figure 6.76: The dimensionless pressure, j
1
, versus the distance, :
1
, at the
dimensionless time, t
1
= 16 (for i = 1, and and decline exponent, / = 2).
6. RATE DECLINE CURVES 249
Figure 6.77: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and the dimensionless distance, :
1
(for i = 1, / = 2).
:
1
= 4 is considered to be within the no-ow moving boundary. Consistently,
all values of pressure, j
1
. for :
1
4 are considered to be outside the no-ow
moving boundary.
In Figure (6.77) to Figure (6.81), we plot the dimensionless pressure, j
1
.
according to Equation (6.26), versus the dimensionless radius, :
1
. and versus
the dimensionless time, t
1
.
Table (6-13) summarises all developed model solutions. In the table solutions
for j
1
and (i). rate inner boundary conditions are given for each variable (with
Arps rate decline exponents of: / = 0.33, / = 0.5, / = 1, and / = 2). Further,
Table (6-13) presents dimensionless pressure, solutions, j
1
, calculated for early
and late dimensionless times, t
1
.
The computations conrm that the production mode must be of variable-
pressure if the rate is to follow a specic value of b, which is in agreement
with the work of Raghavan (1993). The model solutions are analytical and are
available within a certain drainage area and at the wellbore for selected b values.
Furthermore, the model assumes that a vertical well produces under variable-rate
conditions and with a no-ow outward-moving boundary
250 6. RATE DECLINE CURVES
Figure 6.78: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus the dimensionless distance, :
1
(for i = 3, / = 2).
Figure 6.79: The dimensionless pressure, j
1
, versus dimensionless time, t
1
, and
versus the dimensionless distance, :
1
(for i = 5, / = 2).
6. RATE DECLINE CURVES 251
Figure 6.80: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus dimensionless distance, r
1
(for i = 7, / = 2).
Figure 6.81: The dimensionless pressure, j
1
, versus the dimensionless time, t
1
,
and versus dimensionless distance, :
1
(for i = 9, / = 2).
252 6. RATE DECLINE CURVES
Table 6-13: The dimensionless pressure, "p
1
", calculated by the model for "t
1
"
going to 0,and the "t
1
" going to innity, for a variable rate production of Arps
type. The decline exponent, b (b with values of 0.333; 0.5; 1., and 2) denes the
Arps type production decline
6. RATE DECLINE CURVES 253
6.4 Summary
The aim of the multiple-fractured horizontal well model is to provide a tool for
an improved modelling of oil production from horizontal or near-horizontal wells
with induced fractures. This is achieved by implementing original analytical so-
lutions. The state-of-the art model is useful diagnostic and prognostic screening
tool for the oil industry, and its robustness and speed render the programme
for well production optimisation of a horizontal well with induced fractures very
user-friendly. Another benet is the bringing-together of rate-time and pressure-
time analyses, thereby a total package to better characterise the behaviour of a
horizontal well with induced fractures. The oil solutions from the model are also
applicable for gas ows. If pseudopressure, :(j)
1
is employed instead of pres-
sure, j
1
. the corresponding solution will not take into account non-Darcy eects
and can be considered as an approximative gas solution. To include non-Darcy
ow, we should derive additional equations for gas ow for the system when a
well with fractures is coupled to a gas reservoir.
Model solutions for a no-ow moving boundary further investigate the nature
of the wellbore pressure required for the Arps rate decline production, dened
with the xed exponent /. Chapter 5 presented the physical model together with
its analytical solutions. The nature of the pressure proles for several b values
were demonstrated. Since each / value was selected, they may as such be related
to the drive mechanism (i.e., b = 1/3, for the solution-gas-drive, b = 0,5, for
the gravity-drainage, and b = 1, for the multilayered no-cross ow production).
The introduction of the dimensionless pesudo-pressure, :(j)
1
. in the place of
the dimensionless pressure, j
1
.renders possible an extension from oil ow to gas
ow.
254 6. RATE DECLINE CURVES
Chapter 7
CASE STUDIES
The application of horizontal wells and, more specically, that of multi-fractured
horizontal wells for exploiting oil and gas reservoirs, is rmly established within
the industry. Various authors have made signicant contributions to improving
the understanding of the ow behaviour of such wells.
The evaluation of multi-fractured horizontal well performances, or the se-
lection of optimum perforation/stimulation designed for such wells, may be ap-
proached through ne grid reservoir simulation. However, while reservoir simu-
lation is the most advanced method for predicting well performance, it is often
too time-consuming when conducting parametric screening studies. Often, the
required data is unavailable and the eort may not be warranted.
7.1 Model Comparison and Validation
7.1.1 Well Models Comparisons
The multiple-fractured horizontal well model is a tool for an improved modelling
of oil production from horizontal or near-horizontal wells with induced fractures.
The modelling is achieved by implementing original analytical solutions.
The state-of the-art technology model represents a useful diagnostic and prog-
nostic screening tool for the oil industry, and its robustness and speed make this
programme for well production optimisation of a horizontal well with induced
fractures easy to use. Another benet is the bringing-together of rate-time and
pressure-time analyses to provide a total package for an improved characterisa-
tion of the behaviour of a horizontal well with induced fractures.
For a sake of comparison, a fractured-horizontal well code was coupled to
a code of a fractured-vertical well and a partially-perforated-horizontal well.
Thus, one comparison feature consists in available output variables for the three
models. A fractured-horizontal well model is internally developed (as described
in chapter 4) and two other model codes was provided by BP for comparison
255
256 7. CASE STUDIES
Figure 7.1: The pressure-dierence versus time for three models (multi-fractured
horizontal well (MFW-open outer boundary-SLAB), vertical-fractured well
(VFW-open outer boundary-SLAB) and partially perforated horizontal well
(PPHOW-closed, BOX ) [After Cvetkovic et al. (2000)].
purposes only. These two model solutions are partly reviewed in chapters 2, 3
and 4.
Table (7-1) presents input data for the fractured-horizontal well, the fractured-
vertical well, and the partially-perfortated-horizontal well models. The corre-
sponding well response plots are presented in Figure (7.1).
7.1.2 Semi-Analytical versus Numerical Model Valida-
tion
As an alternative to simulation, the application of semi-analytical models can
readily yield wellbore responses to various boundary conditions. This is often
sucient to provide an understanding of which factors that have the most inu-
ence on well performance. If simulation work is warranted, it can then proceed
with the insight obtained from the analytical models. The model is validated to
the numerical simulators (GeoQuest Schlumberger ECLIPSE). Recently, model
solutions were compared to other commercial solutions and the obtained results
were satisfactory. Innite-acting outer boundary and innite conductivity frac-
ture solutions were similar, whereas the nite conductivity varied due to the
fracture ow modelling not being equal between the two models.
7. CASE STUDIES 257
Table 7-1: The input data for fractured-horizontal well (MF), fractured-vertical
well (VFW) and partially perforated horizontal well (PP) model
258 7. CASE STUDIES
Figure 7.2: The semi-analytical model cumulative production compared to the
other models (2 numerical and a semi-analytical models).
A semi-analytical model was also compared to several commercial simulators.
For the innite conductivity fractures (SLAB model), solutions from the semi-
analytical and numerical model were equal, provided that the numerical model
boundary in the x and y directions were large enough (since the outer boundary
of the numerical model was not innite acting). Finite conductivity fracture
solutions varied due to the manner in which the fracture ow was modelled.
A semi-analytical solution was compared to several commercial numerical and
semi-analytical simulators.
Similar results were obtained for the simulator with improved gridding (as
local grid renement and PEBI gridding), and have been presented by Cvetkovic
(2000). The comparison of the recent study with the commercial simulator
demonstrates an almost equal innite conductivity fracture production for a
horizontal well with 5 fractures, whereas the nite conductivity results vary.
Obtained plots cannot be presented due to condentiality reasons.
The validation of the recent model is based on the data provided by the
Amoco Exploration&Production Technology Group (1997). The semi-analytical
model is compared to 3 other model solutions (2 numerical and 1 semi-analytical)
Figure (7.2) demonstrates how the semi-analytical model solution matches other
models (for a fracture conductivity, 1
C
, ranging from 50 to 100 mDft). The
dierence between model solutions is mainly caused by the modelling of a -
nite fracture conductivity, 1
C
. Another dierence is in the availability of the
modelling solutions when handling the early-time production rate with fracture
7. CASE STUDIES 259
Figure 7.3: Comparisons of model-calculated cumulative productions (two nu-
merical models and a semi-analytical single-phase model). The production his-
tory comprises 1200 days.
responses. Despite these early-time dierences in the production proles, it is
possible to match other model proles for the larger time interval. Figure (7.3)
shows that, within 1200 days of production, the semi-analytical model matches
other single-phase model cumulative productions. The cumulative productions
for two numerical models and the single-phase model vary with ow within a
fracture that is not equally modelled. The semi-analytical model is veried by
comparing it to other model production proles as presented in Figure (7.3).
7.2 Fractured-Horizontal Well - Oil Production
7.2.1 Fractured-Horizontal Well (Ekosk Oil Field - North
Sea)
Case history data have been obtained from the Ekosk eld in the North Sea.
Data for the reservoir, a horizontal well and fractures are provided in Table (7-
2). We assume that 8 transversal fractures are equally spaced along a horizontal
well with a length of 2090 ft. Further, each fracture half-length and fracture
height are the same. After performing the sensitivity analysis, we select input
parameters to be ne-tuned. Further, observed well data are compared to data
calculated by the model, as obtained when varying the IBC of pressure and/or
rate.
260 7. CASE STUDIES
Table 7-2: Input data from the Ekosk eld - North Sea (for a horizontal well
with 8 transversal-fractures)
Variable-Pressure IBC
Figures (7.4 and 7.5) compare observed data to data obtained from the model
at a variable-rate IBC. Daily measured wellbore pressures, used as input data in
the model, are divided into 7 pressure intervals, and each interval is averaged.
Matching of the observed rate data with that calculated by the model was
obtained with an IBC of variable-pressure, as given in Figure (7.5). The over-
all time scope of 200 days was divided into 3 time intervals and each interval
comprised the averaged pressure data calculated within the specic increment of
time.
Variable-Rate IBC
Consistently, matching of the observed pressure-dierence data with that cal-
culated by the model was obtained for variable-rate IBC. For identical fracture
half-lengths and fracture heights, the three fracture conductivities were consid-
ered (for a nite conductivity 1
C
= 20, 15 and 5 mDft). The model t is
presented in Figure (7.6).
Thus, selecting a fracture conductivity 1
C
= 20 mDft and choosing two
fracture half-lengths, 1
)
of 50 ft and 25 ft; we plot the pressure-dierence versus
time as in Figure (7.7). The match is not yet achieved, so further tuning is
needed. Due to the initial pressure, 1
i
, not being precisely dened, it can be
7. CASE STUDIES 261
Figure 7.4: A comparison of observed (oil rate) data with that calculated by the
model at an IBC of variable pressure (from 7 selected time intervals).
Figure 7.5: A comparison of observed (oil rate) data with that calculated by
the model at IBCs of variable pressure. The daily measured pressures at the
wellbore are devided into 3 pressure intervals.
262 7. CASE STUDIES
Figure 7.6: The matching of observed and calculated pressure dierences versus
time (for variable-rate IBCs and changes in fracture nite conductivities 1
C
).
varied.
Then, by selecting a lower value of 1
i
(i.e., 3580 psi instead of 4400 psi) and
further maintaining the fracture conductivity and fracture sizes as previously
dened, it become possible to better match the observed data. Observed and
calculated pressure-dierence well data are presented in Figure (7.8).
In a time-interval of only 200 days, calculated data for IBCs of both variable-
pressure and variable-rate were found to be within the range of observed data
for a horizontal well with 8 fractures. These results demonstrate that IBCs of
variable-pressure and variable-rate were integrated in one tool and can be used
for the screening analyses of a producing well with fractures.
7.2.2 Fractured-Horizontal Well (North Sea Oil Field)
Variable-Pressure IBC
The following study compares the model production data to observed well data
for 160 days of production. The inner boundary condition of the model is of
variable pressure, and the fracture conductivity varies as 1
C
= 70, 40 and 20
mDft. Figure (7.9) compares the results from the model to observed production
well rates. Observed well data corresponded to those created by the model for
varying fracture conductivities.
The cumulative production, according to model calculation, for a horizontal
7. CASE STUDIES 263
Figure 7.7: The matching of observed and calculated pressure-dierences versus
time (for variable-rate IBCs and changes in the fracture half-length, 1
)
, from 50
ft and 25 ft, and assuming a maintained fracture conductivity, 1
C
, of 20 mDft).
well with 6 fractures producing oil for 700 days is presented in Figure (7.10).
Each fracture conductivity is 1
Ci
= 70 mDft, and the wellbore IBC is of variable
pressure. With an increase a pressure intervals it would be possible to achieve a
better match.
Variable-Rate IBC
Figure (7.11) presents calculated pressure dierences for several fracture conduc-
tivities, 1
C
, with values of 75 mDft, 100 mDft up to 150 mDft. For the 50 days of
well production, the observed pressure dierence corresponded to that calculated
by the model (for a fracture conductivity of 150 mDft). Between 50 and 140 days
of production the fracture conductivity was reduced to 100 mDft, and after 140
days the conductivity was further reduced to below 50 mDft. This screening was
achieved with variable-rate IBCs and varying fracture conductivities.
By using the fractured-horizontal well features with IBCs of variable-pressure
and variable-rate, it is possible to match the observed wellbore data. The IBCs
of variable-pressure and variable-rate were integrated into one tool that was used
in the screening analyses.
264 7. CASE STUDIES
Figure 7.8: The matching of observed and calculated pressure-dierences versus
time (for variable-rate IBCs and changes in the fracture half-length, 1
)
, from 50
ft and 25 ft, and assuming a maintained fracture conductivity, 1
C
, of 20 mDft).
The initial pressure, 1
i
, is reduced from 4400 psi to 3850 psi.
7. CASE STUDIES 265
Figure 7.9: Observed and calculated rates as functions of time. The IBC of the
model are of variable pressure, and the fracture conductivities, 1
C
, change from
70, 40 down to 20 mDft.
Figure 7.10: Matching of well observed cumulative oil data with a model calcu-
lated.
266 7. CASE STUDIES
Figure 7.11: Observed and model pressure dierences vs. time, in addition to
calculated and observed rates vs. time. The IBC of the model are of variable
rate.
7.2.3 Syd-Arne Oil Field (North Sea)
The Syd-Arne eld, operated by Amerada Hess Corporation, is located in the
Danish part of the North Sea. The eld was originally discovered in 1969 and
its initial reserves estimated by the Danish Energy Agency are 185MMstb of oil
and 434 Bcf of gas. The eld came into production in 1999 and is currently
producing with a total of 19 development wells (from which 12 are fractured-
horizontal oil producers and 7 are fractured-horizontal water injectors). The eld
is an elongated chalk anticline, 12 km by 3 km, with a depth between 2700-2940
m subsea. Over the crest of the eld, the oil column is restricted to the thickness
of the reservoir. A more detailed description of the Syd-Arne eld is provided
by Christensen et al. (2006). The reservoir consists of the Maastrichtian best
reservoir layer (Upper Cretaceus) to the Danian (Paleocene) chalk of the Tor
and Ekosk Formation (Fm) with reservoir parameters presented in Table (7-2).
The reservoir, fracture and well data were provided by Amerada Hess for the
purpose of this thesis in 2008.
A Horizontal Well with 14 Transversal Fractures
A fractured-horizontal well penetrates 14 transversal fractures, as presented in
Figure (7.12). Such a well produces for almost 10 years. Observed well data
7. CASE STUDIES 267
Table 7-3: Some general parameters of the Syd Arne North Sea eld (SPE
103282)
Figure 7.12: The fractured horizontal well, SA-P1 penetrating 14 transversal-
fractures in an oil reservoir (cross section).
are given in Figures (7.13, 7.14 and 7.15), and model input variables include
reservoir, fracture and well data, as presented in Figures (7.16 and 7.17).
Measured Well Data Measured well data are presented in Figures 7.13 to
7.15.
Model Input Data. Model input data for the IBC of constant pressure are
shown in Figures 7.16 and 7.17.
Obtained Match Figure 7.18 shows the obtained match between the calcu-
lated and measured well data.
Additional Model Information A model with inner boundary conditions
of the constant pressure also matches the cumulative production. Individual
fracture rates, fracture cumulative production and productivity index, PI, as a
function of time are additional information available by the model.
For a well that has been producing for almost 10 years, it is with the provided
input data and performed sensitivity analysis, possible to match the observed
well rate production, as given in Figure (7.18). The wellbore IBCs were of
268 7. CASE STUDIES
Figure 7.13: The measured wellbore pressure data, j
&)
(psi), versus time, t (d).
Figure 7.14: The measured oil rate,
c
(bbl/d), the equivalent oil rate,
cc
(bbl/d),
and the gas rate,
j
(Scf3/d), versus time, t (d), for a horizontal well SA-P1 with
14 transversal-fractures.
7. CASE STUDIES 269
Figure 7.15: The measured oil rate,
c
(bbl/d), the equivalent oil rate,
cc
(bbl/d),
and the gas rate,
j
(Scf3/d), versus time, t (d), for a horizontal well SA-P1 with
14 transversal-fractures on a log-lin scale.
Figure 7.16: The well and fracture input data.
270 7. CASE STUDIES
Figure 7.17: The reservoir input data.
Figure 7.18: A comparison of the calculated and measured well data (model data
obtained with an IBC of constant pressure).
7. CASE STUDIES 271
Figure 7.19: The well cumulative production, Q (bbl), and the fracture produc-
tion, Q
)vi
(i = 1. ...14) (bbl), versus time (d). The IBC of the model is of constant
pressure.
constant pressure. Additional model features to the screening analysis consist in
calculating individual fracture rates and cumulative production, as presented in
Figures (7.19, 7.20).
Model Input (IBC of variable rate) An IBC of variable rate matches the
pressure-dierence during the rst interval of the variable-rate IBC given in
Figure (7.21). The rst interval lasted ca. 800 days. More time intervals will
permit an improved match of the observed data, as presented in Figure (7.22).
The programme set- up is limited to only limited interval changes, and the
availability of more intervals would allow a better match for the entire pressure-
dierence history of over 3000 days.
IBC Step-Function A step-function or IBC of constant-rate to constant-
pressure in the same run matches the production history.
The choice of variety with regards to the IBCs makes it possible to fully eval-
uate the model features with the real daily measured well pressures, rates and
GOR values. Such daily measured data were available within the entire produc-
tion history of a well with 14 fractures. Although the details of the production
history were not given, it was possible to achieve a reasonably good match of
the given measured well data. The obtained rate match was generated with the
272 7. CASE STUDIES
Figure 7.20: The model well rate, (bbl/d), and the fracture rates,
)vi
(i =
1. ...14) (bbl/d), versus time (d). The IBC of the model is of constant pres-
sure.
Figure 7.21: The pressure dierence, 1
i
1
&)
(psi), versus time (d) for an IBC
of variable rate. (Well production rates for the rst 800 days are considered as
1 rate-interval).
7. CASE STUDIES 273
Figure 7.22: The pressure dierence, 1
i
1
&)
(psi), versus time (d) for an IBC
of variable rate (The well production rates for the rst 800 days of are devided
into 3 rate intervals).
Figure 7.23: The step function match obtained with an IBC of constant-rate
to constant-pressure processed in a single run. Both pressures and rates are
matched within the single run.
274 7. CASE STUDIES
Figure 7.24: A fractured horizontal well, o 12, penetrating 14 transversal
fractures in an oil reservoir. Cross section view.
step-function, with which the overall time interval was divided into two parts.
In the rst interval the IBCs are of constant rate and in the second, they are
of constant pressure. After the rst interval, the model calculates the wellbore
pressure, maintaining it constant until the end. Thus, within a single run, the
IBC changes from constant-rate to constant-pressure. The step function feature
is applicable to the plateau production (as the production used in North Sea
elds). In the Figure (7.23) match of the pressure-dierence and the rate is
achieved within the single run.
Moreover, for the given rates, the pressure-dierence data were matched with
the IBC of variable-rate. The model is capable of handling IBCs of variable
pressure and rate. Thus, by varying the IBC in this manner the model calculates
well production responses that match the measured well data. A combination
of the model rates and pressures may help to analysis well data. The process
of matching well data is fast, so, overall, the developed tool can be used in the
well production screening analyses as demonstrated. An extension of the model
to naturally fractured reservoir features may be important for the advanced
fractured reservoir study of the Syd-Arne wells.
7.2.4 Syd-Arne Oil Field (North Sea)
A Horizontal Well with 14 Transversal Fractures
Measured Well Data Measured well data are presented in Figures 7.25 to
7.27.
Model Input (IBC of constant pressure) Model input data for the IBC
of constant pressure are shown in Figures 7.28 and 7.29.
7. CASE STUDIES 275
Figure 7.25: The measured wellbore pressure, j
&)
(psi), versus time, t (d).
Figure 7.26: The measured oil rate,
c
(bbl/d), the equivalent oil rate,
cc
(bbl/d),
and the gas rate,
j
(Scf3/d), versus time, t (d), for a horizontal well SA-P2 with
14 transversal fractures.
276 7. CASE STUDIES
Figure 7.27: The measured oil rate,
c
(bbl/d), the equivalent oil rate,
cc
(bbl/d),
and the gas rate,
j
(Scf3/d), versus time, t (d), for a horizontal well SA-P2.with
14 transversal fractures on a log-lin scale.
Figure 7.28: The well and fracture input data.
7. CASE STUDIES 277
Figure 7.29: The reservoir input data.
Obtained Match It is possible to select model-generated well production pro-
les and match them with the well production history as shown in Figures (??,
7.31, 7.32). By varying the fracture half-length and partial fracture penetration,
we are able to match the well production history. Cumulative production proles
were generated from the model by assuming that both the fracture half-length,
1
)
, and the fracture-height, /
)
, (for each of the 14 fractures) were equal. The
model production prole (with 1
)
= 50 and /
)
= 50) matched the well data
during the rst 400 days of production. However, due to changes in the fracture
shape, the last 700 days of production can only be matched by reducing both
the fracture height and the fracture half-length in the model to /
)
= 40 ft. and
1
)
= 34 ft With more well and fracture data, it would be possible to achieve a
better match of the rates.
Nevertheless with an IBC of constant pressure, the model matches the cu-
mulative production data, as shown in Figure (7.31). The two production cu-
mulative rates from the model are dened with the fracture half-length, 1
)
, and
the fracture penetration height, /
)
. The measured production cumulative rate
is matched for the rst 500 days of production and from day 2750 until the end
of the production history. In between the fracture character changes. In order
to investigate such change in fracture character, an additional production pro-
le is created, with an average fracture half-length, 1
)
, of 34 ft and an average
fracture penetration-height of 40 ft. In Figure (7.32), the cumulative production
obtained with the model matches the measured cumulative rate between 550
and 1050 days of a well production history. It is evident from the gure that
278 7. CASE STUDIES
Figure 7.30: A comparison of the calculated and measured well data (model data
obtained with an IBC of constant pressure).
Figure 7.31: The measured versus calculated data for the cumulative rate, Q
(bbl), versus time (d). The calculated data are dened with the half-length, 1
)
,
(of 34 and 50 ft) and the fracture penetration height, /
)
, (of 40 and 50 ft).
7. CASE STUDIES 279
Figure 7.32: The measured versus calculated data for the cumulative rate, Q
(bbl) versus time (d). The calculated data are dened with the half-length, 1
)
(of 20, 34 and 50 ft) and the fracture perforation,/
)
(of 40 and 50 ft).
the fracture character changes. Due to this variation, three models were used
to quantify the fracture closure eects. These models generated the individual
fracture cumulative production and were thus of help in the fracture closure
diagnosis.
Additional Model Information The model provides 14 individual fracture
cumulative production proles in addition to a production prole for a well with
fractures as displayed in Figure (7.33). It is possible to investigate which fracture
rate production causes the reduction in well production.
The individual fracture rates versus time are given in Figure (7.35). Each
fracture half-length and fracture partial-penetration is the same.
It is also possible to include fracture conductivity changes that will reduce
the production prole. In this study, the fractures were considered to be innite
conductive. Further, it is possible to vary the IBC, and the well production
history can be matched accordingly (as presented in the previous well SA-P1
case). With more information on the producing well, it is possible to improve
the well production match and thus obtain a better fracture production prognosis
and diagnosis of the well in question.
Eective Wellbore Parameters and a Fractured Horizontal Well Pro-
ductivity In 2009 Cvetkovic investigated the productivity of a well with frac-
280 7. CASE STUDIES
Figure 7.33: The well cumulative production, Q(bbl), and the individual fracture
cumulative production, Q,:i (i = 1. .... 14), versus time, t (d). (Each fracture
half-length 1
)
=34 ft and the fracture partial penetration height, /
)
=40 ft).
Figure 7.34: The well rate production, q (bbl/d), and the individual fracture
rate production, q
)vi
(i = 1. .... 14), versus time, t(d) (Each fracture half-length
1
)
= 34 ft, and the fracture partial penetration height /
)
= 40 ft).
7. CASE STUDIES 281
Figure 7.35: Fracture rate for individual fractures (1, 6, 7, and 14) for varying
fracture half-lengths, 1
)
(34, 50 ft) and partial penetration heights (40, 50 ft).
tures by means of semi-analytical late-time approximations. These expressions
are related to a vertical and a horizontal well with a single-fracture production.
The ow in a fracture that is coupled to a horizontal well is of uniform ux (or
of innite conductivity ow) and the fractures can be positioned transversally or
longitudinally. To successfully measure the production eciency of a fractured-
horizontal well, the eective wellbore radius and the equivalent single-fracture
half-length were introduced. The eective wellbore radius was dened for a ver-
tical well and could be extended for its horizontal counterpart. The equivalent
fracture half-length was dened for a horizontal well with a single-transversal and
a single-longitudinal fracture. The late-time approximations were developed for
a fractured horizontal well positioned in the middle of an innite oil reservoir.
The developed expressions were employed for a multi-fractured horizontal well in
order to dene an eective wellbore radius and an equivalent fracture half-length
(for both the transversal and the longitudinal fractures). A series of solutions
corresponding to various conditions were given in a multiple-fractured-horizontal
well model. It was possible to verify derived eective parameters of a fractured
horizontal well by using a fast, robust and facile software program, a screening-
tool product, that has been of considerable benet to companies in the petroleum
industry.
Risk Assessments of a Fractured Horizontal Well Cvetkovic (2009) pre-
sented a screening approach with risk analysis considering the fractured well
282 7. CASE STUDIES
Figure 7.36: Valhall eld with several multi-fractured horizontal wells [After
Norris et al. (2001)].
data from Syd-Arne. The analysis of the data obtained for a real reservoir-
fracture-well was carried out by means of parameter changes, selecting objective
parameters as rates and cumulative rates, analysing generated proles and nally
comparing these proles to real data. Furthermore, the veried input informa-
tion was considered as the base data. Such input data were further used in
risk analysis thus generating a wide range of production proles. Such a study
provides a work ow that should be further tested for the optimal fracture posi-
tioning along a horizontal well. A screening procedure is complementary to any
complex simulation since semi-analytical tools are fast and require only limited
reservoir-fracture-well data.
7.2.5 Valhall Oil Field (North Sea)
A Horizontal Well with 5 Transversal Fractures
In this case study, we were able to simulate production well responses with time
for an IBC of constant and variable pressure or rate, as well as constant-rate
to constant-pressure changes; a step-function option. Valhall eld with several
fractured horizontal well is given in Figure (7.36).
Daily production rates and PIs of a horizontal well with 5 fractures and well
7. CASE STUDIES 283
pressures are given in Figures (7.37). The changing GOR with time in Figure
(7.38) conrms that the well was producing below the bubble point. Various
uid, reservoir, fracture and well properties are given in Table (7-5). Unfortu-
nately, we do not have much information concerning the uid characterisation,
the geological model or any obtained simulation results. The choice was made
to match well data (oil production rate, cumulative rate, wellbore pressure and
productivity index, PI) with the data generated by the model.
Measured Well Data For an IBC of constant rate, it is possible to match the
wells PI (with the nite conductivity option in the model), i.e., for early times,
the fracture conductivity is 50 D and for late time it is 2.5 D. Both have fractures
of 0.008 ft. The rate in Figure (7.39) is chosen to be constant at = 7170 bbl.
The next match is carried out for an IBC of variable pressure. The well
pressure as a function of time is split into three intervals, with s specic average
wellbore pressure. This pressure step function is for specic the model input
and the model output is rate. Without any ne-tuning, we are able to obtain a
good match of the rate data during a certain time period, as presented in Figure
(7.41).
For matching wellbore pressure, we use an IBC of variable rate. The time
interval is divided into several smaller intervals, each with a dened average
rate. This rate step function represents input for the model IBC as steps of
rate. Ideally, we should have applied variable rate changes within the entire
time-scale in order to obtain an adequate wellbore pressure response. Even with
an imprecise input of variable rate, as in Figure (7.42), we are able to t the
wellbore pressure data quite well for almost 200 days.
For the basic restart option treated here, we have as input a period of constant
rate followed by a pressure drop maintained constant. Both the SLAB and BOX
models are used in the data analysis. The inner boundary condition changes
from constant rate to constant pressure: the constant rate is kept for the rst
137 days, for which the model calculates the wellbore pressure. For the next
479 days, the calculated wellbore pressure is maintained constant and the model
calculates the rate decline, as shown in Figures (7.43, 7.44).
284 7. CASE STUDIES
Figure 7.37: Rate and PI data versus measured values of the cumulative well
production.
Figure 7.38: The wellbore pressure and GOR data versus time.
7. CASE STUDIES 285
Table 7-4: Wellbore IBCs of variable pressure for the innite conductivity and
nite conductivity, for "F
C
" = 50 (mDft) fractures
286 7. CASE STUDIES
Figure 7.39: Productivity Index versus Time (IBC = Constant Rate). [Model
and Well Data: PI - MATCH].
Figure 7.40: Model and well data PI - match (for an IBC of costant rate - the
fracture permeability and fracture width are constant).
7. CASE STUDIES 287
Figure 7.41: Rate versus time (for an IBC of variable pressure).
Figure 7.42: Calculated wellbore pressure matches model observed data for vari-
able rate IBCs.
288 7. CASE STUDIES
Figure 7.43: The step-function procedure calculates dimensionless pressure for
the IBC of constant-rate and rates for the IBC of constant-pressure. Within the
same run the IBCs are changing from constant-rate to constant-pressure.
Figure 7.44: The match of the models (SLAB & BOX) with the well rates.
7. CASE STUDIES 289
7.3 Fractured-Horizontal Well - Water Injec-
tion
7.3.1 Syd-Arne (North Sea)
A Water Injection Horizontal Well with 16 Transversal Fractures
Figure (7.45) presents the water injection rate and cumulative injection. The
wellbore pressure as a function of time is given a horizontal well with 16 fractures
in Figure (7.46). A well injection lasts ca. 2500 days. The reservoir, fracture
and well data are provided in Figures (7.47 and 7.48).
Measured Data Measured well data are presented in Figures 7.45 to 7.46.
Input Data Input data are shown in Figures 7.47 and 7,48.
Obtained Match Figure (7.49) show an almost perfect match for the water
injection rates with the water cumulative injection rates.
Additional Information The individual fracture injection rates are presented
in Figure (7.50), and the individual cumulative fracture injection production is
given in Figure (7.50). The water injectivity, and the PI, for a horizontal well
with 16 fractures is provided in Figure (7.52). These data represent valuable
additional information and can be complementary to any simulation studies.
290 7. CASE STUDIES
Figure 7.45: The water injection rate and the cumulative injection rate versus
time for a horizontal well with 16 transversal fractures.Well: SA-WI1.
Figure 7.46: The wellbore pressure versus time for the SA-WI1 well.
7. CASE STUDIES 291
Figure 7.47: The well and fracture input data.
Figure 7.48: The reservoir input data.
292 7. CASE STUDIES
Figure 7.49: The model water injection rate,
i
(bbl/d), and the water injec-
tion cumulative production, Q (bbl), versus time, t (d), for the SA-WI1 water
injection well.
Figure 7.50: The fracture water injection rate, q (bbl/d), for a horizontal well
with 16 fractures, and the individual fracture injection rates, q
)vi
(i = 1. .... 16).
7. CASE STUDIES 293
Figure 7.51: The cumulative fracture water injection, Q (bbl), for a horizontal
well with 16 fractures and the individual fracture water injection, Q
)vi
(i =
1. .... 16).
Figure 7.52: The productivity index, PI (bbl/d psi), versus tme, t (d), for the
water injection horizontal well penetrating 16 fractures.
294 7. CASE STUDIES
7.4 Fractured-Horizontal Well - Gas Produc-
tion/Injection
7.4.1 Syd-Arne Synthetic Data
If the pressure, 1. is substitute with the pseudo-pressure, :(j). we should be
able to use model solutions for a horizontal well with fractures producing from
a gas reservoir. Here, we rst assume that solutions are only approximations,
and that no non-Darcy ows exist. Consequently the oil well production data
are converted to an equivalent gas data. Figure (7.53) presents a match of the
overall equivalent gas production history with the model data.
The model pseudo pressure, :(j), is set up as an IBC of constant peudo-
pressure. The cumulative production obtained from the model matches the
reservoir synthetic equivalent gas data. Moreover oil well rates are converted
into equivalent values for gas, thus demonstrating that the model can be ex-
tended also for gas production. In order to include the non-Darcy ow the
model solutions should be developed further.
7. CASE STUDIES 295
Figure 7.53: A horizontal well with 14 transversal-fractures producing from a
synthetic gas reservoir. The cumulative oil production is converted into its cu-
mulative gas equaivalent. The IBCs are either constant or of variable pesudo-
pressures.
296 7. CASE STUDIES
7.5 Vertical Well Exponential Decline with
the Moving Boundary
This case further veries the rate-time analysis concept of a model with a no-
ow moving boundary. It provides new solutions to a diusion problem when
wellbore conditions are of variable rate decline with an exponent b close to zero,
and the no-ow boundary moving outwards from the wellbore axis. Generated
pressure proles comprise both the transient information of a changing drainage
volume and the predened depletion rate decline corresponding to the rate prole
as derived by the Arps exponential decline. The solutions are unique and open
lead to a new approach in solving variable-rate inner boundary conditions with
a no-ow outwards-moving boundary. The variable rate is almost exponential
and the derived expression can thus be compared to the analytical solutions of
depletion decline. One way of determining the velocity of the moving no-ow
boundary is to employ the numerical streamline solutions.
7.5.1 Variable Rate IBCs of b Almost Zero
Variable rate IBCs should be considered for exponent /. almost zero, i.e., the
value : =
1
b
should rst be selected as for instance 100 or 1000. The velocity of a
no-ow outwards-moving boundary, {, should be carefully chosen in order for so
singularity not to appear in the pressure dierence. Thus, for selected values of
{ and :, it would be possible to receive positive and realistic pressure dierence
data. In order words, only certain solutions correspond to physical behaviour
for the selected values of { and :.
The pressure dierence, 1
1
, for the almost exponential decline / = 0 is
dened by the choice of the : value, which is the inverse / exponent, and the
speed of the no-ow moving boundary, {. The expression is given in Equation
(7.1)
j
D
s
_
2

