Sei sulla pagina 1di 260

WATERFLOODING

UNDER
FRACTURING CONDITIONS

E.J.L. Koning

(,< s ^ v
->l*

!,

.f
WATERFLOODING UNDER FRACTURING CONDITIONS

PROEFSCHRIFT
ter verkrijging van de graad van doctor in de technische wetenschappen aan
de Technische Universiteit Delft, op gezag van de rector magnificus,
Prof. Drs. P.A. Schenck, in het openbaar te verdedigen voor een commissie,
aangewezen door het College van Dekanen, op dinsdag 27 september 1988 te
10.00 uur

door
Eric Jan Leonardus Koning
Doctorandus in de Wiskunde en Natuurwetenschappen
geboren te Haarlem

TR diss
1664

- II -

Dit proefschrift is goedgekeurd door de promotor prof.ir. H.J. de Haan

- Ill

Aan mijn ouders

- V -

CONTENTS

Page

CHAPTER ONE

INTRODUCTION

1.1. Waterflooding

1.2. Waterflooding under fracturing conditions

1.2.1. Scope and definition

1.2.2. Objectives of thesis

1.2.3. Previous work

1.2.4. New elements in thesis

1.3. Organisation

References

7
CHAPTER TWO

THE PORO- AND THERMO-ELASTIC STRESS FIELD AROUND A WELLBORE

Summary

10

2.1. Introduction

11

2.2. Solution for the stress field

14

2.2.1. Assumptions

14

2.2.2. Notation and basic equations

14

2.2.3. Solution with Goodier's displacement potential

17

2.2.4. The particular solution in cylindrical coordinates

18

2.2.5. Complete formulation of the problem

20

2.2.6. Method of solution

22

2.2.7. Asymptotic expansion of the stress solution

23

2.2.8. Simplified solution method

27

2.2.9. Numerical method to evaluate Ao

30

T
2.3. Solution for the pressure and temperature fields

30

2.3.1. Temperature field

30

2.3.2. Pressure field

32

2.4. Analytical solution for thermo-elastic stress variations

38

2.5. Analytical solution for poro-elastic stress variations

47

2.6. Fracture initiation pressure

59

2.7. Evaluation of two field cases

64

- VI -

2.8. Other applications

73

2.9. Conclusions

73

List of symbols

75

References

77

Appendix 2-A - Basic equations

78

Appendix 2-B - The particular stress solution in cylindrical coordinates

81

Appendix 2-C - The complete stress solution in cylindrical coordinates

84

Appendix 2-D - Asymptotic expansions of the stress solution

89

Appendix 2-E - Simplified solution method

93

Appendix 2-F - Numerical method to evaluate Ao

97

CTT

Appendix 2-G - Solution for the pressure distribution

101

CHAPTER THREE
ANALYTICAL MODELLING OF FRACTURE PROPAGATION
Summary

103
105

3.1 Introduction

106

3.2 Fracture propagation in an infinite reservoir in the absence


of reservoir stress changes

109

3.2.1 Assumptions

109

3.2.2 One-dimensional leak-off

109

3.2.3 Two-dimensional leak-off - Pseudo-radial solution

114

3.2.4 Two-dimensional leak-off - Elliptical solution

116

3.3 The effect of poro- and thermo-elastic stress changes on fracture


propagation pressure

120

3.3.1 Definition of fracture propagation pressure

120

3.3.2 Analytical calculation of poro-elastic stress


changes at the fracture wall

121

3.3.3 Numerical calculation of poro-elastic stress changes


at the fracture wall

129

3.4 Fracture propagation in an infinite reservoir under the


influence of reservoir stress changes

131

3.4.1 Assumptions

131

3.4.2 Consistency checks

132

- VII -

3.4.3 Two field cases

133

3.5 Fracture propagation in a pattern flood.


Effect on sweep efficiency

139

3.5.1 Zero voidage in the absence of reservoir stress changes

139

3.5.2 Zero voidage with reservoir stress changes

145

3.5.3 General flooding conditions and the use of


a reservoir simulator
3.7 Conclusions

147
148

List of symbols

150

References

152

Appendix 3-A Calculation of poro-elastic stresses in elliptical


coordinates

154

Appendix 3-B A numerical method for calculating poroand thermo-elastic stress changes

163

Appendix 3-C Calculation of thermo-elastic stresses


and of the axes of the elliptical fluid fronts

168

CHAPTER FOUR
A PRESSURE FALL-OFF TEST FOR DETERMINING FRACTURE DIMENSIONS

171

Summary

172

4.1 Introduction

173

4.2 Calculation of the pressure fall-off with a closing fracture

175

4.2.1 Assumptions

175

4.2.2 Integral equation for the dimensionless pressure function

175

4.2.3 Solution for the dimensionless pressure function

179

4.3 Analysis of a pressure fall-off test

183

4.3.1 Four methods to determine fracture length

183

4.3.2 Discussion

191

4.4 Conclusions

193

List of symbols

194

References

196

- VIII -

Appendix 4-A Solution for dimensionless pressure function in Laplace


space

198

Appendix 4-B Relationship between fracture closure constant and


fracture length

202

CHAPTER FIVE
A PRACTICAL APPROACH TO WATERFLOODING UNDER FRACTURING CONDITIONS

207

Summary

208

5.1

Introduction

209

5.2

An example

209

5.3

Conditions for a successful process

210

5.4

Preliminary investigations

210

5.5

Basic data gathering

214

5.5.1 Measurements of in-situ stress, fracture orientation and


elastic moduli

5.6

214

5.5.2 Injectivity test and fall-off testing

215

5.5.3 Matching with propagation model

217

Determination of optimal reservoir pressure, injection rate


and well pattern

217

5.6.1 Determination of maximum reservoir pressure

217

5.6.2 Determination of maximum injection rate

221

5.6.3 Fractured well pattern and reduction in the number of wells

221

5.7 Full-scale implementation

230

5.8 Special applications

232

5.9 Conclusions

233

List of symbols

235

References

237

Appendix 5-A Determination of the ratio of producers to fractured


injectors and the reduction in the number of wells

238

SUMMARY

243

SAMENVATTING

247

ACKNOWLEDGEMENTS

'

251

- 1 -

CHAPTER ONE
INTRODUCTION
1.1. Waterflooding
1.2. Waterflooding under fracturing conditions
1.2.1. Scope and definition
1.2.2. Objectives of thesis
1.2.3. Previous work
1.2.4. New elements in thesis
1.3. Organisation

- 2 -

INTRODUCTION

1.1 WATERFLOODING

When an oilfield is exploited by simply producing oil and gas from a


number of wells, the reservoir pressure, in many cases, drops rather quickly
and so does the production rate.
Therefore, water is often injected into the reservoir to maintain the
reservoir pressure. The injection wells are located at carefully chosen
points so that as much oil as possible is displaced by the water to the
production wells before water starts to break through in the producers. The
process of recovering oil by water-injection is called waterflooding. The
degree to which the water is capable of sweeping the oil to the producers
without bypassing it, is called the sweep efficiency.
Even in the optimal case not all the oil can be produced by
waterflooding. A certain amount of oil always remains trapped inside the
pores of the rock by capillary forces. This so-called residual oil may be as
much as 40% of the total pore volume of the rock.

1.2 WATERFLOODING UNDER FRACTURING CONDITIONS

1.2.1 Scope and definition


A major saving in the exploitation costs of an oilfield can be obtained
by a reduction in the number of wells. To reduce the required number of
injectors, the injection capacity of the well has to be increased. An
effective way of doing this is by rupturing the reservoir rock along a plane
intersecting the wellbore. Such an induced fracture in the rock enlarges the
surface area of injection considerably so that with the same pressure
gradient a dramatic increase in injection rate can be obtained.
Rupturing of the reservoir rock can be effected by injecting water at a
pressure above the formation breakdown pressure. The fluid pressure is then
high enough to overcome the rock stress and the tensile strength that keep
the rock particles together.
Continued injection at this pressure causes the fracture to propagate
into the reservoir. Water injection in this manner is called waterflooding
under fracturing conditions.

- 3-

1.2.2 Objectives of thesis


Although an increase in injectivity is favourable, excessive lateral
growth of the fracture may adversely affect the sweep efficiency of water
injection and result in premature water breakthrough. Moreover, if the
fracture is vertical, excessive vertical growth may establish communication
with other reservoirs resulting in loss of injection water or even worse,
loss of oil.
It is the purpose of this thesis to:
a) investigate the mechanism of fracture initiation and fracture propagation
under the influence of continued water injection,
b) evaluate the effect of fracture growth on sweep efficiency,
c) improve the methods for determining fracture dimensions,
d) give rules for designing the process of waterflooding under fracturing
conditions in such a way that sweep efficiency is unimpaired and vertical
fracture growth is limited.
1.2.3 Previous work
Fracturing of production wells to stimulate productivity is a wellestablished technique. Here, fractures are created with a special injection
fluid that contains proppant particles (e.g. sand) to keep the fracture open
when the well is producing. These fractures thus retain a fixed length after
the fracture has been created. Furthermore, since their length is generally
small with respect to the interwell spacing, sweep efficiency is hardly
affected.
Extensive literature exists on the subject of fracturing for production
stimulation .
For waterflooding under fracturing conditions the effect of continuous
fracture growth on sweep efficiency is of paramount importance. Moreover,
this fracture growth is strongly dependent on changes in reservoir rock
stress and fluid leak-off resulting from changes in reservoir pressure and
temperature. Therefore, the theory developed for the stimulation of
producers is not directly applicable.
Relatively little is known in the petroleum-engineering literature on
the mechanism of fracture propagation during waterflooding and its effect on
sweep efficiency.

- 4-

In the early literature on waterflooding, studies appeared on the


effect of fractures with a fixed length on sweep efficiency using physical
. 2'3
model experiments

4
A major step forward was the construction by Hagoort et al. of a
numerical model that could simulate the growth of a vertical fracture of
constant height in a simple, vertically homogeneous reservoir. They studied
fracture propagation as a function of reservoir and injection/production
conditions. One of the important conclusions of this study was that the
leak-off from the fracture into the reservoir should essentially be modelled
as two-dimensional in the plane of the reservoir. Therefore previously
developed analytical models with a one-dimensional description of leak-off
are generally inadequate for modelling waterflood-induced fractures.
Later, Hagoort presented a broad and innovative study on the subject in
his thesis "Waterflood-induced Hydraulic Fracturing" . Apart from the
numerical simulation model, analytical calculations of sweep efficiency for
a 5-spot containing a fractured injector with a fixed fracture length were
presented. The calculations were also extended to stratified reservoirs. The
effect of reservoir pressure on rock stress and fracture propagation
pressure were discussed using two-dimensional poro-elastic stress
calculations. Finally, a first step towards the. monitoring of fracture
length by pressure fall-off testing was presented. The technique consists of
recording the downhole fluid pressure during an interruption of injection.
The declining fluid pressure with time can be analysed to get an indication
of fracture length.
Recently, the effect of a change in temperature on reservoir rock
stress and on fracture propagation was analysed by Perkins and Gonzalez .
They showed that injection of large amounts of cold water into reservoirs
with a good permeability can eventually result in thermal fracturing of the
rock due to a decrease in rock stress by cooling. With a simple semianalytical model they showed that thermal fractures can reach a considerable
length. They also showed that the common practice of calculating pressure
and temperature-induced changes in rock stress using two-dimensional rather
than three-dimensional poro- and thermo-elasticity, greatly underestimates
the stress changes for large lateral penetration depths of pressure or
temperature.

- 5 -

1.2.4 New elements in thesis


The thesis presented here takes the work of Hagoort and Perkins &
Gonzalez as a starting point. The main new elements are:

- A complete analytical description of the stress field surrounding an


unfractured wellbore. The theory of three-dimensional poro- and thermoelasticity is used to calculate the effect of pressure and temperature
changes on reservoir rock stress.
The calculation of the stress field is used to evaluate the pressure at
which fractures can be initiated.
- An analytical model of fracture propagation with a complete twodimensional description of fluid leak-off into the reservoir.
- A focusing on the mode of propagation for which the fluid flow around the
fracture is pseudo-radial. This means that the pressure transients are
travelling radially into the reservoir.
It is shown that this mode of fracture propagation does not influence the
sweep efficiency.
- An extension of the pseudo-radial model to include a three-dimensional
analytical calculation of the effect of pressure- and temperature-induced
stress changes on fracture propagation.
- The development of a three-dimensional numerical method to calculate poroand thermo-elastic changes in reservoir stress that can be incorporated in
the numerical fracture simulator as developed by Hagoort et al.
- An extension of Hagoort's method to determine fracture length from a
pressure fall-off test by including: a description of fracture closure,
two-phase elliptical fluid flow and different fracture geometries. It is
shown that the pressure fall-off with time is characterised by four
distinct periods, each of which gives independent information on the
fracture size.
- A practical programme for designing the application of waterflooding under
fracturing conditions in a given field.

- 6 -

1.3 ORGANISATION
This thesis is written in five self-contained chapters that can be read
independently of the others.
Chapter 2 deals with the calculation of the stress field around an
unfractured wellbore. The numerical method presented here was published
earlier in Ref. 7. The analytical methods will be published in Ref. 8.
Chapter 3 describes the modelling of fracture propagation and its
effect on sweep efficiency. Most of this chapter was published in Ref. 9.
Chapter 4 describes the method of determining fracture length from a
fall-off test. This chapter was published in Ref. 10. Recently, an extension
of the method together with a field application was published in Ref. 11.
Finally, Chapter 5 concludes this thesis with practical design rules
for field implementation.

- 7 -

1. Howard, G.G. & Fast, C.R., Hydraulic Fracturing.


SPE Monograph Volume 2, 1970.
2. Crawford, P.B. & Collins, R.E., Estimated effect of vertical fractures
on secondary recovery.
Trans. AIME (1954), 201, p. 192.
3. Dyes, A.B., Kemp, C.E. & Caudle, B.H., Effect of fracture on sweep-out
pattern.
Trans. AIME (1958), 213, p. 245.
4. Hagoort, J., Weatherill, B.D. & Settari, A., Modelling the propagation
of waterflood-induced fractures.
SPEJ (Aug. 1980), p. 293.
5. Hagoort, J., Waterflood-induced hydraulic fracturing.
PhD. Thesis, Delft Technical University, 1981.
6. Perkins, T.K. & Gonzalez, J.A., The effect of thermo-elastic stresses on
injection well fracturing.
SPEJ (Feb. 1985), p. 78.
7. Koning, E.J.L. & Niko, H., The effect of reservoir pressure and
temperature variations on fracturing conditions during a waterflood
operation.
Proceedings of the 3rd European Meeting on Improved Oil Recovery
(April 1985), Rome, p. 219.
8. Koning, E.J.L., The poro- and thermo-elastic stress field around a
wellbore.
SPE paper in preparation.
9. Koning, E.J.L., Fractured water-injection wells. Analytical modelling of
fracture propagation.
SPE 14684, 1985.
10. Koning, E.J.L. & Niko, H., Fractured water-injection wells. A pressure
fall-off test for determining fracture dimensions.
SPE 14458, 1985.
11. Koning, E.J.L. & Niko, H., Application of a special fall-off test in a
fractured North-Sea injector.
SPE 16392, 1986.

- 9 -

CHAPTER TWO
THE PORO- AND THERMO-ELASTIC STRESS FIELD AROUND A WELLBORE
Summary
2.1. Introduction
2.2. Solution for the stress field
2.2.1. Assumptions
2.2.2. Notation and basic equations
2.2.3. Solution with Goodier's displacement potential
2.2.4. The particular solution in cylindrical coordinates
2.2.5. Complete formulation of the problem
2.2.6. Method of solution
2.2.7. Asymptotic expansion of the stress solution
2.2.8. Simplified solution method
2.2.9. Numerical method to evaluate Ao
0T
2.3. Solution for the pressure and temperature fields
2.3.1. Temperature field
2.3.2. Pressure field
2.4. Analytical solution for thermo-elastic stress variations
2.5. Analytical solution for poro-elastic stress variations
2.6. Fracture initiation pressure
2.7. Evaluation of two field cases
2.8. Other applications
2.9. Conclusions
List of symbols
References
Appendix 2-A - Basic equations
Appendix 2-B - The particular stress solution in cylindrical coordinates
Appendix 2-C - The complete stress solution in cylindrical coordinates
Appendix 2-D - Asymptotic expansions of the stress solution
Appendix 2-E - Simplified solution method
Appendix 2-F - Numerical method to evaluate Ac
Appendix 2-G - Solution for the pressure distribution

- 10 -

SUMMARY
Changes in reservoir pressure or temperature can change the stress
field in the reservoir rock. Such changes in the stress field influence the
conditions under which fracturing of the reservoir rock can occur. They also
influence the geometry and direction of propagation of induced fractures.
This chapter presents new analytical and numerical methods for
calculating the stress field around a single vertical well in an infinite
reservoir. Poro- and thermo-elastic variations in the stress field resulting
from axisymmetric changes in reservoir pressure and temperature are
incorporated. The methods are general in the sense that no use is made of
the assumption of plane strain. The formulae presented allow application of
arbitrary axisymmetric pressure and temperature fields.
Simple analytical formulae for thermo-elastic stress variations are
presented for temperature distributions with a step profile. Simple
analytical formulae for poro-elastic stress variations are presented for
quasi, steady-state pressure profiles including discontinuities in fluid
mobility.
Analytical expressions for the fracture initiation pressure including
poro- and thermo-elastic effects are presented.
The numerical and analytical methods are used to evaluate the change
in fracturing conditions for two realistic field cases. The first is a
high-permeability reservoir in which a large thermo-elastic reduction in
rock stress occurs following the injection of cold water. Thermal fractures
may be induced under such conditions. These fractures are likely to remain
contained within the cooled reservoir region.
The second case is a low-permeability reservoir for which the
poro-elastic increase in rock stress due to the rise in pore-pressure is
dominant with respect to the cooling effect. The fracture initiation
pressure of the reservoir has therefore increased. The stresses in cap and
base rock have decreased because of cooling. The induced stress gradients
may force created fractures to grow vertically into cap and base rock.

- 11 -

THE PORO- AND THERMO-ELASTIC STRESS FIELD AROUND A WELLBORE

2.1 INTRODUCTION
In this chapter an analysis is made of the factors that influence the
stress field in the reservoir rock surrounding a wellbore. In principle,
knowledge of this stress field allows one to determine the downhole fluid
injection pressure at which tensile failure or fracturing of the rock
occurs.
This so-called fracture initiation pressure is an important parameter
in selecting a suitable downhole pressure during injection. The injection
pressure should be lower if fracturing is to be prevented, for instance to
avoid communication with other reservoir zones. Or, it should be higher if
fracturing is desired, for instance to increase the injection capacity of
the well.
Since the concept of stress is generally considered to be a difficult
one, we start out with some introductory remarks on the definition of
stress. For a more complete introduction see, for instance, Ref. 1.
Consider a slightly deformed material body in equilibrium. Owing to
the deformation internal forces have been generated. Imagine an arbitrary
point inside the body blown up to form an infinitesimal cube. The sides of
this cube are subject to forces which hold it in equilibrium. A cube can be
found with an orientation such that the sides of the cube only experience
forces that are normal to its six surfaces. Since the cube is in
equilibrium, there are three independent forces, one for each pair of
opposite surfaces.
These forces taken per unit area are called the three principal
stresses in the point. If a cube with a different orientation is chosen, for
instance, such that its sides line up with the axes of a given Cartesian
coordinate system, the three principal stresses can be decomposed into their
components parallel to the coordinate axes. This then generally results in
nine nonvanishing components, three for each pair of opposite surfaces.
These nine components are called the elements of the stress tensor in the
point under consideration. Since the cube is prevented from rotating, there
can only be six independent components.

- 12 -

The relation giving the stress resulting from an applied deformation


or the deformation resulting from an applied stress is called the
stress/strain relation. In this work we consider only relations of a linear
form and therefore our analysis falls within the framework of the theory of
linear elasticity.
When fluid is injected from a wellbore into a reservoir a certain
pressure is required to squeeze the fluid into the pores of the reservoir
rock. This fluid pressure inside the wellbore exerts a radially outward
force (or equivalently a radial stress or radial load) on the surrounding
reservoir rock. For simplicity, we consider here an open hole without a
casing. As a result of the radial load the rock is deformed (expanding
radially outward) and stresses are generated in the rock.
Normally the rock is already in a state of stress before the well is
drilled. These so-called tectonic stresses result from the weight of the
overburden and from tectonic movement of the earth's crust. The radial
expansion of the borehole now results in a decrease in the horizontal stress
that is tangential to the borehole circumference. If the fluid pressure is
high enough, this so-called tangential stress may change from a compressive
stress into a tensile stress. If this tensile stress exceeds the tensile
strength of the reservoir rock formation fracturing will occur.
During fluid injection the pressure of the fluid inside the pores of
the rock surrounding the well increases. As a result of this increase in
pore pressure the rock frame or rock matrix tends to expand. Since there is
little room for such a poro-elastic expansion, internal compressive stresses
are generated. An increase in tangential stress results which works against
the decrease induced by the radial loading of the wellbore.
A change in temperature will also affect the stress state of the
reservoir rock. If, for instance, injection of cold fluid cools the
reservoir rock, the rock grains and the rock matrix tend to contract,
resulting in a thermo-elastic decrease in the rock stress. The corresponding
lowering of the tangential stress will facilitate formation fracturing.
Similarly, injection of a hot fluid will tend to impede fracturing.
Since this work is based on the linear theory of elasticity, the
results for the stress field apply only to reservoir rocks that have a
certain degree of consolidation. Furthermore, the stress changes in question

- 13 -

should be small with respect to the in-situ tectonic stresses so that non
linear effects on material constants such as Young's modulus and Poisson's
ratio may be neglected if the latter are taken at their in-situ values.
The linearity of the theory enables us to calculate the stresses
induced by poro-elastic effects, thermo-elastic effects and wellbore loading
separately. The combined effect is then obtained by simply adding the
respective stress components.
In the past poro-elastic effects have been included in the calculation
2-4
of the stress field around a wellbore
. However, these works relied on the
assumption that the induced poro-elastic deformations occur only in the
horizontal plane of the reservoir. This so-called assumption of plane strain
allows the vertical variation of the stresses to be neglected and results in
a considerable simplification of the differential equations involved.
Although in most cases the assumption of plane strain is acceptable
when calculating the stresses induced by the loading of the wellbore
(Section 2.2.7 of this work), this assumption is generally not valid when
poro- or thermo-elastic stresses are considered.
This was first demonstrated by Perkins and Gonzalez . They considered
combined poro- and thermo-elastic stresses for disc-shaped regions of
uniform change in pressure and temperature. It was concluded that as long as
the penetration depth of the pressure or temperature front is small compared
to the reservoir height plane strain conditions would prevail. However, as
the fronts advance during injection the state of rock deformation changes
from horizontal strain initially, to vertical strain when the penetration
depths have become large with respect to the reservoir height. The
horizontal stress may differ by as much as 100% between the case of
horizontal and vertical strain even though the pressure or temperature
inside the disc remained constant.
In their work Perkins and Gonzalez did not provide expressions for the
stresses induced by a more realistic pressure profile. Moreover, the effect
of a wellbore and an analysis of the fracture Initiation pressure were not
included.
This chapter attempts to address these remaining problems. In Section
2.2 new analytical expressions for the tangential stress are presented for
an arbitrary axisymmetric pressure or temperature profile. No assumption of
plane strain has been made. In Section 2.3 simple but realistic expressions
for the pressure and temperature field are derived.

- 14 -

In Sections 2.4 and 2.5 these expressions are used in the general
formulae of Section 2.2 to derive closed form analytical expressions for the
poro- and thermo-elastic change in tangential stress. In Section 2.6 an
analysis of poro- and thermo-elastic effects on fracture initiation pressure
is presented. Both open and cased hole are considered. In Section 2.7 the
change in tangential stress is calculated for two realistic field cases. In
Section 2.8 other applications of the solution for the stress field are
discussed. Finally, in Section 2.9 the conclusions are presented.

2.2 SOLUTION FOR THE STRESS FIELD


2.2.1 Assumptions
In a reservoir of axial symmetry the fracture initiation pressure for
vertical fractures depends on the tangential stress field of the reservoir
rock. The following assumptions have been made in calculating this stress
field.
1. The reservoir has the shape of a horizontal disc of which the axis
coincides with that of the injection well (Fig. 2.1). The disc has finite
height and infinite radius and consists of linearly elastic, isotropic
and homogeneous, permeable bulk reservoir rock.
2. The reservoir is bounded at top and bottom by an infinite, linearly
elastic, isotropic and homogeneous, impermeable cap and base rock with
the same elastic constants as the bulk reservoir rock.
3. Coupling between elastic behaviour and fluid or heat flow is neglected.

2.2.2 Notation and basic equations


The linear stress/strain relations for combined poro- and thermoelastic deformations are :
e. . = - [(1+u) Aa. . - vAo&. .] - a ApS.. - a m AT8..
13
E
l]
ij
p r 13
T
13

(2.1)
i,j=l,2,3

where:
Ap

= change in pore pressure

AT

= change in temperature

- 15 -

GEOMETRY OF A DISC-SHAPED RESERVOIR SURROUNDED BY AN INFINITE MEDIUM.


The reservoir is considered to have infinite radial extent.

FIG. 2.1

- 16 -

Aa. .

= change in total stress tensor with respect to the stress state at


Ap = AT = 0.

c..

3
= trace of stress tensor = Z A a.
X1
1-1
= strain tensor

5..

= Kronecker delta (1 if i=j,


0 if i/j)
J
J

= Young's modulus of bulk reservoir rock

= Poisson's ratio of bulk reservoir rock

= linear thermal expansion coefficient

= linear poro-elastic expansion coefficient

Ao

i]

The linear poro-elastic expansion coefficient is defined as:

% = -3
where c, and c

(c

b-v

ir

(1

-^}

(2 2)

are the compressibilities of the bulk reservoir rock and the

rock grains, respectively.


The total stress is the stress exerted by both the rock matrix and the
fluid in the pores of the rock. The formulation in terms of total stress
offers certain advantages over a formulation in terms of effective stress,
eff
which is defined as: Ao. .
= Ao. . - Ap6...
One of the advantages is
r
i]

ij

i]

continuity of stress across a boundary between permeable and impermeable


media. For further discussion, see Ref. 5.
From (2.1) it is clear that the mathematical treatment of poro- and
thermo-elastic effects is completely equivalent. For compactness, we
introduce the following generic notation for a combined poro- and
*
thermo-elastic deformation :

* The notation for the generic linear expansion coefficient is chosen as a,


in line with the usual notation for the linear thermal expansion
coefficient. It should not be confused with a as used in M.A. Biot's
classic paper: "General Theory of Three-dimensional Consolidation". J.
Appl. Phys., 1941, Vol. 12, p. 155. Biot's a is related to our a
P
l-2u
according3 to: a = a.
p
E

- 17 -

oAT =a Ap + a AT
pr
T

(2.3)

so that Eq. (2.1) becomes:

e.. = [(1+u) Aa.. - uAoS..] - aA 6..


13
E
13
13
13

(2.4)

The sign convention in Eq. (2.1) and (2.4) takes compressive stresses and
strains as positive.
Eq. (2.4) can be inverted to express the stresses in terms of the
strains. If the strains are then expressed in terms of a displacement vector
and the stresses are substituted into the equations of equilibrium, three
differential equations in the three components of the displacement vector
result. For given Ap and AT and for given boundary conditions for the
stresses the solution for the stress field is determined. The development of
the differential equations is given in Appendix 2-A.

2.2.3 Solution with Goodier's displacement potential


It is shown in Appendix 2-A that a particular solution of the
differential equations can be obtained by the introduction of Goodier's
displacement potential. This potential is the solution to Poisson's equation
and is given by:

# = 7- 7 ^ IS! dx'dy'dz' a AT (x',y',z') J

(2.5)

2
2
2 1/2
R = [(x-x') + (y-y') + (z-z) ] '

(2.6)

with

The particular
solution for the stress tensor denoted as Aa. - is obtained
r
13T
from (2.5) by:

a2

Aa. - = - T z 0 + AAT 6. .
13T
1+u 9x.9x.
13
1 3

(2.7)

where the stress consists of poro- and thermo-elastic components

Aa. - = Aa. . + Aa. .m


13T
13P
13T

(2.8)

- 18 -

and the following generic notation has been introduced:


AAT = A Ap + A AT
p
T
A

and A

(2.9)

are called the poro- and thermo-elastic constants, respectively.

They are defined as:


Eft

A = T- 2
p
1-u

12
= T 2 - ^ (1
l-v

c
a
)
c '
b

(2.10)

and
Ea

m = ^
T
1-

(2.11)

2.2.4 The particular solution in cylindrical coordinates


In the rest of this chapter it is assumed that the pressure and
temperature fields possess axial symmetry,
AT = AT(r,z)

(2.12)

Poisson's equation for the displacement potential becomes in cylindrical


coordinates:
3 2 0 + - 3 0 + 3 2 0 = - 7 ^ a AT
r
r r
z
l-v

(2.13)

Because of symmetry the displacement vector has only a radial and a vertical
component, given by, respectively:

u = - 3 0 ;
r

w = - 3 0
z

(2.14)

From Appendix 2-B the particular solution for the stresses is given by:
Ao - =
3 0 + AAT
rT
1+u r v

; Ao - = -~ 3 0 + AAT
zT
1+u z
(2.15)

Ao n - = 77- - 3 0 + AAT ; Aa - = r^- 3 3 0


T
1+u r r^
rzT
l+i r z

- 19 -

Two limiting cases can now be considered. In the first case we assume
that the vertical variation of AT inside the reservoir zone is small.
If furthermore the penetration depth of the pressure and temperature
variations are small with respect to the reservoir height, the only
displacement will be in the horizontal plane and:

w = - 3 0 = 0

(2.16)

z
Then (2.13) can be readily integrated, giving
r
u = - 9 <t> = T ^ a r
l-i) r

dr'r' AT(r' ,z)

(2.17)

r
w

and for the stresses from (2.15):

Aa

rT

= +

A r
~2 '
r r

dr

'r'

AT

< r '' z >

w
A
Ao - = -
r

dr'r' AT(r',z) + AAT(r,z)


r
w
(2.18)

Ao

Ao

= AAT(r,z)

- = 0
rzT

The expressions in (2.18) are the well-known plane strain results for the
1
stresses .
In the second limiting case we assume that the radial variation of AT
is small. When furthermore the penetration depths become very large with
respect to the reservoir height, there will be displacement in the vertical
direction only and therefore:

u = - 9 0 = 0
r
giving from (2.13) and (2.15)

(2.19)

- 20 -

Ac - = Aa.- = AAT(r,z)
rT
0T
'
(2.20)
Aa - = Aa - = 0 .
zT
rzT
We call (2.19) the condition of vertical strain.
From the above it is clear that during fluid injection into a
reservoir the conditions for poro- and thermo-elastic stresses can change
from plane strain to vertical strain. In Sections 2.4 and 2.5 of this
chapter this will be demonstrated with explicit examples.

In Appendix 2-B the full (r,z) dependence of the particular solution


is obtained from (2.13) and (2.15) with the help of modified Bessel fuctions
and Fourier integral transforms.
The particular solution does not generally satisfy all prescribed
boundary conditions on the stresses. By making use of the linearity of the
theory, this can be achieved by adding an appropriate solution of the
homogeneous equations of elasticity to the particular solution.
2.2.5 Complete formulation of the problem
For given AT the stresses are completely determined if stress boundary
conditions are imposed at the wellbore and at infinity.
In the specification of the boundary conditions at the wellbore the
following well model is used (Fig. 2.2). Two packers in a vertical wellbore
are separated by a distance d. Between these packers a pressure Ap , uniform
with height/ is exerted on the surrounding elastic formation. Ap

= p -p. is

the pressure differential needed to sustain a certain fluid injection rate


q. The distance d can be larger than or equal to the reservoir height h. The
change in stress Aa. . is calculated with respect to the initial stress state
at Ap = 0 and AT = 0, and the reservoir at uniform pressure and
w
temperature. Effects of casing or perforations on the change in stress state
are not considered. This model leads to the following set of boundary
conditions:

21

IMPERMEABLE

PERMEABLE

~ri

r-1

GEOMETRY OF WELL WITH PACKERS PENETRATING A DISC-SHAPED


RESERVOIR OF INFINITE RADIAL EXTENT AND SURROUNDED BY AN
INFINITE IMPERMEABLE M E D I U M .

FIG. 2.2

- 22 -

Ao

Aa

r
rz

= Ap

} -v

|.|<f

=0
(2.21)

Aa

= 0
r

Aa

rz

} r.V

|.|f

=0

Since we are considering an infinite elastic medium we have the additional


boundary conditions:
1im Ao.. = 0
r-
^
(2.22)
lint
Aa. . = 0
1J
Izl-

2.2.6 Method of solution


To solve the problem the following decomposition of Ao.. is made:

Aa.. = Aa..- + Aa?.

(2.23)

where Aa. - is the solution to the "traction-free" wellbore problem:


13T
Aa - = 0
rT
}
Aa

- = 0
rzT

lim Aa..- =
r->

r = r , -0 < z < +
w

(2.24)

lim Aa. - = 0
I z I *

and A a . . is a solution to the homogeneous equations of elasticity describing


the s t r e s s e s induced by the radial loading of the w e l l b o r e . From ( 2 . 2 1 ) o
(2.24) Aa.. has to satisfy:

- 23 -

Ao

= Ap

Ao - O
rz
Ao = O
r

(2.25)

} -. | . | f

Ao = O
rz
lim Ao. . =

lim Ao.. = 0

Problem (2.25) is relatively well known and was first solved by Tranter .
7
His result was extended by Kehle to account for the shear stresses exerted
on the formation by the frictional force of the packers. Kehle found that
for realistic cases this effect can be neglected.
The solution to problem (2.24) has, to our knowledge, not been
published in the literature. We have determined Ao. - by decomposing it into
the particular solution and a solution to the homogeneous equations of
elasticity,

Ao. .- = Ao. .- + Ao ljT


ljT
13T

(2.26)

such that (2.24) are satisfied.