1
4
t
n
D
(
r
2
D
t
D
)

1
4
c

r
2
D
8
D
_
cos({
:
_
t
1
)
:
2
sin({
:
_
t
1
)
_
_
:
_
t
1
_
:
1
2

:
4
_
(7.1)
there, constant A is dened asymptotically (for : ) as:
s
:
2
cot
_
{
_
:
:
4
_
(7.2)
We now, choose a large value of : (i.e., a / value approaching zero). The value
of { is, accordingly, related to : by the relation:
7. CASE STUDIES 297
Table 7-5: Input parameters (n and "r
1
") to the semi-exponential decline in
the moving boundary model
{ =
`
_
:
: (7.3)
where ` is an integer. Further, assuming that both : and t are approaching
innity then, Equation (7.1) becomes:
j
1
s
_
2
:
:

1
4
t
a
1
(
:
2
1
t
1
)

1
4
c

r
2
T
8:
T
(7.4)
We note that
a
t
should be very small for calculating j
1
with time t
1
at a distance
:
1
. We plot j
1
versus t
1
for various :
1
values (1, 10, 100). Further we choose
: and {. For :
1
= 1. and n = 100, the dimensionless pressure dierence is
j
1
=
_
2

100

1
4
t
100
(
1
t
)

1
4
c

1
8:
.
Other possible plots should be made by evaluating solutions, signifying that
we should select variables in order to avoid singularities that are present in the
model solutions.
For :
1
= 1.and n = 100, j
1
=
_
2

10

1
4
t
10
(
100
t
)

1
4
c

100
8:
For :
1
= 100.and n = 100, j =
_
2

100

1
4
t
100
(
10000
t
)

1
4
c

10000
8:
For :
1
= 100.and n = 1000, j =
_
2

1000

1
4
t
1000
(
10000
t
)

1
4
c

10000
8:
In Table (7-5), / values are selected as / = 1 (when : = 1), / = 0.1 (when
: = 10), / = 0.01 (when : = 100).
The following Figures (7.54, 7.55, 7.56, 7.57, 7.58, 7.59, and 7.60 ) are plots
of the inverse decline exponent, : = 1, and various dimensionless radii, :
1
, of 1,
10, 20, 50, 100, 200, and 500. For increasing value of the dimensionless radius,
:
1
, the dimensionless pressure, 1
1
, is shifted from low to larger values as the
dimensionless time, t, is increased.
298 7. CASE STUDIES
Figure 7.54: The dimensionless pressure, j
1
, versus the dimensionless time, t,
for an inverse decline exponernt : = 1 and a dimensionless radius :
1
= 1.
Figure 7.55: The dimensionless pressure, j
1
, versus dimensionless time, t, for
an inverse decline exponernt : = 1 and a dimensionless radius, :
1
= 10.
7. CASE STUDIES 299
Figure 7.56: The dimensionless pressure, j
1
, versus the dimensionless time, t,
for an inverse decline exponernt : = 1 and a dimensionless radius, :
1
= 20.
Figure 7.57: The dimensionless pressure, j
1
, versus dimensionless time, t, for
an inverse decline exponernt : = 1 and dimensionless radius, :
1
= 50.
300 7. CASE STUDIES
Figure 7.58: The dimensionless pressure, j
1
, versus dimensionless time, t. for
an inverse decline exponernt : = 1 and dimensionless radius, :
1
= 100).
Figure 7.59: The dimensionless pressure, j
1
, versus dimensionless time, t, for
an inverse decline exponernt : = 1 and dimensionless radius, :
1
= 200.
7. CASE STUDIES 301
Figure 7.60: The dimensionless pressure,j
1
, versus dimensionless time, t, for
an inverse decline exponernt : = 1 and a dimensionless radius, :
1
= 500.
302 7. CASE STUDIES
Figure 7.61: The dimensionless pressure, j
1
, versus the dimensionless time, t,
for an inverse decline exponent, : = 10 and a dimensionless radius, :
1
= 1.
Figures (7.61, 7.62, 7.63, 7.64, 7.65, 7.66, and 7.67) represent plots for a
decline exponent : = 10, which almost corresponds to / = 0, and various dimen-
sionless radii, :
1
, of 1, 10, 20, 50, 100, 200, and 500 and dimensionless pressure,
1
1
, is shifted from low to high values for an increased dimensionless time, t.
Figures (7.68, 7.69, 7.70, 7.71, 7.72, 7.73 and 7.74) are the corresponding
plots for a decline exponent : = 100, which is even closer to the exponential
decline / = 0. The dimensionless pressure displays the same trend as in the two
preceding series of plots.
Figure (7.75 ) plots the dimensionless pressure as a function of dimensionless
time, for a decline exponent : = 1000, which represents an exponential decline
of approximated / = 0, and a dimensionless radius :
1
= 10. 1
1
values for other
dimensionless radii, :
1
, are not available due to the solution singularity caused by
the models limitation to handle only selected values of input model parameters
:, and :
1
.The model is unstable for values of the decline exponent that are
close to zero. Further, the model allows the multiplication of the dimensionles
pressure, 1
1
, so tuning and calibration is needed before data can be validated
with the measured pressure at various distances from the wellbore axis. The
details of the mathematical model are provided in Appendix A.
With reference to other / values, the speed of a no-ow boundary moving out-
wards from a wellbore axis may be further calculated by the streamline model.
Further continuous pressure measurements at various dimensionless radii, :
1
,
when available, may conrm the exact model value and justify the modelling
approach. An additional eort is required in order to model no-ow boundary
moving inwards, i.e., from the drainage radius towards the wellbore axis. This
modelling approach has great potential and provides a basis for further investi-
7. CASE STUDIES 303
Figure 7.62: The dimensionless pressure, j
1
, versus the dimensionless time, t,
for an inverse decline exponent : = 10 and a dimensionless radius :
1
= 10.
Figure 7.63: The dimensionless pressure, j
1
, versus the dimensionless time, t,
for an inverse decline exponent : = 10 and a dimensionless radius :
1
= 20.
304 7. CASE STUDIES
Figure 7.64: The dimensionless pressure, j
1
, versus the dimensionless time, t,
for an inverse decline exponent : = 10 and a dimensionless radius :
1
= 50.
Figure 7.65: The dimensionless pressure, p
1
,versus the dimensionless time, t,
for an inverse decline exponent : = 10 and a dimensionless radius :
1
= 100.
7. CASE STUDIES 305
Figure 7.66: The dimensionless pressure, j
1
. versus the dimensionless time, t,
for an inverse decline exponent : = 10 and a dimensionless radius :
1
= 200.
Figure 7.67: The dimensionless pressure, j
1
, versus the dimensionless time, t,
for an inverse decline exponent : = 10 and a dimensionless radius :
1
= 500.
306 7. CASE STUDIES
Figure 7.68: The dimensionless pressure, j
1
, versus the dimensionless time, t,
for an inverse decline exponent : = 100 and a dimensionless radius :
1
= 1.
Figure 7.69: The dimensionless pressure, j
1
, versus dimensionless time, t, for
an inverse decline exponent : = 100 and a dimensionless radius :
1
= 10.
7. CASE STUDIES 307
Figure 7.70: The dimensionless pressure, p
1
. versus the dimensionless time, t,
for an inverse decline exponent : = 100 and a dimensionless radius :
1
= 20.
Figure 7.71: The dimensionless pressure, j
1
. versus the dimensionless time, t,
for an inverse decline exponent : = 100 and a dimensionless radius :
1
= 50.
308 7. CASE STUDIES
Figure 7.72: The dimensionless pressure, p
1
, versus the dimensionless time, t,
for an inverse decline exponent : = 100 and a dimensionless radius :
1
= 100.
Figure 7.73: The dimensionless pressure, j
1
, versus the dimensionless time, t,
for an inverse decline exponernt : = 100 and a dimensionless radius :
1
= 200.
7. CASE STUDIES 309
Figure 7.74: The dimensionless pressure, j
1
, versus the dimensionless time, t,
for an inverse decline exponent : = 100 and a dimensionless radius r
1
=500.
gations that can also include analyses and possible new methods for creating a
more realistic physical model.
310 7. CASE STUDIES
Figure 7.75: The dimensionless pressure, 1
1
, versus the dimensionless time, t
for an inverse decline exponent : = 1000 and a dimensionless radius r
1
=10.
The singularity case.
Chapter 8
DISCUSSION, CONCLUSIONS,
RECOMMENDATIONS
The purpose of this thesis has been to take into account the production rate
history of wells, which is usually of transient or depletion type, and to improve
the well decline analysis with new modelling and interpretation solutions. When
the recorded rate response is aected by the no-ow outer boundary or when the
reservoir is closed, rates are in the depletion decline. If, on the other hand, the
reservoir is innite, well rates are in the transient decline. This thesis provides
fractured-horizontal-well solutions to a transient rate-decline and analytically
derived pressure-time expressions to vertical wellbore conditions of a variable-
rate decline of Arps type.
8.1 DISCUSSION
8.1.1 Transient Rate Decline
In the transient-decline scope of this work, emphasis has been placed on well rate-
time responses. A well with fractures is coupled to an oil reservoir. Solutions
are full-time solutions and inner boundary conditions are of constant and vari-
able rate and/or pressure. For long times it is possible to create well-equivalent
radii, and fracture equivalent half-length can be obtained for a horizontal well
with fractures. The bringing-together of rate-time and pressure-time analyses
provides a total package to better characterise the behaviour of a horizontal well
with created fractures. The overall modelling work has been intended to con-
tribute to the transient rate decline by including various features of a fractured
horizontal well.
By introducing both pressure and rate full time responses, it was possible
to better characterise wells with fractures as presented in numerous case studies
from North Sea wells. Several case studies with fractured horizontal wells have
311
312 8. DISCUSSION, CONCLUSIONS, RECOMMENDATIONS
illustrated and validated the procedure and applicability of a fractured-horizontal
well model in an oil reservoir. In the history match scenario, due to a range
of input rate data being available, the model calculates well responses that are
similar to observed well data. Even with a limited amount of data being available,
the model predicts the cumulative rate target quite satisfactorily. The model is
also validated with data from a water-injection fractured-horizontal well. In the
case of a fractured-horizontal well producing only gas, the model is only partly
validated. The reason for this is that the model was primarily developed for
the only a liquid ow. Since the obtained matching results were satisfactory,
model features should be benecial in evaluating the fracture behaviour during
the wells production life.
This modelling work illustrates the use of a screening tool in optimising a
horizontal well with fractures located in an oil reservoir. This was achieved by
implementing new analytical solutions into the SLAB and BOX models, with the
aim of providing a useful diagnostic and prognostic tool for the oil industry. In
addition, although semi-analytical modelling requires some trial and error to ne-
tune the matching procedure, it oers a unique exibility in the diagnosis of wells
with fractures. With the special numerical features the implemented model could
be considered ecient in both accuracy and speed. The program was exible and
had many options to accommodate the various wellbore operational properties
and fracture properties. It presents improvements over several published semi-
analytical tools that have been completed and tested as such.
The semi-analytical and mono-phase model oers the following advantages:
both the pressure and rate model IBC are constant and variable; by keeping
the wellbore production rate constant for a certain production time, the model
calculates constant pressure wellbore conditions at the end of this time, and con-
tinue to produce under these conditions until the end; implemented IBC features
are ensured on the well plateau production data; the model includes fracture
features, such as fracture size (half-length, height) and fracture ow character
(uniform ux, innite and nite conductivity fractures); each fracture rate and
cumulative rate is calculated in both transient and depletion mode; the fracture
character is represented by a uniform ux, nite and innite conductivity.
Although a mono-phase model is more simple that any commercial reservoir
simulation model, it has been used to attain complementary and useful results.
These include individual-fracture cumulative-production and rates as well as
well-productivity-indexes.
Moreover, the use of an analytical model implies that there is no need for
static-geology model data. The reservoir properties are restricted to only one
zone, and thus input values such as zone height, porosity and permeability are
mean values or homogenised reservoir model values. As the fracture character is
more or less unknown screening analysis is important for a better understand-
ing of a complex reservoir-fracture-well system. The screening features render
8. DISCUSSION, CONCLUSIONS, RECOMMENDATIONS 313
this model superior to many others, especially commercial simulation models.
It should also be noted that since the aim of the screening analysis has been
to match target characteristics and provide a quick diagnosis to the fracture
behaviour, the program was self-validated.
A series of solutions corresponding to various conditions is given in a general
model that also includes various wellbore conditions. We have demonstrated
that it is possible to match well data with the options implemented in a module.
The implemented method was very fast and provided an excellent history match
on the basis of a well. The individual fracture rate and cumulative rate responses
represents additional valuable information that is available. Fine-tuning of the
variable rate and/or variable pressure would have improved the match of the well
data. Additionally, we showed that the method was ideally suited for solving
a multiple fractured horizontal well in an oil reservoir when available data is
limited and analysis cannot conducted with conventional reservoir simulation
tools. The most used inversion procedure known as "Laplace", dened by the
Stehfest algorithm, was found to not always be stable. New inversion techniques
(as listed in Appendix B) should be thus considered. The selected "Laplace"
inversion techniques should be tested and implemented in the future modelling
work. In all case studies the simulations were very fast. The model has several
advantageous features related to pressure-rate with time modelling. It includes
wellbore friction, chocking eects and the BOX model solutions that need to be
further improved. The gas model should include the non-Darcy ow terms and
should be tested with more well data. As a horizontal well penetrates only one
zone and accordingly becomes fractured, it should be extended to several zones
where each may have a selected number of fractures. The conductivity, angle and
position of each fracture additionally could be included. Nevertheless, from the
results of this study, a number of issues have been identied. First, the program
requires trials and errors to ne-tune a well with fracture responses. Second,
although variable IBC are implemented, only a limited number of time steps are
available in the resent releases. Any increase in the number of time-steps will
improve the matching data and only slightly reduce the processing time. The
time-step function should be extended to more wellbore options, and the step-
function could be improved in order for the constant rate to change in to constant
or variable pressure. The implemented IBC changes from a constant rate to
constant pressure only. The model should also include more zones and each zone
will be penetrated by a certain number of producing fractures. Fracture may
also penetrate several zones. Since the gas model is basic it should comprise
non-Darcy ow and should be fully developed. Moreover, Stehfest algorithm
should be extended to other robust inversion techniques (several of which listed
in Appendix B).
Finally, it could be concluded that the semi-analytical model has been able
to satisfactorily simulate a reservoir-fracture-well system for an oil-producing
314 8. DISCUSSION, CONCLUSIONS, RECOMMENDATIONS
and a water-injection well. With the current implemented and tested features,
advantages of the physical, numerical and programming model were able to
overcame its limited disadvantages. This was especially true when the individual
fracture features of the production such as rate and cumulative rate may be used
in analysing the well production decline with time. In addition, due to the model
versatility, the existing model features could be improved with more heterogenous
features (as layering and a fractured reservoir).
8.1.2 Depletion Rate Decline
The depletion decline presented in Fetkovich type curves is analytically solved
with the specied moving boundary. The model can be extended to also account
for the variation of the speed of the moving no-ow boundary. This approach
has potential in improving knowledge of well behaviour, assuming that pressure
responses can also be continuously measured. A speed should be related to a
driving force that causes a rate-time Arps stem.
Raghavan (1993) discussed conditions under which the decline curvature can
be derived empirically. The inclusion of variable skin was suggested as a means of
analytically deriving the rate-time curvature that denes the decline exponent,/.
Raghavan also mentioned that, as long as the drainage radius and skin remains
unchanged, it is not possible to analytically derive solutions for various rate-time
stems dened by the b exponent. The way to provide the analytical solutions
for rate-time curves dened by the decline exponent b is to introduce a varying
drainage volume that produces the Arps rate at the wellbore. In order to solve
the diusion equation, we specify the moving boundary as being no-ow and
moving outwards from a wellbore axis with the certain speed. Thus, solving
the diusion equation with the specied moving boundary and at the same time
imposing wellbore conditions of variable rate of Arps type is a novel approach
in determining pressure-time solutions. The model input values correspond to
rate-time data dened at the wellbore and specied moving boundary, whereas
the model output values are pressure-time data calculated at any point within
the vertical well drainage area. As soon as these data can be continuously mea-
sured and thus become available this physical model can be tested and validated
accordingly.
A model extension is possible for a specied moving boundary with an in-
wards motion i.e., moving towards the wellbore axis from a vertical well drainage
radius. Further, in order to reduce existing model limitations, new methods
should be considered.
This work should be regarded as complementary to the decline curve analy-
sis. For selected stems, or Arps rates, each dened with a chosen decline
exponent,/.the model generates pressure-time responses within a producing ver-
tical well drainage area. These pressure-time proles represent the solution to
the diusion equation with the specied moving boundary. Since it is a basic
8. DISCUSSION, CONCLUSIONS, RECOMMENDATIONS 315
model, it is limited to the choice of input parameters related to the no-ow mov-
ing boundary, the no-ow boundary speed, i and the coecient : (: =
1
b
). For
the right choice of i and :. it is possible to generate the positive pressure dier-
ence, 1. For certain values of i and :. singularity is avoided, we are thus able
to calculate the 1 expressions at any point within the vertical well drainage
radius. The fact that the model is dened with only particularly chosen para-
meters gives it limitations. The obtained pressures 1 could be normalised.
Further, the physical means of the speed of the no-ow moving boundary, i.
could be further investigated. Other model solutions could also be considered.
The depletion rate-time analysis introduces new analytical pressure-time so-
lutions to the wellbore variable-rate conditions of Arps type. As depletion Arps
rate-time curves are empirical, the analytically obtained solved pressure-time
solutions are innovative. In order to solve the diusion equation, the specied
moving boundary conditions are introduced. These specied moving boundary
condition are no-ow with an outward motion from the wellbore axis with a pre-
dened speed. As the drainage volume changes with no-ow moving boundary
the model is transient. Nevertheless, at the same time the pressure is solved with
the wellbore conditions of Arps rate decline and considered as depletion. De-
spite that the physical model is limited to the specied boundary conditions and
model parameters, this work has great potential and provides a basis for further
investigations that may also include analyses and the possible understanding of
the driving forces that dene the curvature of the rate-time production proles.
8.2 CONCLUSIONS
The overall study considers the transient rate decline and the depletion rate
decline. The transient study brings new solutions and interpretation techniques
that can help in the analysis of a fractured-horizontal well. The depletion study
provides pressure-time solutions for a vertical well producing under variable-rate
conditions (of selected Arps stems).
A general horizontal-fractured well model comprises a series of solutions cor-
responding to various wellbores with fractures conditions. It could be demon-
strated that it was possible to match well data with the options implemented in
a module. The implemented method was very fast and provided a good history
match on the basis of a well. The individual fracture rate, the cumulative rate
and the productivity index responses represented additional valuable information
that was available. A ne-tuning of the variable rate and/or variable pressure
would have improved the tting of the well data. Additionally, the method was
shown to be ideally suited for solving multiple fractured horizontal wells in an oil
reservoir when available data was limited and analysis could not be carried out
with conventional reservoir simulation tools. This thesis illustrates the use of
a screening tool for an improved modelling of oil production from horizontal or
316 8. DISCUSSION, CONCLUSIONS, RECOMMENDATIONS
near-horizontal wells with induced fractures. This was achieved by implement-
ing new analytical solutions into the SLAB and BOX models, thus providing a
useful diagnostic and prognostic tool for the oil industry.
The following is a summary of a fractured-horizontal well that produces from
a SLAB/BOX model:
1. A fast and robust algorithm is developed for calculating the transient
(SLAB model) and the basic depletion (BOX model) responses of a multiple
fractured horizontal well in an anistropic homogeneous reservoir.
2. The bringing-together of rate-time and pressure-time analyses in order
to obtain a total package for an improved characterisation of the behaviour of
horizontal oil wells with induced fractures.
3. The semi-analytical tool developed aided in optimising the well production
of a horizontal oil well with induced fractures.
The following is the summary of a vertical well that produces under wellbore
variable-rate conditions of Arps type:
1. The model provided new solutions to a diusion problem when wellbore
conditions were of variable rate decline and a specied moving boundary (no-ow
outer boundary that moves outwards from the wellbore axis).
2. The model generated pressure proles and comprised the transient infor-
mation of a changing drainage volume, and simultaneously included the depletion
rate decline as derived by Arps.
3. The physical meaning of the speed of a no-ow boundary moving outwards
from a wellbore axis should be further studied by means of a driving force that
creates a rate decline shape.
This work has also further improved the concept of depletion rate-time decline
curve analysis. The model solutions are unique and open for a new approach in
solving diusion equations with variable-rate inner boundary conditions and a
specied-moving outer boundary.
8.3 RECOMMENDATIONS
Both models have potential of being further developed, and a horizontal-fractured
well model can possibly be coupled to other software.
For a horizontal well with fractures the model recommendations are:
1. To be extended to handle any fracture orientation, determined by the
stress orientation of the reservoir, a fracture wall damage or fracture skin and
wellbore skin.
2. To include heterogeneity as layering and naturally fractured reservoir
complexity.
3. To be able to handle gas ow (including the non-Darcy ow)
4. To employ continuous monitored pressure and rate data of a fractured-
horizontal well with the purpose of carrying out fracture closing diagnosis.
8. DISCUSSION, CONCLUSIONS, RECOMMENDATIONS 317
5. To integrate the semi-analytical model into a numerical simulator for an
improved t of calculated and observed data (commercial software).
6. To couple a fractured-horizontal well model to the rate-testing and pressure-
time semi-analytical software (commercial software)
7. To couple screening features of a fractured-horizontal well model to state-
of-technology risking software (commercial software)
8. To improve fracture diagnosis by coupling microseismic modelling devices
to the fractured-horizontal well screening features.
9. To combine a fractured reservoir and a horizontal-well with a prognosis of
the fracture production (commercial software).
10. To verify the numerical model solutions with the semi-analytical solutions
(commercial software).
For a model with the dened variable rate wellbore conditions (of Arps rate
decline) and a specied moving boundary, recommendations are:
1. A continuous monitoring of pressure and rate data for a vertical producing-
well as well as calibration and validation of the existing model.
2. Apossible reservoir production diagnosis based on pressure data calculated
by using values observed at any point within a drainage volume.
3. The physical meaning of the speed of a moving boundary and its relation
to the drive mechanism.
4. The derivation of new modelling solutions with the specied moving
boundary.
318 8. DISCUSSION, CONCLUSIONS, RECOMMENDATIONS
Chapter 9
NOMENCLATURE
= constant
= drainage area of a well, ,t
2
(:
2
)