As discussed earlier and shown in Appendix 2-B the particular solution
Ao. - is obtained from an axisymmetric Goodier potential. In Appendix 2-C it
is shown how the solutions for Ao. - and Ao.. are obtained from an
i]T

13

axisymmetric biharmonic function called the Love potential. The solutions


appear as complicated Fourier integrals over modified Bessel functions.
Table 2-1 summarises the various decompositions and methods of solution.

2.2.7 Asymptotic expansions of the stress solution


Tranter showed how his solution for Ao.. can be evaluated numerically.
Unfortunately this method is rather cumbersome. Moreover, it cannot be
readily extended to evaluate Ao. - and Ao..- since these expressions
contain
r
1

IJT

ijT

additional integrals over AT.


In Appendix 2-D it is shown that the integral transforms of the stress
solution can be evaluated analytically if asymptotic expansions of the

- 24 -

TABLE 2-1 - DECOMPOSITION OF STRESS TENSOR

Aa. . = Aa..- + Aa..


lj
IJT
i]

; Aa. . satisfies complete wellbore boundary

Aa.. = Aa..- + Aa..ljT


1}T
ljT

; Aa.
Aa. -- satisfies
satisfies "tract
"traction-free" wellbore

conditions, Eq. (2.21)


IJT

boundary conditions, Eq. (2.24)

Component

Differential equations (d.e.)

Generating potential

Aa.

homogeneous d.e. of elasticity

Love potential

homogeneous d.e. of elasticity

Love potential

d.e. of poro- and thermo-elasticity

Goodier's displacement

ID
o

Aa i jT
ljT

potential

- 25 -

Bessel functions are made. In the calculation of the integrals it is assumed


that the axisymmetric function AT is constant over the reservoir height and
zero outside of the reservoir, i.e.
AT = AT(r) . H ( | - |z|)

(2.27)

where h is the reservoir height and H the step function defined as:
H(a-x) = 1

x < a
(2.28)

= 0

x > a

If the Bessel functions are expanded in powers of r/h or r'/h (whichever is


the smallest) the particular solution for the tangential stress becomes to
lowest order:

A(J

0T

" 2 ^2
r

' dr'r'AT(r').N(z, j,r) + AAT(r).H(| - |z|)


r
w
(2.29)

+ f dr.r.AT(r.,. {
r

(z

, '+
+ r' )

(z

'" >2 3/2 )


+ r' )

where

N(z,
1

z
z_
-,r) =

, .^ +
~
. -_
' 2
, 2
2,1/2
, 2
2,1/2
(z+ + r )
(z_ + r )

(2.30)

and

z+ = | + z

(2.31)

No precise criteria can be given for the region of validity of (2.29)


as it stands. These criteria depend on the shape of AT and on the ratio of
its penetration depth to the reservoir height. For a specific AT the
accuracy of (2.29) has to be investigated by comparison with a numerical
evaluation of the exact solution for Aa.-. This will be discussed further in
T
Sections 2'. 4 and 2.5.

- 26 -

If in the solution for Ao- the Bessel functions containing r and r


T
* w
are expanded, the result becomes to lowest order in r /h and r/h:

*T

AS

<2'32>

HT

where

AS

HT = 4 ' d r ' r '^( r ') t 2 ' + ,3/2


r
(z
+ r' )
w
+

, 2'
(z

,2,3/2*
+ r' )

<2'33>

As for (2.29) no general criteria can be given for the accuracy of


(2.32). Further discussion appears in Sections 2.4 and 2.5.
If Tranter's solution for Laa is expanded to lowest order in r /h and
a

the term with the modified Bessel function K (kr) (see Appendix 2-D) is
neglected, we find:
Ap
Aa = - ~Y

r 2
(-f-)
N(z, - , r)

(2.34)

where N is the same function as in (2.30) but with h replaced by d. It can


also be written as:
u

d
N(Z

r) =

' 2~'
1

TJl

l//L

[u^ + l]

[u_

(2>35)

/2
+1]

where
d/2 + z
u =
+

(2.36)
'

At the wellbore (r = r ) the expression for Aan can be expanded while


w

retaining the K term, resulting in:


Ap
Aff (r

e w>

" ~f

N(z

' 2' V
(2.37)

where:

P
w

u^ '

2 > V

- 27 -

M(z, - . r) =

~
jji
[u/ + 1] '

+
[u_

In Fig. 2.3 a plot of Aa. /Ap

+ 1]J/2

(2.38)

versus 2z/d is presented for various

values of 2r /d. Poisson's ratio was taken as 0.2. The plot shows Aoa up to
a

first order (Eq. (2.34) at r = r ) and up to second order (Eq. (2.37)). It


is shown that for 2r /d < 0.1 sufficient accuracy is maintained if the
second term is neglected. Furthermore, for 2r /d < 0.01 the function N/2
w
behaves as a step function so that (2.37) becomes:

io (t

' - - *\,

W = f
}

2r
r<

0.01

(2.39)

|.|f
o
8
The result ACT (r ) = - Ap is the same as the well-known result obtained
0 w
w
under plane strain conditions. In the latter case it is assumed that the
loaded interval d is infinitely long.
We have shown that the solution for Aa. can be safely approximated by
the first term in its asymptotic series expansion, as long as 2r /d < 0.1.
However, since we are dealing with an asymptotic series the complete range
of validity of this first term approximation may be substantially larger.
This complete range of validity can only be determined by comparing (2.34)
with a numerical evaluation of the exact expression for ACT.. Since however,

the condition 2r /d < 0.1 is satisfied for most practical cases we have not
investigated this matter any further.
2.2.8 Simplified solution method
If 2r/h < 0.01 then from the previous subsection N(z,-,r) becomes a
step function. Under this condition the particular solution (2.29) becomes
in the neighbourhood of the wellbore (r ~ r ):

Aa

0T

" \

dr r A

' ' T(r') + AAT(r) + ASH-(z)

|z| < |

r r
w
(2.40)
=AS H i (z,
with AS - defined in (2.33).

|z|>|

28

o.o

_-5U=L=

2rw /h 0.01
2r w /h = 0.1
2r w /h= 1.0
-02

UPTOFIRST ORDER
UP TO SECOND ORDER

-0.4

-0.6

-1.0

0.8

1.2

1.6

1.8

2.0

2Z/h

Ao- 0 /Ap w FOR VARIOUS VALUES OF 2r w /h

FIG. 2.:

- 29 -

We now observe that for |z| < h/2 Eq. (2.40) consists of the well-known
plane strain expression (2.18) plus the term AS -.
Taking, for convenience, the distance between the packers as being
equal to the reservoir height (h=d), we have for the complete tangential
stress field near the wellbore:
Aan = Aan- + Ao- + Laa
9
6T
0T
6

= (

r 2
r
S
i ) {AS
Ap }
dr,r,AT r
7
HT " w " ~2
( ')
r r
w

(2.41)

+ AAT(r) + AS H T

|z| < |

r 2
= ()
AS - + AS r
HT
HT

"
|z| > r
' ' 2
for

2r/h < 0.01


r - r
w

In Appendix 2-E it is shown that (2.41) could have been obtained from the
plane strain expression for the particular solution (2.18) and from the
simple plane strain solution to the homogeneous equations of elasticity
provided that the combined plane strain solution is subjected to the
boundary condition
Aa
r
lim
-00

= AS -

-<z<+

(2.42)

From this observation AS - can be interpreted as an apparent change in the


all-round horizontal far field stress.
A similar situation exists for the vertical stress. From Appendix 2-E:
Aa z - = AAT(r) + AS V - (z)

|z| < |
(2.43)

=AS v - ( z)

|z|>|

- 30 -

where
ASv-(z) = - 2ASH(z)

(2.44)

From comparison with (2.18) AS - can be interpreted as an apparent change in


the vertical far-field stress. When plane strain solutions are used this
change must be incorporated as the boundary condition:
lim
r-*oo

A(7

= AS t -

VT

- co < z < +

(2.45)

In Chapter 3 the above interpretation is used in calculating the poroelastic stresses around a vertical fracture. The poro-elastic stresses at
the fracture wall are calculated in plane strain. Eq. (2.42) with a slightly
adapted version of (2.33) is then used to account for deviations from plane
strain.

2.2.9 Numerical method to evaluate Aa^z


0T
In Appendix 2-F a numerical method is presented for the evaluation of
the particular solution Ao.-. This method is useful for investigating the
range of validity of the asymptotic expression (2.29). It can also be used
when AT does not satisfy (2.27) but has a general dependence on the vertical
coordinate. The method consists of the numerical evaluation of certain
integrals. To eliminate the oscillatory behaviour of the integrand, the
solution for Aafl- is represented in terms of complete elliptic functions
rather than in terms of Bessel functions. The integrals are then evaluated
9
using a standard integration routine from the IMSL library .

2.3 SOLUTION FOR THE PRESSURE AND TEMPERATURE FIELDS


2.3.1 Temperature field
The temperature field is determined from Lauwerier's solution in
cylindrical coordinates

. This solution applies to the injection of an

incompressible fluid at constant rate. One-dimensional vertical heat


conduction in cap and base rock is taken into account, horizontal conduction
is neglected and inside the reservoir the temperature is assumed to be
constant in the vertical direction.
Lauwerier's solution in cylindrical coordinates can be written as:

- 31 -

H
T D = erfc Wr

}
2(l-RD )

< 1, |z| < |

2
A
D TD +
D
= erfc {
~

2/(rD(l-RD2,)

< 1, |z| > f

= 0

> 1, -o < z <

(2.46)

where

T - T
res
=
D
T. . - T
mj
res

4a t M
s
s
T
D =
2
2
h M
r
R

(2.47)

r
D = R
c

M
. 1/2
R = l
M h?rJ
c
r

r
and
T

= initial reservoir temperature


res
T. . = injection temperature
in]

= injection time

= reservoir height

= thermal diffusivity of cap and base rock

= heat capacity of cap and base rock

= heat capacity of fluid filled reservoir rock

= heat capacity of injection fluid


= injection rate

is the radius of the temperature front and obeys the simple convective

heat balance:

- 32 -

heat absorbed by injection fluid = heat given off by the reservoir


or
qt
M
n

At R

AT = it R

2
h M AT
c
r

(2.48)
'

the temperature difference tends to zero (Eq. (2.46)).

The assumption of one dimensional vertical heat conduction is justified if:


M R

Pe =

TTT >:> x
s

(2 49

- >

where Pe is the dimensionless Peclet number. Condition (2.49) means that the
radial velocity of the temperature front is much greater than the vertical
velocity of the temperature transients in cap and base rock. Therefore, with
isotropic thermal conductivities and approximately equal thermal
diffusivities in the reservoir and cap and base rock, radial temperature
transients may be neglected. Using the expression for R in (2.47), (2.49)
c
becomes:
M

M ffha
.s
s

(2.50)

Condition (2.50) is satisfied for most field conditions.


In Fig. 2.4 T

inside the reservoir has been plotted vs R

for

different values of r . This plot shows that for r < 0.05 the step function
is a good approximation to the temperature profile inside the reservoir.
This is the convective limit in which the amount of heat given off by cap
and base rock is small with respect to the amount of heat given off by the
reservoir.

2.3.2 Pressure field


For simplicity, we first consider fluid flow without discontinuities
in fluid mobility. The line source solution for injection of slightly
compressible fluid at constant rate is given by :

33

LAUWERIER TEMPERATURE PROFILES INSIDE RESERVOIR

FIG. 2.4

34 -

where
Ap = p ( r , t )

- p
k.

= fluid mobility (X = )
M
= reservoir height

= injection rate

= hydraulic diffusivity (r? = k


0MCfc

= injection

time
00

-g

Ei = exponential integral (- Ei(-x) =


x

ds)

2
For r /47jt < 0.02 ( 2 . 5 1 ) can be approximated by:

2 * ^
q

- - l n f R
e

(2.52)

with
R = 1.5 /(Tjt)
e

(2.53)

Eq. (2.52) has the form of the steady-state solution with a time-dependent
exterior radius. We therefore call (2.52) the quasi steady-state
approximation.
Introducing the dimensionless quantities:

D=Re
=

^D

(2

' 54 >

2jrhXAp.
q

Eq. (2.52) becomes:

A p D = - In r Q

(2.55)

Fig. 2.5 compares (.2.55) with the line source solution (2.51). It is shown
that for r

> 0.6 the accuracy of (2.55) becomes less than 13%. However, for

calculating poro-elastic stresses (2.55) is sufficiently accurate, since


most of the pressure rise and therefore most of the stress build-up occurs
close to the well.
An expression for the pressure field in the presence of
discontinuities in fluid mobility is derived in Appendix 2-G. It is assumed

35

7.0
UNE-SOURCE
QUASI STEADY-STATE

6.0 -

1.2

1.4

UNE-SOURCE AND QUASI STEADY-STATE SOLUTION

FIG. 2.5

- 36 -

that incompressible fluid displaces slightly compressible oil in a piston


like manner (Fig. 2.6). The flooded zone consists of a cold region with
constant fluid mobility and a warm region with constant fluid mobility. The
following exact solution has been obtained:
2
2
27Th .
l n
c
1 _
F
11
,FV.,
Fx
Ap, = r In + r In - ~ r exp (T~7) .Ei(7-)
q
*!
X
r
X
R
2 X
^ 4rjt' v 4rjt'
R

T4p2

x ; l n r - 2 ^ e x p w)>El("^J

(2 56)

R2
2
2?rh .
1 1
. _F .
r

Ap 3 = - - T- exp (T^)-Ei(-T^)

with Ap., X., i=l,2,3 the pressure differences and mobilities in the cold
flooded zone, the warm flooded zone and the oil zone, respectively, TJ is the
hydraulic diffusivity in the oil zone. R
J

and R are the radii of the


F

temperature front and of the flood front, respectively.


If:
-r < 0.02

4rjt (2.57)
2
f - < 0.02
the'following expression is a good approximation to (2.56)
R
X,
R
Xn
R
Ap, = In + r~ In - + r In ID
r
X
R
X_
R
2
c
3
F
X

X,
A

Ap

3D

r3

R
F

R
1

ln

2jrhX
with Ap. =
JD
and R

Ap., j = 1,2,3
*}

given in (2.53).

37

INJECTION FLUID INCOMPRESSIBLE


OIL SLIGHTLY COMPRESSIBLE

SCHEMATIC R E P R E S E N T A T I O N OF P R E S S U R E P R O F I L E
WITH V E R T I C A L D I S P L A C E M E N T F R O N T S .

FIG. 2.6

- 38 -

2.4 ANALYTICAL SOLUTION FOR THERMO-ELASTIC STRESS VARIATIONS


Calculation of stresses for temperature field with step profile
It was shown in Section 2.3.1 that a step profile is a good
approximation to the temperature distribution inside the reservoir provided
that

rD

2
4a t M
= |
1- < 0.05
h M
r

(2.59)

Furthermore, if we neglect the temperature change in cap and base rock due
to conduction, we simply have:
AT(r,z) = AT.H(^ - |z|).H(R -r)

(2.60)

with H the step function defined in (2.28).


If this simple profile is substituted into the thermo-elastic part of
Aa_- the
:he remaining integrals can be easily evaluated. From (2.29) and (2.32)
this thermo-elastic component is given by:

ST

ST

+A

Vr

r 2
= (7*) AS H T

r
dr'ATfr'Jr'.Nfz,-^) + y i ( r ) H ( j - |z|)
2r r
w

z_

dr'r' AT(r'). {

(z 2 + r-V /2

}
(z

2
+

(2.61)

r'2)3/2

where:

AS_ =
HT
4

dr'AT(r')r'
{ r
-r +
^ ^, ]
v
'
, 2
.2 3/2
, 2
,2 3/2n
r
(z
+ r' )
(z + r' )
w
+
-

v(2.62)

'

and N as defined in (2.30).


It is convenient to introduce the following dimensionless variables:

- 39 -

0TD

AAT
T

tc
(2.63)

z
R
c

D ~ 2R

After integration there results:


r

Aa

w
D

*TD " ( ^->

,1
AS

~2
wD

HTD + I <^> ^ D ' W

" N< W

i i
1 )

^V^D'*
r

< 1

(2.64a)

2
wD 2

>

1
AS

wD

-1(r> ^ v w

HTD

D*

(2 64b)

r_

where:
A

S m = T (N(z .h.r ) - N(z,h .1))


HTD
4 v D D wD
D D
z

N(

(z D+2 ++ r \l/2
)
D '
v

(a)
'

z
2 %

(zD- + r D ')1/2

(2

'65)

(b)

(c)
'

z = h + z
D+
D D

At the wellbore (2.64a) becomes:

^TD^wD

"

2AS

The solution for Ao.

HTD

+ H(h

D - lsDl>

(2

incorporates the traction-free boundary condition at

the wellbore. The extra term Aa. due to a load Ap


a

is not included.

'66)

applied at the wellbore


w

- 40 -

A few wellbore radii out into the reservoir Aaflrn = (r ^/r) AS


0TD
wD' D
HTD
becomes very small and Ao.m becomes identical to the particular solution
0TD
Range of validity of asymptotic expressions
To investigate the range of validity of (2.64) we have first compared
the analytical solution for Ao.
evaluation of Ao.

(Eq. (2.64) minus Ao.

) with a numerical

. The latter is obtained using the method described in

Appendix 2-F.
The results are shown in Figs. 2.7a-2.7f. Here Aortm was made
3
0TD
dimensionless with respect to |AT| rather than AT and therefore, since the
dimensionless change in stress is negative for r < 1, a cooled reservoir is
represented. For comparison, the analytical plane strain solution is also
shown (see Eq. (2.73) below).
From the figures there is excellent agreement between the analytical
and numerical solution within the cooled region. For smaller dimensionless
reservoir heights the results become less accurate close to the temperature
front (Fig. 2.7g). Close to the front outside the cooled region the relative
error can become as large as 20%. Within the cooled region the absolute
stress level is higher so that with approximately the same absolute error
the relative error inside is much smaller than the relative error outside
the cooled region.
In Ref. 5 Perkins and Gonzalez investigated the change in tangential
stress induced by a cylindrical inclusion of constant temperature. They did
not consider the presence of a wellbore. They provided an expression for the
change in tangential stress averaged over the cooled region. This expression
was obtained by curve fitting to numerical results.
We calculate the average change in tangential stress by putting
r

= 0 in (2.64a)/ integrating over z and dividing by h. The result is:


A

^ D 5 * 1 " {I* + <1/V 2 1 1 / 2 " 1 / h D }

wD =

(2

- 67)

In Fig. 2.8 the result obtained by Perkins and Gonzalez is compared


with (2.67). The maximum difference is found to be 6%. For completeness
(2.64a) with r = 0 and z = 0 is also shown.
wD
D
At the wellbore we have for the particular solution:
AaamrJr ) = ASm + H(h - I z I )
0TD wD
HTD
D
' D1

(2.68)

41

1.0
NUMERICAL SOLUTION
PLANE STRAIN SOLUTION (ANALYTICAL)
ASYMPTOTIC SOLUTION (ANALYTICAL)

h 0 =10.
z D =0.0

0.8
0.6
0.4

0.2
0.0

0.2

0.4

-0.6
-0.8
-1.0
0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

DIMENSIONLESS RADIAL DISTANCE FROM THE WELLBORE


FIG. 2.7a

CHANGE IN TANGENTIAL STRESS INDUCED BY A STEP PROFILE

1.0

T"

0.8

0.6

NUMERICAL SOLUTION
PLANE STRAIN SOLUTION (ANALYTICAL)
ASYMPTOTIC SOLUTION (ANALYTICAL)

hD = 10
rWD = 0.06
rD =0.1

0.4
0.2

0.0

-0.2
-0.4
-0.6
-0.8

-1.0
0.0

2.0

FIG. 2.7b

4.0

6.0

8.0

10.0

12.0

14.0

16.0

DIMENSIONLESS VERTICAL DISTANCE FROM THE MIDDLE


OF THE RESERVOIR
CHANGE IN TANGENTIAL STRESS INDUCED BY A STEP PROFILE

18.0

20.0

1.0

42

NUMERICAL SOLUTION
PLANE STRAIN SOLUTION (ANALYTICAL)
ASYMPTOTIC SOLUTION (ANALYTICAL)

t.0
0.0

0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
0.8
-1.0 "0.0

_1_

_1_

0.4

0.6

0.2

0.8

1.0

1.2

1.4

1.6

1.8

2.0

DIMENSIONLESS RADIAL DISTANCE FROM THE WELLBORE


FIG. 2.7C

1.0

CHANGE IN TANGENTIAL STRESS INDUCED BY A STEP PROFILE

0.8 h

NUMERICAL SOLUTION
PLANE STRAIN SOLUTION (ANALYTICAL)
ASYMPTOTIC SOLUTION (ANALYTICAL)

h 0 1.0
r w 0 = 0.0'
0.006
0.1

0.6
Q.4
0.2
0.0

j _

-1.0
0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

DIMENSIONLESS VERTICAL DISTANCE FROM THE MIDDLE


OF THE RESERVOIR
FIG. 2.7d

CHANGE IN TANGENTIAL STRESS INDUCED BY A STEP PROFILE

1.8

2.0

43 -

NUMERICAL SOLUTION
PLANE STRAIN SOLUTION (ANALYTICAL)
ASYMPTOTIC SOLUTION (ANALYTICAL)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

J_
1.6

1.8

2.0

DIMENSIONLESS RADIAL DISTANCE FROM THE WELLBORE


FIG. 2 . 7 e

CHANGE IN TANGENTIAL STRESS INDUCED BY A STEP PROFILE

1.0

1
h0

0.8

NUMERICAL SOLUTION
PLANE STRAIN SOLUTION (ANALYTICAL)
ASYMPTOTIC SOLUTION (ANALYTICAL)

0.1

rWD 0.0006
0.6

=0.1

0.4

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

DIMENSIONLESS VERTICAL DISTANCE FROM THE MIDDLE


OF THE RESERVOIR
FIG. 2.7f

CHANGE IN TANGENTIAL STRESS INDUCED BY A STEP PROHLE

0.18

0.20

44

20.0

h =1.0
rw0= 0.006
16.0

12.0

<
Z

LU
O

OC

O
cc
o:
Ld

8.0

<
4.0

0.0
0.0

0.2

0.4

0.6

FIG. 2.7 g

I.U

1i i i i 111

0.8

1.0

1.2

1.4

1.6

1.8

2.0

RELATIVE ERROR OF ASYMPTOTIC SOLUTION

1i i i i 111

1rr M i l ]

co
CO

1 1

CO

0.9

_I

<
zUJ
hO

_i

z
o
CO
z
UJ
2
O

/./

o
z<
X

/ '//
y
i

t A

0.7

/:
'/.'"
'/ *

/ /

0.6

*/
*s

z
UJ

' . /

0.8

z
<
(
en
to
UJ

t:

,''~Ss^

''

'

0.5

FITTED CURVE OF P4G


AV/rOAtTTH AK1AI VTtPAl PT^I II T
AVLKAuCU ANALT llwAL KLOUUI

"

n A

'

,-J

10

10"

'

1 11

10

1 1 1ll

10'

A hi A 1 V T t f * A 1

1 _J - 1 - L U . l l

10

1/hc
FIG. 2.8

D F C I ft T CftO

COMPARISON WITH RESULTS OF PERKINS AND GONZALEZ

l_J_-l. 1 1 1

105

- 45 -

with AS

from (2.65a). The particular solution can be regarded as the

superposition of the solution for a cylindrical inclusion with radius R

and

temperature AT and the solution for an inclusion with radius r and


w
temperature -AT. Since the analytical solution for Aoa__ is accurate at
r > R

for h

> 10 or 2R /h < 0.1, we can similarly conclude that (2.68) is

accurate at least for 2r /h < 0.1.


w
For Aff we have at the wellbore:
0TD
(r
STD
wD> "A W r wD> + Aff TD (r wD ) = 2 A S HTD + ' V ^ D '
From the similarity between (2.69) and (2.68) Aafl

<2'69>

is also accurate at

least for 2 r /h < 0.1. This is confirmed by the fact that the functional
w
relationship between r and h in (2.69) is the same as that between r and d
w
w
in the first term of Ao. (Eq. (2.37)). A conservative criterion for the
o

accuracy of the analytical solution for Aafl is 2r /d < 0.1.

We conclude that when the use of temperature distribution (2.60) is


justified, an accurate description of the thermo-elastic stress change is
given by the analytical expression (2.64) with the restriction that
2r /h < 0.1. Most likely this condition can be relaxed. However, since it is
satisfied for most practical cases we have not attempted to specify it any
further.
The plane strain and vertical strain limits
Following the discussion in Section 2.2.4, two limiting cases can be
considered, h

* which is the plane strain case and h -* 0 which is the

vertical strain case.


For simplicity, we take 2r /h < 0.01 so that according to Fig. 2.3
w
1
"
- N(z .h_,r _) becomes a step function. From (2.64) and (2.65) we have the
1

DO

WO

following results:

- 46 -

1 (plane strain)

AS

HTD

^HTD

lZDl ' h D

(3)

(b)

} h Q > 100
AS

HTD

' 2 DI * h D

S mT , = T
HTD
4

(C)

|zj t h
' D'
D

(2

'701

(d)
'
} 10 < h < 100

S = " 7
HTD
4

0TD
A"

|zj 4. hn
' D'
D

- 0TD " 4

(e)

(a)

D =

I DI

fh

r
2
}(-T2)

(b)

(2.71)

} 10 < h < 100

STD

"

1
4

AaafT,(r ) = 1
0TD wD
A

flmT,(r ) = 1.5
0TD wD

I Z DI * h D

(c)

z = 0
D

(a)

|z ! t h
' D'
D

(b)
}

Aa. (r ) = - ;
0TDv wD'
2

(2.72)

10 < h < 100

|z| * h n
' D'
D

(c)
v

From (2.70a) we have that at z = 0 (2.64) reduces to the plane strain


result:

A^ ( p s ) - A(PS> - i ,fwDx2 * I
0TD
" A<7TD ~ 2 ( ~ > + 2
D

n <

(2.73)
* " 2

2>
r
D

where the superscript (ps) denotes plane strain.

D "

- 47 -

1 (vertical strain)
AS.
HTD

< h.

(2.74)
AS
=0
HTD

> h.

Ao

< h.

Aa

0TD = 1

0TD

wD
( )
D

= 0

(2.75)

> h.

Aartm^(r ) = 2
TD wD

< h.
(2.76)

*WW

> h.

We see from these limits that inside the reservoir the stress changes
double as the state of deformation goes from plane strain to vertical
strain. In the plane strain case the stress change at the vertical reservoir
boundary is 50% larger than a bit further inside the reservoir for
10 < h

< 100.

It is interesting to observe that a few wellbore radii out in the


reservoir the stress discontinuity across the vertical reservoir boundary
equals A AT whereas at the wellbore it equals 2A AT.
In Fig. 2.9 Aa.

(r

) is shown for 2r /h = 0.01 and for various

values of h.

2.5 ANALYTICAL SOLUTION FOR PORQ-ELASTIC STRESS VARIATIONS


Calculation of stresses for quasi steady-state pressure profile
When no discontinuities in fluid mobility are present, the pressure
distribution can be approximated by (from 2.55):

Ap
r
D = - In rDn
with

r =

D t

(2.77)

48

DIMENSIONLESS THERMO-ELASTIC STRESS CHANGE AT WELLBORE

FIG. 2.9

- 49 -

R = 1.5 /(rjt)
e
When (2.77) is substituted into the poro-elastic counterpart of (2.61) the
integrals can be easily evaluated.
We introduce the dimensionless variables:

z
,
h
R ' ftD = 2R
e
e

.-

2ffhX

A-

%>D = a T ^

HpD

qA

<2-79>

Hp

After integration there results:

Q< V V 1 * ~ < V W

= ( r Df > ASHPD + \ N< W V { "

(r

~2
rD

^ DA

D 1}
+

VW

(rD>

(2 80)

'

I> }

1)

where
A

PD!rwD>

AS

HPD = 1

' N( WW
z

Q(z

,h , r

) = asinh

~ ,

D+

r,

D - D

I Q ( Z D' h D' 1} " I V V W

z
+ asinh

(2.81)

- 50 -

/ 2
with asinh x = ln(x + /(x + 1)), N defined in (2.65b) and H the step
function in (2.28).
At the wellbore (2.80) becomes:

S P D (r wD> " A eD (r wD>- H( V K I ' + 2 A S H P D "

< 2 ' 82 >

As for the thermo-elastic case, a few wellbore radii out into the
oir Aaa
_ = (r n_/r_)
_ becomes
become very
reservoir
/r )
As
J small and Aan _ becomes
0pD
wD n D
HpD
0pD
identical to the particular solution Ao,
0pD*
Range of validity of asymptotic expressions
As in the previous section, we have first compared the analytical
solution for Aafl

with the numerical one using the method given in Appendix

2-F. The results are shown in Figs. 2.10a-2.10f. The figures show that there
is excellent agreement between the analytical and numerical solutions both
within and outside the reservoir. In fact, the curves for the numerical and
analytical solution coincide completely in Figs. 2.10a and 2.10b and almost
completely in Figs. 2.10c-2.10f. For comparison the analytical plane strain
solution is also shown (see Eq. (2.87) below).
. o
For Ao. _ we have at the wellbore:
0pD
A o (r \ = AS
0pD
wD
HpD

(2.83)

with AS from (2.81).


HpD
To investigate the properties of (2.83) it is convenient to form the
dimensionless quotient:
Ao
(r 7\
AS n
flpD
wD _ HpD
Ap
fr )
Ap(r
)
r
r
D wD
D wD

Ac. (r )
dp
vi
A Ap(r )
p
w

(2.84)
7 N(z,h,r ) - T Ufz^fh,,^ )
4
D D wD
4
D D wD
where from (2.81) and (2.77):

u(

W r D }

m y y i ) - Q(zD>hD>rwD)
T7T

{2 85)

'

wD
Fig. 2.11 shows the function - U for 2r /h = 0.01 and for various values of
2
1
h . For comparison, the function N(z ,h ,1) is also shown. The latter

51

PRESSURE
STRESS (NUMERICAL)
STRESS (ANALYTICAL)
STRESS (PLANE STRAIN)

rwD 0.003
hD 1.0
z = 0.0

V.

0.00

_i_
0.05

0.10

FIG. 2.10a

0.15

0.20

0.25

_l_

_1_

0.30

0.35

0.40

0.45

0.50

DIMENSIONLESS PORO-ELASTIC STRESS CHANGES

o O

PRESSURE
STRESS (NUMERICAL)
STRESS (ANALYTICAL)

rwD 0.003
1.0
r 0.1

o O

....J.'BJH

M M W W I

0.0

0.2

0.4

FIG. 2.10b

0.6

0.8

1.0

1.2

1.4

DIMENSIONLESS PORO-EUSTIC STRESS CHANGES

1.6

1.8

2.0

52

10.0
9.0

'wD

PRESSURE
STRESS (NUMERICAL)
STRESS (ANALYTICAL)
STRESS (PLANE STRAIN)

0.0003

h =0.1
zD

8.0

=0.0

7.0
6.0
a
a.
b*

5.0

<

4.0
3.0
2.0

1.0
0.0
0.00

0.05

0.10

FIG. 2.10c

0.15

0.20

0.25

0.30

0.35

0.40

0.45

DIMENSIONLESS PORO-ELASTIC STRESS CHANGES

8.0
PRESSURE
STRESS (NUMERICAL)
STRESS (ANALYTICAL)

r w0 = 0 . 0 0 0 3

7.0
6.0

hD

=0.1

rn

= 0.1

5.0

a
a.
b

<

o.o
-1.0
j _

-2.0
0.00

0.02

0.04

FIG. 2.10d

0.06

0.08

0.10

0.12

O.U

DIMENSIONLESS PORO-EUSTIC STRESS CHANGES

0.16

0.18

53

10.0
9.0

r w0 = 0 . 0 0 0 0 3
hD

0.01

8.0 It

zD

= 0.0

PRESSURE
STRESS (NUMERICAL)
STRESS (ANALYTICAL)
STRESS (PLANE STRAIN)

7.0
6.0
5.0
4.0
3.0
2.0

1.0
0.0
0.00

-i_

_1_

_1_

0.05

0.10

0.15

FIG. 2.10e

_1_

_l_

0.20

0.25

0.30

0.35

0.40

0.45

0.50

DIMENSIONLESS PORO-ELASTIC STRESS CHANGES

8.0
7.0 h

PRESSURE
STRESS (NUMERICAL)
STRESS (ANALYTICAL)

rwD 0.00003
0.01

rn =

6.0

0.1

5.0
4.0 h
3.0

-1.0
_i_

-2.0
0.000

_i_

0.002

0.004

0.006

j _

0.008

0.010

j _

0.012

_i_

_L

0.014

0.016

Z
FIG. 2.10f

DIMENSIONLESS PORO-ELASTIC STRESS CHANGES

_l_

0.018

0.020

54

i n

\ * \

08

\ l

^ ~ "* ** -

^ \
\.

^**^^

OS

A\
Ai
Ml

N/2

\ \\
K^-

0.4

0.2

0.4

0.6

h 0 = 01

^ *

'

**-*^

*xs*^

0.0

h 0 * 1.0

""N.

iV

\
\

h0-lO

Mi
1"& t

0.2

"""*-

^ ^*x
"--^^.
> ^ x ^
"""
*

""*--'..