1
= dimensionless drainage area, .
1
= ,1
2
c
/ = Arps decline exponent
/
)
= fracture width, ft (m)
1 = constant
1 = liquid formation volume factor, rb/sb (rcm/scm)
1
0
= oil formation volume factor, rb/sb (rcm/scm)
1
ci
= initial oil formation volume factor, rb/sb (rcm/scm)
1
&
= water formation volume factor, rb/sb (rcm/scm)
1
j
= gas formation volume factor, rb/scf (rcm/scm)
c
)
= formation pore compressibility, 1/psia (1/Pa)
C
)1
= dimensionless fracture conductivity
c
j
= gas compressibility, 1/psia (l/Pa)
c
0
= oil compressibility, 1/psia (l/Pa)
c
t
= total compressibility, 1/psia (l/Pa)
c
t
= average total compressibility, 1/psia (1/Pa)
c
&
= water compressibility, 1/psia (1/Pa)
1
i
= initial decline rate, 1/unit of time
Q
j
= cumulative gas produced, MMscf (scm)
/ = reservoir net pay thickness, ft (m)
1
0
= modied Bessel function of rst kind of order zero
1
1
= modied Bessel function of rst kind of order one
1
c
= modied Bessel function of second kind of order zero
1
1
= modied Bessel function of second kind of order one
/ = average eective permeability, mD (:
2
)
/
o
= absolute permeability. mD (:
2
)
/
)
= fracture permeability, mD (:
2
)
/
j
= eective permeability to gas, mD (:
2
)
/
c
= eective permeability to oil, mD (:
2
)
319
320 9. NOMENCLATURE
/
vj
= relative permeability to gas
/
vc
= relative permeability to oil
/
&
= eective permeability to water, mD (:
2
)
1
C
= system characteristic length, ft (m)
1
1
= dimensionless eective wellbore length, 1
1
= 1
I
,2/
1
I
= eective horizontal well length in pay zone, ft (m)
Q
c
= cumulative oil produced, sb (scm)
Q
ci
. = original oil in place. sbbl (scm)
j
c
= base pressure, lower limit of integration, psia (Pa)
j = pressure. psia (Pa)
j
i
= initial reservoir pore pressure, psia (Pa)
:(j) = real gas pseudopressure, psia
2
/cp (Pa/s) m(p)=2
j
_
j
0
j
0
oj
j

:
j
&1
= "Laplace" space dimensionless wellbore pressure solution
j
&1
=dimensionless wellbore pressure
j
&)
= sandface owing pressure, psia (Pa)

c
= oil ow rate. sb or Mscf/unit of time (scm/s)

i
= initial ow rate. sb or Mscf/unit of time (scm/s)

o
= average well ow rate, STB or Mscf/unit of time (semis)

1o
= dimensionless decline ow rate function

1oi
= dimensionless decline ow rate integral function

1oio
= dimensionless decline ow rate integral derivative function

j
= gas ow rate. sb or Mscf/unit of time (scm/s)

t
= total well ow rate (all uids) sb or Mscf/unit of time (scm/s)

&
= water ow rate. sb or Mscf/unit of time (scm/s)