^-..

~~~

0.8

1.0

1.2

1.4

1.8

Z.0

2z/h

COMPARISON OF FUNCTIONS N/2AND U/2 FOR VARIOUS VALUES OF h D . 2 r w / h =0.01

FIG. 2.11

- 55 -

function is the thermo-elastic counterpart of the function U as can be


seen by comparing (2.84) with (2.69) and (2.65a). From the figure it can be
deduced that the behaviour of the two functions is very similar except that
the function (2.85) has a weaker dependence on dimensionless reservoir
height and resembles therefore more closely a step function. From the
excellent agreement with the numerical calculations and from the similarity
with the thermo-elastic case, we conclude that the analytical expressions
for Ao _ and Aaa _ are accurate for all h_ with the restriction that 2r /h
0pD
0pD
D
w
< 0.1. Again, this restriction may be somewhat conservative.
The plane strain and vertical strain limits
As in the thermo-elastic case two limiting situations can be
considered.

1 (plane strain)

wD

SpD(rwD)
~A7
T~ = 1

~ =

Ap(r )
r
D wD

(2.86)

Away from the vertical reservoir boundary inside the reservoir, the plane
strain expression for the stress is recovered:
pS)
Aai
=i
Ap (r ) - 4^ + K(^"V
0pD
2 F D l D'
t '

{^ Ap
(r ) +4Jl]
* V wD;

(2.87)

( r D < 1)
which at the wellbore becomes:
Al^U
0pD
h

1 = Ap rr )

(2.88)

*D V WD

wD'

1 (vertical

strain)

AS
HpD

A P Jr n ) = 2
rH
A D f wD', = 0

^ W

'V < hD
| z j > hn
D

(2.89)

- 56 -

VD (r wD ) = 2 ,

, .
< h

' Z DI

Ap(r
r
n)
D wD

D
(2.90)

Ap(r
n)
*l> wD

I 2 DI

>h

These limiting values are attained at much smaller dimensionless reservoir


heights than in the corresponding thermo-elastic case (Eq. (2.74) and
Eq. (2.76)). For comparison:

HTD

= 0.45

for h =
D

= 0.1

(a)

ir

(2.91)

c
HpD
Ap

f o r h = = 0.1
D
2R
e

= 0.125

D ( r wD>

where (2.91) was taken at z

(b)

= 0.

Pig. 2.12 shows A. (r l/Ap fr \ = Aa,. (r )/(A Ap(r )) for


0pD wD
D wD
dp w
pr w
2r /h = 0.01 and for various values of hn.
w
D
Calculation of stresses with discontinuities in fluid mobilities
With discontinuities in the fluid mobilities caused by the presence of
a cold-flooded, a warm-flooded and an oil zone the pressure profile can be
approximated by Eq. (2.58).
The corresponding poro-elastic stresses can again be calculated
analytically. Leaving the notation for the coordinates dimensional for
convenience the result is given by:

%1D = <r>'ASHPD

+ A

eiD(r)-H(f - M>

57

ho = 10.
h D = 0.1
h 0 = 0.001

_ ^

"O

0.2

0.4

0.6

0.8

_J_
1.0

1.2

1.4

16

18

2.0

2Z/h
DIMENSIONLESS PORO-ELASTIC STRESS CHANGE AT WELLBORE

FIG. 2.12

- 58 -

r
Aa

0p2D

4S

" T '

HpD

+ A >

t 2D ( "- H ( 2 " 1ZD


-1)}
2

^ Q ( ^ r ) J (^ - ^ ) Q(.f, V
X

R < r < R
c - F
(2.92)

Sp3D

AS

>

HpD

+ A

P 3 D ( r ) ' H ( I - M>

\ -<'!" <- r * <^2-'l ^1D<V * i ' K ' r


X
! Rp 2 X !
l ,
l Xl
+ T
- l4 TX.
4 (v r ) . (' Xr3- - T~)>
X
2

-,
4

R < r < R
F e

% 4 D " <r>

0(^0^)

i^ o , 4v
r

AS

HPD

(z^,r){-^(^)

R 2 Xn

Xn

4 'r

( ^ ) 2 (^Ap 1D (r w ) + )

X,
2

r > R
e

1)

- 59 -

where
2ffhX
AS

AS

HpD

qA

Hp

4 A p iD ( r w J

N(Z

' I ' V ~ Q(Z 't' r w 5

(1

" xj>

<Z'I'V
(2.93)

(
h

+
- < 'I'V
xj " xj> <"'f'V
4 xj
Q(z,-,r) = asinh + asinh

bo.

2irhX
-r. ~
baQ .
0piD
qA
0pi

i = 1,2,3,4.

and the other functions defined in (2.58), (2.31) and (2.30).


Again, at the wellbore we find:
A

oa

,(r ) = Ap,(r ).H(^ - Izl) + 2AS ^


r
plD w
lD w
2
' '
HpD

(2.94)

2.6 FRACTURE INITIATION PRESSURE

In the following we assume:


1) A vertical wellbore coinciding with the vertical in-situ principal stress
axis.
2) An open hole.
3) Unrestricted flow between wellbore and formation, i.e. no mud cake.
4) The least in-situ tectonic stress is in the horizontal plane.
Under these assumptions a vertical fracture is initiated at the wellbore if
the effective tangential stress becomes tensile and satisfies:

- *;"<v p<v - w
where a

* T

is the tensile strength of the rock.

(2 95)

- 60 -

General expression for fracture initiation pressure


To evaluate (2.95) we write:
p(r ) = Ap(r ) + p.
w
w
l

*A W

W
A

(2 96)

8i

v v = V ( r w> + A v r w >

'

ST ( V

with p. the initial reservoir pressure and a. the initial tangential stress
around the wellbore at pressure p. and initial reservoir temperature T
r
"i
res

From Sections 2.2, 2.4 and 2.5 we have:


AffflmU fZ) = AmAT(r ) + 2AS m (z)
0T w
T
w
HT
A. (r , z ) = A Ap(r ) + 2AS ( z )
op w
p
w
Hp

(a)

|z| < J
' '
2

Aa = - Ap(r w )

(b)

(2.97)

(c)

where we have assumed for simplicity that the interval over which the
wellbore is radially loaded by Ap is sufficiently long for (2.97c) to hold
(see Section 2.2.7).
Substitution of (2.96) and (2.97) into (2.95) gives for the fracture
initiation pressure:

, , = 0i + ( 1 - V^P i
pfi(z)

V T * 2AS HP * 2AS HT + T
J^TJ

,|z|
, <h -

(2.98)

where p f . depends on z because the stresses at the wellbore depend on z,


Fracturing occurs if the wellbore pressure exceeds the fracture
initiation pressure:

P(r ) > Pf,(z)


w

11

Two situations can now be distinguished.

(2.99)

- 61 -

I. Sudden change to new injection rate or temperature


injection took place at a constant rate q and a constant temperature AT for
a certain time t. As a consequence, there is an apparent change in far field
stresses by an amount AS

+ AS

. If the injection rate is now increased,

at what pressure will fracturing occur? To answer this question we have to


take a closer look at (2.97). Eq. (2.97b) consists of a rapidly varying part
A Ap(r ) and a slowly varying part 2AS

. That is, if Ap is changed into Ap'

without the pressure profile between r and R

being changed considerably,

AS will remain approximately constant but A Ap will become A Ap'. For


Hp
p
p
practical purposes therefore (2.98) and (2.99) can be used as a fracturing
criterion when changing to a new rate with AS

held constant and evaluated

at the time just before this change is made. ASm and AS can be calculated
HT
Hp
from the expressions in Sections 2.4 and 2.5. The same reasoning applies
when a change is made to a new injection temperature so that AT + AT'. Eq.
(2.98) can be used with the term AAT replaced by AAT* but with AS
T
T
HT
evaluated just before the change and at the old injection temperature. In
time, of course, AS

and As

will be influenced by the new situation so

that their values have to be redetermined.


II. Injection rate and temperature remain unchanged
Injection took place and continues to take place at the constant rate q and
the constant temperature AT. Consequently, AS

and AS

have changed and

continue to change slowly with time. If fracturing has not yet occurred,
will it occur later and at what time?
To answer this question an expression for p(r ,t) has to be entered on
the left-hand side of (2.99). The expressions for AS

and As

from

Sections 2.4 and 2.5 have to be entered into p. and (2.99) has to be solved
for the time t, at which the equality sign holds. Since p(r ), AS

and AS m

depend on time in a non-linear way this must be done with a Newton iteration
procedure. If no zero is found, future fracturing will not occur at the
rate q.

Evaluation of initial tangential stress


If the initial tectonic stresses do not vary considerably over the
height of the reservoir, oa.

may be determined by solving the homogeneous

equations of elasticity under horizontal plane strain conditions. Assuming


the ideal case of a vertical well penetrating the rock that is in an
isotropic state of initial horizontal tectonic stress S. and with the fluid
Hi

- 62 -

in the well at the same pressure and temperature as the reservoir, we have,
using the solution (2-E-7) from Appendix 2-E with C = S . and
1
2
Hi
C

= r

(P

ei

rSHi):
= 2S

Hi " p i

(2

'100)

Substitution into (2.98) gives:


2(S
p

fi

Hi

+ AS

Hp

+ AS

HT) - V i * V

r^j-

+ T

<2-l0l>

If the penetration depths of both the pressure and temperature


variations are small with respect to the reservoir height, we have
AS = ASm = 0 and
Hp
HT
2S
p

fi

Hi - V i * V
V ^

* T
(2 102)

P
For very large penetration depths As
= A AT and AS
= A Ap.
HT
2 T
Hp
2 p
Substitution into (2.101) gives:
2SIT. + A Ap - A p. + 2AmAT + om
P
For arbitrary penetration depths AS

and AS

can be calculated from

the expressions in Sections 2.4 and 2.5.


From these expressions the vertical variation of the fracture
initiation pressure can also be evaluated. If, for instance, the reduction
in stress because of cooling dominates the increase of stress because of the
rise in pore pressure, it can be inferred from Eq. (2.70) that for small a
penetration depth of the temperature front fracturing occurs first near the
top and bottom of the reservoir.
If fracturing is to be prevented at all times an upper limit for the
wellbore pressure can be derived from the minimum of (2.102) and (2.103). In
the case of cooling this becomes:
, . ,

P(r ) < P
w

min)

fl

2S. - A p. + 2A,AT + o m

- ~Si
.

T
A

2 -A
P

(2.104)

- 63 -

For horizontal tectonic stresses of unequal magnitude a. is given


by :

0i

- Shi

+ S

Hi

+ 2(S

hi " SHi>

COS29

(2

" ^

-105>

with SL. the least compressive tectonic stress, S. the intermediate


hi

Hi

tectonic stress and 9 measured from the direction of S..


Hi

The fracture initiation pressure is obtained by substituting the


minimum of (2.105) into (2.98) which results in:
3S. . - S. + 2AS + 2AStim - A p. + A AT + om
_hl
Hi
HP
HT
i
T
T
P

(2<106)

The effect of a casing


In the above analysis the presence of a casing is not considered. If a
casing is present and it is assumed to be rigid so that no load can be
transferred from the wellbore to the formation, the contribution of (2.97c)
must be omitted. The initial stress a. is now determined by the effect of
the cement on the formation. If, for simplicity, we assume that there was no
fluid loss during cementing and that the cement has set without shrinkage or
expansion the analog of (2.100) becomes
oe.
with p

- 2SH. - p s

(2.107)

the pressure of the cement slurry during setting..

The cement exerts a radial compressive stress a

- p

on the

formation. During injection however the pore pressure adjacent to the cement
rises and the point may be reached where the net radial stress exerted on
the cement bond becomes zero or negative and separation of cement and
formation is induced.
If this occurs open hole fracturing conditions again apply. For deeper
reservoirs where vertical fractures can be induced separation of cement and
formation is likely to occur before fracturing of the formation. Thus Eq.
(2.98) can still be used to calculate the pressure at which the formation
parts.

- 64 -

Estimate of horizontal tectonic stress


Most of the time no information on the horizontal and vertical
tectonic stresses is available. If, however, the reservoir is known to be
situated in a tectonically relaxed area, the vertical stress should
approximately equal the pressure of the overburden. Furthermore, the
horizontal stresses should be equal in magnitude. Assuming no lateral
strains we have from Eq. (2-A-l) with e

CS

Hi " (SHi +

Vi)] " a

A =

= e

= 0:

(2

-108)

where S. is the initial vertical tectonic stress.


Vi
Solving for S. gives:
Hi
s. = r 2 - s. + A AT
HI
1-
Vi
(2.109)
= 7^- S. + A p. + Am(T
- T, )
1-y Vi
pi
T res
dep
where T

is the temperature that prevailed during deposition of the

reservoir.
With estimates for T, , u and with S. equal to the pressure of the
dep
Vi
overburden the magnitude of the horizontal stresses can be approximated with
Eq. (2.109). It should be noted however, that Eq. (2.109) may give wrong
answers even in tectonically relaxed areas. The reason is that the burial
history of the reservoir and the associated change in elastic constants
during geologic time is not taken into account. The preferred way of
determining S . is through in-situ stress measurements by creating
microfractures. A more extensive discussion can be found in Chapter 5,
section 5.5.

2.7 EVALUATION OF TWO FIELD CASES


The change in tangential stress has been evaluated for two realistic
field cases. Both cases deal with waterflooding in which the temperature of
the injection water is lower than the original reservoir temperature. The
input data for the calculations are given in Table 2-II.

- 65 -

TABLE 2-1I - INPUT DATA

Injection rate, m /d

Sandstone

Limestone

reservoir

reservoir

8000

60

Time of injection, days

730

730

Reservoir height, m

120

50

2
-15
Effective permeability to water, m *10

250

1.0

1000

4.0

Cold water viscosity, mPa.s

1.0

0.7

Warm

0.3

0.4

0.3

2.6

Connate water saturation

0.12

0.30

Residual oil saturation

0.25

0.25

Porosity

0.24

0.24

-1
-4
Total compressibility, (bar) *10

0.86

0.50

Heat capacity of formation

, kj/m C

2100

2100

"

"
"

"

Oil viscosity

to oil

,
,

"

"

"

of injection water

"

4200

4200

"

"

of cap and base rock,

"

2100

2100

Conductivity of cap and base rock, W/m. C

2.5

2.5

Temperature change at reservoir entry, C

-70

-30

A T , bar/C

1.0

0.9

A , bar/bar
P

0.5

0.4

- 66 -

High-permeability sandstone reservoir


The first case is a high-permeability sandstone reservoir into which
cold water is injected at a temperature 70 C below the reservoir
temperature. The change in tangential stress was first calculated
numerically using the method given in Appendix 2-F. The temperature
distribution is calculated from Lauwerier's solution and the pressure from
expression (2.56).
It should be borne in mind that the numerical method calculates the
particular solution to the tangential stress and therefore the boundary
conditions at the wellbore are not incorporated in this method. The
numerical results are therefore applicable only to the region where
r

w 2
(7V
1.
The results after two years of injection are shown in Figs. 2.13.a2.13c. As the reservoir has good permeability the change in pressure and the
corresponding poro-elastic stress increase are small. The thermo-elastic
stress reduction induced by the large degree of cooling is clearly
dominating. Because of conduction, cooling and a corresponding lowering of
stress have also occurred in cap and base rock. Taking the initial all-round
horizontal stress to be S. = 500 bar, the initial reservoir pressure p. =
Hi
1
450 bar and a tensile strength of zero, we find from (2.101) that at the
beginning of injection (AS
= 0, AS
= 0) fracturing occurs at a
HT
HP
bottomhole pressure of 470 bar. If the penetration depth of the cold front
becomes large with respect to the reservoir height, AS

* - A AT and

fracturing occurs at a bottomhole pressure of 423 bar (AS


- 0). Since this
Hp
pressure is lower than the initial reservoir pressure, fracturing is bound
to occur at some point even when it does not occur at the beginning of
injection. To determine the time when fracturing occurs the analytical
expressions for Ap , AS

and AS

have to be inserted into (2.101) and

(2.101) has to be solved for the fracturing time using a Newton iteration
procedure.
As can be seen from Fig. 13c, the injection of cold water has resulted
in a steep stress gradient at the vertical boundaries of the cooled zone.
Therefore, a fracture is likely to remain vertically contained within this
zone. The figure shows furthermore that also at the radial boundary of the
cooled zone a steep stress barrier has been induced.
The thermo-elastic stress reduction was also calculated using the
assumption of plane-strain. It is clear from fig. 2.13b that the assumption
of plane strain seriously underestimates the magnitude of the stress

67

80.0

20.0

60.0 -

15.0

40.0 -

10.0

20.0

5.0

0.0

-20.0

-5.0

-40.0

-10.0

-60.0

- -15.0

-80.0
0.0

O
JQ

20.0

40.0

60.0

80.0

100.0

120.0

140.0

160.0

180.0

200.0

220.0

-20.0
240.0

RADIAL DISTANCE FROM THE WELLBORE (m)


FIG. 2.13a

SANDSTONE RESERVOIR

80.0

20.0

60.0 -

15.0

40.0 -

10.0

20.0 -

5.0
O

0.0

-20.0

-5.0

-40.0

-10.0

-60.0 -

- -15.0

-80.0
0.0

20.0

40.0

60.0

80.0

100.0

120.0

140.0

160.0

180.0

RADIAL DISTANCE FROM THE WELLBORE (m)


FIG. 2.13b

SANDSTONE RESERVOIR, VERTICAL CENTRE PLANE

200.0

220.0

-20.0
240.0

<

68 40.0

-i

40.0

r
AT

20.0

20.0

0.0

0.0

-20.0

-40.0

-40.0

-60.0 -

-60.0

-80.0
0.0

20.0

40.0

60.0

FIG. 2.13c

80.0

100.0

120.0

140.0

160.0

180.0

_L_
200.0

_1_
220.0

-20.0

-80.0
240.0

VERTICAL DISTANCE FROM THE CENTRE


OF THE RESERVOIR (m)
SANDSTONE RESERVOIR, 20 m FROM THE WELL

40.0
ANALYTICAL
NUMERICAL
20.0

0.0

-20.0

-40.0

-60.0

-80.0
0.0

FIG. 2.14a

20.0

40.0

60.0

80.0

100.0

120.0

140.0

160.0

180.0

_I_
200.0

220.0

VERTICAL DISTANCE FROM THE CENTRE


OF THE RESERVOIR (m)
SANDSTONE RESERVOIR.COMPARISON OF ANALYTICAL AND NUMERICAL RESULTS.

240.0

O
O

69
80.0

I"

-T

r ' - r - -1

ANALYTICAL
NUMERICAL

60.0 -

40.0

20.0 -

0.0

-20.0

if

-40.0

-60.0

11
11

-80.0
0.0

20.0

40.0

60.0

80.0

100.0

120.0

-.. i

140.0

/1
/ 1
/ 1
/
11
*
1

160.0

180.0

.)
200.0

..1
220.0

240.0

RADIAL DISTANCE FT?OM THE WELLBORE (m)


FIG. 2.14b

SANDSTONE RESERVOIR.COMPARISON OF ANALYTICAL AND NUMERICAL RESULTS.

10.0

--i

-i

"r

9.0 8.0
7.0 -

6.0
5.0 -

4.0
3.0

2.0
1.0 1

0.0
0.0

20.0

40.0

60.0

80.0

100.0

120.0

140.0

160.0

1
180.0

L
200.0

1
220.0

RADIAL DISTANCE FROM THE WELLBORE (m)


FIG. 2.14c

SANDSTONE RESERVOIR.COMPARISON OF ANALYTCAL AND NUMERICAL RESULTS.

240.0

- 70 -

reduction. The reason is that after two years of injection the radius of the
cold front (175 m) is quite large with respect to half the reservoir height
(60 m), giving a rather small dimensionless reservoir height of
h/2R

= 0.34.
c
In Figs. 2.14a-2.14c the numerical calculations are compared with the

analytical calculations of Aa.

and Aofl . The agreement is perfect for the

change in poro-elastic stress. There is a difference for the thermo-elastic


stress because the analytical solution uses a temperature step profile
instead of Lauwerier's solution. Therefore, the zone of stress decrease in
cap and base rock, resulting from conduction, is not properly represented.
Overall the agreement is very satisfactory.
Low-permeability limestone reservoir
The second case is a low-permeability limestone reservoir into which
cold water is injected at a temperature 30 C below the reservoir
temperature. The particular solution for the change in tangential stress was
calculated numerically. The results after two years of injection are shown
in Figs. 2.15a-2.15c. The moderate degree of cooling and the sharp rise in
pore pressure due to the low permeability result in a net stress increase of
about 8 bar inside the cooled reservoir region. In common with the highpermeability reservoir, a stress barrier is created at the radial
temperature front. However, because of vertical conduction, the stress in
cap and base rock is now lowered with respect to the reservoir stress.
Depending on the initial stress contrast, the stress in cap and base rock
may become less than the reservoir stress. If the injection well is
fractured, the induced gradients in the horizontal stress could force the
fracture to grow vertically into cap and base rock.
It should be noted that the differences in the two cases above are
mainly related to the different permeabilities, injection rates and
temperature differences. The effect of a different lithology on the elastic
behaviour of the rock is only reflected in the small difference between the
poro- and thermo-elastic constants.

71

120.0

120.0

100.0 \-

100.0

80.0

80.0

60.0 -

60.0

40.0

40.0

20.0

20.0

0.0

0.0

-20.0

-20.0

-40.0

-40.0
0.0

4.0

8.0

12.0

16.0

20.0

24.0

28.0

32.0

36.0

40.0

RADIAL DISTANCE FROM THE WELLBORE ( m )


FIG. 2.15a

40.0

LIMESTONE RESERVOIR

"I

30.0

10.0

20.0

0.0
i

i
i

//
*
-10.0

-20.0

Afftfp
Aa0T
Aff0p+A<T0-j-

-30.0
0.0

4.0

8.0

12.0

16.0

20.0

24.0

28.0

RADIAL DISTANCE FROM THE WELLBORE ( m )


FIG. 2.15b

LIMESTONE RESERVOIR, VERTICAL CENTRE PLANE

32.0

36.0

40.0

72

40.0

30.0

O
XI
V)

(/)
ui

20.0

cc
\
to
10.0

z
ui
o
z
<

0.0

o
z

-10.0

_ 20.0

<

Aff 0 T

Aagp+AffflT
-30.0

0.0

5.0

10.0

1S.0

20.0

25.0

30.0

35.0

40.0

45.0

50.0

55.0

60.0

VERTICAL DISTANCE FROM THE CENTRE


OF THE RESERVOIR (m)
LIMESTONE RESERVOIR, 8 m FROM THE WELL

FIG. 2.15c

- 73 -

2.8 OTHER APPLICATIONS


Although in this chapter the solution for the stress field is applied
to the stress state around a water injection well, other applications are
possible. For instance, it is also possible to use the formulae to calculate
the stress field around a steam injection well. In the latter case the use
of the Marx-Langenheim temperature distribution

is more appropriate than

that of Lauwerier. The stress field around a production well can also be
calculated with these new formulae. Finally, the derived Goodier and Love
potential functions can be used to calculate the displacements rather than
the stresses.

2.9 CONCLUSIONS
1. A simple analytical expression has been derived for the tangential stress
resulting from radial loading of the wellbore over an interval d. The
expression is shown to be valid at least for 2r /d < 0.1 with r the
w
w
wellbore radius. For 2r /d < 0.01 the result becomes equal to the wellw
known plane strain result.
2. Simple analytical expressions have been derived to calculate the change
in stress state of the reservoir resulting from axisymmetric changes in
reservoir pressure and temperature. The new analytical expressions do not
rely on an assumption of plane strain. The boundary conditions at the
injection well are properly incorporated. For realistic pressure and
temperature profiles the new formulae are shown to be accurate for
arbitrary penetration depth and at least for 2r /h < 0.1 with h the
reservoir height. The latter condition is satisfied for most field cases.

3. For 2r /h < 0.01 the new analytical solutions can be derived from
well-known plane strain solutions if the latter are subjected to a
modified boundary condition at infinity. This boundary 'condition may be
interpreted as an apparent change in far-field stress.

- 74 -

4. If the lateral pressure and temperature penetration depths are large with
respect to the reservoir height the poro- and thermo-elastic stress
changes may become twice as large as predicted by the plane strain
solutions.

5. Simple analytical expressions have been derived for the calculation of


the fracture initiation pressure for a vertical fracture. The effects of
poro- and thermo-elastic changes in rock stress have been incorporated.

6. Cooling of the reservoir following the injection of cold water may induce
thermal fracturing. If the radius of the cold front is still small with
respect to the reservoir height fracturing occurs first near the top and
bottom of the reservoir.

7. If the thermo-elastic decrease in reservoir rock stress is dominating


with respect to the poro-elastic increase, steep stress gradients may be
created at the vertical boundaries of the reservoir. Thermal fractures
are therefore likely to remain vertically confined to the reservoir.

8. If the increase in rock stress due to the rise in pore pressure is


dominating with respect to the cooling effect, the pressure at which
fractures can be initiated is increased. If a fracture is initiated, the
induced stress gradients can force the fracture to propagate into cap and
base rock.

- 75 -

LIST OF SYMBOLS

generic elastic constant

poro-elastic constant

thermo-elastic constant

compressibility of bulk reservoir rock

compressibility of rock grains

distance between packers

trace of strain tensor

Young's modulus

reservoir height

heat capacity of fluid-filled reservoir rock

heat capacity of cap and base rock

heat capacity of injection water

radius of cold front

c
R
F
R
e
r
r
w
S
or
S
wc
S.

radius of flood front


effective exterior radius
radial coordinate
wellbore radius
residual oil saturation
connate water saturation
initial maximum horizontal stress

Hi
S, .
hi

initial minimum horizontal stress

S.

initial vertical stress

vertical coordinate

Vi

Greek
a

generic expansion coefficient

linear poro-elastic expansion coefficient

linear thermal expansion coefficient

thermal diffusivity of cap and base rock

6..

Kronecker delta

e. .

strain tensor

i:

Poisson's ratio

AT

generic for combined pressure and temperature change

Ap = p-p.

change in pressure with respect to initial pressure

AT=T-T

change in temperature with respect to initial temperature

Ao. .

change in total stress with respect to initial stress

- 76 -

Aa

trace of stress tensor

tensile strength of reservoir rock

Subscripts
T

generic for combined poro- and thermo-elastic

thermo-elastic

poro-elastic

dimensionless

Superscripts
o

solution to homogeneous equations of elasticity

ps

plane strain

- 77 -

REFERENCES
1. Timoshenko, S.P. & Goodier, J.N., Theory of Elasticity.
McGraw Hill Book Company Inc., New York, 1951.
2. Geertsma, J., Problems of rock mechanics in petroleum production
engineering.
Proc. First Congr. of the Intl. Soc. of Rock Mech., Lisbon 1966,
Vol. I, p. 585.
3. Medlin, W.L. & Masse, L., Laboratory investigation of fracture initiation
pressure and orientation.
SPE-6087, 1976.
4. Rice, J.R. & Cleary, M.P., Some basic stress diffusion problems for
fluid-saturated elastic porous media with compressible constituents.
Reviews of Geophysics and Space Physics, 1_4, No. 2 (May 1976),
p. 227.
5. Perkins, T.K. & Gonzalez, J.A., Changes in earth stresses around a
wellbore caused by radially symmetrical pressure and temperature
gradients.
Soc. Pet. Eng. J. (April 1984), p. 129.
6. Tranter, C.J., On the elastic distortion of a cylindrical hole by a
localised hydrostatic pressure.
Quart. Appl. Math., vol. 4, p. 298, 1946.
7. Kehle, R.O., The determination of tectonic stresses through analysis of
hydraulic well fracturing.
J. Geophys. Res., 1964, vol. 69, p. 259.
8. Hubbert, M.K. & Willis, D.G., Mechanics of hydraulic fracturing.
Trans. AIME, 1957, vol. 210, p. 153.
9. Manual IMSL Library.
10. Prats, M., Thermal recovery.
SPE Monograph Series, vol. 7 (1982), p. 49.
11. Earlougher, R.C., Advances in well test analysis.
SPE Monograph Series, vol. 5 (1977).
12. Nowacki, W., Thermoelasticity.
English edition, Pergamon Press, 1962.
13. Luke, Y.L., Integrals of Bessel functions, p. 317.

- 78 -

APPENDIX 2-A
BASIC EQUATIONS
The linear stress/strain relations for combined poro- and thermoelastic deformations are:
e.. = [(l+u)Ao.. - uAofi. . ] - aAT 6..
i]
E v
l]
i]
i]

(2-A-l)

where Ao. . is the change in total stress with respect to the initial stress
state at AT = 0. The trace Ao is given by:

Ao = Aa..
i-1
"
The combined pressure and temperature effect has been generically denoted
as:

aAT = a Ap + a m AT
p *
T

(2-A-2)

where a m is the linear thermal expansion coefficient and a is the linear


T
^
p
poro-elastic expansion coefficient. The latter is defined as:

. =2
p

(1

- S,

(2.A_3)

c, '
b

(2-A-l) can be inverted into:


Aa.. = r^- [e.. + r^r- e 6..] + 7^7- AT 6..
13
l+o 13 l-2u i]
l-2u
13

(2-A-4)
v

with e = Z e...
i=l X1
Expressing the strain tensor in terms of the displacement vector:
e..=-r{9.u.+3.u.}
i]
2 12
31
with 0. = -
1 ox.
1

(2-A-5)

- 79 -

gives:
3 u + 3 u
3 u 6
+
Aa..
-3- 1+ T^T"
^ k
- ]
13 = - T77I"T
2(l+o) 1 -]
l-2o ^ k
13 T^T"
l-2uAT 8. .13 (2-A-6)

where summation over repeated indices is understood.


Substitution into the equilibrium conditions for the stresses:

3.Aa. . = 0
1 ID
results

(2-A-7)

in:

- ( l - 2 u ) 3,3, u. - 3.3, u, + 2(1+0) a 3.AT = 0


k

(2-A-8)
v

This is a set of three differential equations in the three components of the


displacement vector. A particular solution to (2-A-8) can be found by
12
introducing Goodier's displacement potential :

u. = - 3.0
1

(2-A-9)

Substitution into (2-A-8) gives:


3.(2(1-0) 3. 3.0 + 2(1+0) OAT} = 0

(2-A-10)

which has as a solution:


3, 3, 0 = - 7 ^ aAT
k k
l-o

(2-A-ll)

This equation for the displacement potential is Poisson's equation which has
the well-known particular solution:
<t> = T~ 7 ^ a dx'dy'dz' AT(x',y',z') ^
4ff 1 0

with

R = ((x-x') + (y-y')2 + (z-z')2)

(2-A-12)

- 80 -

The volume dilatation generated by this potential is from (2-A-ll)


e = 3. 3. 0 = - 7 ^ aAT
k k
1-v

(2-A-13)
'

Substitution into (2-A-4) gives as a particular solution for the stresses:


Ao. - = 77- e.. + AAT 6..
ljT
1+u l]
l]

(2-A-14)

where the following generic notation has been introduced:


AAT = A Ap + AmAT
(2-A-15)
pr
T
'
Ea
Ea
p
T
with A = - c , A = - . A and Am are called the rporo- and thermo-elastic
p
1-u
T
1-v
p
T
constants, respectively.
Expressing (2-A-14) in terms of the displacement potential gives:

Ao. - = - 7 - 3.3.0 + AAT 6..


IJT
1+u 1 2
ID

(2-A-16)

From (2-A-2), (2-A-12), (2-A-15) and (2-A-16), the particular solution


consists of two components:

Ao..- = Ao.. + Ao..


13T
ijp
13T
where Ao.. is determined by Ap and Ao.. by AT.
jp
IJT

(2-A-17)

- 81 -

APPENDIX 2-B
THE PARTICULAR STRESS SOLUTION IN CYLINDRICAL COORDINATES
When AT has axial symmetry, i.e.
AT = AT(r,z)

(2-B-l)

the displacement vector has only a radial and a vertical component:

= U

' U2

' U 3

= W

(2-B-2)

which can be expressed in terms of the displacement potential as:

u = - 3 0 ; w
r

= -30
z

(2-B-3)

Using the relationship between displacement components and strains in


12
cylindrical coordinates , we find:
2
e = - 3u = 3 0
r
r
r

2
; c = - 3w = 3 0
z
z
z

6 a = - _ = - 3 0
6
r
r r^

; e
rz

(2-B-4)

=--{3u+3w}
2
z
r
= 330
r z

From (2-A-14) we have for the stresses:


E
2

Ao - = t 3 0r + AAT
rT
l+i) r

E
2

Aa - = - 3 0 + AAT
zT
1+u z
(2-B-5)

Ao,
m
fl
0T

= 7T~
1+u r 3 r0 + AAT ;

AorzT
- = 7 - 3r 37.0
1+v

Poisson's equation for the displacement potential (2-A-ll) becomes in


cylindrical coordinates:
320 + - 3 0 + 320 = - 7 ^ oAT
r
r r^
z
1-u

"

(2-B-6)
v

- 82 -

A solution to (2-B-6) can be obtained with modified Bessel functions in the


form:
j-

0(r,z)

OS

+00

= - dr'
ir
r
w

dz* dk A T ( r ' , z ' )


-o
o

r'cos k(z-z')

(kr')

K (kr)
o
(2-B-7)

CD

+ - dr'
n
r
with m = -
l-o

+00

00

dz'
-e

dk A T ( r ' , z ' ) r '


o

cos k ( z - z ' ) K ( k r ' )


o

(kr)

a.

Using the following properties of the modified Bessel functions:

KQ(X)

= - K^x)
K (x)

K,(x) = - K (x)
1

l'(x) = I (x)
o
l

(2-B-8)
Ix(x)

I.(X) = I (X) -

"1

I (x)K(x) + In(x)K (x) = O

where the prime means differentiation with respect to x, it can be shown by


direct substitution that (2-B-7) satisfies (2-B-6).
From (2-B-5) and (2-B-7), we obtain:
r
+
oo
_
K (kr)
Ao - = - dr' dz' dk AT(r',z')r'k cosk(z-z') I (kr')[K (kr) + ]
r
-oo
0
w

o
+oo
oo
I (kr)
+ - dr' dz' dk AT(r',z')r'k cosk(z-z') K (kr')[I (kr) - ]
r
-oo
o
(2-B-9)

- 83 -

r
+

d r ' d z ' dk A T ( r ' , z ' ) r ' k c o s k ( z - z ' )


ir
r
-o
o
w

Aa- =
&1!

+ ff

K (kr)
I (kr') -^-j
o
kr

I,(kr)
d r ' d z ' dk A T ( r ' , z ' ) r ' k c o s k ( z - z ' ) K ( k r ' ) - i ;
o
kr
r
-o
o
+ AAT(r,z)

rzT

d r

'
r

'

* dz'
-o

'

dk

+00

00

AT(r',z')r'k2sink(z-z')

(2-B-10)

I ( k r ' ) K (kr)

w
OD

(2-B-ll)

d r ' d z ' dk A T ( r ' , z ' ) r ' k


r
-oo
o

sink(z-z')

K (kr') I (kr)

From (2-B-5) and (2-B-6) we have:

Aa = Ao - + Aofl- + Aa - = 2AAT
rT
0T
zT
so that Aa - can be obtained from (2-B-12), (2-B-9) and (2-B-10).

(2-B-12)
*
'

- 84 -

APPENDIX 2-C
THE COMPLETE STRESS SOLUTION IN CYLINDRICAL COORDINATES
For given AT the stresses are determined by the boundary conditions at
the wellbore and at infinity. The wellbore-model and the corresponding
boundary conditions are discussed in Section 2.2.5. The problem is solved by
decomposing the stress tensor as follows:

Ao.. = Aa. - + Ao?.


13
13T
i]

(2-C-l)

where Ao..- is the solution to the "traction-free" wellbore problem and Ao. .
r
13T
13
is the solution to the homogeneous equations of elasticity satisfying
boundary conditions of uniform radial wellbore loading over a limited
interval.