o1
= dimensionless wellbore ow rate

o1
= Laplace space dimensionless wellbore ow rate
Q
1o
= dimensionless cumulative production
Q
j1o
= dimensionless decline cumulative production function
:
1
= dimensionless radius, :
1
= :,1
c
= :,:
&
:
c
= eective reservoir drainage radius, ft (m)
:
c1
= dimensionless eective drainage radius .
:
&
=.wellbore radius, ft (m)
:
&o
=apparent or eective wellbore radius, ft (m)
:
&1
= dimensionless wellbore radius, :
&1
= :
&
,/
: = "Laplace" space parameter
o
j
= reservoir gas saturation, fraction of pore volume
o
c
= reservoir oil saturation, fraction of pore volume
o
cv
= residual oil saturation, % of pore volume
o
&
= water saturation, % of pore volume
o
&i
= initial water saturation, fraction or % of pore volume
o
&iv
= irreducible water saturation, % of pore volume
9. NOMENCLATURE 321
t = time, hours, days, months, years (s)
t
0
= time value parameter of integration
t
1
= dimensionless time
t
jcc
= time to reach pesudo-steady-state or boundary-dominated ow,
units of time (s)
j =pressure dierential, psia (Pa) j = j
i
j
&)
c = reservoir average eective porosity, fraction of bulk volume
`
t
= average total system mobility function, 1,cj (1,1c :)
j =reservoir uid viscosity, cp (1c :)
j
c
=oil viscosity, cp (1c :)
j
j
=gas viscosity, cp (1c :)
j
&
= water viscosity, cp (1c :)
9.1 Functions
cos = cosine function
cosh = hyperbolic cosine
e = exponential function
exp = exponential function
In = natural logarithm
sin = sine function
sinh = hyperbolic sine
9.2 SI Metric Conversion Factors
mD x 9.869 233 E - 04 = m2
D x 9.869 233 E - 07 = m2
psi x 6.894 757 E + 00 = kPa
psi x 6.894 759 E - 02 = bar
in x 2.54* E - 02 = m
in2 x 6.4516* E - 04 = m2
ft x 3.048* E - 01 = m
ft2 x 9.290 304* E - 02 = m2
ft3 x 2.831 685* E - 02 = m3
bbl x 1.589 873 E - 01 = m3
cp x 1.0* E - 03 = Pa s
*Conversion factors are exact
322 9. NOMENCLATURE
Chapter 10
REFERENCES
Aanonsen, S. (1985). Application of Pseudotime To Estimate Average Reservoir
Pressure. SPE Annual Technical Conference and Exhibition. Las Vegas, Nevada.
Abdat, T. (2000). Transient Behaviour of Horizontal Well - Selected pa-
pers. Texas A&M University, Department of Petroleum Engineering, Reservoir
Simulation Group.
Abramowitz, M.,& Stegun, I.A. 1972. Handbook of Mathematical Functions.
Dover Publications.
Agarwal, R. G., Gardner, D. C., Kleinsteiber, S.W., and Fussell, D.D. (1999).
Analyzing Well Production Data Using Combined-Type-Curve and Decline-Curve
Analysis Concepts. SPE Reservoir Evaluation & Engineering 2(5), pp. 478-486.
Agarwal, R. G., Carter, R. D. and C.B. Pollock. (1979). Evaluation and
Performance Prediction of Low-Permeability Gas Wells Stimulated by Massive
Hydraulic Fracturing. Paper SPE 6838-PA , SPE JPT 31(3), pp. 362-372.
Ahmed, T. (2006). Reservoir Engineering Handbook. Elsevier Science &
Technology Books, Edition Number: 3.
Al-Hussainy, R., J. Ramey, H.J. and P.B. Crawford. (1965). The Flow of
Real Gases through Porous Media. Paper SPE 1243-A, Fall Meeting of the
Society of Petroleum Engineers of AIME. Denver, Colorado, USA, Oct., 3-6.
Al-Hussainy, R. and H. J. Ramey. (1966). Application of Real Gas Flow
Theory to Well Testing and Deliverability Forecasting. SPE JPT 18(5), pp.
637-642.
Al-Khalifah, A.-J. A., Aziz, K. and R.N. Horne. (1987). A New Approach
to Multiphase Well Test Analysis. Paper SPE 16743-MS, SPE Annual Technical
Conference and Exhibition. Dallas, Texas, USA, Sept., 27-30.
Al-Kobaisi, M., Ozkan, E. and H. Kazemi (2006). AHybrid Numerical/Analytical
Model of a Finite-Conductivity Vertical Fracture Intercepted by a Horizontal
Well. Paper SPE 92040-PA, SPE Reservoir Evaluation & Engineering 9(4),
Aug., pp. 345-355.
Anderson, D.M., Stotts, G.W.J. , Mattar, L., Ilk, D. and T.A. Blasingame.
(2006). Production Data Analysis-Challenges, Pitfalls, Diagnostics. Paper SPE
323
324 10. REFERENCES
102048-MS, SPE Annual Technical Conference and Exhibition, San Antonio,
Texas, USA, Sept., 24-27.
Ansah, J., Knowles, R.S., and T.A. Blasingame. (2000). A Semi-Analytic
(p/z) Rate-Time Relation for the Analysis and Prediction of Gas Well Perfor-
mance. Paper SPE 66280-PA, SPE Reservoir Evaluation & Engineering, Dec.,
pp. 525-533.
Araya, A. and E. Ozkan (2002). An Account of Decline-Type-Curve Analy-
sis of Vertical, Fractured, and Horizontal Well Production Data. SPE Annual
Technical Conference and Exhibition. San Antonio, Texas, USA.
Arnold, R., and R. Anderson. (1908). Preliminary Report on Coalinga Oil
District. U. S. Geol. Survey Bull. 357, 79.
Arps, J.J. (1945). Analysis of Decline Curves. Trans. AIME 160, 228-247.
Arps, J.J. (1956). Estimation of Primary Oil Reserves. Trans., AIME 207,
182-191.
Azar-Nejad, F., Tortike, W. S., and S.M. Farouq Ali. (1996). 3-D Ana-
lytical Solution to the Diusivity Equation for Finite Sources with Application
to Horizontal Wells. Paper SPE 35512-MS, European 3-D Reservoir Modelling
Conference. Stavanger, Norway, April 16-17.
Aziz, K.; and A. Settari. (1979). Petroleum Reservoir Simulation.
Ayan, C. and W. J. Lee (1988). Multiphase Pressure Buildup Analysis: Field
Examples. SPE California Regional Meeting. Long Beach, California, USA.
Babu, D.K. and A.S. Odeh. (1988). Productivity of a Horizontal Well.
Paper SPE 18298, SPE 63rd Annual Technical Conference and Exhibition of
SPE, Houston.
Babu, D.K., Odeh, A.S. (1989). Productivity of a horizontal well, SPE Reser-
voir Engineering: 417-421.
Barker, B. J. and H. J. Ramey. (1978). Transient Flow to Finite Conductiv-
ity Vertical Fractures. SPE Annual Fall Technical Conference and Exhibition.
Houston, Texas, USA.
Bellman, R., Kalaba, R. E., and J.A. Lockett. (1966). Numerical Inversion
of the Laplace Transform:Applications to Biology, Economics, Engineering, and
Physics. American Elsevier, New York.
Berard, M., (2007). Internal Schlumberger Report. Schlumberger Research
Moscow, SMR.
Besson, J. (1990). Performance of Slanted and Horizontal Wells on an Anisotropic
Medium. European Petroleum Conference. The Hague, Netherlands.
Blasingame, T.A., Johnston, J.L., and W.J. Lee (1989). Type Curve Analysis
Using the Pressure Integral Method. Paper SPE 18799, SPE California Regional
Meeting, Bakerseld, CA, USA, April 5-7.
Blasingame, T. A., McCray, T. L. and W.J. Lee (1991). Decline Curve
Analysis for Variable Pressure Drop/Variable Flowrate Systems. Paper SPE
21513-MS, SPE Gas Technology Symposium. Houston, Texas, USA, Jan., 22-
10. REFERENCES 325
24.
Blasingame, T.A. and B.D. Poe. (1993). Semi-Analytic Solutions for a Well
with a Single Finite-Conductivity Vertical Fracture. Paper SPE 26424, 68th
Annual SPE Technical Conference and Exhibition, Dallas, USA, TX, Oct., 3-6.
Brekke, K. and L. G. Thompson (1996). Horizontal Well Productivity and
Risk Assessment. SPE Annual Technical Conference and Exhibition. Denver,
Colorado.
Brill, J.P. and D. H. Begs (1979). Two phase ow in pipes. Textbook, Univ.
of Tulsa.
Borisov, Y.P. (1964). Oil Production Using Horizontal and Multiple Devia-
tion Wells. Nedra, Moscow.
Bourdarot, G. (1998). Well Testing: Interpretation Methods. Editions Tech-
nip, Paris, France.
Buchholz, H. (1969). The conuent hypergeometric Function. Springer
Tracts in natural Phylosophy, Vol. 15., Springer Verlag.
Busswell, G., Banerjee, R., Thambynayagam, R.K.M. and J. Spath. (2006).
Generalized Analytical Solution for Reservoir Problems With Multiple Wells and
Boundary Conditions. Paper SPE 99288-MS, SPE Intelligent Energy Conference
and Exhibition, Amsterdam, Netherland, April 11-13.
Busswell, G.S. (2006). Generalized Analytical Solutions for Single Layer
Problems with Multiple Wells and Boundary Conditions. Paper SPE 99288,
SPS Intelligent Energy Conference, Amsterdam, 8-11 April.
Be, A., Skjveland, S.M., and C.H. Whitson. (1989). Two Phase Pressure
Analysis. SPE Formation Evaluation, 4(4), Dec., pp. 604-610.
Callard, J. G. and P. A. Schenewerk (1995). Reservoir Performance History
Matching Using Rate/Cumulative Type-Curves. SPE Annual Technical Confer-
ence and Exhibition. Dallas, Texas.
Camacho-V. R. G. (1987). Well Performance under Solution-Gas Drive
Reservoirs. PhD Dissertation, U. Tulsa, Tulsa, Oklahoma, USA.
Camacho-V., R. G. and R. Raghavan, (1989). Boundary-Dominated Flow in
Solutions-Gas-Drive Reservoirs. SPE RE, Nov., 4(4), pp. 503-512.
Camacho-V., R. G. and R. Raghavan (1989). Performance of Wells in Solution-
Gas-Drive Reservoirs SPE FE 4(4), pp. 611-620.
Camacho-V., R. G. and R. Raghavan (1991). Some Theoretical Results Use-
ful in Analyzing Well Performance Under Solution Gas Drive (Supplement to
SPE 16580).
Camacho-Velazquez, R. and A. Galindo-Nava, (1996). Optimum Position for
Wells Producing at Constant Wellbore Pressure." SPE JPT 1(2), pp.155-168.
Carslaw, H.S., and J.C. Jaeger (1947). Conduction of Heat in Solids. Oxford
at the Clarendon Press.
Carslaw, H.S., J:C. Jaeger. (1959). Conduction of Heat in Solids. 2nd
edition, Clarendon Press, Oxford, New York.
326 10. REFERENCES
Carter, R. D. (1985). Type Curves for Finite Radial and Linear Gas-Flow
Systems: Constant-Terminal-Pressure Case. APE JPT 25(5), pp. 719-728.
Caudle, B.H. (1967). Fundamentals of Reservoir Engineering. Lecture Notes
for a video tape Course. SPE AIME.
Chang, M.-M., Tomutsa, L.,and M.K. Tham (1989). Predicting Horizon-
tal/Slanted Well Production by Mathematical Modeling. Paper SPE 18854-
MS, SPE Production Operations Symposium. Oklahoma City, Oklahoma, USA,
March 13-14.
Chen, C-C. and R. Raghavan. (1996). An Approach To Handle Discontinu-
ities by the Stehfest Algorithm. SPE JPT 1(4): 363-368.
Chen, Z., (1991). A Detecting Technique for Production Rate Decline-Curve
Analysis with Residual Plots. Paper SPE 22313; Dallas, Texas, USA.
Cheng, Y. (2003). Pressure Transient Testing and Productivity Analysis for
Horizontal Wells. PhD Dissertation, Texas A&M University.
Cheng, Y., Lee, W. J. and D.A. McVay. (2003). A Deconvolution Technique
Using Fast-Fourier Transforms, Paper SPE 84471-MS, SPE Annual Technical
Conference and Exhibition, Denver, Colorado, USA, Oct. 5-8,
Chen, C.-C., E. Ozkan, et al. (1991). A Study of Fractured Wells in Bounded
Reservoirs. Paper SPE 22717-MS, SPE Annual Technical Conference and Exhi-
bition. Dallas, Texas, USA, Oct., 6-9.
Chen, H.-Y. and L. W. Teufel (2000). A New Rate-Time Type Curve for
Analysis of Tight-Gas Linear and Radial Flows. SPE Annual Technical Confer-
ence and Exhibition. Dallas, Texas, USA.
Cheng, Y., Lee, W. J., and D.A. McVay. (2003). A Deconvolution Tech-
nique Using Fast-Fourier Transforms. SPE Annual Technical Conference and
Exhibition. Denver, Colorado, USA.
Chien, T. and S. H. Caudle (1994). Improved Analytical Solution for Real
Gas Flow in Porous Media. Permian Basin Oil and Gas Recovery Conference.
Midland, Texas, USA.
Chien, T. (1993). Modeling of Real Gas Flow Behavior in Porous. Media.
Ph.D. Dissertation. U. of Texas, Austin, USA.
Chien, T. and S. H. Caudle (1994). Improved Analytical Solution for Real
Gas Flow in Porous Media. Permian Basin Oil and Gas Recovery Conference.
Midland, Texas, Society of Petroleum Engineers.
Christensen, S.A., Ebbe-Dalgaard, T.E., Rosendal, A. Christensen, J.W.,
Robinson, G. and A.M Zellou. (2006). Seismically Driven Reservoir Character-
ization Using an Innovative Integrated Approach: Syd Arne Field. SPE Annual
Technical Conference and Exhibition. San Antonio, Texas, USA.
Chu, W.-C., Reynolds, A.C. and R. Raghavan. (1986). Pressure Transient
Analysis of Two-Phase Flow Problems. Paper SPE 10223-PA, SPE Formation
Evaluation 1(2), April, pp. 151-164.
Cinco-Ley, H., Samaniego, F.V., and A.N. Dominguez (1978). Transient
10. REFERENCES 327
Pressure Behavior for a Well with a Finite-Conductivity Vertical Fracture. SPE
JPT, 18(4), Aug., pp. 253-264.
Cinco-Ley, H. and F. V. Samaniego (1981). "Transient Pressure Analysis for
Fractured Wells." SPE Journal of Petroleum Technology 33(9): 1749-1766.
Cinco-Ley, H. (1982). Evaluation of hydraulic fracturing by transient pres-
sure analysis methods. Paper SPE 10043, SPE International Petroleum Exhibi-
tion and Technical Symposium, Beijing, China, Mar. 18-26.
Cinco-Ley, H., Ramey H.J., Samaniego, F. V., and F. Rodriguez. (1987).
Behavior of Wells with Low-Conductivity Vertical Fractures. Paper SPE 16776-
MS, SPE Annual Technical Conference and Exhibition. Dallas, Texas, USA,
Sept. 27-30.
Cinco-Ley, H. and H-Z. Meng (1988). Pressure Transient Analysis of Wells
With Finite Conductivity Vertical Fractures in Double Porosity Reservoirs. SPE
Annual Technical Conference and Exhibition. Houston, Texas, USA.
Clegg, M.W. (1967). Some Approximate Solutions of Radial Flow Problems
Associated with Production at Constant Well Pressure. SPE JPT, 7(1), pp.
31-42, March.
Clenshaw, C. W. (1955). A Note on the Summation of Chebyshev Series.
Math. Comp. 9, 118 - 120. MR 17:194e.
Clonts, M. D. and H. J. Ramey. (1986). Pressure Transient Analysis for
Wells with Horizontal Drainholes. SPE California Regional Meeting. Oakland,
California, USA.
Closmann, P. J. and N. W. Ratli (1967). Calculation of Transient Oil
Production in a Radial Composite Reservoir. SPE JPT 7(4), pp. 355 - 358.
Cox, D. O. (1979). The Solution Of Problems Associated With Constant
Well Pressure. Paper SPE 8386-MS, SPE Annual Technical Conference and
Exhibition, Las Vegas, Nevada, USA, Sept., 23-26.
Cox, D. O., Kuuskraa, V. A. and J.T. Hansen. (1996). Advanced Type Curve
Analysis for Low Permeability Gas Reservoirs. Paper SPE 35595-MS, SPE Gas
Technology Symposium. Calgary, Alberta, Canada, Apr. 28. -May 1.
Crank, J. (1984). Free and Moving Boundary Problems. Clarendon Press,
Oxford.
Crump, K. S. (1976). Numerical Inversion of Laplace Transforms Using a
Fourier Series Approximation. Journal of the ACM (JACM), v.23 n.1, pp. 89-
96, Jan.
Cryer, C. W. (1978). The interrelation between moving boundary problems
and free boundary problems, Moving Boundary Problems. Proc. Sympos., and
Workshop.Gatlinburg, Tenn., Academic Press, New York, pp. 91-107.
Cutler, W.W.Jr. (1924). Estimation of Underground Oil reserves by Well
Production Curves. U.S. Bur. Mines Bull. 228.
Cvetkovic, B. (1992). Fundamental Equations and Techniques Used in Decline
Curve Analysis. University of Trondheim, January.
328 10. REFERENCES
Cvetkovic, B. (1992). Modelling and Solution Methods for Layered Reser-
voirs. University of Trondheim, Norway, June.
Cvetkovic, B., and J.S. Gudmundsson. (1993). Analytical Models for Rate
Decline in Oil and Gas Reservoirs. University of Trondheim, Norway, December.
Cvetkovic, B. and G. Halvorsen. (1995) Horizontal Well Induced Fractures
Production Optimization. The Pre-study work for Norwegian Research Council,
NFR. December.
Cvetkovic, B., Halvorsen, G. and J. Sagen. (1996). HOWIF-OIL SLAB
Model. The 1996 Final Report and Half-Year Report for BP, CONOCO, PHILLIPS
(Condential Report).
Cvetkovic, B., Halvorsen, G and J. Sagen. (1997). HOWIF-OIL SLAB Im-
proved Model. The 1997 Final and Half-Year Report Report for BP, CONOCO,
PHILLIPS (Condential Report).
Cvetkovic, B., Halvorsen, G. and J. Sagen. (1998). HOWIF-OIL SLAB
Advanced Model. The Final and the Half-Year Report for BP, CONOCO,
PHILLIPS (Condential Report).
Cvetkovic, B., Halvorsen, G. and J. Sagen. (1999). HOWIF-OIL BOXModel.
The Final and the Half-Year Report for BP, CONOCO, PHILLIPS (Condential
Annual Report).
Cvetkovic, B., Halvorsen, G. and J. Sagen. (2000). A Multiple Fractured-
Horizontal Well Case Study. Paper SPE 65503-MS, SPE/CIM International
Conference on Horizontal Well Technology. Calgary, Alberta, Canada, Nov.,
6-8.
Cvetkovic, B.,and G. Halvorsen. (2001). Multi-fractured horizontal well
study with HOWIF from a Vallhal Field. A Selected Software Day Seminar in
BP-AMOCO, Stavanger, Norway, July 26.
Cvetkovic, B., Halvorsen, G., Sagen, J., and E.N. Rigatos (2001). Modelling
the Productivity of a Multifractured-Horizontal Well. Paper SPE 71076, SPE
Rocky Mountain Petroleum Technology Conference, Keystone, Colorado, USA,
May 21-23.
Cvetkovic, B., Halvorsen, G., Sagen,J., and R. Banerjee. (2001). Eective
Wellbore Radius for a Multi-fractured Horizontal Well Simulation. GeoQuest
Schlumberger, FORUM, London, UK, March 6-9.
Cvetkovic, B. (2007). A Vertical/Horizontal Well with Laterals Simulation.
Reservoir Symposium Schlumberger (EUREKA), Moscow, Russia, June.
Cvetkovic, B. (2009). Semi-Analytical Modelling of a Horizontal Well with
Fractures in an Oil Reservoir - Screening Approach with Risk Analysis. SPE
Applied Tecnology Workshop, Penang, Malaysia, March 8-11.
Cvetkovic, B. (2009). Eective Wellbore Parameters and a Multi-fractured
Horizontal Well Productivity. Paper SPE 120826, SPE Production and Opera-
tions Symposium, Oklahoma City, Oklahoma, USA, April, 4-8.
Da Prat, G ., Cinco-Ley, H. and H.J. Ramey. (1980). Decline Curve Analy-
10. REFERENCES 329
sis Using Type-Curves for Two-Porosity Systems. Paper SPE 9292, SPE 55th
Annual Conference, SPE of AIME, Dallas, Texas, USA, Sept. 21-24.
Da Prat, G.: (1981) Well Test Analysis for Naturally Fractured Reservoirs.
Ph.D. Dissertation, Stanford University, USA.
Da Prat, G., Cinco-Ley, H. and H. Ramey. (1981). Decline Curve Analysis
Using Type Curves for Two-Porosity Systems. Paper SPE 9292-PA, SPE JPT
21(3), pp. 354-362.
Dake, L.P. (1978). Fundamentals of Reservoir Engineering, Elsevier Science
Publishers B.V., Amsterdam.
Dake, L.P. (2001). The Practice of Reservoir Engineering. Elsevier, Amster-
dam.
Darcy, H. P. G. (1856). Les Fountaines Publiques de la Ville de Dijon. Victon
Dalmont, Paris, France.
Daviau, F., G. Mouronval, et al. (1988). Pressure Analysis for Horizontal
Wells. SPE Formation Evaluation 3(4), pp. 716-724.
Davies, B. and B. Martin. (1979). Numerical Inversion of the Laplace Trans-
form: a Survey and Comparison of Methods. J. Computational Physics, 33, pp.
1-32.
de Carvalho, R. S. and A. J. Rosa (1988). Transient Pressure Behavior for
Horizontal Wells in Naturally Fractured Reservoir. Paper SPE 18302-MS, SPE
Annual Technical Conference and Exhibition. Houston, Texas, USA, Oct., 2-5.
Demarchos, A.S., Porcu, M.M. and M. Economides (2006). Transverse Mul-
tifractured Horizontal Wells: A Recipe for Success. Paper SPE 102262-MS, SPE
Annual Technical Conference and Exhibition, San Antonio, Texas, USA, Sept.,
24-27.
Ding, Y. (1999). Using boundary integral methods to couple a semi-analytical
reservoir ow model and a wellbore ow model. Paper SPE 51898, SPE Reservoir
Simulation Symposium, Houston, USA, Feb. 14-17.
Ding, Y., P. Lemonnier, P., Estebenet, T. and J-F. Magras. (1998). A Con-
trol Volume Method for Flow Simulation in Well Vicinity for Arbitrary Well
Congurations. Paper SPE 48854-MS. SPE International Oil and Gas Confer-
ence and Exhibition in China. Beijing, China. Nov., 2-6.
Ding, Y., Renard, G. and L. Weill. (1998). Representation of Wells in Nu-
merical Reservoir Simulation. Paper SPE 29123-PA, SPE Reservoir Evaluation
& Engineering 1(1), Feb., pp. 18-23.
Dou, H, Chen, C. and Y. Chang. (2009). Analysis and Comparison of Decline
Models: A Field Case Study for the Intercampo Oil Field, Venezuela. Paper SPE
106440-PA, SPE Reservoir Evaluation & Engineering 12(1), Feb., pp. 68-78.
Doublet, L.E. and T.A. Blasingame. (1995a). Evaluation of Injection Well
Performance Using Decline Type Curves. Paper SPE 35205, SPE Permian Basin
Oil and Gas Recovery Conference, Midland, Texas, USA, March, 27-29.
Doublet, L.E. and T.A. Blasingame. (1995b). Decline Curve Analysis Using
330 10. REFERENCES
Type Curves: Water Inux/Waterood Cases. Paper SPE 30774, SPE Annual
Technical Conference and Exhibition, Dallas, USA, Oct., 22-25.
Dranchuk, P.M. and D. Quon. (1967). Analysis of the Darcy Continuity
Equation. Producers Monthly, Oct. 25-28.