The boundary conditions for the "traction-free" wellbore problem are
given by:
Ao - = 0, Ao - = 0
rT
rzT

r = r , Izl <
w ''

(2-C-2)

lini Ao. - = lim


Ao. - = 0
r-
^T
|z|+
*3 T
We solve problem (2-C-2) by making the decomposition
Ao..- = Ao. .- + Ao?.IJT

IJT

(2-C-3)

IJT

Ao. - is the particular solution obtained from Goodier's displacement


potential as described in Appendix (2-B) and Ao. - is a solution of the
homogeneous equations of elasticity such that (2-C-2) are satisfied.
12
It can be shown
that a solution to the homogeneous equations of
elasticity having axial symmetry can be obtained from a potential function,
the Love function, which satisfies the biharmonic equation:
(32+-3 +32)(32+-3 +32)*=0
r
r r
z
r
r r
z
The corresponding stresses are obtained through:
2
2
o = 3 (i>V - 3 ) *
r
z
r

(2-C-4)

- 85 -

aa = 3 (uV2 - - 3 ) *
B
z
r r

(2-C-5)
v

a
= 3 ((l-u)V2 - 32) $
rz
r
z
with

V 2 = 3 2 + - 3 + 32
r
r r
z

(2-C-6)
v

Since (2-C-2) must be satisfied for arbitrary axisymmetric functions


AT(r,z),
), it is clear that Ao. - also depends
<
on AT(r,z). To isolate this
l jT
dependence it is convenient to write:
00

Ac

+00

- = dr' dz' A T ( r ' , z ' ) AS - ( r , r ' , z , z ' )


-*

00

(a)

(2-C-7)
go

+00

Ao - = dr' dz' A T ( r ' , z ' ) Ao" - ( r , r \ z , z ' )


J

jW

(b)

-"

_00

The problem is solved if Aa. - can be obtained from a Love function such
IJT

that:
Af

rT(Vr''Z'Z,)

+A f

rT (r w' r '' Z ' Z ' ) =


} r = rw, | SB | < (2-C-8)

rzT (rw'r''z'z')

+ A2f

rzT (r w' r '' Z ' Z,)

Mathematically, Ao. - + Ao. .- can be regarded as a Green's function solution


13T
IJT
to the general differential equations of poro- and thermo-elasticity for the
stresses. The Green's function satisfies the boundary conditions (2-C-2) .
Since two equations (2-C-8) have to be satisfied, a Love function is sought
that consists of two independent solutions to (2-C-4). We proceed from
modified Bessel functions of the second kind to ensure that the stresses
vanish as r * <*>. A formal solution to (2-C-4) is:

* In Ref 12 Ao. is called the nucleus of strain solution since it gives


13T
*
the stress at (r,z) as a result of a 'nucleus' of strain induced by a unit
temperature change (AT = 1) at (r',z').

- 86 -

00

$(r,r',z,z') = dk[B(r',k)K (kr) + C(r',k)krKn(kr)]


o
l

^-r
, o

(2-C-9)

Using the properties of modified Bessel functions:

(3
v

+ - 3 ) K (kr) = k 2 K (kr)
r r' o
o

K*(x) = - K,(x)
o
1
K l( x) = - Ko(x) -

(2-C-10)
K (x)
- l r

where the prime means differentiation with respect to x, Eq. (2-C-9) can be
verified by direct substitution into (2-C-4).
From (2-C-5) we obtain:

Ao- = - dk[B(r',k)F (kr) + C(r',k)P (kr)] cosk(z-z')


o
(2-C-ll)
Aff - = + / dk[B(r',k)F,(kr) + C(r',k)F,(kr)] kr sink(z-z')
rzT
j
4
o
with
K (kr)
F.(kr) = K (kr) +
1
o
kr
F 2 (kr) = -(l-2u)KQ(kr) +krK 1 (kr)
(2-C-12)
K (kr)
F

3(kr)

---TT-

K (kr)
F 4 (kr) = 2(1-0)

Q(

kr

and

K (kr)
^ m = ! dk[B(r',k)
+ C(r ' ,k) (l-2u)K (kr)] cosk(z-z')
CTT
kr
o
o

From (2-B-9), (2-B-ll) and (2-C-7a), Aa - and Aa

(2-C-13)

- are determined.

Substitution of these expressions and those of (2-C-ll) into (2-C-8) gives:

- 87 -

V, V ]

B F nv( k r ) + C F ( k r ) = - r ' k K ( k r ' H l ( k r )


l
w'
2
w
ir
o
o
w

kr

w
(2-C-14)
B F3(krw)

A
2
CF4(krw, = - r'k K ^ k r ' )

(kr

Solving for B and C gives:

I {kr

)
l-, w
F(kr )}

V-. V ] F ( k r )

r
B = - r ' k K ( k r ' ) ,, t {
[I (kr )
ir
o
D(kr )
o
w
w

kr
w

w'

kr
w

A
2
1
,
^ ^ ' w '
C = - - r ' k K (kr*) r - { [ I (kr )
] F.(kr )
ir
o
D(kr )
o
w
kr
3
w
w
w
w

w'

l(krw)
F n ( k r )}
kr
1 w

(2-C-15)

with

D(kr ) = F n ( k r ) F . ( k r ) - F . ( k r ) F , ( k r )
w
1
w 4
w
2
w
3 w
= - K ^ ( k r ) + K?(kr ) [ 1 + 2 ( l - u ) ( k r ) " 2 ]
o
w
1
w
w
(2-C-15),

( 2 - C - 1 3 ) and ( 2 - C - 7 a ) r e s u l t i n :

o>
+oo
CD
A a - = - - d r ' d z ' dk A T ( r ' , z ' ) r ' k
T

(2-C-16)

it

-co

>

K (kr' )

cosk(z-z')
v

'

2nt.

(kr ) D(kr )
w
w

(2-C-17)

K(kr)
*

kr

W(kr

}
w

"

( 1 - 2 )

K
0

(kr)^

with
W(kr ) = ( k r ) 2 [ I ( k r )K ( k r ) + I ( k r )K ( k r ) ]
w
w
l w l w
o w o w
+ 2(l-w)[I1(kr
l

)K,(kr
l

) - 1]

From Egs. (2-B-10) and (2-C-17) the tangential stress field for the
'traction-free'

wellbore problem is determined by

(2-C-18)

- 88 -

ST

= Ao

+ Aa

ei

(2-C-19)

The remainder of the problem consists of solving the homogeneous


equations of elasticity for a wellbore subjected to a radial load Ap

over

an interval d.
The boundary conditions are:

Aa

= Ap , Aa
=0
r
w
rz

Aa = 0

, Ao z = 0

r = r ,
w

IzI< r
' '- 2

r - rw,

|z| > f

(2-C-20)

lim Aa.. = lim Aa. . = 0


r- x3
|z|-x 1D
Problem (2-C-20) was solved by Tranter

in a manner analogous to that

used above. Using our notation and taking compressive stresses as positive,
his result for the tangential stress becomes:

AP

W -

*e - ~ -r '

dk F(k z d)

l(krw)

l(kr)

' ' Hifc-y t^ -) " V - - V - -

w
K (kr)

KQ(krw)

kr

w
K

kr

i< w
+ (l-2u)

. KQ(kr)}

(2-C-21)

w
where
d
d
sink ( + z) + sink ( - z)
F(k,z,d) =
(2-C-22)
and D(kr ) as defined in (2-C-16).
w
Finally, the complete solution for the tangential stress field can be
obtained by combining (2-B-10), (2-C-17) and (2-C-21) into:
Loe

= Aae-

+ Aa =_Ao&- + *o-

+ Ao

(2-C-23)

- 89 -

APPENDIX 2-D
ASYMPTOTIC EXPANSIONS OF THE STRESS SOLUTION
Asymptotic expansion for the particular solution
We consider the case that AT is constant over the reservoir height and
zero outside of the reservoir:
AT(r,z) = AT(r) . H(| - |z|)

(2-D-l)

with H the step function as in (2.28).


The integral over z' in the expression (2-B-10) for Ac,- can then be
carried out, resulting in:
K (kr)
Ar
2
l
Ao^- = - J dr' dk r' AT(r') F(k,z,h)k I0(kr') k f
r
o
w

I (kr)
+ - dr' dk r' AT(r') F(k,z,h)k2K (kr*) ,
it
o
kr
r
o

(2-D-2)

+ AAT(r,z)
with

F(k,z,h) =

sin(kz ) + sin(kz_)
;

(2-D-3)

and

z+

h
= - + z

h
; z_ = - - z

, _ _ .,
(2-D-4)

Using the asymptotic expansions:

Io(x) -> 1
}

(2-D-5)

Il(x) - i x
we expand the first term in (2-D-2) to lowest order in r'/h and the second
term to lowest order in r/h which results in:

- 90 -

r
A(7

0T

" ff

dr

r
w

'

S dk

7" A ^ r ' >

(k'z'h) ^ ( k r )
(2-D-6)

as

+ \ - dr' dk r' AT(r') F(k,z,h)k2K (kr') + AAT(r,z)


2
ff
O
r
o

The integrals over k can now be effected, resulting in:

Aa

" 2 ^2 5 d r ' r ' AT(r').N(z, | ,r)


r r
w
(2-D-7)
Z

oo

f / dr'r' AT(r') {_ , ^ ^

zA

/2 +

^2

A A
, }^ ++ AAT(r,z)
z + r ' )

+ r' ) '

'

with

N(z, ^
2 ,r)
' '

z
. 2 A 2,1/2
[z+ + r ]

z
. 2 A 2.1/2
[z_ + r ]
(2-D-8)

z+/r

z_/r

[(z + /r) 2 + 1 ] 1 / 2

[(z_/r)2 + 1 ] 1 / 2

The same analysis can be carried out for Ao - in (2-B-9). Integrating


over z' and expanding to lowest order in r'/h and r/h results in:

rT

2 H S dr'r'
r r
w

A
+4

(r').{N(z, | ,r) + M(z, | ,r)}

dr'r' AT(r') {
j
u u r , ,.
r

+
,2.3/2

(z+r')

}
, 2A
(z

,23/2

+ r' )
(2-D-9)

where

h ,r) =
M(z, -

Z A

V -r
2

S/Z

((z / r ) + 1)
and N defined in (2-D-8).

--

((z / r ) 2 + 1 ) J / 2

(2-D-10)

- 91 -

Asymptotic expansion for Ag.Using (2-D-l) the integral over z' in (2-C-17) can be carried out,
resulting in:

<
">
k K (kr' )
9
Ao- = - J dr' dk AT(r')r' F(k,z,h)
-
r
o
(kr ) D(kr )
w
w
w
K (kr)
( kr
W(krw) - (l-2i>) KQ(kr)}

(2-D-ll)

with W defined in (2-C-18) and F in (2-D-3).


Apart from (2-D-5) we need the additional asymptotic expansions:

Ko(x) * - in - - 7
}

X 1

(2-D-12)

K.(x) - 1
x
where 7 = 0.5772 is Euler's constant.
If the Bessel functions containing ;r

are expanded to lowest order in r /h

we have, using (2-D-5) and (2-D-12)


(kr )
w
,1

v ,i
(krw )
D(krw ,)

2(1-0)

(a)

'
(2-D-13)

W(kr )
w
(kr ) D(kr )
w
w

- - 2 <kr w') 2

(b)

If, furthermore, we expand the Bessel functions containing r to lowest order


in r/h, we can neglect the term with K (kr). Eq. (2-D-ll) then becomes to
lowest order in r /h and r/h:
w
r

Ao- = ^

(^)

"

dr' dk AT(r')r' F(k,z,h) k KQ(kr')


r
o
w

(2-D-14)

- 92 -

Evaluation of the k-integral results in

Aff

0T

= (

r 2
7 i ) ASHT(Z)

(2-D-15)

where

ASH(z) = f dr' AT(r')r' { 2 ' +


r
(zL + r' )
w
+

(z
-

'~ 2 3/2 >


+ r' )

(2-D-16,

Asymptotic expansion for Ao.


If in (2-C-21) we expand the Bessel functions containing r to lowest
order in r /h and if furthermore the term with K (kr) is neglected we have,
w
o
using (2-D-13.a):
A
Ap
r
j!L
Acr = - -

it

dk F(k,z,d)k K,(kr)
1
o
(2-D-17)

Ap

r 2
(T)
.H(, j ,r)

with N as in (2-D-8) but with h replaced by d


At the wellbore (2-C-21) can be expanded while retaining the K

term,

resulting in:
Ap
Aa(r=r ) = - - dk (F(k,z,d)k K (kr )
a
w
IT
1
w
o
Ap
+ rr

ir

1-v

>
r 2 dk F(k,z,d)k2 K (kr )
w
o
w
o
(2-D-18)

Ap

v
+ ;
l-

w
N(Z

d
' 2 ' r w'

w
d
r~ M(z, - ,r )
2 v
2 w'

With M as in (2-D-10) but with h replaced by d.

- 93 -

APPENDIX 2-E
SIMPLIFIED SOLUTION METHOD

According to Section 2.2.7 if 2r/h < 0.01 the function N/2 becomes a
step function and the function M becomes negligibly small. N and M are
defined in (2-D-8) and (2-D-10) respectively. Under this condition and close
to the wellbore (r - r ) we have from (2-D-7) and (2-D-9) for the particular
solution:

Ao

m = T 2 / dr'r' AT(r') + AS HT
-(z
rT
r r
w

|z| < 2
(2-E-l)

1*1 >2

HTV

Aaz = - ^r dr'r' AT(r') + AAT(r) + AS -(z)


2
0T
HT'
r rw

|z| <
(2-E-2)
i

"

AS

W >2

HT(Z)

with AS - independent of r and defined in (2-D-16). We see that for


|z| < h/2 the stresses are equal to their well-known plane strain solutions
(2.18) plus the term AS -.
If AT(r) is zero beyond a certain radius or behaves as r , with o < 0,
from (2-E-l) and (2-E-2) we have:

Aa
rT
lira {
r-*

Ao 0T

AS -(z)
HTV '

|z| < J

(2-E-3)

Therefore, AS - can be interpreted as an apparent all-round change in


far-field stress which has to be incorporated as a boundary condition at
infinity when solving the equations of poro- and thermo-elasticity in plane
strain.
From Section 2.2.7 we have that the stresses induced by the radial
loading of the wellbore also exhibit plane strain behaviour as long as

- 94 -

2r /h < 0.01, where for simplicity the loaded interval is taken as being
equal to the reservoir height (h=d). This suggests that we can solve the
complete stress problem using plane strain solutions only and (2-E-3) as a
boundary condition at infinity. In the following we demonstrate that this is
indeed the case.
The complete solution for the stress field consists of the sum of the
particular solution and a solution to the homogeneous equations of
elasticity.
We now write this as:
Ac. . = Aa. .- + Ao?.
i]

(2-E-4)

ij

13T

If we solve the problem in plane strain and incorporate (2-E-3) we have the
following boundary conditions:
Ao

Ac

= Ap
w

=V H

= 0

" V I2' >2

r
Ao
lim {,
= AS HT
r+ L9

os <

(a)

(b)

z < + oo

(2-E-5)

(c)

Note that because we are solving in plane strain, the boundary condition
Ao

(r ) = 0 is assumed to be automatically satisfied. The plane strain


rz w
solution for Ao. - is 3given by:
J
13T
A r
Ao r - = dr'r' AT(r')
r r

(a)

^am

h
r

M <|

^ dr'r' AT(r') + AAT(r)


r r
w

Aafl- = A a - = 0
T
rT

(b)

|z| > X
' ' 2

(2-E-6)

(c)

and the plane strain solution for Aa.. is given by :


C
Aa r = -j + C 2
r

(2-E-7a)

- 95 -

(2-E-7b)
r
where the constants C

and C- are determined by the boundary conditions.

Boundary condition (2-E-5c) gives

C
2 = A SHT
and application of boundary condition (2-E-5a) to (2-E-6a) and (2-E-7a)
gives:
: = r
1
w

(Ap - AS -)
*w
HT

(2-E-8)

whereas (2-E-5b), (2-E-6c) and (2-E-7a) give:


C_ = - r AS 1
w
HT

(2-E-9)

1*1 > 2

Adding the solutions together we have finally:


r 2
r
ACTr = ()
(Ap
AS
-)
+
^r
dr'r' AT(r')
vr i
v fw
/ + AS
HT/
2
H T
r r
w
r 2
= - ()
AS - + AS k
r '
HT
HT

r 2
r
Ao& = - (-p) (Apw - ASH-) - dr'r' AT(r')
r r
w
+ AAT(r) + AS H -

'|z|
'

< 2

\z\

>~

(2-E-10)

<
(2-E-ll)

r 2
= K( ) AS - + AS t '
HT
HT

|z| > 2

Result (2-E-ll) is exactly the same as (2.41).


From (2-B-12) we have:
Ao - + Aafl- + Aa - = 2AAT(r,z)
rT
0T
zT
Using (2-E-l) and (2-E-2) we obtain:

(2-E-12)

- 96 -

Aaz-

= AAT(r) + ASy-(z)

|z| < |
(2-E-13)

= ^v?(z)

|z|>

where
AS v -(z) = - 2AS H (z)

(2-E-14)

Here, As - can be interpreted as an apparent change in the vertical farfield stress which has also to be incorporated as a boundary condition at
infinity (r -* ) when solving in plane strain.

- 97 -

APPENDIX 2-F
A NUMERICAL METHOD TO EVALUATE Ao.0T
In Appendix 2-A it was shown that the particular solution Aaa- could
0T
be obtained from the displacement potential:
dx'dy'dz' QT(X''Y''z'?

0(x,y,z) = ^

(2-F-l)

with
2
2
2 1/2
R = f(x-x') + (y-y') + (z-z') ] '

(2-F-2)

If aT has axial symmetry (2-F-l) can be transformed into cylindrical


coordinates and the integral over the azimuth can be evaluated leading to:
OO

0(r,z) = f-
Z7T

+00

dz' AT(r',z') {^~)1/2

X K(X)

(2-F-3)

r -<*>

with

X - [
^
- ]
(z-z') + (r+r')

1 / 2

(2-F-4)
K(X) =

da
. 2 .1/2
rn .2
o [1-X sin a]

K is the complete elliptic integral of the first kind.


From (2.15) we have:
La

am = 7T~ - 3 0 + AAT
T
1+v r r

(2-F-5)

Using

M _ EiXi . MM

/?--i

dX

<

X V

2 F 6

>

- 98 -

2
2
where X' = 1-X and E the complete elliptic integral of the second kind,
(2-F-5) becomes:
OP

Aff

0T

A
" Ti

+OP

! dr

' ' dz'


-o

AT

r' 1/2

< r ' ' z ' ) (~)


M K ( X ) - u E(X)} + AAT(r,z)
r

w
(2-F-7)
with
=

2 2
2
r' -r + S(z-z')
2
'2
(r-r') + (z-z')

(2-F-8)

The integrand in (2-F-7) has a singularity in r=r', z=z'. In order to


evaluate (2-F-7) numerically this singularity has to be isolated and
evaluated analytically. For this purpose, we write the solution to Poisson's
equation (2.13) in terms of ordinary Bessel functions:
OB

09

00

#(r,z) = - d z ' dr' dk AT(r',z')r' e


i _
-o
r
o
w

'

' J (kr) J (kr')


o
o

(2-F-9) '

where

z-z'

= z-z'

z > z'
(2-F-10)

= z'-z

z < z'

That (2-F-9) solves Poisson's equation can be verified using:

dk k J (kr) J (kr') =
o o r '
o

5 r-r')
(2-F-ll)

(3

+ - 3 ) J (kr) = - k J (kr)
r r
o
o

From (2-F-5) and (2-F-9) we have:

Aa

ow

A
~ ! dr'
r
w

+
!
dz

2 -k I z-z' I J l ( k r )
' / dk AT(r',z')r'k e '
' -
J (kr') + AAT
-oo
o
(2-F-12)

- 99 -

From (2-F-12) the contribution of the singularity to the integrals in


(2-F-7) can now be written:

6(.,_) = - f

r+e

r_6

dr'

z+e
_
-klz-z'l J l ( k r )
dz' dk AT(r,z')r'k e K | Z z ' -^r
J (kr')
Z

'62
(2-F-13)

If the region of integration is sufficient small so that AT is essentially


constant in this region, the integral over z' can be evaluated.
Using:
CO

dk J,(kr) J (kr') = 1
o
r
o

r' < r

1
2r'

r' = r

=0

(2-F-14)

r' > r

we find:
r

5(

V 2

} =

"

AAT r z

r'

' dr' ~i

( ' ^
r

"el

(2-F-15)
r+e
+ AAT(r,z)
r-e

-ke
dr'r' dkk e
o

J (kr)

J (kr')
Kr
o

and since the remaining integrand is continuous in r' this becomes:


C2
6(e1,e2) = - 2AAT(r,z)

{- - dk e
o

J^k) JQ(k)}

(2-F-16)

From Ref. 13 we have:

oo

I = dk e
o

J (k) J (k) = - \

-^
K (
-^
) + \
/(e. + 4r )
/(e. + 4r )

(2-F-17)

- 100 -

Using
lim {K(X) - In -i-} = 0
2

x +i

"

(2-F-18)

we have:
lim
-*n

I = ; - - ( ; In 16 - x lnx}
2

IT

(2-F-19)

with

x =

/2
2
(2 + )

(2-F-20)

Finally we have to lowest order in e e :

5(.,_) = ^ ~ ~ x In fr

(2-F-21)

For given AT, Ao.- - 8 can be evaluated numerically with a simple


integration scheme. We have used a standard routine from the IMSL library
called Dblin (Ref. 9 ) .

- 101 -

APPENDIX 2-G
SOLUTION FOR THE PRESSURE DISTRIBUTION
We assume that in an infinite reservoir slightly compressible oil is
displaced piston-like by incompressible injection fluid. The flooded zone
consists of two regions of different constant mobility. In the cold region
the mobility is determined by the viscosity of injection fluid at the
injection temperature and in the warm region by the viscosity of injection
fluid at the original reservoir temperature (Fig. 4). This leads to the
following set of differential equations,

- 3 (r3 ) Ap, = 0
r r r l

r < r <R
w
- c

- 3 (r3 ) Ap = 0
r r
r
*2

R < r < R
c - P

- 3 (r3 ) Ap,
= - 3 Ap.
r r
r r3
17 t 3

R < r <
F -

(2-G-l)

and the following boundary and interface conditions:

2hX, r3 Ap, = q
1
r 1
lim Ap

Ap

= Ap

r =r
w

= 0

2
}

X,r3 Ap, = Xrd Ap


1 r *l
2 r v2

r = R
c

(2-G-2)

Ap 2 = Ap 3

= *F

Xr3 Ap_
= Xr3 Ap^
2 r e2
3 r *3
where Ap = p-p^ and Ap.,X., j=l,2,3 are the pressures and mobilities,
3

'

respectively, in the-cold flooded, warm flooded and oil zone and rj = - is


*ct
the hydraulic diffusivity in the oil zone.
The cold front and flood front are given by a heat and a volume balance,

- 102 -

respectively:
M

= C

w t
M~ hjr]
r

1/2

(2-G-3]
R

1
at,l/2
0(1-S -S ) hffJ
or wc

where S
is the residual oil saturation in the flooded zone, S
is the
or
wc
connate water saturation and 0 is the porosity. The other symbols are
defined in Section 2.3.1.
Solutions to (2-G-l) are sought in the form:
Ap

= A lnr + B; Ap, = C lnr + D


(2-G-4)
2

Ap =F

' Ei (" t^> + G

with Ei the exponential integral:


o s
-Ei(-x) = - ds
x

(2-G-5)

The constants F and G must be independent of time.


Applying (2-G-2) we find the following exact solution:
2

Ap
^

2Th Ap.

= j - in r
1

= 1 ln ,

+ j - in r - - 5- exp ()
2
c
3
Fexp 1 1

r2 r-ir3

T"

AP

- 2 X^

6XP

2
, F ,E l (. ,

W-

W> *

-^>

. Ei(- )
'

(2, G 6)r,

Bl(

" 4^>

Note that because R


i s rproportional to t
F
therefore Ap

2
Fv

is of the required form.

1/2

, -
i s a constant and
Arjt

""

- 103 -

CHAPTER THREE
ANALYTICAL MODELLING OF FRACTURE PROPAGATION
CONTENTS
Summary
3.1 Introduction
3.2 Fracture propagation in an infinite reservoir in the absence
of reservoir stress changes
3.2.1 Assumptions
3.2.2 One-dimensional leak-off
3.2.3 Two-dimensional leak-off - Pseudo-radial solution
3.2.4 Two-dimensional leak-off - Elliptical solution
3.3 The effect of poro- and thermo-elastic stress changes on
fracture propagation pressure
3.3.1 Definition of fracture propagation pressure
3.3.2 Analytical calculation of poro-elastic stress
changes at the fracture wall
3.3.3 Numerical calculation of poro-elastic stress changes
at the fracture wall
3.4 Fracture propagation in an infinite reservoir under the
influence of reservoir stress changes
3.4.1 Assumptions
3.4.2 Consistency checks
3.4.3 Two field cases
3.5 Fracture propagation in a pattern flood.
Effect on sweep efficiency
3.5.1 Zero voidage in the absence of reservoir stress changes
3.5.2 Zero voidage with reservoir stress changes
3.5.3 General flooding conditions and the use of a reservoir
simulator
3.6 Conclusions
List of symbols
References

- 104 -

Appendix 3-A Calculation of poro-elastic stresses in elliptical coordinates


Appendix 3-B A numerical method for calculating poroand thermo-elastic stress changes
Appendix 3-C Calculation of thermo-elastic stresses
and of the axes of the elliptical fluid fronts

- 105 -

SUMMARY
Analytical modelling of waterflood-induced fracture propagation is
discussed. A model is presented with a complete two-dimensional description
of fluid leak-off into the reservoir. A dimensionless injection rate is
defined and it is shown that for values of this number smaller than 0.61 the
fracture propagates with pseudo-radial leak-off. This means that the
pressure transients are travelling radially in the plane of the reservoir.
An approximate three-dimensional calculation of poro-elastic stress changes
at the fracture face is performed analytically for a fracture surrounded by
a pseudo-radial pressure profile including elliptical discontinuities in
fluid mobility. The results are compared with numerical calculations and are
shown to be correct for ratios of fracture half-length to reservoir height
smaller than 10. The numerical method can be easily incorporated into
numerical fracture/reservoir simulators such as developed in the past. If a
thermal simulator is used, the same method can be employed to calculate
thermo-elastic stress changes. An analytical model for waterflood-induced
fracture growth with pseudo-radial leak-off under the influence of poro- and
thermo-elastic changes in reservoir stress is presented. Two realistic
examples are given to illustrate the model. The effect of fracture growth on
pattern flood sweep efficiency is discussed. It is shown that for balanced
injection and production, fracture growth with pseudo-radial leak-off does .
not influence sweep efficiency, regardless of fracture orientation.
Analytical methods are presented for calculating the final fracture length
and injection pressure at water breakthrough. Finally, the use of numerical
fracture/reservoir simulators is discussed.

- 106 -

ANALYTICAL MODELLING OF FRACTURE PROPAGATION

3.1 INTRODUCTION
In the previous chapter a study was made of the conditions under which
fractures are created at an injection well. This chapter investigates the
factors that influence the growth of such an induced fracture. Knowledge of
these factors may help us to decide whether fracturing of injection wells is
beneficial or detrimental to the performance of the displacement process in
a given reservoir.
Unlimited fracture growth can have a number of undesirable
consequences. First, if a fracture grows rapidly into the reservoir it will
distort the geometry of the displacement fronts. Depending on the position
of the production wells this may result in premature breakthrough of
injection fluid and consequently in a poor sweeep efficiency of the process.
Second, if a fracture propagates vertically into another reservoir, loss of
injection fluid may occur. Third, the successful application of a tertiary
follow-up process may be impaired by the presence of large fractures.
However, if fracture growth is limited, fracturing offers the advantage
of a large increase in injection capacity of the well without jeopardising
the efficiency of the displacement process. The increased injectivity
permits the reduction of the number of injectors. This can result in a
considerable reduction in well costs. In addition, the favourable
injectivity may obviate the need for expensive equipment to filter the
injection fluid.
Over the last two decades an overwhelming amount of literature has been
published dealing with many different aspects of hydraulically induced
fractures (for a recent review article see Ref. 1). The majority of these
papers, however, deals with fractures that are created with the specific
purpose of stimulating a production well. Such fractures are induced by
pumping a special fracturing fluid at high rates into the formation. The
fluid is designed to minimise leak-off from the fracture into the reservoir
and to act as a carrier for propping material. The proppant is injected to
ensure that the fracture remains open during the production stage of the
well. The available literature chiefly focuses on aspects such as well
productivity, fracture geometry, vertical fracture containment, fracturing
fluid rheology and proppant transport. The leak-off from the fracture into

- 107 -

the formation during pumping is always modelled as one-dimensional


perpendicular to the fracture surface.
Fractures extending from an injection well are normally not created
with specially designed fluids. They are usually induced with the same fluid
that is injected for the purpose of controlling the displacement process in
the reservoir; most of the time this is just water. Quite often, fracturing
occurs unintentionally when the well is brought to its required injection
potential.
Since here the injection rate of the fluid and its viscosity are
usually much lower than in a typical stimulation job, and since no additives
are present to minimise the leak-off, an injection well fracture grows more
slowly than a fracture in a stimulation job. As a result, the leak-off has
an essentially two-dimensional flow pattern in the plane of the reservoir
rather than one-dimensional. Furthermore no special fluid rheology needs to
be considered and the viscous pressure drop along the fracture is
negligible.
In contrast to stimulation job fractures, which retain a fixed length
after the job has been completed, injection well fractures keep propagating
as injection at fracturing conditions continues. As the fracture propagates,
changes in reservoir pressure and temperature may take place. The resulting
change in reservoir rock stress can have a profound influence on fracture
propagation. Furthermore, after some time the influence of nearby production
wells on fluid leak-off becomes important and needs to be taken into
account.
This clearly illustrates that fractures, continuously propagating from
an injection well, have to be modelled differently from stimulation job
fractures.
An important study of the propagation of a fracture induced at a water
2
injection well was presented by Hagoort et al. . They coupled an analytical
fracture model with a numerical reservoir simulator and studied fracture
propagation as a function of reservoir and injection/production conditions.
They found that when the velocity of the leak-off pressure transients
through the reservoir is of the order of the propagation velocity of the
fracture, the leak-off is essentially two-dimensional in the plane of the
reservoir. They also presented a value of dimensionless time below which the
fracture propagates as if an infinite reservoir. At a later stage the
fracture growth is controlled by the voidage replacement in the pattern

- 108 -

element, the fracture becoming stationary for balanced injection and


production.
Hagoort et al. did not consider the effect of fluid mobility ratios
different from unity on fracture propagation and sweep efficiency.
Furthermore, the effect of thermo-elastic stresses on fracture propagation
were not included, while the effect of poro-elastic stresses were included
only in a very simplistic way.
Recently, Perkins and Gonzalez

presented a semi-analytical model of a

fracture extending from a single injection well into an infinite reservoir.


Leak-off was modelled in two dimensions. In addition, thermo-elastic changes
in reservoir rock stress and their effect on fracture propagation pressure
were incorporated. It was shown that cooling of the reservoir rock following
injection of cold water may cause fractures to become very long. In their
work Perkins and Gonzalez did not provide a satisfactory method of
calculating the effect of poro-elastic stress changes on fracture
propagation, nor did they consider the effect of nearby producers on
fracture growth.
4
Hagoort, in his thesis , calculated the poro-elastic stress changes at
the fracture face for an exponentially decaying elliptical pressure profile.
This calculation, however, was performed assuming plane strain conditions.
It was shown in Chapter 2 that the assumption of plane strain is not valid
when the pressure penetration depth becomes of the order of the reservoir
height. Moreover, the pressure profile that he considered is not applicable
to the flow of fluids with different mobilities.
In this chapter we try to fill in some of the gaps discussed above. In
Section 3.2 an analytical model for fracture propagation in an infinite
reservoir in the absence of reservoir stress changes is presented. Depending
on the dimensionless injection rate the leak-off ranges from one-dimensional
to two-dimensional pseudo-radial. In Section 3.3 an analytical calculation
of poro-elastic stress changes at the fracture face is performed. Deviations
from plane strain and effects of different mobilities are incorporated. In
addition, a simple numerical method is presented for calculating poro- and
thermo-elastic changes in reservoir rock stress. A'comparison with the
analytical results is made. In Section 3.4 an analytical model for fracture
propagation in an infinite reservoir under the influence of poro- and
thermo-elastic changes in rock stress is presented. The model is applied to
two realistic field cases. In Section 3.5 the effect of fracture propagation

- 109 -

on sweep efficiency in a pattern flood is considered. The use of a numerical


reservoir simulator and the incorporation of the numerical method of Section
3.3 for calculating changes in rock stress is discussed. Finally, in Section
3.6 the conclusions are presented.

3.2 Fracture propagation in an infinite reservoir in the absence of


reservoir stress changes
3.2.1 Assumptions
In the construction of our simple analytical fracture propagation model
we use the following assumptions:
1. A vertical fracture with a rectangular surface area extends laterally
from a single well in an infinite reservoir. The fracture height is
constant and is the same as the reservoir height. The fracture shape is
elliptical in horizontal cross-sections (Fig. 3.1).
2. The fracture has infinite conductivity, i.e. the fluid pressure drop
along the fracture can be neglected.
3. The total leak-off rate from the fracture into the reservoir equals the
constant injection rate from the well into the fracture. That is, the
rate of change of fracture volume is negligible with respect to the total
leak-off rate.
4. The injection fluid has the same mobility and compressibility as the
reservoir fluid.
5. Rock mechanical changes in fracture propagation pressure are neglected.
Therefore, the fracture propagates with a constant fluid pressure inside
the fracture.
3.2.2 One-dimensional leak-off
In the following we assume that the leak-off is one-dimensional in the
direction perpendicular to the fracture wall (along y-axis, see Fig. 3.1).
The constant pressure solution for the one-dimensional diffusivity equation
is given by:
y
P(yt) = p.+Ap erfc (2)/(T>t))

(3.1)

- 110 -

GEOMETRY OF FRACTURE

FIG. 3.1

- Ill -

where
p.

= initial reservoir pressure

Ap
V

= P f -p. with p
k
0MCfc

fluid pressure inside fracture

= permeability

- viscosity

= total compressibility

4>

= porosity

= time since start of injection

2
erfc = complementary error function (erfc(x) = -j-

2
s
e ds)
x

Applying Darcy 's law to (3.1) gives, for the one-dimensional leak-off per
fracture area:

k 3E
9y y = o

k _Ap_

1_

M /(ff*,)

y/t

i<^,

From assumption 3 we have:

q = q^
where q is the constant injection rate and q

(3.3)
is the total leak-off rate.

By integrating (3.2) over the fracture area and accounting for the time
since leak-off began at a particular distance x along the fracture length,
(3.3) becomes:

4khApL(.t) . 1
~~T, 7 ' 771 TTT d x
uV(irr\)
/[t-r(x)]

(3.4)

where r is the time at which the fracture tip arrived at x.


Suppose the fracture propagates according to:

L = a i/t

(3.5)

- 112 -

x2
with a an, as yet, unknown constant. This gives r(x) = . Substituting this
a
into (3.4) and integrating gives
a

(3

2jrkhAP

'6)

confirming that a is indeed constant and that (3.5) is correct.


Defining the dimensionless quantities:

Ti^y and

q
D

(3 7)

= i^z;

(3.5) and (3.6) give, for the fracture propagation:


L D = H qD

(3.8)

We shall call (3.8) the Carter solution since the above analysis is similar
to that given by Carter in Ref. 5. The difference is that Carter considered
the complete volume balance:
dV
<

3 =^

(3 9)

'

dT"

where V f is the fracture volume.


If we assume a constant fracture half-width w at the wellbore, the
fracture volume is given by:

V f = jrwhL

(3.10)

and from (3.5) and (3.6) the dimensionless rate of change of fracture volume
is given by:
dV.
AV
=
/q = - ^ - ./f1-)
V
/q
fD
dt
4 kAp "ljrt'
-3
Since w is very small (of the order of 10 m), AV

(3 11)
(J-J--U
rapidly becomes much

smaller than unity. Using (3.3) instead of (3.9) is therefore justified.


This means that the fracture propagation model becomes independent of
fracture mechanics except for the value of p , the fracture propagation
pressure. If effects of a critical stress intensity factor or changes in

- 113 -

reservoir rock stress are neglected, this pressure is constant and equals
the in-situ horizontal reservoir rock-stress S . Rock mechanical changes in
propagation pressure will be discussed in Section 3.3.
The assumption of one-dimensional leak-off perpendicular to the
fracture is justified if

3E / / IE 1
3x 3y

(3.12)

where p = p(x,y) is the pore pressure distribution around the fracture.


By using (3.1) and accounting for the time since leak-off began, the
surrounding pressure profile becomes:
p(x,y,t)-pi= Ap erfc (

2/(r?(t-^))
a

L
= Ap erfc(^

y
2)

(3.13)

where (3.7) has been used together with the following definitions:

xD = I

and

yD = J

(3.14)

The pressure penetration front, defined as the line where the pressure
difference (3.13) is 1% of its maximum Ap, is given by the ellipse:

* ( S ' - ^ ' < l -h. 2 >

<>">

Along an equipotential line we have, from (3.12) and (3.13):


x
0 = 7
TD /(l-xD )

with 0 <, a 3.6

C o n d i t i o n (3.12) i s t h e r e f o r e s a t i s f i e d

(3.16)

if:

(3.17)

- 114 -

In Fig. 3.2 the pressure penetration front (3.15) is plotted for


various values of L . Condition (3.17) means that the velocity of fracture
propagation is much greater than the velocity with which the pressure
disturbance travels into the reservoir.
3.2.3 Two-dimensional leak-off -_Pseudo-radial solution
If the fracture propagates much more slowly than the pressure
disturbance, we have

LD

(3.18)

and Carter's model is certainly not valid.


When condition (3.18) is satisfied, it may be assumed that the fracture
behaves quasi-statically. That is, the pressure profile may be determined by
treating L as a constant and making the substitution L * L(t) after the
pressure solution has been found. In addition, the pressure penetration
front may be taken to move radially outward into the reservoir with respect
to the relatively slowly growing fracture. If it is assumed that Eq. (3.3)
is valid, that the injection rate is constant and that the pressure in the
fracture is uniform (infinite conductivity), the pressure in the fracture is
given by the late-time approximation of the transient elliptical flow
6
solution :
_gu_n
P

f" i i 5 k T

,3.0|/(T?t.
ln(

>

,,

ln4

(3 19)

'

This late-time approximation is commonly called the pseudo-radial limit.


By making the substitution L * L(t) and assuming that the fracture
propagates at the constant pressure difference Ap = pf-p., (3.19) can be
solved for L to give:

L = 3.0 exp(- ^ ^

Ap) /(?t)

(3.20)

or with (3.7):
L

= 3.0 exp(-)

(3.21)

- 115 -

1.0
LD=5.
L0=10.
LD=50.
Lo=10O.

LD=L/SQRT(77t)

0.8

0.6

0.4

0.2

0.0
0.0

0.2

0.4

0.6

0.8

1.0

SHAPE OF PRESSURE PENETRATION FRONT

FIG. 3.2

- 116 -

Again L behaves according to Eq. (3.5), but now with a much smaller a. We
shall call (3.21) the 2-D pseudo-radial solution. From (3.10) and (3.20),
the dimensionless rate of change of fracture volume becomes:

AV

fD = f * f '<?> eXP(-^5

(3 22)

3.2.4 Two-dimensional
_ _
_ leak^off - Elligtcal_solution
Hagoort et al. observed that, in an infinite reservoir, L is always
proportional to the square root of time, even in the region intermediate
4
between (3.17) and (3.18). In Ref. 2 and also in Hagoort's thesis , a graph
is presented of L

vs. q . This graph was calculated numerically with a

numerical reservoir simulator coupled to an analytical fracture model. In


the following we give an analytical solution of L

as a function of q

that

also holds in the intermediate region.


7
Gringarten et al. gave the following analytical expression for the
constant rate solution of the pressure in a stationary infinite conductivity
fracture:
A

! J. >. r
, . 0 . 1 3 4 . , , 0 . 8 6 6 . , -_ . ,
D= 2 V ( i r t D ) l e r f < - ~ J l ) + e r f ( _ 7 i ) } - 0 . 0 6 7 Ei(

0 . 0 1 8 0 . . 0 . 7 5 0 .
)-0.433 E i (
)

D
(3.23)

where

Ap
D-

^r12*kh

and

s = It

(3 24)

Prom (3.23) we extract a relation of L

vs. q

in the same way as (3.21) was

obtained from (3.19). Thus, making the following substitutions in (3.23):


Ap^
q
D

and

t * -^D
LD2

(3.25)

we have,
=
a~~
2 L2 *erf(-134 V
q

erf(0.866 LD)}

- 0.067 Ei (- 0.018 L

) " 0.433 Ei(- 0.750 L 2 )

(3.26)

- 117 -

Using the properties of the error function and the exponential integral:
erf(x) - 1
} x 1

(3.27)

Ei(-x) - 0
erf(x) - ^

} x 1
Ei(-x) -* -

Jn(yx)

where 7 = 0.5772, it can be shown from (3.26) that Carter's solution (3.8)
is retrieved for L
retrieved for L

1 and that the 2-D pseudo-radial solution (3.21) is

1.

In Fig. 3.3, Eq. (3.26), which we call the 2-D elliptical solution, is
shown together with the Carter and the 2-D pseudo-radial solution. In Fig.
3.4 the 2-D elliptical solution is compared with the graph in Hagoort's
thesis on p. 132. The agreement is everywhere within 10%.
7
According to Gringarten et al. , the linear flow period is valid for
t < 0.016, which gives, from (3.25), for the range of validity for the
Carter model:
L

D > Tb^?

= 7 9

(3 28)

or from (3.8)

q D > 4.5

(3.29)

Also according to Gringarten et al., the pseudo-radial flow period


begins at t

> 3.0, which gives as the range of validity for the 2-D pseudo-

radial model:

<

7O

= 0 58

(3 30)

or from (3.21)

qD <

0.61

(3.31)

- L18 T

1 I I I II

1I

M I N I

1 l/l I I |J

L D =L/SQRT(77t)
q D =q/x/27rkhAp

2 - D ELLIPTICAL
2-DPSEUOO-RADIA
1-D (CARTER)
' i i iI

.-1

10

' '

10"

10'

I I I 11

10*

qD
FRACTURE PROPAGATION FOR VARIOUS LEAK-OFF MODELS

FIG. 3.3

10

119

1r-irTTrr

_
_
,.
-

i yy

i i i

S'

s/s'

sf'

# / ' #
9

Iff r_

.*
*

#
#

Sf

//

//
//

//
//
II
1
II

--1

10 :r

2-D ELLIPTICAL"
HAGOORT
1-D (CARTER)

~
,-2

10

_L
,-1

10

,
-

i. i_ 1.1 L i 1.1

icP

i i i i

10'

COMPARISON OF ANALYTICAL AND NUMERICAL RESULTS

FIG. 3.4

- 120 -

Fig. 3.3 clearly demonstrates the importance of the way leak-off is


modelled to the prediction of fracture length. For

< 0.61 the correct

leak-off model is 2-D pseudo-radial. If, instead, the 1-D Carter model is
used, a fracture length is predicted that is too large by 86% or more. The
difference becomes rapidly larger as q decreases.

3.3 The effect of poro- and thermo-elastic stress changes on fracture


propagation pressure
3.3.1 Definition of the fracture grogagation gressure
In the following we analyse the factors that influence the fracture
propagation pressure. Again we consider the wedge-shaped fracture of
Fig. 3.1.
If the fluid pressure in the fracture and the horizontal stress in the
reservoir rock are uniform over the fracture surface, the propagation
pressure is given by :

pf - SH =

(3 32)

fer

with S the in-situ horizontal total rock stress and K


the critical stress
H
XC
intensity factor.
The stress intensity factor, K , is a quantity that characterises the
stress behaviour near the fracture tip. For the conditions given above, it
is given by K = (p,-S ) /(JTL). The fluid pressure inside the fracture
I
r H
provides enough energy for the fracture to propagate if this stress
intensity factor reaches the critical value K

.K

is a material constant.

This stress energy criterion is expressed in Eq. (3.32).


-1/2 *
K
is typically of the order of 10 bar.m
. Since S is typically
XC
H
several hundred bar, the right-hand side of (3.32) becomes small with
respect to S

for fractures with a length of some 10 metres or more. For

simple models such as in Section 3.2 the fracture propagation pressure can
therefore be approximated by:

* For less brittle rocks K

may be considerably larger. See Ref. 4, p. 86.

- 121 -

Pf = S H

(3.33)

If was shown in Chapter 2 that changes in reservoir temperature or


pressure affect the state of stress of the reservoir rock. When the
reservoir is cooled, the rock tends to contract and a thermo-elastic
decrease of S_ results. Similarly, heating causes S to increase. When the
H
H
reservoir pressure rises the rock matrix tends to expand, resulting in a
poro-elastic increase in S .
H
One of the important results of Chapter 2 is that the variation in
horizontal stress for a given wellbore pressure or temperature depends on
the ratio of the reservoir height and the penetration depth of the pressure
or temperature front. The state of deformation or strain of the reservoir
rock ranges from strain only in a horizontal plane (plane strain) for small
penetration depths to strain only in a vertical direction (vertical strain)
for large penetration depths. For a certain wellbore pressure or temperature
the change in horizontal stress under vertical strain conditions may be
larger than the change in stress under plane strain conditions by as much as
100%.
The effect of poro- and thermo-elastic stress changes on the
propagation pressure (3.32) can be incorporated by substituting:
S = S. + Aa
+ Aa m
H
Hi
yp
yT

(3.34)

where S. is the initial horizontal stress. Aa


and Aa are the poroand
c
Hi
yp
yT
thermo-elastic changes in total stress at the fracture face in a direction
perpendicular to it. By using the principle of superposition it can be shown
(see, for instance, Ref. 8) that substitution of (3.34) into (3.32) is
permissible if Aa
and Aa m are calculated subject to the boundary
yp
yT
condition of zero displacement perpendicular to the plane of the fracture.
3.3.2 Analy_t^cal calculation of_goro-elastic_stress_changes at the_fracture
wall
~
3
Perkins and Gonzalez calculated Aa for an elliptical disc with
r
yT
uniform change in temperature. The elliptical temperature discontinuity was
taken to be confocal with the fracture tips. They provided analytical
expressions that were obtained by curve-fitting to the results of numerical
calculations. They used the same approach for calculating Aa

, i.e. they

- 122 -

assumed that an elliptical area in which the change in pressure is uniform


surrounds the fracture. While a temperature profile with the shape of a step
function in a lateral direction is quite realistic under certain conditions,
such a profile is highly unrealistic for a pressure field created by fluid
flow. In the following we calculate the poro-elastic stress changes at the
fracture face using a more realistic approach.
*
Employing the elliptical coordinate system (Fig. 3.5) :
x = L cosh | COSJJ
(3.35)
y = L sinh sinrj
the following expression for the steady-state pressure profile surrounding
g
an infinite conductivity fracture was derived by Muskat ,

Ap(

*> * i S h

ln

<L coshg + I sinh i>

(3

' 36 >

with a = L cosh { and b = L sinh being the major


and minor axes
J
e
e
e
e
'
respectively of the ellipse bounding the area that is influenced by the
change in pressure. If it is assumed that the pressure penetration front
moves radially outward with respect to the slowly growing fracture, an
effective time-dependent exterior radius may be defined from the late-time
behaviour of the radial constant rate solution (see Chapter 2, section
2.3.2):
Re(t) = 1.5 /(ijt)
By substituting a

= b

(3.37)
= R

into (3.36), the elliptical pressure

distribution surrounding the fracture is then approximated as:


A

m
P(i

'

_3M_ l n
2jrkh

3.0 / W
)
L cosh + L sinh$'

(3.38)

At the fracture face ( 1 = 0 ) the pressure Ap(o) is again given by (3.19).

* The periodic elliptical coordinate TJ is not to be confused with the


hydraulic diffusivity for which we have used the same symbol. The intended
meaning of the symbol should be evident from the context.

- 123 -

4 = 1.5

ELLIPTICAL COORDINATE SYSTEM THAT IS


CONFOCAL WITH FRACTURE TIPS

FIG. 3.5

- 124 -

The corresponding poro-elastic stress changes at the fracture face are


derived in Appendix 3-A. When the assumption of plane strain is made, the
dimensionless stress changes become:
A a ( P ^ = J Ap ro) - 7
ypD
2 *D
4

(3.39a)

Ao(p^
xpD

(3.39b)
'

= J Ap fo) + 7
2 *D ' 4

where
Aa

, .
xp 2)rkh
= .
xpD
A
qu
with an analogous definition for the y-component and

(3.40)

A
Ao

Ap(o) =
r
D

P (

* 2jrkh
qu

(3.41)

The superscript ps denotes plane strain. A

is the poro-elastic constant as

defined in the same appendix.


After some time the pressure penetration depth becomes of the order of
the reservoir height and plane strain conditions no longer apply. In the
following we modify (3.39a) and 3.39b) to take this into account.
In Chapter 2, section 2.2.8 it was shown that, for axisymmetric
variations in pressure or temperature, the deviation from plane strain
conditions could be accounted for by subjecting the plane strain solution to
a modified boundary condition at infinity. This boundary condition
prescribes that at infinity the horizontal stresses become equal to'an
apparent change in far-field stress AS

. If the axisymmetric pressure

distribution is approximated by:


(axial)
R
Ap D (r)
= In -fwith R

(3.42)

given in (3.37), it was shown in Chapter 2, section 2.5 that this

apparent change in far-field stress is given by:

(axial)
AS

where

HPD

(axial)
=2

PD(V

Q (

- I

< V

(3

-43)

- 125 -

Q(u) = 2 asinh j -

(3.44)

with h the reservoir height. The superscript (axial) denotes axisymmetrie.


Equation (3.43) gives the apparent change in far-field stress at the
vertical centre of the reservoir.
The dimensionless steady-state pressure drop between two concentric
circles with radii r and r

is given by:

Apaxial) = l n ^

(3.45)

The dimensionless steady-state pressure drop between two confocal ellipses


with major axes a , a
replacing r

and minor axes b , b

can be obtained from (3.45) by

and r with an "average radius"

a. + b.
r. - - ^ -

i = 1,2

(3.46)

so that we have
a

_llh

(elliptical, =
r

(3.47)

+ b

We propose to obtain the apparent change in far-field stress for the


elliptical case from that for the axial case by the same substitution.
However, for a strongly elliptical exterior radius the apparent change in
far-field stress will be different for the x- and y-direction. Only for a
pseudo-radial pressure profile will the apparent change in far-field stress
be approximately the same in both directions. This means that the proposed
method only applies when a = b = R . Therefore, R in (3.43) remains
e
e
e
e
unchanged. Considering the fracture to be an ellipse with a minor axis of
zero we have from (3.46):
rw-

(3.48)

By making this substitution in (3.43) the apparent change in far-field


stress for the pseudo-radial case becomes:

- 126 -

AS

HpD

2 APD(0)

4Q

Q(

(3

f>

'49)

with
2R
ApD(o) = In - ^

(3.50)

As is discussed in Appendix 3-A the additional boundary condition at


infinity is satisfied if AS
is simply added to (3.39). The modified
HpD
poro-elastic stress changes at the fracture face therefore become:
Aa

yPD

Ao

2 APD(0) - + A S H P D

(3

= T; Ap(o) + 7 + AS
xpD
2 *DV
4
HpD

w i t h ASIT from

'51a)

(3.51b)

(3.49)

HpD

Using asinh(x) = ln(x + /(x + 1)) we can show that taking the plane
strain limit of (3.49) results in
limAS

HPD=

(3

'52)

2R
e
so that in this limit the plane strain solution (3.39) is indeed recovered
from (3.51).
The stress changes can also be calculated in the presence of a zone of
cold injection fluid, a zone of warm injection fluid and an oil zone. If it
is assumed that the fronts separating these zones are ellipses confocal with
the fracture tips and that the fluid mobilities are constant within each
zone, the generalisation of (3.38) becomes:
a + b
2irh A , v%
1 . ,
c
c
,
Ap U ) = l n (
g-iut)
q
l
K
L cosht + L sinh

, ,
(a>

F
(3.53)

27rh .

,..,

Ap2 a) - r me

F *

127

L coshi + L s i n h | )

. + _,_ 1

2
2jrh .

. ,v

1 .

,3.0i/(T?t),

r in( y+^ >)


F

,.v

(b)

3.0 (nt)

,%

Ap3({) = - ln( L cosh J ^ s J n h )

(c)

with Ap.,X., i=l,2,3 the pressure change and fluid mobility in the cold
flooded, warm flooded and oil zone respectively, TJ is the hydraulic
diffusivity1 in the oil zone, a

and b _ are the major


and minor axes of
J
c,F
c,F
the elliptical cold and flood front, respectively.
It is shown in Appendix 3-A that the corresponding stress changes in

plane strain become:


<P }
AoypD
! - }
Pin (o) - R
2 A*1D

(a)
(3.54)

(P

A o ^ = \ Ap fo) + R
xpD
2 'ID

(b)

where
e
R =

1
2

c
(
l+e '
c

, X,
e
e
1 J.
F
c
2X 'l+e
" l + e 1
2
F
c

2 X3 ( 2

(3.55)

i + ep}

and

'F

c,F
(3.56)

Aa

y,xpD

Ao
T^E
A

2ffXnh
L_ . A
q

Ap,2jrXnh
L_
= _1
ID
q

In Chapter 2, section 2.5 it was shown that the apparent change in farfield stress for the axisymmetric equivalent of (3.53), i.e. with circular
zones of different constant mobility, is given by:

- 128 -

(axial)

i,

= i Ayp ( a x i a l ) V(r 1 ) - ^uvQ(r )


2 lD
w
4
w'

AS
HpD

(3.57)
X

+ 4 (i - T2) Q(c> 4 (^ " xj) Q(RP) 4 ^


with R

defined in (3.37), Q in (3.44) and R ,R

Q(Re)

the radius of the cold and

flood front respectively. Following (3.46) we make the substitutions:

+b
R .-> C ^ F , CfF
c,F
2
Ap

(3.58)
'

DlXial,(rw) " ApDl(0)

fr

(3

'53a)

The modified poro-elastic stress changes at the fracture face become:


Ac

= - Apn Vfo) - R + AS
ypD
2 ^1D '
HpD
(3.59)

Aa , = - Ap,
fo) + R + AS
xpD
2 ell>
HpD
with

AS

HPD " \

AP

1D(0) " \

Q(

2>

(1

" xj> Q ( ! T - ^ )
(3.60)

1
+

I X" ' X^

F * bF

< ~ > I X^ Q < V

Again, in the limit that h/2R


e

* , we have AS * 0 as it should be.


HpD

- 129 -

3.3.3 Numerical alculation_of_goro-elastic stress_changes_at the fracture


wall
We have developed a very simple numerical method for verifying the
analytical expressions for the poro-elastic stress changes at the fracture
wall. The method consists of dividing the reservoir into Cartesian grid
blocks with a constant pressure within each block. Analytical expressions
can be derived for the stresses exerted by an individual block. The stresses
at the fracture face are calculated by summing the contributions of all
blocks. The method is outlined in Appendix 3-B. In Table 3-1 the results of
numerical calculations are presented together with those of analytical
calculations from expressions (3.39) and (3.51). The range of dimensionless
reservoir heights considered is 0.01 h S 1.0 with h
values of h

= h/2R . For larger

the plane strain expressions are sufficiently accurate.

It is shown that the analytical results for Ao are within 1% and for
ypD
Ao ^ within 5% of the numerical results for L/h 1.0. For L/h > 1.0 the
xpD
analytical results become less accurate, the difference with the numerical
results being 10% for Ao

_ and 7% for Ao n when L/h = 10 and Ir = 0.01.


xpD
ypD
D

The reason for this loss of accuracy lies in the procedure by which the
analytical expressions (3.51) were obtained from the plane strain
expressions. This procedure was originally developed for the axisymmetric
case in Chapter 2, section 2.2.8 and was shown to be valid for r /h < 0.01.
It is therefore to be expected that the analytical results become less
accurate for larger values of L/h, but it is quite surprising that they are
good within 10% for L/h up to as much as 10. In line with expectations the
plane strain results increasingly underestimate the stress changes with
decreasing3 h. For li = 0.01 and L/h = 1.0 the correct result for Ao is
D
D
ypD
1.9 times larger than the plane strain result.
We conclude that for practical purposes the analytical formulae for
Ao

and Ao are sufficiently accurate provided that L/h ^ 10..


xpD
ypD
Using the same numerical method as above we have checked the results
3
obtained by Perkins and Gonzalez for Ao and Ao induced by an
1
J
xT
yT
elliptical inclusion of uniform change in temperature.
They provided analytical expressions for Ao

and Ao

(see Appendix

3-C) that were obtained by curve-fitting to the results of numerical


calculations. We found that our numerical calculations were everywhere
within 10% of the results from their empirical equations.

130

TABLE 3 - 1 - CALCULATION OF PORO-ELASTIC STRESSES AT FRACTURE WALL


COMPARISON OF NUMERICAL AND ANALYTICAL RESULS

Num.

h
2R
e

0.01

L/h

1.0

Anal.

Num.

Anal.

Anal.

xpD

Ao<P^
xpD

ypD

ypD

Aa(P^
ypD

ApD(o)

0.01

6.82

6.82

4.85

6.32

6.32

4.35

9.21

0.1

5.66

5.66

3.70

5.16

5.16

3.20

6.91

1.0

4.32

4.42

2.55

3.93

3.92

2.05

4.60

2.28

2.51

1.40

2.17

2.01

0.901

2.30

0.01

4.56

4.56

3.70

4.06

4.06

3.20

6.91

0.1

3.40

3.41

2.55

2.90

2.91

2.05

4.60

1.0

2.06

2.16

1.40

1.67

1.66

0.901

2.30

0.01

2.65

2.65

2.55

2.15

2.15

2.05

4.60

0.1

1.49

1.49

1.40

0.996

0.994

0.901

2.30

10

0.1

xpD

Anal.

- 131 -

The numerical method used in Ref. 3 differs from the one we used in
that it divides the reservoir into small adjacent cylinders rather than into
small parallelepipeds. The analytical results for the stresses exerted by
one cylinder is known and the calculation is performed by summing over all
contributing cylinders.
Since our method uses Cartesian grid blocks it has the advantage
that it can be easily incorporated into numerical fracture/reservoir
simulators such as developed in Ref. 2 and refined in Ref. 10. The
application for such simulators will be discussed further in the next
section.

3.4 Fracture propagation in an infinite reservoir under the influence of


reservoir stress changes
In the following we construct an analytical fracture propagation
model that incorporates the effect of poro- and thermo-elastic changes in
reservoir rock stress on the fracture propagation pressure.
3.4.1 Assumptions
The first three assumptions are the same as assumptions 1, 2 and 3
in Section 3.2.1 for the propagation model where the effect of reservoir
stress changes are neglected. The following additional assumptions are made.
4. The fracture propagates sufficiently slowly with respect to the velocity
of the pressure transients that the fluid leak-off from the fracture into
the reservoir can be described as 2-b pseudo-radial (Section 3.2.3).
Slightly compressible oil is displaced piston-like by incompressible
injection fluid. The shape'of the fronts that separate the cold-flooded,
the warm-flooded and oil zones are approximated by ellipses that are
confocal with the fracture tips at all times.
6. The poro-elastic stress changes at the fracture face can be calculated
from expressions (3.59). The thermo-elastic stress changes at the
fracture face can be calculated from the expressions given by Perkins and
Gonzalez in Ref. 3 , which were derived under the assumption of an
elliptical inclusion of uniform change in temperature. These expressions
are given in Appendix 3-C for completeness.
7. The fracture propagation pressure is given by Eq. (3.32) with S from
H
Eq. (3.34).

- 132 -

These assumptions enable p. in (3.32) to be calculated from Ap (o)


in (3.53a). S can be calculated from (3.34) with Aa
from (3.59) and Aa
'
H
yp
yT
from the expression in Appendix 3-C. At every time t the major and minor
axes of the elliptical cold and flood fronts can be calculated from a heat
and volume balance respectively as was shown in Ref. 3 and is again repeated
in Appendix 3-C. Substitution of p

and S

into (3.32) then leads to a non

a.

linear algebraic equation for the fracture half-length for every time t.
This equation is solved with a simple Newton iteration procedure.
This model is similar to the one proposed by Perkins and Gonzalez in
Ref. 3. The main difference is in the calculation of the poro-elastic
stresses. Perkins and Gonzalez have calculated Aa
from the formulae for
yp
Aa

in Appendix 3-C after making the substitutions AT * Ap and A

-* A .

They do not state explicity how Ap is defined. This procedure, however,


implies that the fracture is assumed to be surrounded by an elliptical zone
of uniform change in pressure Ap. We have taken the poro-elastic stresses
from the analysis in the previous section. Here the poro-elastic stresses
are calculated from an elliptical pressure field that closely approximates
the actual pressure field corresponding to the assumed flow of fluids.
If the parameters A , Am and K,. are set to zero, and if the
v
p
T
IC
mobilities of the cold and warm injection fluid are taken to be the same as
that of the oil, our model reduces to the one in (3.21) with p = S ..

HI

3.4.2 Consistency checks


The following three checks on the consistency of the model can be
c a r r i e d out.
1. Leak-off model
Assumption 4 in the previous subsection is valid if criterion (3.30) is
satisfied. For the calculation or L

the hydraulic diffusivity of the oil

zone should be used. Alternatively, (3.31) can be inspected; for a


conservative check the smallest of the three fluid mobilities can be
taken to calculate q .
2. Rate^of_chjmge_of_fracture_volume
Assumption 3 in Section 2.3.1 is valid if (3.22) is small, q

and rj can

be calculated as for the first check.


3. Steepness of_temperature front
In the previous chapter, section 2.3.1 it was shown that for axial
symmetry a steep temperature front prevails if:

- 133 -

4a t M
r D = " f " -*j < 0.05
h
M
r

where a

(3.61,

is the thermal diffusivity of the cap and base rock and M , M

are the volumetric heat capacities of the fluid-filled reservoir rock and
of the surrounding formation, respectively.
We define (3.61) also as a criterion in the case of elliptic symmetry.
3.4.3 Two_field_cases
Fracture propagation has been evaluated for two realistic field
cases. Both cases deal with waterflooding in which the temperature of the
injection water is lower than the original reservoir.temperature. The two
cases correspond to the ones considered in Chapter 2, section 2.7. The input
data are given in Table 3-II.
Hicjh-germeabilHy sandstone_reservoir
The first case is a high permeability sandstone reservoir into which
cold water is injected at a temperature 70 C below the reservoir
temperature. The development in time of the bottomhole pressure (BHP), the
fracture half-length and the axes of the fluid fronts are shown in Figs. 3.6
and 3.7. The combined poro- and thermo-elastic stress changes Aa = Aa
+
y
yp
Aa _ and Ao = Aa
+ Aa m are also shown. Since the reservoir has good
3
yT
x
xp
xT
permeability, the change in pressure and the corresponding poro-elastic
increase in stress are small. The large degree of cooling causes an overall
decrease in horizontal stress and results in thermal fracturing. The actual
fracture initiation has not been modelled here. Instead, it was assumed that
a fracture with half-length L = 0.3 m was already present. The fracture
starts propagating when the pressure is high enough for the fracture
propagation criterion (3.32) to be satisfied. As the fracture propagates the
cold front becomes more elliptical and the thermo-elastic change Aa
becomes larger than Aa _. It was discussed in Ref. 3 that this creates the
yT
possibility that at some moment the horizontal stress in the x-direction
becomes lower than that in the y-direction so that the direction of fracture
growth becomes indeterminate.
To check whether the model is consistent we first apply criterion
(3.30). The fracture length after 100 days is 46 m. Taking the hydraulic

- 134 -

TABLE 3-II - INPUT DATA

Injection rate, m /d

Sandstone

Limestone

reservoir

reservoir

8000

60

Time of injection, days

730

730

Initial horizontal reservoir stress, bar

500

220

Initial reservoir pressure, bar

450

160

Reservoir height, m

120

50

2
-15
Effective permeability to water, m *10
oil f

250

1.0

1000

4.0

Cold water viscosity, mPa.s

1.0

0.7

Warm

0.3

0.4

0.3

2.6

Connate water saturation

0.12

0.30

Residual oil saturation

0.25

0.25

Porosity

0.24

0.24

-1
-4
Total compressibility, (bar) *10

0.86

0.50

, kJ/m . C

2100

2100

"

"

Oil viscosity

,
,

"

Heat capacity of formation


"

"

of injection water

, "

4200

4200

"

"

of cap and base rock , "

2100

2100

Conductivity of cap and base rock, W/m. C


o
Temperature change at reservoir entry, C

2.5

2.5

-70

-30

A . bar/ C

1.0

0.9

0.5

0.4

10

10

fp *

A , bar/bar
P
1/2
Critical stress intensity factor, bar.m

135

500

100
INITIAL PRESSURE = 450 bar

490
o

a:
(A

INITIAL HORIZONTAL STRESS = 500 bar

gup
FRAC LENGTH
MAJOR AXIS,COLD FRONT
MINOR AXIS, COLD FRONT

80

INJECTION RATE = 8000 m V d

480

60

470

40 5

CO

Ui
IX.
OL

yo
I

O
O

m 460

20

450

20
FIG. 3.6

40
60
TIME OF INJECTION (days)

80

100

FRACTURE GROWTH IN SANDSTONE RESERVOIR

-100

500
INITIAL PRESSURE = 450 bar

-80

INITIAL HORIZONTAL STRESS = 500 bar

Ac

INJECTION RATE = 8000 m 3 / d

MAJOR AXIS

o
.a
<g - 6 0

400 J,
hZ

MINOR AXIS

300 g

z<
xo

to
:

lo
- 4n 0u
10

200

Ld

100

Ld
CO

-20

20
FIG. 3.7

40
60
TIME OF INJECTION (days)

80

RESERVOIR STRESS CHANGES FOR SANDSTONE RESERVOIR

100

- 136 -

diffusivity in the oil zone gives for L

after 100 days: L

= 0.004, hence

(3.30) is satisfied. As an additional check we can apply (3.31). From


Fig. 3.6 the average bottomhole pressure, p., is 460 bar, thus during
propagation Ap = pf-p. =s 10 bar. For a conservative check, it may be assumed
that the reservoir contains only cold water. This gives q

= 0.49 and 3.0

exp(-l/q ) = 0.39. Assumption 4 is therefore justified. Assuming a typical


fracture half-width w of 10

m, then, from (3.22), the time at which

AV

= 0.01) is 3.4 hours. Assumption 3 is therefore

= 0.01 is given by t(AV

justified. From (3.61) we have that at t = 100 days, T

= 0.003. The

assumption of a sharp temperature front is therefore also consistent.

Low-germeabilitv^ limestone_reservoir
The second field case that we have evaluated is a low permeability
limestone reservoir into which cold water is injected at a temperature 30 C
below the reservoir temperature. The results are shown in Figs. 3.8 and 3.9.
As, for the sandstone reservoir the presence of a fracture with L = 0.3 m was
taken as the initial condition. For the present case, however, the injection
pressure at the beginning of injection is sufficiently high for the fracture
to propagate. The moderate degree of cooling and the sharp rise in pore
pressure results in a net increase in horizontal reservoir stress. As can be
seen from Fig. 3.9, the horizontal stresses initially grow rather quickly in
magnitude. This is because the penetration depth of the pressure front
travels faster than the penetration depth of the cold front and therefore
the poro-elastic increase in stress grows faster than the opposing thermoelastic decrease in stress. Although the reservoir stresses increase with
time, the bottomhole pressure decreases slightly. This is because the
right-hand side of Eq. (3.32) decreases faster than the stress increases.
Again we apply (3.30) to check whether the model is consistent. The
fracture length after 100 days is 7.2 m. Taking the hydraulic diffusivity in
the oil zone gives for L

after 100 days: L

= 0.022, hence (3.30) is

satisfied. Since the mobility of the cold water and the oil are almost the
same we may assume for the calculation of q

that the reservoir contains

only cold water.-From Fig. 3.8 the average bottomhole pressure during
propagation is 228 bar, thus Ap = p-p. = 68 bar. This gives q = 0.22 and
3.0 exp(-l/q )= 0.037. Equation (3.31) is therefore satisfied. From (3.22)
-3
we have t(AV = 0.01) = 1.04 hours if it is asssumed that w = 10 m. At

- 137
234

20
BHP
FRAC LENGTH
MAJOR AXIS.COLD FRONT
MINOR AXIS.COLD FRONT

INITIAL PRESSURE = 160 bar

232

INITIAL HORIZONTAL STRESS = 220 bar

o
INJECTION RATE = 60

XI

a:
V)

(/>
ce
a.

16

nf/d

230

12

Ul

228

tO

CD

226

224

40
60
TIME OF INJECTION (days)
FIG. 3.8

80

100

FRACTURE GROWTH IN UMESTONE RESERVOIR

10

50
INITIAL PRESSURE = 160 bar

iO"\i

INITIAL HORIZONTAL STRESS = 220 bar

40

INJECTION RATE = 6 0 m / d
o
XI
V)

MAJOR AXIS

MINOR AXIS

a
o
o
ei

6 -

ui

oz
<
I
o

(/)

V)

u.
O

to

20

40

60

TIME OF INJECTION (days)


FIG. 3.9

RESERVOIR STRESS CHANGES FOR UMESTONE RESERVOIR

80

100

- 138 -

t = 100 days we have from (3.61) r = 0.017. The model is therefore


consistent under the given conditions.
Monitoring of injectivity_index
For the sandstone reservoir S continuously decreases because of
H
cooling. As a result the injection pressure required to propagate the
fracture decreases from 477 bar initially to 450 bar after 100 days. The
injectivity index (II), defined as q/Ap with Ap = pf-p., correspondingly
3
3
increases from 296 m /d.bar to 1000 m /d.bar, a factor of 3.4.
For the limestone reservoir the combined effect of an increasing S and