Duda, J. R. and K. Aminian (1989). Type Curves for Predicting Production
Performance From Horizontal Wells in Low-Permeability Gas Reservoirs. Paper
SPE 18993-MS, Low Permeability Reservoirs Symposium. Denver, Colorado,
USA, March, 6-8.
Duong, A.N. (1989). A New Approach for Decline-Curve Analysis. Paper
SPE 18859, presented at the 1989 SPE Production Operations Symposium, Ok-
lahoma City, Oklahoma, USA, March 13-14.
Earlougher, R.C.,Jr., and H.J. Ramey.(1968). The Use of Interpolation to
Obtain Shape Factors for Pressure Buildup Calculations. SPE JPT, May, pp.
449-450.
Earlougher, R.C. Jr. (1977). Advances in Well Test Analysis. Monograph
Series 5., Society of Petroleum Engineers, Dallas, USA.
Economides, M., Deimbachor, F. X., Brand, C. W., and Z. E. Heinemann.
(1991). Comprehensive Simulation of Horizontal-Well Performance. Paper SPE
20717-PA, SPE Formation Evaluation 6(4), Dec., pp. 418-426.
Economides, M.J., Hill, D.A., Ehlig-Economides, C. (1994). Petroleum Pro-
duction Systems. Prentice Hall, Englewood Clis, NJ, USA.
Economides, M. J., C. W. Brand, et al. (1996). Well Congurations in
Anisotropic Reservoirs. SPE Formation Evaluation 11(4), pp. 257-262.
Economides, M.J., Valko, P.P., and X. Wang. (2001). Recent Advances in
Production Engineering., JCPT, 01-10-01.
Ehlig-Economides, C.A. 1979. Well Analysis for Wells Produced at a Con-
stant Pressure. PhD dissertation, June, Stanford U., Stanford, CA. USA.
Ehlig-Economides, C.A., and H.J. Ramey. (1981). Transient Rate Decline
Analysis for Wells Produced at Constant Pressure. Paper SPE 8387-PA, SPE
JPT Volume 21, Number 1, Feb., pp. 98-104.
Ehlig-Economides, C. A. and J. Joseph (1987). "A New Test for Determina-
tion of Individual Layer Properties in a Multilayered Reservoir." SPE Formation
Evaluation 2(3), pp. 261-283.
Ehlig-Economides, C. A. (1993). Model Diagnosis for Layered Reservoirs.
SPE Formation Evaluation 8(3), pp. 215-224.
El-Banbi, A.H. (1998). Analysis of Tight Gas Well Performance. Ph.D.
dissertation, May, Texas A&M University, College Station, TX, USA.
Fair, P. S. (1992). Investigation of Diusivity Equation With General Pressure-
Dependent Rock and Fluid Properties as Applied to Well Testing.
Feitosa, G. (1993). Well test analysis for heterogeneous reservoirs. PhD
thesis, University of Tulsa.
Fetkovich, M.,(1973). Decline Curve Analysis Using Type Curves.Paper.
10. REFERENCES 331
SPE 4629.
Fetkovich, M. J. (1980). Enlarged Type Curves Supporting Paper SPE 4629.
Fetkovich, M. J. (1980). Decline Curve Analysis Using Type Curves. SPE
JPT 32(6), pp. 1065-1077.
Fetkovich, M.J. (1980). Decline Curve Analysis Using Type Curves. Paper
SPE-4629-PA, SPE JPT 32 (6), pp. 1065-1077.
Fetkovich, M.J., Vienot, M.E., Johnson, R.D. and B.A. Bowman. (1985).
Case Study of a Low-Permeability Volatile Oil Field Using Individual-Well Ad-
vanced Decline Curve Analysis. Paper SPE 14237-MS, SPE Annual Technical
Conference and Exhibition, Las Vegas, Nevada, USA, Sept., 22-26.
Fetkovich, M.D., Guerrero, E.T., Fetkovich, MJ.. and L.K. Thomas (1986).
Oil and Gas Relative Permeabilities Determined From Rate-Time Performances
Data. Paper SPE 15431, SPE Annual Technical Conference and Exhibition, New
Orleans, Oct. 5-8.
Fetkovich, M.J., Vienot, M.E. Bradley, M.D. and U.G. Kiesow. (1987).
Decline Curve Analysis Using Type Curves: Case Histories. Paper SPE 13169
PA, SPE FE 2 (4), pp.637-656.
Fetkovich, M.J., Fetkovich., E.J., and M.D. Fetkovich (1996). Useful Con-
cepts for Decline Curve Forecasting, Reserve Estimation, and Analysis. Paper
SPE-28628-PA, SPE RE 11 (1), pp. 13-22.
Fetkovich, M.D., Petrosky, G.E. , Hughesman, C.B., and R.P. Sawatzky
(2006). Rate-Time Flow Behavior of Heavy Oil From Horizontal and Multi-
lateral Wells. Paper SPE 100065-MS, SPE/DOE Symposium on Improved Oil
Recovery, Tulsa, Oklahoma, USA, April, 22-26.
Fokker, P.A., Brouwer, G.K., Verga, F. and D. Ferrero. (2005). ASemi_Analytical
Model for Productivity Testing of Multiple Wells. Paper SPE 94153-MS, SPE
Europec/EAGE Annual Conference, Madrid, Spain, June 13-16.
Fokker, P. A. and F. Verga (2008). ASemianalytic Model for the Productivity
Testing of Multiple Wells. SPE Reservoir Evaluation & Engineering 11(3), pp.
466-477.
Fraim, M. L., W. J. Lee, and Gatens, J.M.(1986). Advanced Decline Curve
Analysis Using Normalized-Time and Type Curves for Vertically Fractured Wells.
SPE Annual Technical Conference and Exhibition. New Orleans, Louisiana,
U.S.A., Oct., 5-8.
Fraim, M. L., and R. A. Wattenbarger (1987). Gas Reservoir Decline-Curve
Analysis Using Type Curves with Real Gas Pseudopressure and Normalized
Time. Paper SPE FE, Vol.2, No. 6, Dec., pp. 671-682.
Gentry, R.W and A. W. McCray. (1978). The Eect of Reservoir and Fluid
Properties on Production Decline Curves. Paper SPE-18562-PA, SPE JPT 30
(9), pp. 1327-1341.
Gilchrist, J. P., Busswell,G., Banerjee, R., Spath, J., and M. Thambynayagam.
(2007). Semi-analytical Solution for Multiple Layer Reservoir Problems with
332 10. REFERENCES
Multiple Vertical, Horizontal, Deviated and Fractured Wells. Paper SPE 11718-
MS, International Petroleum Technology Conference, Dubai, U.A.E. Dec., 4-6.
Giger, F. (1983). Reduction du Nombre de Puits par LUtilisation de Forage
a Horizontaux. Revue de IFP, vol. 38, No. 3, May-June.
Giger, F.M. (1985). Horizontal Wells Production Techniques in Heteroge-
neous Reservoirs. Paper SPE 13710, SPE Middle East Oil Technical Conference
and Exhibition, Bahrain, March 11-14.
Golan, M. and C.H. Whitson. (1986). Well Performance. IHRDC.
Goode, P. A. and Thambynayagam, M. (1987). Pressure Drawdown and
Buildup Analysis of Horizontal Wells in Anisotropic Media. SPE Formation
Evaluation 2(4): 683-697.
Goode, P. A. and F. J. Kuchuk (1991). Inow Performance of Horizontal
Wells. SPE Reservoir Engineering 6(3): 319-323. Aug..
Gringarten, A.C., and H,J, Ramey. (1972). A Comparison of dierent solu-
tiond to the radial ow problem. Paper SPE 3817-MS.
Gringarten, A.C., and H.J Ramey (1973). The Use of Source and Greens
Functions in Solving Unsteady-Flow Problems in Reservoirs. SPE Journal, pp.
285-296, Oct..
Gringarten, A.C.. Ramey, H.J., and R. Raghavan (1974). Unsteady-State
Pressure Distributions Created by a Well with a Single Innite-Conductivity
Vertical Fracture. Paper SPE-4051-PA, SPE JPT 14 (4): pp.347-360, Oct. 8
-11.
Gringarten, A. C. (2006). From Straight Lines to Deconvolution: The Evolu-
tion of the State of the Art in Well Test Analysis. Paper SPE 102079-MS, SPE
Annual Technical Conference and Exhibition. San Antonio, Texas, USA, Sept.
24-27.
Gringarten, A. C. (2008). From Straight Lines to Deconvolution: The Evo-
lution of the State of the Art in Well Test Analysis. SPE Reservoir Evaluation
& Engineering 11(1): pp. 41-62.
Gringarten, A. C. and H. J. Ramey (1973). The Use of Source and Greens
Functions in Solving Unsteady-Flow Problems in Reservoirs. 13(5): 285-296.
Guo, G. and R. D. Evans (1993). Inow Performance and Production Fore-
casting of Horizontal Wells With Multiple Hydraulic Fractures in Low-Permeability
Gas Reservoirs. SPE Gas Technology Symposium, Calgary, Alberta, Canada,
June 28-30.
Guo, G., R. D. Evans, et al. (1994). Pressure-Transient Behavior for a
Horizontal Well Intersecting Multiple Random Discrete Fractures. Paper SPE
28390-MS, SPE Annual Technical Conference and Exhibition, New Orleans,
Louisiana,USA. Sept. 25-28.
Guo, B. and D. S. Schechter (1997). Use of a Simple Mathematical Model
for Estimating Formation Damage in Wells Intersecting Long Fractures. Paper
SPE 38178-MS.
10. REFERENCES 333
SPE European Formation Damage Conference. The Hague, Netherlands,
June 2-3.
Gurley, J. (1963). A Productivity and Economic Projection Method- Ohio
Clinton Sand Gas Wells. SPE JPT 15(11), pp1183-1188.
Hareland G. and P. R. (Rampersad. (1995). Design and Performance of
Fractured Horizontal Wells. Published in the Norwegian Petroleum Directorate
- Near Wellbore Flow - Summary Reports, Stavanger, Norway.
Hegre, T.M., and , L. Larsen. (1994). Productivity of Multifractured Hor-
izontal Wells, Paper SPE 28845-MS, Source European Petroleum Conference ,
London, UK, 25-27 Oct..
Hegre, T. M. and Larsen, L (1995). Modelling Horizontal Fractured Well.
Published in the Norwegian Petroleum Directorate - Near Wellbore Flow - Sum-
mary Reports, Stavanger, Norway.
Horne, R. N. and K. O. Temeng, (1995). Relative Productivities and Pressure
Transient Modeling of Horizontal Wells with Multiple Fractures. Paper SPE
29891-MS, Middle East Oil Show, Bahrain, March 11-14.
Horne, R.N. (1995). Modern Well Test Analysis A Computer-Aided Ap-
proach. Palo Alto, CA, USA, second edition.
Hurst, W. (1934). Unsteady ow of uids in oil reservoirs. Physics, 5, 20-30.
Ikoku C. U.(1984). Natural Gas Production Engineering. John Wiley &
Sons, N. Y.
Ilk, D., P. P. Valko, and T.A. Blasingame (2006). Deconvolution of Variable-
Rate Reservoir Performance Data Using B-Splines. Paper SPE 95571-PA, SPE
Reservoir Evaluation & Engineering 9(5): pp. 582-595.
Issaka, M.B., and Ambastha.A.K. (1998). Decline Curve Analysis for Com-
posite Reservoirs. JCPT, June, Volume 37, No. 6.
Jacob, C.E., and S.W., Lohman. (1952). Nonsteady Flow to a Well of
Constant Drawdown in an Extensive Aquifer. Trans., AGU, 33, Aug.,559-569.
Jaeger, J.C. (1941). Conduction of Heat in Regions Bounded by Planes and
Cylinders. American Mathematical Society. Bulletin., 47, pp. 734-741.
Jaeger, J.C., and M.E. Clarke. (1942). A Short Table of Mathematical
Formula. Royal Society of Edinburgh. Proceedings., A, 61, pp. 229-230.
Jasti, J.K., Penmatcha, V.R., and D.K. Babu. (1997). Use of Analytical
Solutions in Improvement of Simulator Accuracy. SPE Annual Technical Con-
ference and Exhibition. San Antonio, Texas, USA.
Javandel, I. and P. A. Witherspoon (1969). A Method of Analyzing Transient
Fluid Flow in Multilayered Aquifers. Water Resour. Res., Vol. 5 (4), pp. 856-
869.
Jelmert, T. A. and H. Selseng (1997). Pressure Transient Behavior of Stress-
Sensitive Reservoirs. Latin American and Caribbean Petroleum Engineering
Conference. Rio de Janeiro, Brazil.
Jongkittinarukorn, K. and D. Tiab (1998). Development of the Boundary El-
334 10. REFERENCES
ement Method for A Horizontal Well in Multilayer Reservoir. SPE Rocky Moun-
tain Regional/Low-Permeability Reservoirs Symposium. Denver, Colorado, So-
ciety of Petroleum, Engineers.
Johnson, R.H., and .A.L. Bollens. (1927). The Loss-ratio Method of Extrap-
olating Oil Well Decline Curves. Trans. A.I.M.E. 27, 771.
Jordan, C. L., Jackson, R. and C. Smith. (2008). Making the Most of
Conventional Decline Analysis. CIPC/SPE Gas Technology Symposium 2008
Joint Conference. Calgary, Alberta, Canada, Society of Petroleum Engineers.
Jordan, C. L., Jackson, R. A., and C. Smith. (2008). Simplifying Gas Produc-
tion Modeling. Paper SPE 114954-MS, CIPC/SPE Gas Technology Symposium,
Joint Conference, Calgary, Alberta, Canada, 16-19 June.
Joshi, S.D. (1986). Augmentation of Well Productivity Using Slant and Hor-
izontal Wells. Paper SPE 15375, SPE JPT Volume 40, Number 6, pp. 729-739.
Joshi, S. D. (1987). A Review of Horizontal Well and Drainhole Technology.
SPE Annual Technical Conference and Exhibition. Dallas, Texas, USA.
Joshi, S.D. (1991). Horizontal Well Technology. Penn Well Publishing Com-
pany.
Juan-Camus, I. (1977). Determination delas Propiedades de un Yacimiento
Mediante Pruebas de Gasto en un Pozo a Presion Constante. M.S. Report,
University of Mexico.
Kacir, T.P. (1990). A Streamline Simulator for Depletion Drive Gas Reser-
voirs. M.Sc. Thesis, U. of Texas, Austin.
Karcher, B. J., Giger, F. M., and J. Combe. (1986). Some Practical Formulas
to Predict Horizontal Well Behavior. Paper SPE 15430 presented at the SPE
61st Annual Technical Conference and Exhibition, NewOrleans, Louisiana, USA,
October 5-8.
Karimi-Fard, M., Aziz, K., Durlofsky, L.J., and B. Gong. (2009). Modeling
Well Performance in Naturally Fractured Reservoirs. SPE Applied Technology
Workshop (ATW), Maximizing the Value of Horizontal and Multilateral Wells:
New Challenges, Technologies, and Approaches, 8-11 March, Penang, Malaysia.
Kato, E. T. and K. V. Serra (1991). Inuence of Multiphase Flow in Well Test
Analysis of Nonowing Wells. SPE Annual Technical Conference and Exhibition.
Dallas, Texas, 1991.
Kikani, J. and R. N. Horne (1993). Modeling Pressure-Transient Behavior
of Sectionally Homogeneous Reservoirs by the Boundary-Element Method. SPE
FE 8(2), pp. 145-152.
Kleppe, J. and H. M. Cekirge (1980). Constant Pressure Well Tests in Innite
Radial Reservoirs, Paper SPE 9029-MS.
Koh, L. S. and D. Tiab (1993). 3D Boundary-Element Model for Predicting
Performance of Horizontal Wells. Paper SPE 26101-MS, SPE Western Regional
Meeting, 26-28 May 1993, Anchorage, Alaska, USA.
Kozeny J. (1933). Theorie und Berechnung der Brunnen. Wasserkraft und
10. REFERENCES 335
Wasserwirtschaft 28, 88-92, 101-105, 113-116.
Kuchuk, F. and W. E. Brigham (1979). Transient Flow in Elliptical Systems.
SPE JPT 19(6), pp.401-410.
Kuchuk, F. and Ayestaran, L. (1985). Analysis of Simultaneously Measured
Pressure and Sandface Flow Rate in Transient Well Testing (includes associated
papers 13937 and 14693 ). SPE JPT, 37(2), pp.323-334.
Kuchuk, F.J. (1987). New Methods for Estimating Parameters of Low Per-
meability Reservoirs. Paper SPE 16394, SPE/DOE Low Permeability Reservoirs
Symposium, Denver, CO, USA, May 18-19.
Kuchuk, F.J., Goode, P.A., Wilkinson, D.J., and R.K.M. Thambynayagam.
(1988). Pressure Transient Behavior of Horizontal Wells With and Without Gas
Cap or Aquifer. Paper SPE 17413, SPE California Regional Meeting, Long
Beach, CA, USA, Mar. 23-25.
Kuchuk, F. J. and T. M. Habusky (1994). Pressure Behavior of Horizontal
Wells with Multiple Fractures. University of Tulsa Society of Petroleum Engi-
neering Symposium. Tulsa, Oklahoma, USA.
Kuchuk, F.J., and A.A. Kader. (1994). Pressure Behavior Of Horizontal
Wells in
Heterogeneous Reservoirs. JCPT, Paper No. HWC94-25.
Kuchuk, F. J., Hollaender, F. , Gok, I.M., and M. Onur. (2005). Decline
Curves from Deconvolution of Pressure and Flow-Rate Measurements for Pro-
duction Optimization and Prediction. SPE Annual Technical Conference and
Exhibition, Dallas, Texas, USA.
Kuppe, F., and A. Settari. (1996). Practical Method For Determining the
Productivity of Multi-Fractured Horizontal Wells. JCPT, ATM 96-22.
Lam, G. and B.P. Clapeyron. (1981). Mmoire sur la Solidication par
Refroidissement dun Globe Liquide. Ann. Chimie Physique , 47, pp. 250-256.
Larsen, L. and T.M. Hegre. (1991). Pressure-Transient Behavior of Hor-
izontal Wells With Finite-Conductivity Vertical Fractures. Paper SPE 22076,
International Arctic Technology Conference in Anchorage, Alaska, USA, May
29-31.
Larsen, L. and T.M. Hegre (1994). Pressure Transient Analysis of Multi-
fractured Horizontal Wells. Paper SPE 28389, SPE Annual Technical Conference
and Exhibition, New Orleans, U.S.A., 25-28 Sept..
Larsen, L. (1997). Productivities of Fractured and Non-Fractured Deviated
Wells in Commingled Layered Reservoirs. Paper SPE 38912, SPE Annual Tech-
nical Conference and Exhibition, San Antonio, Texas, USA, 5-8 Oct..
Larsen, L. (1998). Pressure-Transient Behavior of Multibranched Wells in
Layered Reservoirs. SPE Annual Technical Conference and Exhibition. New
Orleans, Louisiana, USA.
Larsen, L. (1998). Productivities of Fractured and Nonfractured Deviated
Wells in Commingled Layered Reservoirs. SPE JPT 3(2): 191-199.
336 10. REFERENCES
Larsen,L. (2000). Pressure-Transient Behavior of Multibranched Wells in
Layered Reservoirs. Paper SPE 60911, Feb., SPE RE, Volume 3, Number 1, pp.
68-73.
Lefkovits, H. C. and C. S. Matthews (1958). Application of Decline Curves
to Gravity-Drainage Reservoirs in the Stripper Stage.
Lefkovits, H. C., Hazebroek, P., and C.S. Matthews.(1961). "A Study of
the Behavior of Bounded Reservoirs Composed of Stratied Layers." 1(1), pp.
43-58.
Levitan, M. M. (2005). Practical Application of Pressure/Rate Deconvolu-
tion to Analysis of Real Well Tests. SPE Reservoir Evaluation & Engineering
8(2), pp. 113-121.
Levitan, M. M. (2006). Deconvolution of Multi-Well Test Data. SPE Annual
Technical Conference and Exhibition. San Antonio, Texas, USA.
Levitan, M. M., Crawford, G. E. and A. Hardwick. (2006). Practical Consid-
erations for Pressure-Rate Deconvolution of Well-Test Data. SPE Journal 11(1):
pp. 35-47.
Liu, J.S. (2009). Reservoir Simulation of Horizontal and Multilateral Wells.
SPE Applied Technology Workshop (ATW), Maximizing the Value of Horizontal
and Multilateral Wells: New Challenges, Technologies, and Approaches, 8-11
March, Penang, Malaysia.
Locke, C. D. and W. K. Sawyer (1975). Constant Pressure Injection Test
in a Fractured Reservoir-History Match Using Numerical Simulation and Type
Curve Analysis. Fall Meeting of the SPE of AIME. Dallas, Texas,USA.
Lolon, E., R., Archer, A., Ilk, D., and T.A. Blasingame. (2008). New Semi-
Analytical Solutions for Multilayer Reservoirs. CIPC/SPE Gas Technology Sym-
posium Joint Conference. Calgary, Alberta, Canada.
Loucks, T. L. (1961). Pressure Distribution in Systems With Continuously
Varying Permeability. Society of Petroleum Engineers of AIME.
Luke, Y.L. (1969) The Special Functions and Their Approximations, vol. 1-2,
Academic Press. New York.
Luke, Y.L. (1971). Miniaturized Tables of Bessel Functions, Math. of Com-
putation, Vol. 25, pp. 323-330.
Luke, Y. L. (1978). Error Estimation in Numerical Inversion of "Laplace"
Transforms Using Pade Approximation. J. Franklin Inst., 305, pp. 259-273.
Marhaendrajana, T. and T. A. Blasingame (2001). Decline Curve Analysis
Using Type Curves - Evaluation of Well Performance Behavior in a Multiwell
Reservoir System. SPE Annual Technical Conference and Exhibition. New
Orleans, Louisiana, USA.
Martin, J. C. (1959). Simplied Equations of Flow in Gas Drive Reservoirs
and the Theoretical Foundation of Multiphase Pressure Buildup Analyses. Pe-
troleum Transactions, AIME, Volume 216, 1959, pp. 321-323.
Mattar, L., and D.M. Anderson. (2003). A Systematic and Comprehensive
10. REFERENCES 337
Methodology for Advanced Analysis of Production Data. Paper SPE 84472-MS,
SPE Annual Technical Conference and Exhibition, 5-8 Oct., Denver, Colorado,
USA.
Mavor, M. J. and H. Cinco-Ley (1979). Transient Pressure Behavior of Nat-
urally Fractured Reservoirs. SPE California Regional Meeting. Ventura, Cal-
ifornia, American Institute of Mining, Metallurgical, and Petroleum Engineers
Inc..
Medeiros, F., Ozkan, E., and H. Kazemi. (2006). A Semianalytical, Pressure-
Transient Model for Horizontal and Multilateral Wells in Composite, Layered,
and Compartmentalized Reservoirs, Paper SPE 102834-MS, SPE Annual Tech-
nical Conference and Exhibition, 24-27 Sept., San Antonio, Texas, USA.
Medeiros, F., Kurtoglu, B., Ozkan, E., and H. Kazemi (2007). Pressure-
Transient Performances of Hydraulically Fractured Horizontal Wells in Locally
and Globally Naturally Fractured Formation. Paper Number 11781-MS Inter-
national Petroleum Technology Conference, 4-6 Dec., Dubai, U.A.E.
Medeiros, F., Kurtoglu, B., Ozkan, E. and H. Kazemi. (2007). Analysis
of Production Data from Hydraulically Fractured Horizontal Wells in Tight,
Heterogeneous Formations. SPE Annual Technical Conference and Exhibition.
Anaheim, California, U.S.A.
Merkulov, L.P. (1958). Flowrate of a Horizontal Well. Neft. Khoz., vol. 6,
51.
Michelevichius, D., and A.B. Zolotukhin. (2002). Evaluating Productivity
of a Horizontal Well. Paper SPE 79000-MS. SPE/PS-CIM/CHOA International
Thermal Operations and Heavy Oil Symposium and International Horizontal
Well Technology Conference, 4-7 Nov., Calgary, Alberta, Canada.
Moore, T. V., Schilthuis, R. J. and W. Hurst. (1933). The Determination of