a decreasing resistance to fracture propagation (right-hand side of (3.32))
results in an almost constant injection pressure during the 100 days of
propagation. Consequently the II is almost constant over this period with a
slight tendency to increase.
If the injection well is unfractured and the injection fluid has a
lower mobility than the reservoir fluid the II declines continuously. This
is a consequence of more injection pressure being required as the low
mobility fluid bank advances.
It follows that a constant or increasing II is an important indication
of fracture propagation. The II can be monitored at surface by recording the
injection pressure.
Vertical containment
For both field cases it is assumed that the fracture is contained
vertically inside the reservoir. As was pointed out in Chapter 2, section
2.7, cooling creates a steep stress gradient across the boundaries between
the reservoir and the cap and base rock. In the sandstone reservoir, where
the poro-elastic increase in reservoir stress is small compared to the
thermo-elastic decrease, vertical fracture containment is therefore very
likely. For the limestone reservoir, however, the horizontal reservoir
stress increases, whereas the stress in cap and base rock decreases as a
result of conductive cooling. The penetration depth of the conductive
cooling is / (Jra t), which is 5.7 m after 100 days. Therefore, if the
initial stress in the cap and base rock exceeds the initial reservoir stress
by at least Ao , which is about 4 bar, the vertical propagation will be of
the order of 6 m at most, after 100 days.

- 139 -

3.5 FRACTURE PROPAGATION IN A PATTERN FLOOD. EFFECT ON SWEEP EFFICIENCY


3.5.1 Zero voidage in the absence of reservoir stress changes
Calculation of stable fracture length
Initially, the fracture propagates as in an infinite reservoir. During
this stage the fracture propagation depends on the dimensionless injection
rate and on the hydraulic diffusivity in the reservoir (Section 3.2).
After a while, however, the transients of the injection and production
wells will meet and the fracture growth will be mainly determined by the
voidage replacement in the pattern. For unit mobility ratio the fracture
becomes stable if injection and production are in balance (zero voidage).
With a voidage replacement ratio greater than one (overinjection) the
fracture continues to grow whereas with a voidage replacement ratio less
than one (underinjection) it will stop growing and close.
For zero voidage and unit mobility ratio, the injection and production
transients meet at a constant pressure boundary where the pressure remains
at the initial reservoir pressure.
For a confined inverted 5-spot the constant pressure boundary is given
by a square. For a confined inverted 9-spot it can be approximated by a
square, if the corner and side producers have a production rate that equals
one third of the injection rate (Fig. 3-10).
For a non-fractured well the pressure rise in the injector in the
constant pressure square A is given by
^ ^ E

= in 0 . 5 4 ^

(3.62)

with Ap = p -p., p is wellbore pressure and r wellbore radius.


w i
w
w
The producers are surrounded by a square of the same size and the
pressure drop in the producer can also be obtained from (3.62). The total
pressure drop between injector and producer can be found by adding up the
two differences. For a 5-spot the result is exact (see injectivity formulae
in Ref. 12). For a 9-spot a small error is made. The largest error is in the
pressure drop between injector and corner producer (5% for d/r

= 1000).
w

This error arises because the producers are not equidistant from the
injector.

- 140 -

(h

-f)

/ \
/

/d

/
\

/
/
\

X
\

/
/

/
\ /

<

X = INJECTOR

= PRODUCER

CONFINED 5-SP0TAND9-SP0T AND CORRESPONDING


CONSTANT PRESSURE SQUARES.

FIG. 3.10

- 141 -

For a fractured injector we use (3.62) and substitute L/2 for r :

^ ^ E > . ln1>08 f
qu

(3<63)

where /A was replaced by its equivalent, d, the distance between injector


and nearest producer.
We can now solve (3.63) for the stable fracture length at zero voidage
and unit mobility ratio:

J = 1.08 exp (- ZT)


d

(3.64)

where q , the dimensionless injection rate, is defined as in (3.7):

<3 = o , u^A
^D
2ffkh Ap
where Ap = pf-p. and p

(3.65)
v

is the fracture propagation pressure which for

longer fractures, according to Section 3.2, equals S , the horizontal rock


H
stress.
If the fracture propagates with pseudo-radial leak-off before it
reached its stable length, then, according to Section 2.2.3, the pressure
during propagation is given by:
2khA
qu

p^jAzt
L

where 77 is the hydraulic diffusivity.


Comparing (3.66) and (3.63), the fracture reaches its stable length at
a time:
d2
t
= 0.13
ss
r?

(3.67)

where the subscript denotes steady-state.


Effect on sweep efficiericy
According to Section 3.2.3, the fracture propagates in the pseudoradial mode when the dimensionless injection rate q

is smaller than 0.61,

Since the pressure transients are travelling radially into the reservoir

- 142 -

prior to the onset of steady-state, this mode of fracture propagation is not


expected to influence the sweep efficiency.
To prove this we look at the situation for a non-propagating fracture
with a constant length throughout injection.
In Ref. 7 it is shown that for a fracture with constant length pseudoradial flow is reached at time:
t

= 3.0 L2/TJ

(3.68)

where the subscript pr denotes pseudo-radial flow.


According to the conjecture above there is no effect on sweepefficiency if pseudo-radial flow is reached before the onset of steady-state
or if t
< t . From (3.67) and (3.68) we then find:
pr
ss
L/d < 0.21

Hagoort

(3.69)

calculated sweep efficiencies for a 5-spot with a fracture of

constant length and unit mobility ratio. He considered various values of L/d
for two extreme fracture orientations. Pig. 3.11a shows the case for
diagonal fractures (pointing towards a producer) and Fig. 3.11b shows the
case for parallel fractures (pointing in between two producers). Both
figures are reproduced from Hagoort's thesis. The figures show that sweep
efficiency at water breakthrough is not affected, regardless of fracture
orientation, for L/d < 0.25. This is in accordance with (3.69).
For a fracture propagating in the pseudo-radial mode, pseudo-radial
flow is already present almost from the moment of fracture initiation.
Therefore, in any pattern, pseudo-radial flow prevails before the onset of
steady-state. We conclude that for this mode of propagation sweep efficiency
is not influenced by the fracture, regardless of its orientation.
Non-unit mobility ratios
Suppose a low mobility waterbank displaces the oil. If the wells are
producing at a fixed bottomhole pressure, the average reservoir pressure
will rise as the low mobility fluid bank advances. Therefore the fracture
continues to grow, even with zero voidage. It will become stable if the
fluid with the lowest mobility, i.e. the cold water, breaks through in the
producers.

- 143 -

0.8

0.4
3
O

1
3
O

0.2.

0.4
0.6
0.8
1.0
cumulative Injccllon , W|

1.2

1.4

.8

1.6

2.0

SWEEP EFFICIENCY FOR Dl AGON AL-FRACTURED FIVE-SPOT


(after Hagoort)

FIG. 3.11a

144

l:U

Cl ft
\J.O

N
\

o c >U. .

A
ui/

,\
-\

A\
\x

0 4

U.*T

0.2

~" m'

-,; , .

0
1.0
J-*~*

^<^X^'

n o
U. o
ex

0.6

o
u
w
S
.>

fractured

r\ A
0 . 't
L /C

1 = 0 ? ?

f
Lf/c i = 0 . 6 5

E
3

0.2

0.2

0.4

0.6
0.8
1.0
cumulative Injection , VY.

._

1.2

1.4

1.6

1.8

SWEEP EFFICIENCY FOR PARALLEL-FRACTURED FIVE-SPOT


( after Hagoort)

FIG. 3.11b

2.0

- 145 -

The fracture length at (warm) water breakthrough can still be


calculated from (3.64) provided the correct reservoir pressure and
mobilities are used.
Let us denote the fixed bottomhole pressure in.the producers by p . For
the producers to deliver warm water at rate q

at breakthrough the reservoir

pressure has to be (Eq. 3.62):

p = p + - 2 - (f) In 0.54
^
*w
2irh k'ww
r
w

v(3.70)

where q = q for a 5-spot and q = 1/3 q for a 9-spot with equal rates for
P
P
corner and side wells, q is the injection rate. The subscript denotes warm
water.
If at water breakthrough the constant pressure square surrounding the
injector is mainly filled with cold water, the fracture length at
breakthrough can be obtained from (3.64) provided q
9

q^ =

is taken as:

()

(3.71)

2jrh(pf - p)
with p from (3.70) and the subscript cw denoting cold water properties.

3.5.2 Zero voidaqe with reservoir stress changes


As the average reservoir pressure rises from p. at the beginning of
injection to p at water breakthrough, an overall poro-elastic increase in
reservoir stress takes place, given by:

= S

Hi

+ A

(P P

(3

- i>

'72)

with S. the initial horizontal rock stress at reservoir rpressure rp. and A
Hi

the poro-elastic constant.


For longer fractures, the propagation pressure, p f , equals the
horizontal rock stress at the fracture face. Therefore, we have at
breakthrough (compare Section 3.3.1):

p = S + ha
+ Ao _
f
H
yp
yT

(3.73)

- 146 -

with Aa , Aa _ the poro- and thermo-elastic changes in rock stress at the


e
yp
yT
fracture face.
Aa
can be calculated with the methods in Section 3.3.2. It is
yp
approximately given by:
Aa
= c.A .(p.-p)
yp
p
f

(3.74)

where c has a value in the range 0.5 < c < 1.0 depending on the ratio of
h/d with h the reservoir height. In most cases c will be close to 0.5.
Taking c = 0.5, substituting into (3.73) and solving for p , we find:

2S

Pf = ~

+ 2Aa

- A p

2^H

(c = 0>5)

*"

(3.75)

If we assume that the cold-water front has become more or less circular by
the time warm water breaks through, Aa

can be calculated from Eq. (2.67)

in Chapter 2:

Aa
S

A AT
T

= 1 - 7 4 + v( l / h j 2 ] 1 / 2 - 1 / h J
4

' D'

' D

(3.76)

where h = h/2R , R the radius of the cold front, A the thermo-elastic


D
c
c
T
constant and AT the difference between injection and reservoir temperature.
After the propagation pressure at breakthrough has been determined from
(3.75) and (3.76), q

at breakthrough can be determined from (3.71). The

fracture length at breakthrough is then determined from (3.64).


Of course, the above method is only approximate since it assumes that
c = 0.5 and that the cold-water front has become circular. The fracture
length at breakthrough can more accurately be determined using the
analytical fracture propagation model of Section 3.4.1. The input values for
the reservoir pressure and the initial horizontal rock stress should be
taken as p from Eq. (3.70) and S

from Eq. (3.72) respectively. The pressure

penetration depth, R = 1.5/(rjt), should be fixed at i t s steady-state v a l u e .


e
Comparing Eqs. (3.63) and (3.66) this steady-state value is given by
R

= 0.54 d. If the input value of the injection time is taken as the time

that is expected for water to break through in the producers the analytical
model then gives the fracture length at breakthrough.

- 147 -

Of course, the model should be modified such that the calculation of


the pressure rise between the steady-state value for R

and the fracture

takes only the discontinuities in fluid mobility contained with R


account. For example, no pressure rise between R

into

and the flood front R

should be calculated if R _ > R .


F
e
In a similar way the model can be used to determine the maximum
fracture propagation pressure that will occur in the period from fracture
initiation until water breakthrough. This is important for the evaluation of
vertical fracture containment (see Chapter 5, section 5.6.1).
3.5.3 General flooding conditions and the use of a reservoir simulator
To analyse fracture growth and sweep efficiency for larger
dimensionless injection rates or a voidage replacement ratio different from
2
one, a numerical simulator is required. Hagoort et al. were the first to
couple an analytical model of a propagating fracture to a numerical
reservoir simulator. The effect of reservoir stress changes on fracture
propagation were not taken into account except for a very much simplified
way of calculating poro-elastic stresses. The simulator could only handle
fluids with unit mobility ratio. A confined inverted 9-spot was studied
under various injection strategies. It was found that for zero voidage the
fracture becomes stationary for t ~ 0.2 where t

= Jjt/A, TJ is the

hydraulic diffusivity and A the area of a quarter symmetry element. This is


in line with Eq. (3.67).
As was shown in the previous sections reservoir stress changes and
non-unit mobility ratios can have a significant effect on fracture
propagation and need to be accounted for. Regarding the calculation of
reservoir stress changes, the numerical method developed in Appendix 3-B is
of special importance. Since this method uses Cartesian gridblocks, it can
be easily incorporated into fracture/reservoir simulators such as developed
by Hagoort. The same gridblocks that are used to calculate the fluid
pressure can then be used to calculate the poro-elastic stresses. Moreover,
if a thermal reservoir simulator is used, the thermo-elastic stresses can be
calculated with the same method, but replacing Ap with AT and A

with A .

The use of fracture/reservoir simulators is further discussed in


Chapter 5, section 5.8.

- 148 -

3.6 Conclusions
1. An analytical model of fracture growth in an infinite reservoir and in
the absence of reservoir stress changes has been presented. The only rock
mechanical parameter incorporated in this model is the horizontal
reservoir stress. It is shown that leak-off from the fracture into the
reservoir is one-dimensional perpendicular to the fracture face for
dimensionless injection rates greater than 4.5. The leak-off is 2-D
pseudo-radial for dimensionless injection rates smaller than 0.61. In the
latter case a one-dimensional description will lead to an overestimation
of fracture length by a factor of two or more.
2. The three-dimensional poro-elastic stress changes at the fracture face
induced by a 2-D pseudo-radial pressure profile that includes elliptical
discontinuities in fluid mobility have been calculated analytically. For
small ratios of reservoir height to pressure penetration depth the 3-D
stress changes are significantly larger than those calculated under an
assumption of 2-D plane strain.
3. A numerical method has been presented for calculating poro- and thermoelastic changes in reservoir rock stress. A comparison with the
analytical results for the poro-elastic stresses shows that the
analytical results are applicable for ratios of fracture half-length to
reservoir height smaller than 10.
4. Characteristic features of waterflood-induced fracture growth in an
infinite reservoir under the influence of three-dimensional poro- and
thermo-elastic reservoir stress changes have been modelled analytically.
Dimensionless numbers pertaining to leak-off distribution, rate of change
of fracture volume and steepness of the temperature front have been
introduced as a check on the consistency of the model.
5. The injection of cold water into permeable reservoirs can induce
fractures of considerable length. Thermal fracture propagation induced by
cooling can be recognised from a continously increasing injectivity index
of the injector.
6. For balanced injection and production, sweep efficiency at water
breakthrough is not affected by fracture growth for dimensionless
injection rates smaller than 0.61, regardless of fracture orientation. In
the definition of dimensionless injection rate the effects of reservoir
stress changes and non-unit mobility ratios can be taken into account.

- 149 -

7. The numerical method for calculating poro- and thermo-elastic stresses as


presented here can be incorporated into a numerical fracture/reservoir
simulator. For the study of fracture propagation under more general
flooding conditions such a simulator can be a powerful tool.

- 150 -

List of symbols
a

major axis of ellipse

area of constant pressure square

A
P
A

poro-elastic constant

compressibility of bulk rock

compressibility of rock grains

total pore compressibility

e
e
c

trace of strain tensor


b /a
c c

VaF

thermo-elastic constant

erf

error function

erfc

complementary error function

Young's modulus

Ei

exponental integral

g. .

metric tensor

reservoir height

permeability

critical stress intensity factor

fracture half-length

heat capacity of fluid-filled reservoir rock

heat capacity of cap and base rock

heat capacity of injection water

p.

initial reservoir pressure

fluid pressure in fracture

q
q.
4
r
w
R

injection rate
total leak-off rate from fracture into reservoir
wellbore radius
radius of temperature front

S
H
S
or
S
wc
t

horizontal reservoir rock stress

T. .

injection temperature
r

AT =

T-T

fracture volume

fracture half-width

in]

residual oil saturation


connate water saturation
injection time

, change in temperature with respect to initial temperature

- 151 -

Greek
a

linear poro-elastic expansion coefficient

thermal diffusivity of cap and base rock

linear thermal expansion coefficient

5..

Kronecker delta

JO

c..

strain tensor

ID

7}

hydraulic diffusivity

fluid mobility

viscosity

Poisson's ratio

porosity

elliptical coordinate

a. .

stress tensor

dimensionless heat injection time

Subscr ipts
1

cold fluid zone

warm fluid zone

oil zone

dimensionless

cold front

flood front

P
T

poro-elastic
thermo-elastic

- 152 -

REFERENCES
1. Veatch, R.W.Jr., Overview of current hydraulic fracturing design and
treatment technology - Part I.
JPT (April 1983), pp. 677-687.
2. Hagoort, J., Weatherill, B.D. & Settari, A., Modelling the propagation
of waterflood-induced fractures.
SPEJ (Aug. 1980), pp. 293-303.
3. Perkins, T.K. & Gonzalez, J.A., The effect of thermo-elastic stresses on
injection well fracturing.
SPEJ (Feb. 1985), pp. 78-88.
4. Hagoort, J., Waterflood-induced hydraulic fracturing.
Ph.D. Thesis, Delft Technical University, 1981.
5. Carter, R.D., Appendix to "Optimum fluid characteristics for fracture
extension" by G.C. Howard and G.R. Fast,
Drill, and Prod. Prac, API (1957), p. 267.
6. Kucuk, F. & Brigham, E.W., Transient flow in elliptical systems.
SPEJ (June 1981), pp. 309-314.
7. Gringarten, A.C., Ramey, H.J. & Raghavan, R., Unsteady-state pressure
distribution created by a well with a single infinite conductivity
vertical fracture.
SPEJ (Aug. 1974), pp. 347-360.
8. Olesiak, Z., On a method of solution of mixed boundary-value problems of
thermoelasticity.
Journal of Thermal Stresses (1981), 4, pp. 501-508.
9. Muskat, M., The flow of homogeneous fluids through porous media.
McGraw Hill (1946), p. 185.
10. Nghiem, L.X., Forsyth, P.A. & Behie, A., A fully implicit hydraulic
fracture model.
JPT (July 1984), pp. 1191-1198.
11. Kumar, A. & Ramey, H.J., Well-test analysis for a well in a constant
pressure square.
SPE 4054, 1972.
12. Sneddon, I.N. & Berry, D.S., The classical theory of elasticity.
Encyclopedia of Physics, ed. S. Flgge, vol. IV, Springer-Verlag
(1958), p. 12 and p. 85

- 153 -

13. Weinberg, S., Gravitation and cosmology - Principles and applications of


the general theory of relativity.
Wiley & sons, New York (1972), chapter IV.
14. Adler, R., Bazin, M. & Schiffer, M., Introduction to general relativity.
McGraw-Hill (1975), second edition, p. 149.
15. Nowacki, W., Thermoelasticity.
English edition, Pergamon Press (1962), p. 50.

- 154 -

APPENDIX 3-A
CALCULATION OF PORO-ELASTIC STRESSES IN ELLIPTICAL COORDINATES
The poro-elastic linear stress - strain relations in a general
curvilinear coordinate system are given by:

E<1
E
Aa. . = 7T~ ( +7~T~ e g. .) + T H ? - Ap g. .
i]
1+u
i] l-2o
'13
l-2u
'13

where a

is the linear poro-elastic expansion coefficient given by:

o = (^J
p
E
and c

(3-A-l)
'

(1 - - 2 )
c,
b

(3-A-2)

and c, are the compressibilities of the grains and of the bulk matrix

of the rock respectively, g. . is the metric tensor and e is the contraction


of the strain tensor.
The covariant equations of equilibrium are given by

12
:

Ao l j .= 0

(3-A-3)

where summation over repeated indices is understood. Sub- and superindices


denote co- and contravariant transformation properties respectively. We take
the notation for covariant differentiation as used, for instance, in Ref.
13. The co- and contravariant stress tensor are connected through:
vj

Aa. . = g.,g. . Ao
13
ik ] ^

(3-A-4)

A particular solution to (3-A-3) can be obtained from


E
Aa. . = 7 4> . . + A Ap g. ,
ljp
1+v ;i3
p
13
where A is the poro-elastic constant defined as
P

(3-A-5)

- 155 -

Ea
A = E
p
1-

(3-A-6)

The scalar 0 is called Goodier's displacement potential and satisfies:


k

0.

with m = *-

= (9

kJ

* f / ) . k = " m Ap

(3-A-7)

*- a .

(1-u) p
The covariant derivative .. is given by
;i]
2

0 . . = 7 : <t> - T. . ~
; i ]J
- i
:
i] - k
3x a3x
3x

<t>

(3-A-8)

k
with T.. being the affine connection. The latter is related to the metric
13
13
tensor by :

kJ

^-\J

rK. = % 3g *(-^
i]
2
' ]
ox

^iJ

^ii

+ - r 1 )
,i
. k '
3x
3x

(3-A-9)

*
The displacements generated by the potential <f> are given by
u.
= -^rx <t>
1
a
ox

(3-A-10)

3
In the following we will consider plane strain conditions (u =0).
This reduces the problem of determining 0 to two (horizontal) dimensions. In
1
2
a two-dimensional orthogonal curvilinear coordinate system (x , x ) the
infinitesimal line element is given by:
ds

ii
2 1 2
= g..dx dxJ = h (dx ) + h

2 2 2
(dx )

(3-A-ll)

* The strain tensor- is given by e..= 0 ... To prove that this tensor
together with (3-A-7) and (3-A-5) is a solution to (3-A-3) and (3-A-l)
requires the results that <j> .

. = 6

. .. This result follows if the

elements of the Riemann tensor are zero (see Ref. 14). This is the case
for Euclidean spaces.

- 156 -

and the elements of the affine connection are given by

12
:

_m _ 1
m
mm
h _m
m dx
(3-A-12)
, 3h
_m _ _,m _ 1_
m
mn
nm
h .n
i 9x

_n _
mm

h
3h
_ra
m
, 2 ,n
h
dx
n

where m,n are different and repeated indices are not summed. In a confocal
elliptial coordinate system

(,TJ)

the metric tensor is given by

L2

i3

(cosh2{ -

COS2TJ)

"

0
,
L

hence h = h
From (3-A-12):

2
L
= [ (cosh2 -

I _ 2 _ 2 _
II " 12 " 21

_1 _ _1 _ 2 _
12
21 ~ 22
As is customary

12

(cosh2 - cos2rj)

(3-A-13)

1/2
COS2TJ)]

= h.

1 _ L2sinh2S
22
u2
2h

(3-A-14)

2 _ L sin2n
11
2
2h

, we define the physical components of the stresses and the

displacement as:
Aau =

Ao

*
A

(
=

l l

777

Aa

22

(h2)

%*

A0

12

(3 A 15)

" "

- 157 -

* ^

"i

\-t2"2
From (3-A-5), (3-A-8), (3-A-14) and (3-A-15) we have

A. _ _J_ ri_ 3_ _ k_ sinh2j 3


{p~ 1+u l h 2 ^ 2
2
3*
h4

L_ sin2q M i . a A n
2 h4
3r,J +
VP

Aa

_ _ I _ f l_ jfi . L ! sinh2$ 30. _ l sinh2r? 30,

Afl

OT

" 1+ V

.
A<

Bv2

ft

,1

h4

h4

"

L 2 s i n h 2 { 30,

sin2?? 30

W TTV (~2 Jfr, - T "^4

if " F ~

9?

,
(3 A 16)

~ -

Equation (3-A-7) becomes:


.2

g2
0 + ft = - mh
3*?

Ap

(3-A-17)

We will first calculate the poro-elastic stress changes induced by


the steady-state pressure profile in (3.36). This pressure satisfies the
steady-state equation:
d2
r Ap(|) = 0

(3-A-18)

<sr
Equation (3-A-17) becomes:
2
2
3 0
3 0
r ^ + r ^ = - mh2Ap
...2
.2
3
3^7

0 <, I I

3202
320 r - + -r- = 0
-v2
. 2
3*
3r)

f > *
e

(3-A-19)
e

(3-A-20)

- 158 -

The ellipse with coordinate

separates the infinite reservoir into two

regions. The inner region forms the area that is affected by the change in
pressure. We have to define a separate potential in the outer region to
ensure continuity of the displacements across .
We seek a solution in the form

*!

2
mr2 *
*i
mT
J d
" 2
l ' <3{2cosh2$2Ap(2)d{2- =Ui- Ap(ncos2rj
o
o

+ K
0

= K

-2{
e
cos2rj + K cosh2{ COS2TJ
e~2'cos2r? + K

(3-A-21)
(3-A-22)

By using (3-A-18) it can be shown by direct substitution that (3-A-21) is a


solution of (3-A-19).
According to the discussion leading to (3.34), the following boundary
conditions must be imposed on the displacements:
uf

= 0

= 0, 0 ^ TJ < 2JT
(3-A-23)

=0

$ > 0, rj = 0 A n = ir

Furthermore, Aaf and Aaf

must be continuous at { =

and the stresses must

vanish at infinity. Inspection of (3-A-16), (3-A-21) and (3-A-22) shows that


continuity of Ao,. and Aa

is guaranteed if the displacements are continuous

at = i . Eq. (3-A-22) guarantees that the stresses go to zero at infinity.


Boundary condition (3-A-23) together with continuity of the displacements at
{

determine the four unknown constants {K.).


l

Boundary condition (3-A-23) gives:


2

- ~

i ? A p IJ=0 " 2K1

Continuity of the displacements at = gives:

(3-A-24)

- 159 -

r2

" T i dT

AP

U=0

1^

AP

h =l

COsh2

*e

(3-A-25)

e
K4 = -

/ d{ COsh2{ A p U ) d*
o

From (3-A-16) and (3-A-17) we have:


A

%P

" Aip

(3-A-26)

We obtain from (3-A-16) and (3-A-21):

Aa

p ( * = 0 )

2 V P ( 0 ) " ^ f K l " fc ?2 K2

<3"A"27>

0 r> < 2ff

From (3.36) we have:

i A p l{=o = f ? A p . = " i ? l

(3 A 28)

" "

e
Putting Aa ({O) = Aa

and using the definitions for dimensionless

P
yp
quantities in (3.40) and (3.41), we obtain:

bo{PV
ypD

= Apfo) - 7 (1 - e
2

(3-A-29)

The major and minor axes of the ellipse separating the parts of the
reservoir affected by pressure and unaffected by pressure are given by
respectively:
a = L cosh
e
e
(3-A-30)
b = L sinh st
e
e
b
Defining e =
we have
e
a
e

- 160 -

-2{

(1

"

S)=

(3_A 31)

4"

e
(p

Putting
({=0) we finally
have from (3-A-26), (3-A-29) and
J
3 Aa ^ = Aa
xpD
rjpD s
(3-A-31):

Aa ( p s ) = X Ap
(o) - ^ (*-)
ypD
2 y D v ' 2 vl+e '
(3-A-32)
Aa ( p S ) = i Ap (o) + l l(-^-)xpD
2 * V ' 2 l+e ;
c
e
Since in (3.38) it was assumed that a = b = R (t), with R given in
e
e
e
e
(3.37), we have that e = 1 and (3.39a) and (3.39b) result from (3-A-32).
e
As discussed in the text, we must modify (3-A-32) to account for
deviations from plane strain conditions. Following Chapter 2, section 2.2.8,
this is done by imposing the additional boundary condition:
lim (Aa _, Aa _)
= AS
.
H D
-*
*P
'JP0
P
Since AS

is a uniform stress field this boundary condition is satified if

AS
is simply added to the solutions obtained from (3-A-16), (3-A-21) and
HpD
(3-A-22). Boundary condition (3-A-23) is still satisfied since, from the
argument of symmetry, the superposition of a uniform stress field does not
affect the displacement perpendicular to the line {=0, O^rj <2tr and >0, TJ=0
Arpff}.
We note here that whereas the solution in (3-A-32) is valid for a
e
and b corresponding to any ellipse confocal with the fracture, the
correction term AS
is only applicable in the case that the pressure
profile is pseudo-radial, i.e. a = b .
^
*
e . e
In the presence of elliptical discontinuities in fluid mobility, the
pressure distribution is given by (3.53). To determine the poro-elastic
stresses for this case we define 4 potential functions:
_L2
<t>i = -

I
2
d cosh2* Ap()d -

Ap(|) COS2TJ

+ K ^ e " 2 * cos2rj + K^'cost^J COS2TJ

i = 1,2,3

(3-A-33)

- 161 -

0 4 = K < 4 ) e"2cos2rj + K^ 4) J

where

O { <, l
c

*c " * * *F

Ap

Ap

(3-A-34)

Ap = {

and

Ap3

p < | ^

f * e

are the elliptical coordinates of the elliptical temperature and

flood fronts respectively and i=l,2,3 for in the cold water, warm water
and oil zone, respectively. The potential functions satisfy:
2
3 0.
at2

2
3 0.

+
9r?

3204

3204

3?

= - mh Ap.

i=l,2,3
(3-A-34)

Boundary conditon (3-A-23) and continuity of the displacements at { = { ,


...
c
_, { , d e t e r m i n e t h e 8 unknown c o n s t a n t s {K.
} . The s o l u t i o n i s g i v e n b y :
r e
1
K

(l,
1

_ rail
16

2
( D = _ SU^
2
16
^2

^ 1
d

{
l

dAp
dAp,
_^2, _ _ ^ 1 , }
d* ' {
d{ l{ i
c
c
dAp3
dAp^

16
2 dAp

+, mil

d* ' { F i

d* l { F
-2{

3i

i?"^ru

-2
e
e

-2JF
6

- 162 -

K <2) =

- ir ir'o" ir ^ir1*c - ir1*c } cosh2*c


mL2 **1.

(3)
K

mL2

dAp

dAp

2,

**!.

dAp

dAp

-mL2 f d{A p 2 , |

2 dAp

-2

i - - ir ir o ir -ir i - ir 1 ?, } c o,s h 2 | c
mL 2

i ,1

-2|

*Pl,

^2

dAp 3

dAp 2

-ir -ir iF-ir|iF}cosh2^


2 dAp
K

16

-2*
6

d 'l

e
,,.
dAp
T 2 dApn
T2
(4)
mL
lli
mL ,__12i
K
T = ~ 77
77" ~ 77" t~TT"L
3

16

d{ 'o

mL2

dAp 3

16

dj
dAp 2

-1

" IT "dT C "dl !.

dAp,
*l i -.
._..
7 7 L J cosh2{
d{

1 COsh2

'%
c
mL2

c
dAp 3
l

V IT ~W i

2
K^4)= - *p- Je d{ cosh2 cosh2 Ap({)

COSh2{

e
(3-A-36)

From (3.53) we have


dAp.

i f = " 1^x7

i - 1.2,3

(3-A-37)

When (3-A-37) is substituted into K*1' and K*1* and with

- i (1 - s"2t'P>
c,F

13*3.,

(3.54a) results from (3-A-27) and the definition of dimensionless quantities


in (3.56). (3.54b) results form (3.54a) and (3-A-26).

- 163 -

APPENDIX 3-B
A NUMERICAL METHOD FOR CALCULATING PORO- AND THERMO-ELASTIC STRESS CHANGES
In Cartesian coordinates (xi'x2'x^

t n e e<

Juation

for tne

displacement

potential (3-A-7) becomes

a2

a2
1 +

(
9x

2
3x

a2
+

2 } * = " "^P
3X

(3-B-l)

which has a solution


=

4 ' d x i d x 2 d x 3 Ap(x^,x',x^) |

(3-B-2)

where
2
2
2 1/2
R =-[(x1-x|) + (x2-x^) + x3-x^) ] '

(3-B-3)

From (3-A-5) the stresses are given by:

*ijP= y i ? '

W*

A
3

P(xi'x2'x3} aTaxT I
1 3

+ Ap 6 . . }

(3-B-4)

13

where 8.. is the Kronecker delta.


If the coordinate space is divided into parallelepipeds with edges of
length 2a , 2a , 2a

in the x ,x ,x direction respectively and if Ap is

taken to be constant within such a parallelepiped, (3-B-4) becomes:

l'X2
X

i^ a i
. dk
X

i" a i

X +a
2 +a 2
3 3
dk2
dk3

2 "a2

3 "a3

- 164 -

. girlr K* r v 2
+

+ ( k )2 +

v2

(v k 3 )2] " 1/2}

V P ( X 1 ' X 2 ' X 3 } 5ij

(3

"B_5)

The integrals in (3-B-5) can be shifted to obtain:

3 l

.
-aL

dk x

*2
*3
dk 2

-a2
-a3

a2
dk 3

i r 5 r

(x3-x3-k3)2]"1/2}

The

2
2
(
x
x
k
)
+
2
2

[ x ^ - k ^
D

+ A Ap(X;L,x2x3)

(3-B-6)

integrals
a

<

,u > =

"ai

dk x

dk 2

"a2

dk 3

2
g]rgir

"a3

2
2
2 -1/2
. [(u1-k1)z + (u2-k2)z + (u3-k3T] x/

can be integrated analytically. The results are given in Ref. 15. To


calculate Ao

xp

and Aa

,
Va2
I,,
1 1 = atan
Vai

U
r

yp

we only need I,, and I... Ref. 15 gives:


11
22

3~ 3 3

+l,+2,+3

atan

2"a2

Vai

3 +a 3

r
+l,+2,-3

, U 2 + a 2 U3~a3
, U 2 +3 2 U 3 +3 3
- atan
+ atan
U
U
l~ a i r+l,-2,+3
l"ai r+l,-2,-3

(3-B-7)

- 165 -

.
Va2
- atan
Vai

. U2_a2 U3+a3
+ atan
r
-l,+2,+3
V a i r-l,+2,-3
V

U +a
,
2 2
+ atan

3'a3

, U2+a2
atan

-l,-2/+3

3+a3

'-1,-2,-3

where
r

l,2,3 "

[ (

V V

+ {

V 2

can be obtained from I


u

-* u

2
)

2 1/2

<V 3

(3 B 9)

" "

by making the interchanges:

2
(3-B-10)

a *-* a
1
2
If the pressure distribution Ap is symmetrical in the line x =0, then
this line becomes a boundary of no displacement in the x -direction for the
displacements generated by (3-B-2). According to the discussion leading to
(3.34), it follows that Ao

( = Aff ), calculated from (3-B-5) along

2 =0,

can

be used directly to calculate changes in fracture propagation pressure or in


fracture width if the fracture plane coincides with the plane x =0. This
method can be very easily incorporated into numerical fracture/reservoir
simulators such as in Refs. 2 and 10. Here we have used this method to check
the 3-D analytical results of Appendix 3-A.
Since the analytical pressure profile Ap is constant over the
reservoir height and zero outside the reservoir, the summation over x ' in
(3-B-6) can be dropped and a

can be set to - where h is the reservoir

height. Since the pressure profile has elliptical symmetry, the boundary
condition of zero displacement in the x -direction at the fracture face is
automatically satisfied if the summations in (3-B-6) are carried out across
the complete horizontal reservoir plane. The values for Ap(x ,x ,x,) are
calculated in the centre of each grid block from the formulae in (3.36).
These formulae can be expressed in Cartesian coordinates through

cosh + sinh = 1

v S 1
(3-B-ll)

- 166 -

2
1/2
cosh{ + sinhg = v+ (v -1)

v > 1

where

[(,W,
v = {

x A V / 2 +x22+Xl2+L2 1/2
\
2L

In the calculation we have multiplied the edges a,#a

(3-B-12)

of the grid

blocks by a constant factor in a direction away from the fracture, hence


across each grid block there is approximately the same pressure drop.
In Fig. 3-B-l a schematic representation of the gridblock layout is
shown. The dots represent the gridblock centres at which Ap is calculated
from (3.36). For illustration purposes the edges of the gridblocks in the
figure are multiplied by a factor of 1.5 in a direction away from the
fracture.

- 167 -

SCHEMATIC REPRESENTATION OF GRIDBLOCK


OF PORO-ELASTIC STRESSES

ARRAYS FOR THE CALCULATION

FIG. 3.B-1

- 168

APPENDIX 3-C
CALCULATION OF THERMO-ELASTIC STRESSES AND OF THE AXES OF THE ELLIPTICAL
FLUID FRONT
We define the following dimensionless quantities:

xTD

xT
AJT '
T

yT
yTD " A AT
J
T

(3-C-l)

where A is the thermo-elastic constant defined as:


T
A

with a

Eo
m = ,
T
1-U

(3-C-2)

the linear thermal expansion coefficient. AT is the temperature

difference between injection temperature and reservoir temperature. For an


elliptical inclusion of uniform AT, we have from Ref. 3:

e
A

VD=

(1 + e

^e" +

1 + 0 . 5[1 . 45(h

c)_1

0.9+

xTD

1
l+ec

(1
c

+
+

0.9+
D

0.77 4]

<3"C-3}

DC

* ec)~1
h

2
D

1.36
C

(3

"C-4:

where h = h/2b , e = b /a and b ,a are the minor and major


axes,
J
D
c
c
c
'
c
c
c
respectively, of the elliptical temperature front, h is the reservoir
height.
In the remainder of this appendix it is shown how the major and minor
axes of the temperature and flood fronts can be obtained following the
method of Ref. 3.
The elliptical coordinates of the temperature and flood fronts can be
obtained from a heat or volume balance according to:

- 169 -

2
V. = ir L h s i n h . cosh.

i=c,F

(3-C-5)

and
M
v
c

sr

qt

3 c 6)

- '

r
V

F * I-B'-S )
or wc

qt

(3

-C"7)

Since a. = L cosh. and b . = L s i n h . r i t follows t h a t


i

2 " * i * 7?7'
i=c,F

(3-C-8)

"i I <*! " 7FT'


where
2V.
2V. 2
.
F. = y - + - [(f~)
+ 1]X/Z
JTL

ffL

i=c,F

(3-C-9)

- 170 -

- 171 -

CHAPTER FOUR
A PRESSURE FALL-OFF TEST FOR DETERMINING FRACTURE DIMENSIONS

Summary
4.1 Introduction
4.2 Calculation of the pressure fall-off with a closing fracture
4.2.1 Assumptions
4.2.2 Integral equation for the dimensionless pressure function
4.2.3 Solution for the dimensionless pressure function
4.3 Analysis of a pressure fall-off test
4.3.1 Four methods to determine fracture length
4.3.2 Discussion
4.4 Conclusions
List of symbols
References
Appendix 4-A Solution for dimensionless pressure function in Laplace space
Appendix 4-B Relationship between fracture closure constant and fracture
length

- 172 -

SUMMARY

Pressure transient theory for a water-injection well in the presence of


a closing fracture and a discontinuity in fluid mobility is developed in
elliptical coordinates. Solutions are obtained using the Laplace transform
and numerical inversion. From the results, it is concluded that a pressure
fall-off test with a closing fracture in principle provides four different
methods for determining the fracture length. The first method is based on
rock mechanical principles only. The second method makes use of formation
linear flow. The third method analyses the transition of the pressure
transients from the inner fluid region to the outer region in conjunction
with a heat or volume balance. The fourth method analyses the skin as seen
during pseudo-radial flow in the outer fluid zone.

- 173 -

A PRESSURE FALL-OFF TEST FOR DETERMINING FRACTURE DIMENSIONS

4.1 INTRODUCTION
In the previous chapter the propagation of waterflood-induced fractures
was studied with relatively simple models. Two important conclusions from
this study are a) the conventional Carter model of one-dimensional leak-off
perpendicular to the fracture is generally inadequate, b) changes in
reservoir pressure and temperature can have a significant effect on the
reservoir rock stress and therefore on the fracture propagation pressure.
A quantitative prediction of fracture length with these models should
be treated with some care. This is because, on the one hand, they rely on
many simplifying assumptions and, on the other hand, a great number of input
data are required. A method of determining the dimensions of a growing
fracture at certain moments during its propagation is therefore useful in
gauging these models. After such a calibration, a reliable assessment of
growing fractures on the sweep efficiency of the pattern-flood in question
can be made.
In injection wells, pressure fall-off tests are conducted to obtain
information on reservoir properties and possible well damage. Such a test
consists of shutting in the well after a suitable injection period and
recording the subsequent drop in pressure as a function of time. The
required information is then extracted by comparing the measured pressure
profile with the theoretically predicted one.
For fractured injection wells, fall-off tests can be used to determine
fracture length. The early time data immediately after shut-in can be
related to fracture length since initially fluid flow and pressure behaviour
are completely dominated by the presence of the fracture.
1 2
The literature on fall-off tests for fractured wells ' deals with
stationary propped fractures which do not exhibit any closure behaviour
during the pressure fall-off.
For waterflood-induced fractures such closure behaviour may well
dominate the early time pressure behaviour and therefore a suitable
theoretical model must be applied to analyse these pressure data..

- 174 -

In Ref. 3 a method was proposed for analysing the dimensions of


minifracs from a recording of the pressure data immediately after pumping
has stopped. Since this method relies on the Carter leak-off model both
during fracture propagation and during fracture closure and since,
furthermore, the fluid leak-off is assumed to be independent of pressure,
this method is in general not suited to analyse the dimensions of
waterflood-induced fractures.
An important first step towards the analysis of a fall-off test for
4
closing waterflood-induced fractures was made by Hagoort in his thesis . By
applying a simple rock mechanical model for the fracture he was able to
relate fracture closure to early-time fluid flow. He presented an analytical
expression for the theoretical pressure response which is valid as long as
formation fluid flow is still linear perpendicular to the fracture and
before the fracture has completely closed.
This chapter presents an extension of Hagoort's model to account for
different fracture geometries, transition from early time linear to late
time pseudo-radial flow, pressure response during and after closure and the
effect of an elliptical discontinuity in fluid mobility.
In Section 4.2 the theoretical model for the pressure response is
developed. The results are presented in a number of dimensionless
type-curves. In Section 4.3 the analysis of a fall-off test is discussed. It
is shown that, in principle, the test provides four ways of determining the
fracture length. Finally, the conclusions are presented in Section 4.4.

- 175 -

4.2 CALCULATION OF THE PRESSURE FALL-OFF WITH A CLOSING FRACTURE


4.2.1 Assumptions
The transient pressure response in the fractured injection well is
calculated using the following assumptions.
1. The fracture has infinite conductivity during propagation, during closure
and after closure. The assumption of infinite conductivity after closure
means that a channel of very high permeability remains where the fracture
occurred. This could be caused by erosion or by a mismatch of the
fracture faces. The effect of this condition and deviations from it will
be discussed in the analysis of the results.
2. During the pressure decline the length and height of the fracture remain
constant. The fracture volume diminishes according to:
dV

dp

-dT

= C

where V

(4

f ^

' X)

is the fracture volume, p f is the fluid pressure in the fracture


*

and C f is called the fracture closure constant .


3. The isotropic, homogeneous, horizontally infinite reservoir with uniform
properties contains two zones of slightly compressible fluid. A different
constant mobility and diffusivity is assigned to each zone.
4. The discontinuity in fluid mobility has the shape of an ellipse that is
confocal with the fracture tips.

4.2.2 Integral equation for the dimensionless pressure function


One of the surprising results of Chapter 3, section 3.2.3, is that the
fracture growth as a function of time can be obtained from the constant rate
pressure solution for an infinite conductivity fracture with fixed half-

4
* Hagoort called C f the fracture storage constant in line with the
mathematical analogy with wellbore storage. However, since C may be
confused with the actual fracture storage constant for a non-closing
propped fracture, we adopt a different terminology here.

- 176 -

length by substituting L = L(t), pf(t) = p f = constant and solving for L(t)


(L = fracture half-length, t is injection time). This method was shown to be
correct by comparison with the results from a numerical fracture/reservoir
simulator as presented in Ref. 4.
Conversely, the pressure distribution surrounding a propagating
fracture at time t can be obtained by evaluating the constant rate/fixed
fracture length solution for time t and fracture half-length L(t).
Suppose the fracture has a half-length L(t . ) when the well is shut-in
SO

at time t .. Generalising the above result to the case of varying rates


sh
gives for the pressure in the fracture at t :

fi

pf(tsh> ' - f "


o

6p

(t

"X)dX

.
sh

(4.2)

where 5p. = p,-p. and q. is the total leak-off from the fracture into the
f
f
4
er
reservoir. 5p_
is the constant rate solution for an infinite conductivity
fracture of half-length L(t . ) and for an injection rate q . Duhamel's
J
*
sh
o
superposition principle

for varying rates has been used in Eq. (4.2).

The basic volume balance for the fracture is:


q=q^

"

dVf

(4.3)

d T

where q is the injection rate from the well into the fracture. In Chapter 3
it is shown that during propagation dVf/dt usually becomes negligibly small
after a few hours of injection. If the injection rate into the well is
constant, Eq. (4.3) therefore becomes approximately

q =s q

= constant

(propagation)

(4.4)

If at t L the flow from the well into the fracture is shut-off


sh
instantaneously so that q = 0, then according to assumption 2 and from
Eq. (4.3)

d Pf
= - C f r-

(during closure)

(4.5)

For times t > t . , Eq. (4.2) therefore becomes (with the constant rate
sh
solution for q = q):

- 177 -

t
.
t C
sn
r
Spf(t) = J |^ 5p^ (t-X)dX - [-| ~
o
t u

5p (X)] f^ 5p^r(t-X)dX

(4

sh
which can be rewritten as:
t .
t
C
Pf(t) = f^ 6p"(t-X)dX - [1 + ~ ~ 8p (X)] f^ 5p"(t-X)dX
o
t u
sh

(4

The first term on the right-hand side of Eq. (4.7) can be integrated to
give:
5 p " (t-X)dX = 8p"(t)

(4

o
We now introduce the following dimensionless variables:
2irX h
p

fD

(A

~T~

2TTX

Ap (At) =

2jrX h
P

fD(V

-T"

~T~

(p

sh " Pf ( t sh

+ At)

>

2JTX h
6p (t)

~T~

(P r(t)

" Pi>

t^, . , = dimensionless time at which well is shut-in


D(sh)
17 At
At

' T
JJ

n
- ~2 ^sh5
L

At, , = dimensionless time at which fracture closes


D(cl)

fD

2
2*rL h ( 0 C t ) 1

where:
X,
1
k

= (-), = fluid mobility in inner region


Ml
= permeability

- 178 -

u
h

= viscosity
= reservoir height

p . = pressure in the fracture at the moment of shut-in (p . is a constant)


sh
,
sh
n, = (), = hydraulic diffusivity in inner region
1

0MC

= porosity

= total compressibility
cr
We multiply Eq. (4.7) with -1 and add Sp. ( t ) = p . - p. to both
-

sides. Using Eq. (4.8) and

5p = -

Sli

SO

Ap, we have, after the introduction

of the dimensionless variables


P

fD (At D>

'

PfD^Dfsh)' " PfD(tD(sh) +

D[1 c

- fD axl p f D ( V ] i c f r r PfD(AtD-xD)dXD

D
( 0

A t

<

At

D(cl))

If t h e a n a l y s i s i s r e s t r i c t e d t o t h e time region where At

(4 10)

'

we have

fD

(t

D(sh)> - P f D ( t D ( s h ,

+ A

( 4

-U)

(4

-12)

and Eq. (4.10) becomes:

fD<AV "

jt

[1 C

" fD d f DP fD (X D )[ iiTT
v
D
( 0 < A t

<

fD(AtD-VdXD
^(Cl)

If the fracture closes at At, , . with assumption 1 we can still use


D(cl)
^
superposition and substitute C,

= 0 for At

> At

into Eq. (4.12).

After integration this results in:


P

fD<AV " fD(Cl) [1"CfD 5^ PfD<V] ^ 7 ^ W ^ D


+ *GfllttD-AtUlcl)>

(At

D>AtD(cl)>

<4'13)

- 179 -

4.2.1 Solution for the dimensionless pressure function


If the constant rate solution is known, Eqs. (4.12) and (4.13)
determine the wellbore pressure response during and after closure,
respectively. An analytical solution for the constant rate infinite
conductivity fracture was presented by Gringarten et al. in Ref. 2. In Ref.
6 the same problem was solved in elliptical coordinates in Laplace space
7
using Mathieu functions, Recently , this approach was extended to include
the presence of an elliptial discontinuity in fluid mobility. It is this
method that we have adopted in solving Eqs. (4.12) and (4.13). Some details
are given in Appendix 4-A. For an elliptical coordinate system, see
Fig. 4.1. Eq. (4.12) is solved using the Laplace transform. The solution in
g
Laplace space is inverted numerically with the Stehfest algorithm .
Eq. (4.13) is solved by determining -rr p,_, ..A,r p._ in Laplace space and
n
' D
inverting with the Stehfest algorithm for 0 < At < At
,.. The integral in
the first term on the right-hand side of Eq. (4.13) is then integrated
numerically. The second term is again calculated in Laplace space and
inverted numerically with the Stehfest algorithm. The results are shown in a
number of type-curves in Figs. 4.2-4.4. The following additional constants
were introduced: .
\
\
- /

(c t *) 1
(cfc0)2

K =

\
K = r
A
2

diffusivity ratio (inner to outer)

mobility ratio (inner to outer)

(4.14)

position of mobility discontinuity (elliptical


coordinates)

In Fig. 4.2 a log-log plot of p

vs At

is given for when K = 5 = 0.33

and = 5 , for several values of C. and for arbitrarily assumed values of


o
fD
the dimensionless closure time At , ,.. The first flow regime is fractureD(cl)
closure-dominated and changes abruptly to formation linear flow for small
At

D(cir
In Fig. 4.3 a derivative type-curve plot is shown of d(p

)/d(ln At )

vs At /C f . Fracture closure shows up clearly as high peaks in this plot.

180

ELLIPTICAL COORDINATE SYSTEM THAT IS


CONFOCAL WITH FRACTURE TIPS

FIG. 4.1

- 181 10 i

At D

FIG. 4.2

TYPE CURVE FOR A PRESSURE FALL-OFF TEST WITH A CLOSING FRACTURE

FRACTURE CLOSURE

TRANSITION PERIOD

MOBILITY DISCONTINUITY AT | 0 = 5
MOBILITY (DIFFUSIVITY) RATIO:0.33

"-

102

10'

" " i

103

1<-

10"

At D /C f D

FIG. 4.3

DERIVATIVE-TYPE CURVE PLOT FOR PRESSURE FALL-OFF TEST


WITH CLOSING FRACTURE

10s

- 182 -

d(PD)

d (In At n )

4n= 0.5

AtD (cl) = 0.1 (ASSUMED)


x = = 0.33

10'

i 11111|

i 11111 I

10'

i i 11 111

i i 111 11

10

10

i 111iiI

10"

Atr
DERIVATIVE-TYPE CURVE PLOT FOR VARIOUS POSITIONS OF THE
ELLIPTICAL MOBILITY DISCONTINUITY

111|

i
b

10

i i 111111

10b

- 183 -

Pseudo-radial flow is reached inside the inner zone when the derivative
reaches the value 0.5. After transition to the outer zone the constant value
0.5 K is reached. Fig. 4.4 shows a log-log plot of the logarithmic
derivative of p, vs At for various values of { . This plot shows that if
r
fD
D
o
the fracture is close to the front (small ), pseudo-radial flow is not
attained in the inner region.

4.3 ANALYSIS OF A PRESSURE FALL-OFF TEST


4.3.1 Four methods to determine fracture length
Based on the conceptual results given in Figs. 4.2-4.4 the following
philosophy emerges for a pressure fall-off test. First, shut the well in
with a downhole shut-off tool to avoid the fracture closure being masked by
wellbore storage. Since early-time' data are important, a pressure gauge with
a high sampling rate should be used to record the pressure. Then make a
semilog plot of the late-time data to determine kh/M and skin in the outer
fluid zone in the usual way. Estimate the mobility ratio < from fluid and
relative permeability data. The fracture length is then determined using the
following four methods.
Method 1 Closure flow
During the very early part of the closure-dominated flow regime p
varies linearly with At. In this period the leak-off into the formation is
still approximately equal to the injection rate before shut-in. The fracture
has to close to provide the necessary fluid volume. Therefore,

q=

dp

(4 15)

- f^f

and the slope of a linear plot of p

'

vs At is equal to ~~ , from which C.

can be determined. Note that this situation is analogous to the very


early-time fall-off behaviour of an unfractured well with wellbore storage.
In Appendix 4-B it is shown how C

can be related to fracture length

for various fracture geometries. The fracture models considered are the
9
Christianovic-Geertsma-de Klerk model (CGK) , applicable to fractures for

- 184 -

which 2L/hf
2L/h

9
1, the Perkins-Kern-Nordgren model (PKN) , for fractures with

1 and, ellipsoidal fractures

for the intermediate case, h

is the

fracture height at the wellbore. Fig. 4.5 shows the CGK and PKN geometries,
Fig. 4.6 shows the ellipsoidal geometry.
The fracture length can be determined graphically as follows. Estimate
the fracture height h , form a new dimensionless closure constant,

fD

(4

. 3,n 2,
hf (l-o )

'16)

where v and E are Poisson's ratio and Young's modulus, respectively. The
superscript (rm) denotes rock mechanical and serves to distinguish this
definition of the dimensionless closure constant from that in (4.9).
dp
f
C, = -q/ - TT is determined from the linear plot of p vs At. Enter Fig. 4.7
t
at
and determine 2L/h for a particular fracture geometry. This then gives a

value for L.
Alternatively, if h

is not known, a certain value for the ratio 2L/hf

may be assumed and Fig. 4.7 then gives a value for c!l

from which h

can

be calculated. This method may, for instance, be applied if it seems


plausible that the fracture area is circular (penny-shape fracture).
Evaluation of the curve for the ellipsoid at 2L/h
C

fn m ) '

the value

= 1, then gives, for

of 0.67.

Method 2. Linear formation flow

Method 2a
Linear formation flow becomes visible if: (i) the fracture closes at a
small dimensionless closure time At

and (ii), owing to erosion or a

mismatch of the fracture walls, a channel of very high permeability remains


where the fracture occurred. The latter condition justifies the use of the
infinite conductivity solution after closure. Linear formation flow can be
recognised by a straight line with slope 1/2 on a log-log plot of Ap vs At
2
of i
of the field data. In the usual way , a plot of p f vs /At then gives a slope
of,

185

CGK-MODEL

PKN-MODEL
FRACTURE MODELS WITH RECTANGULAR FRACTURE AREA
FIG. 4.5

- 186 -

Z-AXIS

FRACTURE MODEL WITH ELLIPTICAL FRACTURE AREA (ELLIPSOID

- 187 -

10'

-1

r ir-n-n

I I

r"

i ' i

i > i i |

CGK
PKN
ELLIPSOID

10

r
/..'

10"

10

10"

-iU.'

10"
10"

'

'

IQ"

I I I I I

10"
2L/hf

10'

10z

DIMENSIONLESS FRACTURE CLOSURE CONSTANT VS. FRACTURE LENGTH

FIG. 4.7

- 188 -

From the permeability of the semi-log analysis and from the fluid and
relative permeability data ( M A 0 C ) can be estimated. Then from Eq. (4.17)
the quantity 2hL is obtained. Although Eq. (4.17) is conventionally derived
for a rectangular fracture area, it can be used for a fracture area of any
geometry if 2hL is replaced by the corresponding expression for this area.
We therefore have

h = h

(CGK and PKN)

(4.18)

h = | hf

(ellipsoid)

(4.19)

and

If the fracture closes very tightly, formation linear flow cannot


occur. The dimensionless pressure will behave differently in this case than
indicated in our type-curves and the analysis described above cannot be
applied.
Method 2b
For longer closure times At

a combination of closure flow and

D(C1)

linear formation flow takes place before the fracture closes. In his thesis
Hagoort presented an analytical expression for p f

in this flow regime. It

is given by:
pf

= /(ffAt
/(ffAt D) -
rrfD

,2,
" e"
e*3 A t DD erfc
))
(1 erfc (0/At
(0/Atg))

(4.20)

where
& = Z^T^fD

(4.21)

The other quantities are defined in Eq. (4.9). If the analysis is restricted
to the time that formation flow is still linear, type-curves can be
generated either from Eq. (4.20) or by using the method given in Appendix
4-A. A match with the field plot gives values for L and Cf . Since only
linear formation flow is considered, again either Eq. (4.18) or Eq. (4.19)
can be used, depending on the geometry being studied.

- 189 -

Note that Eq. (4.20) does not consider the presence of an elliptical
discontinuity in fluid mobility. It can therefore be used only when i
large enough for linear flow to occur in the inner region.

Method 3. Transition flow


The dimensionless derivative plot of Fig. 4.4 can be compared with a
similar plot of the field data to estimate the position of the mobility
discontinuity . Since the onset and length of the transition zone and its
height above or below the line with constant value 0.5 K depend on the
eccentricity of the mobility continuity, a match with the field data in this
region gives a value for In the usual way, a match also provides a value
for L since the corresponding At and At

are known.

If | has been determined, a heat or volume balance, depending on the


type of mobility discontinuity, can then be used to obtain an additional
estimate of L. Following Chapter 3, Appendix 3-D, L can be obtained from:
V.
L =

f f hsinh{

1/2

cosh*
o

i = C

'

<4'22)

where for a temperature or flood-front mobility discontinuity, V. is given


by, respectively,
M
Vn = rr* qt
C
M
r
F

0(1-S -S )
or we

(4.23)

qt

(4.24)

Method 3 is applicable in the following circumstances.


a) After closure, if a channel of very high permeability remains where the
fracture occurred. Whether this condition is satisfied may be inferred
from the fact that method 2a was applicable or from type-curve matching.
b) Before closure, if the fracture closes sufficiently slowly that the
transition from the inner to the outer fluid region occurs while the
fracture is still open.

- 190 -

Method 4. Pseudo-radial flow


If pseudo-radial flow in the outer fluid zone has been reached, the
skin as seen in the outer zone, can be determined from:
1.5 ( * At)
A

*f

(At

{ln

> 5* r

+ s}

<4-25>

where Xh is determined from the slope of a Horner plot. The complete


expression for the pressure drop in the fracture is given by:
n

At

*f < > * <t

ln

+ b

^T^

1
where a , b

ln
2

3.<V(T),At)

a+b
o

< 4 ' 26 >

are the major and minor axes, respectivey, of the elliptical

discontinuity in fluid mobility. It follows that the skin is related to


fracture half-length according to:

X_
a + b
2 ,
o
o

=r

ln

ln

a + b
o
o

r~ ~ -J7

,. __,

(4 27)

In Chapter 3, Appendix 3-C, we showed that, for a temperature or a floodfront elliptical discontinuity in mobility, the sum of the major and minor
axes is given by:
1/2
a + b = L F. '
o
o
i

i = c,F

(4.28)

with
2V.
2V
2
1/2
F. = j - + [ ( Y )
+ 1]
JTL h
JTL h

(4.29)

where V. is defined in Eqs. (4.23) and (4.24). L can be determined from the
non-linear algebraic equation that results after combining Eqs. (4.27)(4.29). If the front is sufficiently far away from the fracture, such that
it is almost circular, the solution for L simplifies F. to:
\
L = 2R exp C- r [S + ln(R /r )]}
o
X_
o w

(4.30)

- 191 -

with R , the radius of the mobility discontinuity given by:

V. 1/2
= (-J)

i = c,F

(4.31)

jrh

Of course, if pseudo-radial flow has already been reached in the inner


fluid zone, then a Horner plot also gives X h and the skin as seen in the
inner zone can be determined. This skin is related to fracture length in the
usual way:

L = 2r

exp(-S)

(4.32)

As for method 3, method 4 is applicable before closure or after


closure, if a channel of very high permeability remains where the fracture
occurred.

4.3.2 Discussion
To obtain a measure of the fracture height h, at the wellbore is
sometimes difficult. If cold water is injected a temperature log can be run
in the hole. From this log, the height of the cooled region can be
determined. If, because of cooling, the reservoir rock stress has decreased
significantly, the height of the cooled region may then be used as an upper
limit for h

(see Chapter 3). A more extensive discussion on the

determination of fracture height can be found in Chapter 5, section 5.5.2.


The assumption of a uniform fluid pressure inside the fracture during
closure may not be strictly valid if viscous pressure keeps the fracture
open. However, from Muskat's analysis

of the steady-state pressure profile

for a rectangular infinite-conductivity fracture, it can be shown that the


leak-off distribution in the fracture is given by:

(L -x )
From the discussion at the beginning of Section 4.2.2 it follows that this
is also the leak-off distribution at the time of shut-in. As discussed in
method 1, the leak-off will not change appreciably during the early-time
since shut-in. The condition of zero fracture flow at the wellbore and high

- 192 -

leak-off at the tip will produce an appreciable pressure gradient only in


the fracture near the tip. Therefore, the pressure at the wellbore does not
differ very much from the average pressure in the fracture and the foregoing
closure analysis is valid. For a similar conclusion/ see Ref. 3.
The choice between substitution of Eq. (4.18) or Eq. (4.19) in methods
2a and 2b applies when the fracture does not extend over the whole reservoir
height. For this reason the early-time type-curve analysis must be
restricted to the period of linear flow perpendicular to the fracture. If
the fracture covers the whole reservoir height, h is always equal to the
reservoir height. Then the type-curve analysis in methods 2a or 2b may be
extended beyond the linear flow regime into the elliptical flow regime. The
types curves must then be generated with the method given in Appendix 4-A.
The analysis of methods 3 and 4 relies on a sharp elliptical
temperature or flood front. If the temperature of the injection fluid is
different from that of the reservoir, according to Chapter 3, Section 3.4.2,
a sharp front prevails if:

4o t
%
h

M2
"f < 0.05
M
r

(4.34)

The fracture closes if the fluid pressure in the fracture has become
equal to the in-situ horizontal rock stress. The absolute pressure at which
closure is observed therefore provides a value for the in-situ stress.

If a fracture starts influencing the sweep efficiency of a waterdrive,


injection at a lower pressure or at a higher temperature may be considered.
This action will neutralise the unfavourable effect of the fracture if it
can close tightly. Information on whether or not the fracture behaves as an
infinite conductivity fracture after closure can therefore be very valuable.
Finally, if more than one of the above methods can be applied to
determine the fracture length, consistent results will enhance the
reliability of the interpretation.

- 193 -

4.4 CONCLUSIONS
In principle, a pressure fall-off test provides four methods for
determining fracture length.
1. In the first flow period the fracture closes according to its
compressibility. The pressure varies linearly with time. By using rock
mechanical principles, the slope of the straight line gives a value for
the fracture length.
2. The second flow period is determined by either linear formation flow or a
combination of fracture closure and linear formation flow. In the first
case the usual square-root time analysis and in the second case typecurve matching result in a value for the fracture length.
3. The third flow period is determined by the transition from the inner
fluid region to the outer fluid region. The length, pressure level and
onset of this transition flow can be used to obtain a type-curve match
with a logarithmic derivative plot. The match provides a value both for
the elliptial coordinate, , of the fluid front and for the fracture
length. A heat or volume balance together with { gives an additional
estimate for the fracture length. This value should be in agreement with
that from the type-curve match.
4. The fourth flow period is determined by pseudo-radial flow in the outer
fluid zone. From a Horner plot the mobility in this fluid zone and,
subsequently, the skin can be determined. The skin can be related to
fracture length if the presence of an elliptical inner zone with
different mobility is properly taken into account.
If more than one of the four methods can be used, consistent results
should give a reliable estimate of the fracture length. Additionally, the
in-situ horizontal rock stress can be determined from the test. It is given
by the absolute pressure at which closing of the fracture is observed.
Finally, the test may provide information on whether the fracture closes
tightly or whether it remains highly permeable after closure. This
information is important when injection below the fracture opening pressure
is considered in order to minimise the effect of the fracture on sweep
efficiency.

- 194 -

LIST OF SYMBOLS
A

o
c

Fourier coefficient of Mathieu functions


total pore compressibility

Ce_2n
C,

modified Mathieu function of the first kind

Young's modulus

E(k)
Fek
zn
h

complete elliptic integral of the second kind


modified Mathieu function of the third kind

permeability

fracture half-length

heat capacity of fluid-filled reservoir rock

heat capacity of cap and base rock

heat capacity of injection fluid

p.

initial reservoir pressure

p,

fluid pressure in the fracture

p ,

pressure at shut-in

fracture closure constant

formation thickness

6 Pf
P f -P.
cr
opf
constant rate solution for infinite conductivity fracture
* p f = psh " p f
q

injection rate

qj

total leak-off rate from fracture into reservoir

q.

leak-off rate as function of position along fracture

Laplace time variable

S__

horizontal reservoir stress

t
t .
sh
At

injection time
time at which the well is shut-in
t-t

u
sh
time since shut-in at which the fracture closes

At. ..
(cl)
AT

temperature difference between injection fluid and reservoir

V,

fracture volume

fracture half-width

- 195 -

Greek
a

thermal diffusivity of cap and base rock

ij

diffusivity

mobility ratio

mobility of inner and outer fluid region resp.


1/2

s/4

tf

viscosity

Poisson's ratio

porosity

elliptial coordinate

diffusivity ratio

Subscripts
1

inner region

outer region

temperature front

dimensionless

fracture

flood front

sh

shut-in

- 196 -

REFERENCES
1. Clark, K.K., Transient pressure testing of fractured water injection
wells.
JPT, (June 1968), pp. 6.
2. Gringarten, A.C., Ramey, H.J.Jr. & Raghavan, R., Unsteady-state pressure
distribution created by a well with a single infinite conductivity
vertical fracture.
SPEJ (Aug. 1974), pp. 347-360.
3. Nolte, K.G., Determination of fracture parameters from fracturing
pressure decline.
SPE 8341, 1979.
4. Hagoort, J., Waterflood-induced hydraulic fracturing.
Ph.D. Thesis, Delft Technical University, 1981.
5. Van Everdingen, A.F. & Hurst, W., The application of Laplace
transformation to flow problems in reservoirs.
Transactions AIME (Dec. 1949), pp. 305-324.
6. Kucuk, F. & Brigham, E.W., Transient flow in elliptical systems.
SPE (June 1981), pp. 309-314.
7. Obut, S.T. & Ertekin, T., A composite system solution in elliptical flow
geometry.
SPE 13078, 1984.
8. Stehfest, H., An algorithm for numerical inversion of Laplace
transforms.
Communications ACM (Jan. 1970), pp. 47-49.
9. Geertsma, J. & Haafkens, R., Comparison of the theories to predict width
and extent of vertically hydraulically induced fractures.
Transactions ASME (March 1979), pp. 8-19.
10. Green, A.E. & Sneddon, I.N., The distribution of stress in the
neighbourhood of a flat elliptical crack in an elastic solid.