Permeability from Field Data. Proc. API Bull., 211, pp. 1-8.
Morse, R. A., and W.D. von Gonten, (1972). Title Productivity of Vertically
Fractured Wells Prior to Stabilized Flow. Paper SPE 3631-PA, July, SPE JPT,
Volume 24, Number 7, pp. 807-811.
Mukherjee, H., and M.J. Economides. (1988). A parametric Comparison
of Horizontal and Vertical Well Performance. Paper SPE 18303. SPE Annual
Technical Conference and Exhibition, Houston, TX, Oct. 2-5.
Muskat, M. and M.W. Meres. (1936). The Flow of Heterogeneous Fluids
through Porous Media. Physics, 1936. 7 (Sept.): pp. 346-363.
Muskat, M.(1937). The Flow of Homogeneous Fluids through Porous Media.
McGraw-Hill Book Co. Inc., New York City 55.
Muskat, M., Wycko, R. D., Botset, H. G, and M.W. Meres. (1937). Flow
of Gas-liquid Mixtures through Sands. Published in Petroleum Transactions,
AIME, Volume 123, pp. 69-96.
Muskat, M. (1945). The Production Histories of Oil Producing Gas-Drive
Reservoirs. Journal of Applied Physics. March, pp. 147-159.
338 10. REFERENCES
Muskat, M., (1949). Physical Principles of Oil Production. McGraw-Hill.
Norris, S. O., J. L. Hunt, and M.Y. Soliman. (1991). Predicting Horizontal
Well Performance: A Review of Current Technology. SPE Western Regional
Meeting. Long Beach, California, USA.
Norris, M. R., L. Bergsvik, L., Teesdale, C., and B.A. Berntsen (2001). Multi-
ple Proppant Fracturing of Horizontal Wellbores in a Chalk Formation: Evolving
the Process in the Valhall Field. SPE Drilling & Completion 16(1). Pp. 48-59.
Oberhettinger, F. (1957). Tabellen zur Fourier Transformation, Springler
Verlag.
Oberhettinger, F., and L. Badii. (1973). Tables of Laplace Transforms,
Spronger Verlag.
Odeh, A. S. and D. K. Babu (1990). Transient Flow Behavior of Horizontal
Wells: Pressure Drawdown and Buildup Analysis. SPE FE, 5(1), pp. 7-15.
Oguztoreli, M. and D. W. Wong. (1998). VERTEX: A New Modeling
Method to Direct Field Development. Paper SPE 39806 presented at the Per-
mian Basin Oil and Gas Recovery Conference, Midland, Texas, USA, Mar. 22-25.
Okuszko, K.E., Gault B.W., and L. Mattar. (2008). Production Decline
Performance of CBM Wells, JCPT 2007-078.
Olarewaju, J. S. and W. J. Lee (1987). An Analytical Model for Composite
Reservoirs Produced at Either Constant Bottomhole Pressure or Constant Rate.
SPE Annual Technical Conference and Exhibition. Dallas, Texas, USA.
Olarewaju, J. S. and J. W. Lee (1989). A Comprehensive Application of
a Composite Reservoir Model to Pressure-Transient Analysis. SPE Reservoir
Engineering, 4(3): 325-331.
Olarewaju, J. S., J. W. Lee, and D. E. Lancaster (1991). Type- and Decline-
Curve Analysis With Composite Models. Paper SPE 17055-PA, SPE Formation
Evaluation 6(1), March, Pp. 79-85.
Olarewaju, J. S. and W. J. Lee (1991). Rate Behavior of Composite Dual-
Porosity Reservoirs. SPE Production Operations Symposium. Oklahoma City,
Oklahoma, USA.
Oliver, D. S. (1990). The Averaging Process in Permeability Estimation From
Well-Test Data. SPE FE, 5(3), pp. 319-324.
Onur, M., K. V. Serra, et al. (1991). Analysis of Pressure-Buildup Data
From a Well in a Multiwell System. SPE FE, 6(1), pp.101-110.
Ouyang, L-B., Arbabi, S., and K. Aziz (1998). "General Wellbore Flow
Model for Horizontal, Vertical, and Slanted Well Completions." SPE JPT 3(2),
pp.124-133.
Ozkan, E. (1988). Performance of Horizontal Wells. PhD Dissertation. The
University of Tulsa, Tulsa, USA.
Ozkan, E., T. Yildiz, and F. J. Kuchuk. (1998). Transient Pressure Behavior
of Dual-Lateral Wells. SPE JPT 3(2), pp. 181-190.
Ozkan, E. and R. Raghavan. (2000). A Computationally Ecient, Transient-
10. REFERENCES 339
Pressure Solution for Inclined Wells. SPE Reservoir Evaluation & Engineering
3(5), pp. 414-425.
Padilla, S.R., and V.R. Camacho. (2004). Reservoir Performance Under
Solution Gas - Drive and Gravity Drainage. Paper SPE 92186 presented at the
International Conference in Mexico, Puebla, 8-9 Nov..
Palacio, J.C. and T.A. Blasingame. (1993). Decline-Curve Analysis Using
Type Curves -Analysis of Gas Well Production Data. Paper SPE 25909 presented
at the Rocky Mountain Regional / Low Permeability Reservoirs Symposium,
Denver, CO, April 12-14.
Papatzacos, P. (1987). Exact Solutions for Innite-Conductivity Wells. SPE
RE, 2(2), pp. 217-226.
Pechera, R. and J. F. Stanislav. (1977). Boundary Element Techniques in
Petroleum Reservoir Simulation. SPE JPT, Volume 17, Issues 3-4, May, pp.
353-366.
Penner S.S. and S. Sherman. (1947). Heat Flow Through Composite Cylin-
ders. J Chem Phys.;15, pp. 569 -. 574.
Perrine, R. L. (1955). Well Productivity Increase from Drain Holes as Mea-
sured by Model Studies.
Piessens, R. (1969). IEEE Trans. Automatic Control AC-14. pp. 299-301.
Piessens, R. (1975, 1976). "Laplace" transform inversion J. Comp. Appl.
Maths., 1, 115; 2, 225.
Poe, B.D. (2005) Method and Apparatus for Eective Well and reservoir
Evaluation without the Need for Well Pressure History. US Patent 6,842,700.
Poon, D.C. (1991). Decline Curves for Predicting Production Performance
from Horizontal Well. JCPT 91-01-06.
Postone, S.W., and B.D. Poe. (2008). Analysis of production Decline Curves,
Society of Petroleum Engineers, Richardson, TX, USA.
Prats, M. (1961). Eect of Vertical Fractures on Reservoir Behavior - In-
compressible Fluid Case. Paper SPE 1575-G, June, Volume 1, Number 2, pp.
105-118.
Prats, M., Hazebroek, P., and W.R. Strickler. (1962). Eect of Vertical
Fractures on Reservoir Behavior-Compressible-Fluid Case. Paper SPE 98-PA,
June, Volume 2, Number 2, pp. 87 - 94.
Prats, M. (1990). Letter correspondence with Caudle, B.H.
Prijambodo, R., Raghavan, R, and , A.C. Reynolds. (1985). Well Test
Analysis for Wells Producing Layered Reservoirs With Crossow. Paper SPE
JPT 25(3): 380-396.
Raghavan, R. (1976). Some Practical Considerations in the Analysis of Pres-
sure Data. SPE JPT 28(10), pp. 1256-1268.
Raghavan, R. (1976). "Well Test Analysis: Wells Producing by Solution Gas
Drive." 16(4): 196-208.
Raghavan,R.(1993). Well Test Analysis. Monograph. Prentice Hall Petro-
340 10. REFERENCES
leum Engineering Series.
Raghavan, R. and S. D. Joshi (1993). Productivity of Multiple Drainholes or
Fractured Horizontal Wells. SPE FE 8(1): 11-16.
Raghavan,R. and E. Ozkan. (1994). A Method for Computing Unsteady
Flows in Porous Media. Pitman Research Notes in Mathematics Series, 318:118.
Raghavan, R., Chen, C.C., and B. Agarwal (1994). An Analysis of Horizon-
tal Wells Intercepted by Multiple Fractures. SPE Permian Basin Oil and Gas
Recovery Conference, Midland, Texas, March 16-18.
Raghavan, R. and E. Ozkan (1995). A Method For Computing Unsteady
Flows in Porous Media. Chapman & Hall / CRC.
Raghavan, R. S., Chen, C.-C. and B. Agarwal. (1997). An Analysis of
Horizontal Wells Intercepted by Multiple Fractures. SPE JPT 2(3), pp. 235-
245.
Ramsay Jr., H. J. and E. T. Guerrero (1969). The Ability of Rate-Time
Decline Curves to Predict Production Rates. SPE JPT 21(2), pp. 139-141.
Renard, G.I. and J.M. Dupuy. (1990). Inuence of Formation Damage on
the Flow Eciency of Horizontal Wells. Paper SPE 19414, presented at the
Formation Damage Control Symposium, Lafayette, Louisiana, Feb. 22-23.
Riley, M. F., Brigham, W. E., and R.N. Horne (1991). Analytic Solutions
for Elliptical Finite-Conductivity Fractures. SPE Annual Technical Conference
and Exhibition. Dallas, Texas, USA.
Roberts, B. E., van Engen, H., and C.P.J.W., van Kruysdijk. (1991). Produc-
tivity of Multiply Fractured Horizontal Wells in Tight Gas Reservoirs. Oshore
Europe. Aberdeen, United Kingdom.
Rodriguez, F. and H. Cinco-Ley (1993). ANewModel for Production Decline.
SPE Production Operations Symposium. Oklahoma City, Oklahoma, USA.
Rodriguez-Roman, J. and R. Camacho-Velazquez (2002). Decline Curve
Analysis Considering Non Laminar Flow in Two Porosity Systems. SPE Interna-
tional Petroleum Conference and Exhibition in Mexico. Villahermosa, Mexico.
Rosa, A. J. and R. de S. Carvalho (1989). A Mathematical Model for Pressure
Evaluation in an lnnite-Conductivity Horizontal Well. SPE FE 4(4), pp. 559-
566.
Roumboutsos, A. and G. Stewart (1988). A Direct Deconvolution or Con-
volution Algorithm for Well Test Analysis. SPE Annual Technical Conference
and Exhibition. Houston, Texas, 1988 Copyright 1988 Society of Petroleum
Engineers.
Russell, D. G. and M. Prats (1962). Performance of Layered Reservoirs with
CrossowSingle-Compressible-Fluid Case. SPE JPT 2(1): 53 - 67.
Russell, D. G. and M. Prats (1962). "The Practical Aspects of Interlayer
Crossow." SPE JPT 14(6): 589-594.
Russell, D. G. and N. E. Truitt (1964). Transient Pressure Behavior in Ver-
tically Fractured Reservoirs. SPE JPT 16(10): 1159-1170.
10. REFERENCES 341
Sabet, M.A. (1991). Well test analysis: Gulf Publishing Co., Houston.
Samaniego V.F. (1974). An Investigation of Transient Flow of Reservoir
Fluids Considering Pressure Dependent Rock and Fluid Properties. PhD disser-
tation, Stanford University, Stanford, Ca., USA.
Samaniego-V, F., Brigham, W. E., and F. G. Miller. (1976). Performance-
Prediction Procedure for Transient Flow of Fluids Through Pressure-Sensitive
Formations. Paper SPE 6051 presented at the SPE-AIME 51st Annual Fall
Technical Conference & Exhibition, New Orleans, Oct., 3-6.
Samaniego V., F., Brigham, W. E., and F.G. Miller. (1977). An Investigation
of Transient Flow of Reservoir Fluids Considering Pressure-Dependent Rock and
Fluid Properties. SPE JPT 17(2): 141-150.
Samaniego, F.V., and H. Cinco-Ley. (1980). Production Rate Decline in
Pressure-Sensitive Reservoirs. JCPT 80-03-03
Sato, K. and R. N. Horne. (1993). Perturbation Boundary Element Method
for Heterogeneous Reservoirs: Part 1 - Steady-State Flow Problems. Paper SPE
FE, 8(4), pp. 306-314.
Sato, K. and R. N. Horne. (1993). Perturbation Boundary Element Method
for Heterogeneous Reservoirs: Part 2 - Transient-Flow Problems. Paper SPE
FE, 8(4), pp. 315-322.
Schapery, R.A. (1962). Approximate methods of transform inversion for vis-
coelastic stress analysis. Proceeding of the 4th U.S. National Congress of Applied
Mechanics, Vol. 2, ASME, pp. 1075-1085.
Schroeter, T., Hollaender, v., F., and A.C. Gringarten. (2002). Analysis
of Well Test Data From Permanent Downhole Gauges by Deconvolution. SPE
Annual Technical Conference and Exhibition. San Antonio, Texas, Copyright
2002, Society of Petroleum Engineers Inc.
Schulte, W.M. (1986). Production from a Fractured Well With Well Inow
Limited to Part of the Fracture Height. Paper SPE 12882-PA, Journal SPE PE,
Volume 1, Number 5, Sept., pp. 333-343.
Scott, J. O. (1963). The Eect of Vertical Fractures on Transient Pressure
Behavior of Wells. SPE JPT 15(12): 1365-1369.
Sheng-Tai, L., and J. R. Brockenbrough. (1986). A New Approximate Ana-
lytic Solution for Finite-Conductivity Vertical Fractures. Paper SPE 12013-PA,
Journal SPE FE, Volume 1. Number 1, Feb..
Shih, M.Y., and T.A. Blasingame (1995). Decline Curve Analysis Using Type
Curves: Horizontal Wells. Paper SPE29572, presented at the Joint Rocky Moun-
tain Regional/Low Permeability Reservoirs Symposium, 20-22 March, Denver,
CO, USA.
Silverberg, M. (1970). Electron Lett. 6, pp 105-106.
Slichter, C.S. (1898). Theoretical investigation of the motion of ground wa-
ters. In U.S. Geol. Surv. Annu. Rep. Part 2. U.S. Gov. Print. Oce,
Washington, DC, pp. 295-384.
342 10. REFERENCES
Soliman, M. Y, Hunt, J. L., and EI Rabaa, W. 1988. On Fracturing Horizon-
tal Wells. Paper SPE 18542 presented at the SPE Eastern Regional Conference
and Exhibition, Charleston, West Virginia.
Soliman, M. Y. (1990). Interpretation of Pressure Behavior of Fractured,
Deviated, and Horizontal Wells. SPE Latin America Petroleum Engineering
Conference. Rio de Janeiro, Brazil, 1990 Copyright 1990, Society of Petroleum
Engineers Inc.
Soliman, M. Y., Hunt, J. L. and W.. El Rabaa.(1990). Fracturing Aspects
of Horizontal Wells. Paper SPE JPT, Aug., pp. 966-973.
Soliman, M.Y. (1998). Stimulation and Reservoir Engineering Aspects of
Horizontal Wells. Solution Through Learning, Halliburton edition.
Spath, J.B., Ozkan, E., and R. Raghavan. (1994). An Ecient Algorithm
for Computation of Well Responses in Commingled Reservoirs. SPE Formation
Evaluation, June.
Spivey, J. P., Gatens, J. M., Semmelbeck, M.E., and S.A. Lee. (1992). In-
tegral Type Curves for Advanced Decline Curve Analysis. SPE Mid-Continent
Gas Symposium. Amarillo, Texas, USA.
Stehfest, H. 1970. Numerical Inversion of Laplace Transforms (Algorithm
368 with correction). Communications of the ACM 13 (1): 47-49.
Stewart, P. R. (1970). Low-Permeability Gas Well Performance At Constant
Pressure. SPE JPT 22(9): 1149-1156.
Tariq, S M.,and H. J. Ramey. (1978). Drawdown Behavior of a Well with
Storage and Skin Eect Communicating with Layers of dierent Radii and Other
Characteristics, Paper SPE 7453-MS, Annual Fall Technical Conference and Ex-
hibition, 1-3 October. Houston, Texas.
Tarzia, D.A. (1988). A bibliography on moving-free boundary problems for
the heat-diusion equation. Prog. Naz. "Equazioni di evoluzione e applicazioni
sico-matematiche" , Firenze.
Thambynayagam, R.K.M. (2006). Diusion: A Compendium of Analytical
Solutions. TBP.
Theis, C. V. (1935). The relation between lowering of the piezometric sur-
face and rate and duration of discharge of a well using groundwater storage.
Transactions of the American Geophysical Union, 16, 519-524.
Thompson, L.G., Marique,J.L., and T.A. Jelmert. (1991). Ecient algo-
rithms for computing the bounded reservoir horizontal well pressure response,
paper SPE 21827 presented at the SPE Rocky Mountain Regional Meeting and
Low-Permeability Reservoir Symposium, Denver, 15-17 April.
Towler, B.F., and S. Bansal. (1993). Hyperbolic Decline Curve Analysis
using Linear Regression. Journal of Petroleum Science and Engineering 8, 257-
268.
Tranter, C.J. (1951). Integral Transforms in Mathematical Physics. John
Wiley & Sons, Inc., New York.
10. REFERENCES 343
Tsarevich, K.A., and I.F. Kuranov. (1956). Calculation of the Flow Rates
for the Center Well in a Circular Reservoir Under Elastic Conditions. Problems
of Reservoir Hydrodynamics, Part I, Leningrad, 9-34.
Turki, L. (1986). Decline Curve Analysis in Composite Reservoirs. PhD
Stanford University, Stanford, California, December.
Turki, L., Demski, J.A., and A.S. Grader. (1989). Decline Curve Analysis
in Composite Reservoirs. Paper Number 19316-MS , SPE Eastern Regional
Meeting, 24-27 October 1989, Morgantown, West Virginia.
Umnuayponwiwat, S., Ozkan, E. and R. Raghavan. (2000). Pressure Tran-
sient Behavior and Inow Performance of Multiple Wells in Closed Systems. SPE
Annual Technical Conference and Exhibition. Dallas, Texas.
Uraiet, A.A. (1979). Transient Pressure Behavior in Cylindrical Reservoir
Produced by a Well at Constant Bottomhole Pressure. PhD dissertation, U. of
Tulsa, Tulsa, OK.
Uraiet, A.A., and R. Raghavan. (1980). Unsteady Flow to a Well Producing
at a Constant Pressure. Paper SPE 8768-PA, JPT Volume 32, Number 10,
October 1980, Pages 1803-1812.
Valk, P. and M.J. Economides. (1995). Hydraulic Fracture Mechanics.
Wiley, NY and Chichester.
Valko, P. and M. J. Economides (1996). Performance of a Longitudinally
Fractured Horizontal Well. SPE Journal 1(1): 11-19.
Valko, P. P., Doublet, L.E., and T.A. Blasingame. (2000). Development and
Application of the Multiwell Productivity Index (MPI). Paper SPE 51793-PA,
Journal SPE Journal, Volume 5, Number 1, Pages 21-31.
van Everdingen, A.F., and W. Hurst. (1949).The Application of the Laplace
Transformation to Flow Problems in Reservoirs. Trans., AIME, 186 Dec., 305-
324.
van Kruijsdijk, C.P.J.W.(1988). Semianalytical Modelling of Pressure Tran-
sients in Fractured Reservoirs. Paper SPE18169, SPE 63rd Annual Technical
Conference and Exhibition of SPE, Houston, October 2-5.
van Kruijsdijk, C.P.J.W. and G.M. Dullaert. (1989). A Boundary Solution
to the Transient Pressure Respose of Multiply Fractured Horizontal Wells. 2nd
European Conference on Mathematics of Oil Recovery, Cambridge.
von Schroeter,T., Hollaender, F., and A.C. Gringarten. (2004). Deconvolu-
tion of Well-Test Data as a Nonlinear Total Least-Squares Problem. Paper SPE
77688-PA, Journal SPE Volume 9, Number 4, December, Pages 375-390.
Vicente, R., Sarica, C., and T. Ertekin. (2000). A Numerical Model Coupling
Reservoir and Horizontal Well Flow Dynamics: Transient Behavior of Single-
Phase Liquid and Gas Flow. Paper SPE 65508-MS, SPE/CIM International
Conference on Horizontal Well Technology, 6-8 November 2000, Calgary, Alberta,
Canada.
Villegas, M.E., Wattenbarger, R.A., Valko, P., and M.J. Economides. (1996).
344 10. REFERENCES
Performance of Longitudinally Fractured Horizontal Wells in High-Permeability
Anisotropic Formations. Paper SPE 36453-MS, SPE Annual Technical Confer-
ence and Exhibition, 6-9 October, Denver, Colorado.
Villegas, M.E. (1997). Performance of Fractured Vertical and Horizontal
Wells. PhD Dissertation. Texas A&M University.
Zhang, W-X, Lifu, C. and A.S. Grader. (1993). Rate Decline of an Arbitrarily
Positioned Well in Composite and Irregular Reservoirs. Paper SPE 26913-MS,
SPE Eastern Regional Meeting, 2-4 November 1993, Pittsburgh, Pennsylvania.
Zheng, S.-Y. and W. Fei (2008). Application of Deconvolution and Decline-
Curve Analysis Methods for Transient Pressure Analysis. Europec/EAGE Con-
ference and Exhibition. Rome, Italy, Society of Petroleum Engineers.
Wan, J. and K. Aziz (1999). Multiple Hydraulic Fractures in Horizontal
Wells. SPE Western Regional Meeting. Anchorage, Alaska, Society of Petroleum
Engineers.
Wattenbarger, R. A. and H. J. Ramey Jr. (1969). Well Test Interpretation
of Vertically Fractured Gas Wells. SPE JPT, 21(5): 625-632.
Wei,Y., and M.J. Economides (2005). Transverse Hydraulic Fractures from a
Horizontal Well. Paper SPE 94671-MS, SPE Annual Technical Conference and
Exhibition, 9-12 October, Dallas, Texas.
Weller, W. T. (1966). Reservoir Performance During Two-Phase Flow. SPE
JPT, 18(2): 240-246.
Wilkinson, J. P., Smith, J. H., Stagg, T.O., and D. A. Walters (1980). Hori-
zontal Drilling Techniques at Prudhoe Bay, Alaska (1980) Society of Petroleum
Engineers, pp. 1-12.
Wilkinson, D. J. (1989). New Results for Pressure Transient Behavior of Hy-
draulically Fractured Wells. Low Permeability Reservoirs Symposium. Denver,
Colorado.
Wimp, J .1980. Computation with recurrence relations. Boston, Pitman.
Wimp, J. (1984). Computation with recurrence relations, Applicable Math-
ematics Series, Pitman (Advanced Publishing Program), Boston, MA.
Appendix A
No-Flow Moving Boundary
Model Solutions
In Chapter 7 we presented dimensionless pressure j
1
and constant . In this
Appendix we provide solutions to the diusion equations with variable rate of
Arps type. These solutions are of variable rate IBC close to an exponential
Arps decline. To solve diusion equation we dene variable rate almost equal to
exponential decline / s 0. Now we introduce : as an inverse of decline exponent,
b so, we get
(t
1
) = t
(a)
1
(A.1)
We should consider multiplying right hand side with the constant C in order to
normalize the parameters : and {, so rate becomes now
(t
1
) = Ct
(a)
1
(A.2)
Further, the "Laplace" inverse transform is
1
1
() =
/
a1
(:)
(A.3)
and dimensionless pressure, j
1
is now
j
1
=
t
(a)
1
(:)
1
_
0
n
a1
c
&
_
J
0
(
:
_
t
1
_
n)
:
2
1
0
(
:
_
t
1
_
n)
_
dn (A.4)
from which by requiring that no-ow boundary moves with distance r from a
wellbore as
Jj
1
J:
1
= 0. for :
1
= n
_
t
1
(A.5)
and dening constant as
345
346 A. No-Flow Moving Boundary Model Solutions
=
:
2
1
_
0
n
a
1
2
c
&
1
1
({
_
n)dn (A.6)
Further, referring to Obertrettinger and Badii (1973), page 155, we get the fol-
lowing expression
1(t
a1
1
_
J
0
(c
_
t
1
) + i1
0
(c
_
t
1
)

) = (A.7)
2
:c
[(:)]
2
j
(
1
2
&)
c
(
a
2
8
)
\1
2
a,0
_
c
2
4j
c
i
_
(A.8)
where \1
2
a,0
_
o
2
4j
c
i
_
is the Wittaker function, and 1 is the "Laplace" trans-
form
1(,(t)) =
1
_
0
c
jt
,(t)dt (A.9)
For the inverse decline exponent, : , referring to Buchholz.(1969), page.99,
we have the following asymptotic formulas when :
\1
2
a,0
_
c
2
4
c
i
_
s 2

1
2
_
c
2
4(:
1
2
)
_1
4
(
c
:
1
2
)
a
1
2
c
i(o
_
a
1
2

r
4
)
(A.10)
In addition we need the Stirlings formula:
(:) =
_
2:(
:
c
)
a
:

1
2
(A.11)
Now, asymptotically when, : we get
j
D
s
_
2

1
4
t
n
D
(
r
2
D
t
D
)

1
4
c

r
2
D
8
D
_
cos({
:
_
t
1
)
:
2
sin({
:
_
t
1
)
_
_
:
_
t
1
_
:
1
2

:
4
_
(A.12)
Where constant, , is now dened asymptotically (,o: : ) and considering
(c = {) as:
s
:
2
cot
_
{
_
:
:
4
_
(A.13)
Appendix B
"LAPLACE" Inversion
Transforms
There are many types of partial dierential equations describing uid ow phe-
nomena in porous medium whose solution may be found in terms of a "Laplace"
transform. Layered reservoirs equations are too complicated for inversion us-
ing the techniques of complex analysis. Numerous methods have been devised
for the numerical evaluation of the "Laplace" inversion integral as presented by
Piessens (1975).
The main diculty in applying "Laplace" transform technique is the deter-
mination of the original function ,(t) from its transform:
1 (:) =
1
_
0
c
ct
, (t) dt (B.1)
Inverse "Laplace" transform methods are given in Table B-1.
Three basic types of "Laplace" transform functions F(s), which should be
evaluated in the comparison analysis and they are not particularly the "Laplace"
transform functions solution of the one phase ow model:
-Functions which are continuous and for which 1(:) :
c
as : ;
-Functions which are continuous and for which there is not value c for
which 1(:) :
c
as : . and
-Functions which have discontinuities.
A case of a 1(:) about which little is known and for which an all-purpose
method is initially most convenient to apply is dicult to evaluate. An 1(:) for
which the form of the solution f(t) is known are presented in Table (B-2).
Numerical methods used in the numerical "Laplace" transform comparison
are classied into methods which compute a sample, methods which expand
function ,(t) in exponential functions, methods based on Gaussian quadrature,
methods based on a bylinear transformation and methods based on Fourier series.
Extensive results comparison are presented by Davies and Martin (1979).
347
348 B. "LAPLACE" Inversion Transforms
Table B-1: The inverse "Laplace" transform methods [after Davies and Martin
(1979)]
B. "LAPLACE" Inversion Transforms 349
A number of dierent numerical inversion techniques presented in Table (B-2)
were evaluated by Davies and Martin (1979), according to the following criteria
which are not fully independent: applicability to a variety of common types of in-
version problems, numerical accuracy, relative computation times, programming
and implementation diculties.
Examples include problems with numerical data at arbitrary points, problems
with transforms in the form of rational fractions, problems with noisy data, or
problems for which the solution is known to be of particular form should be
treated by a special available method. For many problems an all purpose method
for which numerically inverting the "Laplace" transform may be inappropriate.
It is usually worth using more than one method as recommended by Bellman
et al. (1966) on any unknown function as a check against peculiar behaviour of
the function or of the numerical method, and against programming and imple-
mentation errors. The dierent methods produce dierent results in the trou-
blesome case. By trying the methods on a function whose analytical "Laplace"
transform is known and is similar in form to the troublesome function, we can
evaluate the benet of the used method. In the Table (B-2), there are test
functions which are covering a wide range of functional forms which are covering
well known and understandable functional behaviours, and because of the simple
analytical solutions they have.
Numerical accuracy can be presented for each function and each method by
two measures. By the used measures, the presentation of the comparison data
results for each t value would not take up a large amount of space. Measure 1
gives the root mean-square deviation between the analytical ,(t) and numerical
,
o
(t) solutions for various t values. 1 gives a fair indication of the success of a
method for large t and is given by the following expression:
1
c
=
_
30

i=1
(, (i,2) ,
o
(i,2))
2
,30
_
12
(B.2)
Measure 1
c
is a similar measure weighted by the factor c
t
and is used for rela-
tively small t. 1
c
measure is given as:
1
c
=
_

(, (i,2) ,
o
(i,2))
2
c
ij2
,
_
30

i=1
c
i2
__
12
(B.3)
By the expressed measures we can evaluate the accuracy for each method
and for each function. It is also important to evaluate computation time require-
ment of dierent functions and dierent numerical algorighms implemented. A
programming eort should also be evaluated. Errors in programming can be
detected by comparing results on test problems with results in the papers orig-
inally describing the method. The determination of the function 1(:) which is
350 B. "LAPLACE" Inversion Transforms
Table B-2: The comparrison of the inverse "Laplace" transform methods
dedicated to the particular uid ow phenomena is easy for real values of s. For
complex values may be a source of error.
A large number of numerical methods for inverting the "Laplace" transform
are variants of other methods. If the original method is successful, it is reasonable
to test variants. Methods that show great promise are Dubner and Abate (1968),
Silverberg (1970) and Crump (1976). On the opposite rather poor methods are
those of Bellman et al. (1966) and Piessens (1969).
Appendix C
Regression Techniques
The approaches that have been presented in the literature in decline curve
analysis are statistical method, least-square methods, log-log type-curve over-
lays methods and the computer-automated curve tting methods. These meth-
ods are based on the general hyperbolic (exponential or harmonic is a special
case) decline curve equation developed by Arps (1945). The hyperbolic decline-
curve has three unknowns initial decline rate 1
i
, initial production rate
i
and
decline exponent b which have to be determined from the unknown measured
rate-time production data.
The least square method assumes the type of decline to be as exponential,
hyperbolic or harmonic before curve tting can be performed. The type curve
analysis also has disadvantages due to the non-uniqueness problem in type-curve
matching. The computer-automated curve tting is performed for decline curve
analysis by the linear multiple regression of the selected variables. The approach
is based on the Arps (1945) general form equation and it is applied to the rate-
time data.
C.1 Linear Regression
The various graphical and type curves approaches are less accurate and less
reliable than the statistical approach. Towler and Bansal (1993) proposed lin-
ear regression method to determine the decline parameters 1
i
,
i
and b from
measured rate-time data. A maximum in the regression coecient was used to
indicate the straight-line t. The two iterative methods to nd the three un-
known parameters in hyperbolic decline-curves are presented by Cvetkovic and
Gudmundsson (1993) in Table 6-1 (page 104). The regression coecient plotted
versus chosen parameters shows when optimum linearity is achieved. The square
of regression coecient is dened as:
351
352 C. Regression Techniques
r
2
= 1
n

i=l
(j
i
1r
i
)
2
n

i=1
(j
i
1r
i
)
2
+ 1
n

i=1
(r
i
r) (j
i
j)
where (r
i
.
i
) is for the rst method and second method dened as:
(r
i
, j
i
) = (|oq(1 + /1
i
t), |oq
t
)
(r
i
, j
i
) = (
1b
t
,
p
)
Field examples were presented and discussed regarding decline exponent big-
ger than 1 related to the fractured reservoir and the negative decline exponent
b is believed to be caused by mechanical problems in the well. The analysis is
based on hyperbolic decline and by making comparison of both methods, we can
guess harmonic b equal one decline.
C.2 Linear Multiple Regression
A linear multiple regression can be performed for the rate-time variables , `
j
,
and qt for optimum variables of
i
, 1
i
and b in empirical Arps (1945) equations.
The time-cumulative relationship is derived from the hyperbolic solution of q(t)
and can be expressed as:

p
=
t
_
0
(t) dt =
[ (t) (1 + /1
i
t)
i
]
[(/ 1) 1
i
]
It can be rearranged to give general form, (t) for the exponential (/ = 0),
hyperbolic (0 < / < 1) and harmonic solutions (/ = 1) of Arps equation in the
form expressed as:
(t) =
i
+ (/ 1) 1
i
`
j
/1
i
(t) t (i) + `
j
+ 1 (t) t
Duong (1989) suggested a procedure for the calculation of the unknown coe-
cients q
i
,A and B by using the linear multiple regression for variables (t), `
j
and (t)t. Decline rate 1
i
= ( + 1) and decline exponent / = 1,1
i
.
The cumulative production data can be calculated from rate-time data by the
equations:
`
j
= 0.5
a

)=1
_
(t)
)
(t)
)1
_
_
t
)
t
)1
_
The production rate can be calculated by the expression:
C. Regression Techniques 353
(t) =
[
i
+ `
j
]
[1 1t]
and the future production can be expressed as:
(t) =
[
i
+ (`
j
) , 1]
[1 (t t
)1
) 1t]
The future production rate can be calculated from the initial production rate, q
i
,
previous cumulative production, (`
j
)
)1
, and constants A and B and expressed
as:
(t) =
i
+ (`
j
)
)1
+ (t) [t t
)1
] + 1
)
t =
_

i
+ (`
j
)
)1
_
[1 (t t
)1
) 1t]
Chen (1991) introduced the relationship between cumulative production, N
j
rate,
q and product of rate and time, qt in the form:
`
j
= + 1 + Ct
Where, . 1. C are dened as:
=

i
1
i
1
1 /
1 =
1
1
i
1
/ 1
C =
1
/ 1
and decline exponent b, initial decline 1
i
and initial rate
i
are related to the
coecients A, B and C through the following expressions:
/ =
C
C 1
1
1
=
1
1
(C 1)

i
=

1
Again, N
j
, q and qt are not fully independent.
354 C. Regression Techniques
These approaches are related only to data that have been aected by the
boundary conditions. The linear multiple regression method is based on the
assumption of full independency of the variables rate q, cumulative production
N
j
, and time t. The variable dened as a rate-time, qt can cause some diculties
for certain sets of production data. Coecients in multiple linear regression are
also sensitive to the cumulative production term `
j
. If the values `
j
are not
measured, it is suggested to integrate rate versus time with the trapezoid method.
This term can change the coecient values so it is important to nd the most
appropriate integration method for the given values of rate versus time.
The Chen (1991) and Duong (1989) expressions were tested for various pro-
duction decline data. The interactive systemfor symbolic computation, MAPLE,
was used for the linear multiple regression. The program can easily be extended
to the linear multiple regression calculation.
C.3 Weighted Residuals Regression
The extrapolation of the decline curves is a method of predicting the perfor-
mance of the production wells. Ramsey and Guerrero (1969) used the least
square methods and could not reduce the eect of large terms in the calculation.
Least squares tting expressed by Ramsey and Guerrero (1969) results in the
shortcomings which can be eliminated by the use of weighted residuals. The oil
production q(t) can be dened as:
(t) =
i
_
1 +
/t
c
_

1
2
where decline exponent, / controls the degree of curvature of the decline curves
and has no units,
c
is the initial oil production at time zero and parameter a is
the reciprocal initial of the initial decline rate 1
i
with the same units as time t.
Appendix D
Relevant Reports and Papers
D.1 SPE Papers
Cvetkovic, B. (2009). Eective Wellbore Parameters and a Multi-fractured Hor-
izontal Well Productivity. Paper SPE 120826. SPE Production and Operations
Symposium, Oklahoma City, Oklahoma, USA, 48 April.
Cvetkovic, B. (2009). Semi-Analytical Modelling of a Horizontal Well with
Fractures in an Oil Reservoir - Screening Approach with Risk Analysis. SPE
Applied Technology Workshop. Discussion Leader and Section Manager of the
Reservoir modelling Section, March 8-11, Penang, Malaysia.
Cvetkovic,B., Halvorsen, H., Sagen, J. and E.N. Rigatos (2001). Modelling
the Productivity of a Multifractured-Horizontal Well. Paper SPE 71076, SPE
Rocky Mountain Petroleum Technology Conference, Keystone, Colorado, May
2123.
Cvetkovic,B., Halvorsen, G., and J.Sagen.(2000). Multiple-Fractured Hori-
zontal Well Case Study. SPE-65503 presented on the "4th International Con-
ference and Exhibition on Horizontal Well Technology, November 4-8, Calgary,
Canada, (http://www.petsoc.org/hw2000.html).
D.2 Presentations
D.2.1 Conferences/Forums
Cvetkovic, B., and G. Halvorsen, G. (2001). Multi-fractured horizontal well
study with HOWIF from a Vallhal Field. A Selected Software Day Seminar in
BP-AMOCO, July 26, 2001, Stavanger.
Cvetkovic, B., Halvorsen, G., Sagen,J., and R. Banerjee. (2001). Eective
Wellbore Radius For a Multi-fractured Horizontal Well Simulation. GeoQuest
Schlumberger, FORUM 2001, London, UK, March 6-9.
355
356 D. Relevant Reports and Papers
Cvetkovic, B., Halvorsen, G. and J. Sagen. (1999). A Horizontal Well with
Induced Fractures - Real Case Study. Presentation, The Heriot-Watt & Stanford
University Reservoir Description and Modelling Forum, Crie (Skotland).
Cvetkovic, B., Halvorsen, G., and E. Lw (1955). Modelling of a Horizontal
and a Vertical-Fractured Well, Mathematical Modelling of Fluid Flow Through
Porous Media Conference, Poster Session, St. Etienne, France, May 22-26
D.2.2 Schlumberger Internal EUREKA Presentations:
Cvetkovic, B. (2007). A Vertical/Horizontal Well with Laterals Simulation.
Reservoir Symposium 2007 Schlumberger (EUREKA), Moscow, June.
Cvetkovic, B. (2007). Unconventional Stimulation Studies. Reservoir Sym-
posium Schlumberger (EUREKA), Kula Lumpur 2007
Cvetkovic, B. (2006). Fractured Carbonate-dolomite Gas-condensate Reser-
voir Simulation Studies. Reservoir Symposium 2006, Schlumberger (EUREKA),
Bucharest, September.
Cvetkovic, B (2006) Unconventional Stimulation Challenges, Schlumberger
Internal Seminar on Unconventional Reservoir Studies, Engineering and Man-
ufacturing Russia and Schlumberger Moscow Research, SMR Novosibirsk, 6 th
November.
D.3 Industry Reports
Cvetkovic, B., Halvorsen, G. and J.Sagen (1999). HOWIF-OIL BOX Model.
(Final and Half-Year Report for PHILLIPS- Condential).
Cvetkovic, B., Halvorsen, G. and J. Sagen (1998) HOWIF-OIL SLAB Ad-
vanced Model. (Final and Half-Year Report for BP, CONOCO, PHILLIPS -
Condential);
Cvetkovic, B., Halvorsen, G.and J. Sagen (1997). HOWIF-OIL SLAB Im-
proved Model. (Final and Half-Year Report for BP, CONOCO, PHILLIPS -
Condential);
Cvetkovic, B., Halvorsen, G.and J. Sagen. (1996). HOWIF-OIL SLAB
Model (Final Report and Half-Year Report for BP, CONOCO, PHILLIPS -
Condential);
Cvetkovic, B. and G. Halvorsen (1977) Horizontal Well Induced Fractures
Production Optimization. Pre-study work for Norwegian Research Council, NFR
(December 1995-February 1996) IFE-Kjeller, March.
D.4 NTNU Faculty Reports
Cvetkovic, B. and J.S. Gudmundsson (1993). Analytical Models for Rate Decline
in Oil and Gas Reservoirs. University of Trondheim, December.
D. Relevant Reports and Papers 357
Cvetkovic, B. (1992). Modelling and Solution Methods for Layered Reser-
voirs. University of Trondheim, June.
Cvetkovic, B. (1992). Fundamental Equations and Techniques Used in Decline
Curve Analysis. University of Trondheim, January.
D.5 Other Related Presentations and Reports
Cvetkovic, B. (2008). Risk Analysis of a Tri-Lateral Well Producing the Gas
Condensate Trym Reservoir with MEPO-ECLIPSE. Bayerngas internal presen-
tation.
Cvetkovic, B. (2003). A Horizontal Well Production Optimization, simula-
tion study for Petrobaltic, 2003.
Cvetkovic, B. (2001). A Multiple-Fractured Horizontal Well Case Study",
Invited presentation to Croatian SPE Section- January 30, INA-Naftaplin Oil &
Gas Co., Zagreb, Croatia.
Le Turdu,C., Cvetkovic,B., and S. Gacesa.(2001). Geological Modelling (Zu-
tica Field, Croatia) by PETREL & HOWIF. Poster presentation, International
Petroleum Engineering Conference, October 2001, Zadar, Croatia.
Cvetkovic, B., Halvorsen, G., and J. Sagen, (1999). A Horizontal Well with
Multiple Fractures: Real Case Study. Poster presentation, Petroleum Associa-
tion Meeting, Stavanger.
Cvetkovic, B. (1999). Horizontal well with Induced Fractures Production
Optimization in Oil Reservoir-HOWIF Programme. SPE, Croatian Section, Za-
greb, May.
CvetkovicB., Halvorsen, G., Sagen, J., and Y. Bhushan (1999). Wellbore Re-
sponses of a Horizontal Well with Created Fractures Coupled to an Oil Reservoir.
Petroleum Engineering Summer School, Workshop 4, Dubrovnik, 1999.
Cvetkovic, B., Halvorsen, G., Sagen, J., and Y. Bhushan (1998). Coupling
of a Reservoir to a Wellbore with Created Fractures. Presentation of Semi-
analytical software tool and reservoir simulation results, Reservoir Simulation
Users Meeting, Stavanger.
Cvetkovic , B et al.(1999).Optimal Massive Gas Injection Conditions for Oil
Recovery Enhancement by Diusion in Fractured and Heterogeneous Reservoirs,
MAGIC OR. Final Report 1996-1999, IFE-Kjeller, June (EU Research Project
JOULE Final Report)
Bekken C., Lw E., Cvetkovic,B. and G. Halvorsen.(1996). Thermal In-
duced Fracturing model ALFAFRAC Output Upgrade. IFE Internal Report,
IFE/KR/F-95/230, IFE-Kjeller, March.
Cvetkovic, B., SPE, Halvorsen, H. and J. Sagen (2001). History Matching a
Multiple-Fractured-Horizontal-Well Responses with Step Function of Rates and
Pressures. Canadian International Petroleum Conference, June 12-14, Calgary,
Alberta, Canada (Abstract accepted not realized).
358 D. Relevant Reports and Papers
Cvetkovic, B. and G. Halvorsen.(1998) Horizontal Well with Created Frac-
tures Production Optimization, Lectures with the HOWIF Programme Demon-
stration, SPE Annual Technical Conference and Exhibition, Horizontal Well
Seminar, New Orleans, LA. (Abstract accepted not realized).

S-ar putea să vă placă și