Proceedings Cambridge Phil. Soc. (1950), Vol. 46, pp. 159-163.
11. Muskat, M., The flow of homogeneous fluids through porous media.
McGraw Hill (1946), p. 185.
12. McLachlan, N.W., Theory and applications of Mathieu functions.
Oxford University Press, London, 1947.
13. The Group Numerical Analysis, On the computation of Mathieu functions.
Journal of Engineering Math. (1973), vol. 7, pp. 39-61.

197 -

14. Palmer, I.D. & Carroll, H.B.Jr., 3D hydraulic fracture propagation in


the presence of stress variations.
SPE/DOE 10849, 1982.

- 198 -

APPENDIX 4-A
SOLUTION FOR DIMENSIONLESS PRESSURE FUNCTION IN LAPLACE SPACE

Solution to diffusivity equation in elliptical coordinates


*
Employing a confocal elliptical coordinate system !
x = L cosh cos 17
(4-A-l)
y = L sinh sin r\
the dimensionless diffusivity equation becomes:

32p

32p

J- + Y ~ 2
3

9P

(cosn 2

*~

cos2r

) J~

3rj

(4-A-2)

where

2irkh ,

"qTT

(p p J

- i

(4-A-3)
t

.-*-*<j>uc.

T2

Lt

Defining the Laplace transform of p as:

(p

n>

Pn ( s )

(4-A-4)

Eq. (4-A-2) becomes in Laplace space


,2-

0 p

,2-

0 p

-2 + - ~

ar

= f (cosh 2| - COS2J7) P n

(4-A-5)

St)

* The periodic elliptical coordinate J? should not be confused with the


hydraulic diffusivity, for which we have also used the symbol t) in the
text.

- 199 -

This equation has the following general solution that is periodic in 77


with period JT:

PD(*,r?) = Z C2nCe2n(,-X) ce2n(7j,-X) + F 2 n P e k 2 n ( *'~ X)Ce 2n (T? '~ X,}


n=0
where X = -, C.

and F

f 4 ^" 6 )

are constants that are determined by the boundary

conditions. Ce , ce_ , Fek. are Mathieu functions with the following


, 2n
2n
2n
definitions :

ce (7/,X) = Z A
*n
r=0

(X) cos2rrj
r

OB

ce_ (TJ,-X) = (-l)n Z (-l)rA^n(X) cos2rr?


2n
_
2r
r=0
ce (0,X)ce (f,X)
T _
Ce.U,-X) = (-i)-2^
| S - 2 _ Z <-l)rA2n(X).
2n
2r
[A 2n (X)] 2
r=0
I r (Ae" { ) I r (Ae { )

,
Fek
2n

ce (0,X)ce_ (f ,X)
,
,v ,
, ,^n
2n
2n 2
_ ,2n.N.
($,-X)
= (-1)

Z
A X .
r 2n/% ,2
. 2r
it[h
(X)]
r=0
I (Ae~*) K (Ae*)

(A-4-7)

(X) are Fourier coefficients. We have calculated these coefficients with

the algorithm of Ref. 13. I and K are modified Bessel functions of the
r
r
first and second kind, respectively. Note that Fek
Ce

goes to zero whereas

diverges as goes to infinity.


The periodic Mathieu functions ce_

satisfy the orthogonality

relationships:
2ir

ce. (i|,-X) ce. (77,-X)d77 = {


2n
2m
0
ir
and have the integral:

m? n

(4-A-8)
m=n

- 200 -

2ir
i
ce
0

(j,-X) dr? = (-1) 2JT A n (X)

(4-A-9)

Solution to the composite problem in elliptical coordinates

The Laplace transform of the constant rate solution p

in the presence
7
of an elliptical discontinuity in fluid mobility is given by :

PfD = " { 1 6 X 2 * ^ ^
n=0

on(X,[C2nCe2n^w'-X>

+ F

2nFek2n<*w'"X>]^

<4"A-10>

where X = s/4. The functions Ce_ and Fek_ are defined in (4-A-7). is
v
2n
2n
' sw
the coordinate of the inner boundary. The prime denotes differentiation with
respect to .
The constants C. and P. are obtained from:
2n
2n
A
m

^n
A
=

2n

22 A 33 - A 23 A 32
det(A)

(4 A

11}

23 A 31 " A 21 A 33
det(A)

with

det(A) = A n ( A 2 2 A 3 3 - A 2 3 A 3 2 )

+ A12(A23A31-A21A33)

(4-A-12)

and
A
il

= [ ( - l ) n ( 2 X ) / A ^ n ( X ) ] Ce ( | , - X )
o
^n w

= [ ( - l ) n ( 2 X ) / A ^ n ( X ) ] Fek
O

Li.

21

Ce

22

Fek

(|
t-TL

,-X)
W

2n^o'"X)
2n^o'"X)
(4-A-13)

23

-Fek2n(o'-q)

A . . = icCe' ( { , - X )
31
2n o

- 201 -

A 3 2 = KFek'n({o,-X)

33

"

Fek

2n ( o'- q )

where q = XJ and K, $ and are given in Eq. (4.14). In Ref. 7 the


constants A

to A

are multiplied by a periodic Mathieu function. Since

these periodic functions drop out in Eq, (4-A-ll). we have omitted them
here.
Solution with closing fracture
Equation (4.12) is a convolution-type integral equation that also
occurs in the analysis of ordinary wellbore storage problems . Taking t
Laplace transform of Eq. (4.12) and solving for p. gives:

PfD< S)

~cr. v
PfD<s>
,
2-cr, |
1+C
fD S P fD ( S )

(4

"A"14)

Substitution of (4-A-10) into (4-A-14) then determines Pfn We have


evaluated (4-A-10) at = 0 and (4-A-14) was inverted numerically using the
W
8
Stehfest algorithm .

- 202 -

APPENDIX 4-B
RELATIONSHIP BETWEEN FRACTURE CLOSURE CONSTANT AND FRACTURE LENGTH
Assuming a uniform pressure p f inside the fracture, we have:
dVf
dt~

dVf
=

dpf
(4-B-l)

dp^ ' dt~

so that from Eq. (4.1),


dVf
Cf =

(4-B-2)

For a fracture of the CGK type (Fig. 4.5) the volume is given by,

= h

it w(o) L

(4-B-3)

9
with w(o) the half-width at the wellbore given by :

w(o) = 2 ( 1 ~"

L (Pf-SH)

(4-B-4)

Assuming that the height and length of the fracture remain constant during
the pressure decline and assuming that the pressure decline is sufficiently
fast that poro-elastic changes in S can be neglected, Eqs. (4-B-2), (4-B-3)
H
and (4-B-4) give:
CCGK = 2ff

1 1 ^

h f L2

(4.B.5)

For a fracture of the PKN type (Fig. 4.5) the volume is given by:
h

f
= 2L ir j=- w(o)

(4-B-6)

9
where the half-width at the wellbore is now given by :

w(o) = *

2
' hf (Pf~SH)

(4-B-7)

- 203 -

giving,

PKN
(1-u ) T . 2
Cf
= ff -*- * L h f

,,,
(4-B-8)

For an ellipsoidal fracture (Fig. 4.6) the volume is given by

- 1 '^

its " f V

with a and b the major and minor axis respectively of the elliptical
fracture area. E(k) is the complete elliptical integral of the second kind
and
2
b2
k = 1 - Sj

(4-B-10)

a
From Eqs. (4-B-9) and (4-B-10) we have:
i

elpsd
U
f

.i

2 ilzv}
3 *
E

.2 L h -

f
E(k)

2
K

h, 2

_f
1(
2L'

h
f
L >

elpsd _ 4
f
" 3

ff

(1-u) 2 V _
E
E(k)

(4-B-ll)

.2 _
2L 2
k - l ( )
h

f
Y >L

(4-B-12)

Expressions (4-B-4), 4-B-7) and (4-B-9) are valid for a uniform pressure p f
and a uniform rock stress S.
H
When the fracture length or height extends across a discontinuous change in
S, the expressions (4-B-4) and (4-B-7) for the half-width become different.
H
Suppose, for example, that the fracture extends vertically into cap and base
rock so that h

> h with h the reservoir height. If furthermore,

H = SH1

204

M
(4-B-13)

= S

Izl > -

H2

' '

the width equation (4-B-7) becomes

14
,

w(o) = = i hf.
(4-B-14)
^Pf-Sni) " \ ( S ^ - S ^ t c o s - V j - u I n C 1 ^ 1 - " >)}]
where
u = z~
h
f

(4-B-15)

However, from (4-B-14),

gal. nail hf

(1.B.U)

If it is assumed that the fracture retains approximately an elliptical shape


in spite of the stress contrast, Eq. (4-B-8) is still valid.
The same reasoning holds for Eq. (4-B-5) in the presence of a
horizontal stress contrast. It is therefore reasonable to assume that
Eqs. (4-B-ll) and (4-B-12) can also be used in the presence of stress
contrasts.
As discussed in the text, during the early part of the pressure
fall-off, we have:
d Pf
q = - Cf

(4-B-17)

The following dimensionless pressure decline coefficient can be formed:

(rm)
fD

__V
. 3 2X
hf (1-u )

d Pf
where C f = - q/^T-

(4-B-18)
'

- 205 -

From Eqs. (4-B-5), (4-B-8), (4-B-ll) and (4-B-12) a plot of C** ' vs
2L/h, can be made for the various fracture geometries (Fig. 4.7). This plot
can be used to determine L or h

as described in method 1 in the text.

- 207 -

CHAPTER FIVE
A PRACTICAL APPROACH TO WATERFLOODING UNDER FRACTURING CONDITIONS

Summary
5.1 Introduction
5.2 An example
5.3 Conditions for a successful process
5.4 Preliminary investigations
5.5 Basic data gathering
5.5.1 Measurements of in-situ stress, fracture orientation
and elastic moduli
5.5.2 Injectivity test and fall-off testing
5.5.3 Matching with propagation model
5.6 Determination of optimal reservoir pressure, injection rate
and well pattern
5.6.1 Determination of maximum reservoir pressure
5.6.2 Determination of maximum injection rate
5.6.3 Fractured well pattern and reduction in the number of wells
5.7 Full-scale implementation
5.8 Special applications
5.9 Conclusions
List of symbols
References

Appendix 5-A Determination of the ratio of producers to fractured injectors


and the reduction in the number of wells

- 208 -

SUMMARY

A practical approach to waterflooding under fracturing conditions is


presented. It is shown that injecting under fracturing conditions has a large
potential for a reduction in the total number of wells. The process can, in
many cases, be designed in such a way that sweep efficiency is unimpaired and
vertical fracture containment is ensured. Design formulae are presented for
calculating the optimum reservoir pressure, injection rate and well pattern.
The case of a reservoir at hydrostatic pressure with a typical gradient for
the horizontal reservoir rock stress of 0.17 bar/m is investigated in detail
for unit mobility ratio. With the stress in cap and base rock exceeding that
in the reservoir by 14% or more, the most attractive options are a fractured
7-spot and a fractured 9-spot with a maximum reduction in the number of wells
of 40% and 33%, respectively. For more moderate stress contrasts the reservoir
pressure has to be lowered to ensure vertical containment. With the horizontal
rock stress in cap and base rock exceeding that in the reservoir by only 9% a
fractured 13-spot at a lower reservoir pressure results in a 16% reduction in
the number of wells. If both injectors and producers are fractured and if the
fractures at the production wells are of the same length as the stabilised
fracture at the injector, the reduction in the number of wells can reach a
maximum of 67%. In-situ stress measurements should be performed in the
reservoir and cap and base rock. The stress data are essential for an optimal
design of the process. An injectivity test should be carried out to establish
a trend in fracture propagation pressure. Regular fall-off testing can be used
to determine the change in fracture dimensions with time. A match with the
analytical propagation model results in values for the thermo- and poroelastic constants. From the model the maximum fracture propagation pressure
can be determined that will occur during fracture propagation under balanced
conditions. This quantity is important for determining the maximum reservoir
pressure for which fracturing occurs under the condition of vertical
containment. In the field, the value for the in-situ stress in cap and base
rock should serve as a maximum for the downhole injection pressure. Lateral
uniformity of this stress should be checked with sonic logging and, if
possible, with leak-off tests. It is shown that, with the injection rate and
injection pressure constrained to their appropriate maximum values, fracture
growth will occur in a completely controlled manner.

- 209 -

A PRACTICAL APPROACH TO WATBRFLOODING UNDER FRACTURING CONDITIONS

5.1 INTRODUCTION
In the previous chapters the physics of fracture initiation and
propagation were described in detail. A method for monitoring fracture length
by conducting a fall-off test was discussed.
This chapter shows how the concepts developed so far can be applied in
practice.
It is outlined what the advantages may be of waterflooding under
fracturing conditions and how the process should be designed in order to
control the fracture growth.

5.2 AN EXAMPLE
Let us consider a tight chalky limestone reservoir with a permeability of
a few millidarcies. The reservoir contains light oil with a viscosity of the
order of a few centipoise. There is a bottom water layer but almost no aquifer
support. The latter circumstance, together with the favourable water-oil
mobility ratio, makes the reservoir a good candidate for waterflooding.
For this low-permeability reservoir hydraulic fracturing would be an
effective method of increasing the flow capacity of injection and production
wells significantly. Such an increase' could lead to a reduction in the number
of wells at a certain total field offtake or to accelerated oil production at
a certain number of wells.
Fracturing of the production wells, however, entails the risk of creating
a fracture that extends into the bottom water layer, which would result in
increased water production . Moreover, the limestone may be rather soft, so
that the fracture would eventually close around the proppant.

This is, of course, dependent on how much a poor vertical permeability


suppresses coning in the unfractured case.

- 210 -

These problems do not occur with fracturing of the injectors. Injection


into the water layer is generally no problem and injection at the fracture
propagation pressure ensures that the fracture stays open.
Suppose that at the prevailing reservoir pressure the maximum
production rate for a typical producer equals approximately the maximum
injection rate for a typical (non-fractured) injector. Therefore, to avoid
pressure depletion of the reservoir, as many injectors are required as there
are producers. This makes an arrangement of confined five-spots the most
suitable well pattern (Fig. 5.1).
If, by injecting under fracturing conditions, the injection rate can be
increased by, for example, a factor of three, an inverted nine-spot would be
possible (Fig. 5.1). For a certain total field offtake only a third of the
injectors would be required compared to the five-spot pattern (for
simplicity's sake we discard deviations from symmetry near the field
boundaries).

5.3 CONDITIONS FOR A SUCCESSFUL PROCESS

For waterflooding under fracturing conditions to be a successful


process two conditions must be met:

1) The fracture should grow sufficiently slowly for the waterflood sweep
efficiency not to be adversely affected.
2) The fracture propagation pressure should be low enough to keep the
fracture vertically contained within the reservoir zones of interest.

5.4 PRELIMINARY INVESTIGATIONS

A preliminary assessment of the maximum injection rate for which


fracture propagation does not adversely affect the sweep efficiency can be
made by looking at the dimensionless injection rate as defined in Chapter 3,
section 3.5.1. It is given by

T>

2jrkh(p - P i )

(5.1)

v<p

J/p

"VS*

211 -

%'AP

Vo

.V*P

/4P

'*p

*?

CONFINED 5-SPOTS
BALANCED CONDITIONS : L= p
X = INJECTOR
= PRODUCER

v,

Vjpg

CONFINED INVERTED 9-SPOTS


BALANCED CONDITIONS: L = 3p
L = INJECTION RATE
P = PRODUCTION RATE

RATIO OF INJECTION AND PRODUCTION RATES


FOR TWO ALTERNATIVE WELL PATTERNS

FIG. 5.1

- 212 -

where:

= injection rate

= reservoir height

= fracture propagation pressure

p. = prevailing reservoir pressure


k/u = mobility of injection fluid
If q

is less than 0.61, the pressure transients travel radially into the

reservoir before a situation of steady-state is reached. Therefore, fracture


growth will have no effect on sweep efficiency, regardless of fracture
orientation. It is assumed that, at the prevailing reservoir pressure p.,
injection and production are balanced so that after the onset of steadystate the fracture hardly grows any further.
jnay

The maximum injection rate, q

, for injecting under fracturing

conditions at reservoir pressure p., thus becomes:


q m a X = 0.61 2rrh jjj ( P ^ )
In a first estimate of q

(5.2)

, the propagation pressure, p f , may be

approximated by the horizontal reservoir rock stress, S . For some areas


correlations are available for an initial estimate of S 1). S lies
H
H
typically in the range of 0.16-0.19 bar/m.
The maximum injection rate for a non-fractured well, q

, may be

estimated from Eq. (3.62) in Chapter 3, section 3.5.1,


~max
k
1
q
= 2^h - (SH-Pi) . l n ( 0 . 5 4 . d / r , +
w

(5.3)

where:
d

= distance from injector to producer

r = wellbore radius
w
S = skin of injector

* In the following p. will denote the prevailing reservoir pressure, i.e.


the reservoir pressure at the moment of analysis. Therefore, if the field
has been on production for some time, this pressure may be different from
the original (i.e. initial) reservoir pressure.

- 213 -

Initially, of course, the well spacing still has to be determined.


However, the logarithmic term in (5.3) has a typical value of about 7. In a
limestone reservoir the wells.are mostly stimulated with an acid treatment
giving a typical skin of S = -2. On the other hand the well may have some
mechanical damage giving for instance a skin of S = +1.
Prom (5.2), with p = S , and (5.3), a typical improvement in

H
injectivity by injecting under fracturing conditions is found to be:
max

s= 2

< ->

3 K

^ziz K 5

< s = +1 >

<5-4>

~max
q
where the logarithmic term was taken as 7.
We see from (5.4) that injecting under fracturing conditions, as far as
unimpaired sweep efficiency is concerned, results, in most cases, in a
considerable improvement in injectivity.
With regard to vertical fracture containment the situation is more
difficult. If cap and base rock consist of thick shales, there is a good
chance of vertical containment. However, as was shown in Chapter 3,
section 3.4.3, the reservoir rock stress may increase considerably, during
injection into low-permeability reservoirs, as a result of poro-elastic
effects whereas the stress in cap and base rock may be lowered as a result
of conductive cooling. The altered stress gradients may eventually result in
vertical fracture growth through cap and base rock.
A special case in which vertical containment can be assumed with
reasonable certainty is for permeable reservoirs where fracturing is
preceded by the injection of large amounts of cold water. In this way the
cooling can induce a considerable lowering of the initial stress with
respect to cap and base rock whilst at the same time the poro-elastic
increase is small owing to the good permeability (see Chapter 2,
section 2.7).
If the initial conditions are not optimal for vertical fracture
containment it may be considered to lower the reservoir pressure in order to
lower the reservoir rock stress level. Of course, a lower reservoir pressure
results in a reduced productivity of the wells. However, waterflooding under
fracturing conditions at a lower reservoir pressure may still be very
advantageous. As will be shown in the following, a proper assessment can
only be made after all basic data have been gathered.

- 214 -

5.5 BASIC DATA GATHERING


5.5.1 Measurements of in-situ stress, fracture orientation and elastic
moduli
To be able to predict fracture propagation as a function of injection
rate the in-situ stress in the reservoir needs to be determined. In order to
investigate vertical fracture containment the in-situ stress in cap and base
rock must be measured.
At present/ the only reliable way of determining in-situ stresses is by
2 3
creating microfractures ' . The part of the formation to be tested is
straddled off with a dual packer assembly and subsequently pressurised by
the injection fluid at a low rate. After formation breakdown has occurred
the created fracture is propagated for a short period of time. Pumping is
then stopped and the pressure is allowed to decline as a result of fluid
leak-off into the formation. The propagation and leak-off sequence is
repeated for several cycles. An analysis of accurately recorded downhole
pressure data allows the pressure at which the fracture closes to be
determined. This closure pressure equals the in-situ stress.
From the difference between the fracture propagation pressure and the
in-situ stress, together with an estimate of the fracture radius, the
critical stress intensity factor, K

, can be determined. This quantity may

be important in determining the expected maximum propagation pressure when


waterflooding under fracturing conditions (see below).
If the microfracture is created in open hole, the fracture orientation
can be determined by running an impression packer with a soft rubber sleeve.
Inflation of the packer imprints a trace on the rubber of the intersection
of the fracture with the wellbore. Acoustic logging with a borehole
televiewer or detection with geophones of the seismic signals created during
propagation are alternative methods. For a survey of these techniques see
Ref. 4. This reference also discusses wellbore ellipticity measurements
together with sonic velocity and differential strain measurements on
oriented cores for a field-wide determination of the direction of minimum
horizontal rock stress and thereby, indirectly, of fracture orientation.
Recently , the possibility of determining fracture orientation with the
formation micro scanner (FMS) has been discussed. This high-resolution
resistivity log should be able to detect the intersection of the fracture
with the borehole wall as a channel of increased conductivity.

- 215 -

With the array sonic logging tool the dynamic elastic moduli of the
rock can be determined . Ideally, one should like to use these for a direct
calculation of the in-situ stress. It should be noted however, that
generally these constants are different from those measured under static
conditions. The latter are more relevant for investigating the stress state
in a rock. Quite apart from this difference, it is not recommended to
compute the in-situ stresses using the in-situ values of the elastic moduli
since this ignores the burial history of the formation (see Chapter 2,
section 2.6). However, if in the same field the difference in the dynamic
elastic moduli for two adjacent layers is measured as uniform across the
field, then, most likely, the same holds for the difference in stress.
Therefore, sonic logging can be a powerful method for investigating the
lateral uniformity of stress contrasts across the field.
5.5.2 Injectivity test and fall-off testing
In the same well in which the stress measurements were made an
injectivity test should be carried out to determine the characteristics of
fracture propagation. Preferably the well is a new well in a virgin part of
the reservoir, so that the measured stress is the actual initial stress
unaltered by changes in pressure or temperature.
The well should be perforated over the full reservoir interval in order
to initiate a fracture across the total reservoir height. This facilitates
the interpretation of the injectivity test and the accompanying fall-off
tests considerably.
First, a reference case for fall-off testing should be established by
performing a fall-off test after injection below the fracture initiation
pressure. This fall-off can be analysed for kh and skin in the usual way.
Before the well is shut in, a base case spinner survey should be made to
determine the flow distribution along the reservoir interval.
Then an injectivity test should be performed at the rate that is
desired in the actual waterflood. Of course, this rate should not exceed
q

as given in Eq. (5.2). The water should have the quality and

temperature that it would have in the full-scale project. The injection


under fracturing conditions should last sufficiently long so that a trend in
propagation pressure can be established (typically from two to six weeks).
Several fall-off tests may be carried out during this period in order to
determine the change in fracture length and, possibly, the change in
horizontal rock stress with time.

- 216 -

A day or so before a fall-off test will be carried out injection is


stopped to lower a high-resolution pressure gauge and spinner into the hole.
After injection has resumed a spinner survey is carried out to see whether
the flow distribution is different or the same as the base case indicating
partial or complete extension, respectively, of the fracture towards cap and
base rock.
The pressure fall-off can be analysed with the method of Chapter 4 to
determine the dimensions of the induced fracture. If the actual closing of
the fracture is observed as a peak on a logarithmic pressure derivative
plot, the horizontal reservoir rock stress can be determined. A comparison
with the initial stress as determined in the microfracture test then
indicates possible poro- and thermo-elastic changes.
If kh has significantly increased compared to the reference fall-off
case, the fracture has most likely propagated into another permeable zone.
If the period of afterflow from the wellbore into the reservoir is
expected to be large with respect to the period of fracture closure one
should ideally use a downhole shut-off device to shut the well in. As an
alternative, however, simultaneous pressure and spinner data together with
the technique of deconvolution can be used to eliminate the effect of
afterflow. This modification of the method of Chapter 4 was recently
presented in Ref. 7, where the analysis of a fall-off test in a thermally
fractured injector is discussed.
If from the results of the in-situ stress measurements there appears to
be a chance that, during the injectivity test, the fracture will propagate
extensively into cap and base rock, an attempt to monitor vertical fracture
growth may be considered. Possible methods include high resolution
o

temperature logging , gamma-ray logging after the injection of a r a d i o g

active tracer and array sonic logging. It has been recently suggested that
the latter method should be able to detect fracture height as a result of
9
the interference of the shear-wave propagation by the fracture . For all
three methods reference logs should be run before fracturing. During the
test the logging should be repeated at regular intervals. In some cases it
may be possible to perforate the casing at a suitable depth opposite the
caprock. Propagation of the fracture upto this level would then be

visible

from a change in annulus pressure.


The advantages of monitoring the vertical fracture growth are twofold:
a) it may be possible to terminate the test before unwanted communication
with other reservoirs is established and

- 217 -

b) the maximum injection pressure may be established for which no such


communication will occur.

5.5.3 Matching with propagation model


If the fracture half-length obtained from the fall-off test at the end
of the injection test satisfies:
L = /, L,. T < 0.58

(5.5)

/(Vsh>

where n is the hydraulic diffusivity in the outer fluid zone and t . the
2
sn
total injection time, then according to Chapter 3, aection 3.2.4, flow
around the fracture was pseudo-radial. This means that no effect on sweep
efficiency is expected if injection and production in the pattern are
balanced. It also means that it should be possible to match the propagation
pressure as a function of time with the analytical model of Chapter 3,
section 3.4. The prediction for fracture length should, of course, agree
with the results of the fall-off tests.
A match with the propagation pressure in time should give reliable
values for A

and A , the thermo- and poro-elastic constants respectively.

If possible, these should be compared with laboratory measurements on


reservoir core material.

5.6 DETERMINATION OF OPTIMUM RESERVOIR PRESSURE, INJECTION RATE AND WELL


PATTERN
5.6.1 Determination of maximum reservoir pressure
With all the basic data available the maximum reservoir pressure can be
determined for which the fracture propagation pressure is still low enough
to guarantee vertical fracture containment.
According to Chapter 3, section 3.3.1, the fracture propagation
pressure, p , at a certain reservoir pressure p is given by:
Pc = STI + Ao
f

yp

IC

+ Ao m +
yT

(*,)

(5.6)

- 218 -

where:

S
H
La
yp
Ac
yT
K
L

= horizontal reservoir rock stress at reservoir pressure p


= poro-elastic stress change at fracture face
= thermo-elastic stress change at fracture face
= critical stress intensity factor
= fracture half-length

The poro-elastic stress change is given by:

Ao

yP

where A

=c

' V (p f " p)

(5>7)

is the poro-elastic constant and c is a number in the range

0 < c < 1.0.


The actual value of c depends on the ratio of the pressure penetration
depth to the reservoir height and on the ratio of the pressure penetration
depth to the fracture length (see Chapter 3, section 3.3.2).
When the penetration depth of the pressure transients is much smaller
than the reservoir height and the fracture length, c is close to zero. When
the pressure penetration depth becomes very large with respect to the
reservoir height and the fracture length, c is close to 1.0. In most cases
La

reaches a maximum when steady-state is reached in the pattern and the

yp
fracture becomes stable. At that stage the pressure penetration depth, R ,
e
is of the order of half the well spacing, d. In Table 3-1, Chapter 3,
section 3.3.2, the poro-elastic stress change at the fracture wall is given
as a function of h/(2R ) with h the reservoir height. The value of c can be
e
inferred from this table by calculating the ratio La /Ap (o) where La
1
ypD *D
ypD
is the dimensionless poro-elastic stress change and Ap (o) is the
dimensionless pressure rise in the fracture respectively. When the fracture
becomes stable we have h/(2R ) ~ h/d and for typical values of this ratio
e
between 1.0 and 0.1, c ranges between 0.4 and 0.7.
From (5.6) and (5.7) the pressure propagation pressure at reservoir
S
H " C V + AvT + K T C / , / ( , r L )
pressure p can be Pwritten:
5 8

*t = -

i-

<->

- 219 -

If in the in-situ stress test the magnitude of the horizontal reservoir


rock stress was determined to be S . at the prevailing reservoir pressure
p., we have for the reservoir rock stress at reservoir pressure p:

H " SHi

+ A

AP

(5 9

' >

where Ap = p - p..
The fracture propagation pressure, p. at reservoir pressure p. can be
obtained from (5.8) by replacing S with STI. and p with p.. We thus find
H

Hi

from (5.8) and (5.9) for the difference in propagation pressure at reservoir
pressure p and p.:
(1-c) A

Pf

=p

f T^T^

Ap

(5 10)

P
Since cA

< 1, according to (5.10), a lower reservoir pressure results in a

lower fracture propagation pressure.


For the fracture to be vertically contained, the maximum propagation
-max

pressure, p.

. . . . , ,

, that will occur during injection at reservoir pressure, p,

must be lower than a certain threshold value.


Using the analytical fracture propagation model, the behaviour of the
propagation pressure with time as observed in the injectivity test can be
matched (see previous section). The model can subsequently be used to
j

i.

determine the maximum propagation pressure p.

J.

at reservoir pressure p. .

This should be done as follows. Estimate the maximum penetration depth that
the pressure transients will reach. As discussed above this will typically
be half the well spacing. Modify the model such that after the pressure
penetration depth, R , has reached this maximum, R is kept constant at this
e
e
value. Run the model for the period of time that is expected for water
breakthrough to occur in the pattern. Of course, the modification of the
model should be such that, after steady-state has been reached, the
calculation of the pressure difference between the fracture and R , takes
e
only the discontinuities in fluid mobility contained within R into account.
e
It may be, therefore, that at the end of the run only cold water is
contained within R . This way of using the analytical model is similar to
that described in Chapter 3, section 3.5.2 where it was used to determine
the fracture length at water breakthrough.

- 220 -

Initially/ of course, the well spacing is not known and therefore the
model should be run for an expected range of well spacings.
It should be noted that the calculated maximum propagation pressure,
pf

, may never be reached during the actual injectivity test, for instance

because vertical propagation into another reservoir occurred. On the other


hand p

may also be exceeded in the test because no situation of balanced

injection and production could be created and therefore no steady-state


could occur.
When the maximum propagation pressure is determined from the model the
corresponding value for c can also be obtained. From (5.10) the expected
-max

i.

i.1.

maximum propagation pressure, p. , at reservoir pressure p can then be


obtained from p
according to:
-max
P
f

= P

(1-c) A Ap
p
1-C A

max
f +

, c .. n ,
'11}

(5

As was discussed above, c typically ranges between 0.4 and 0.7. For c = 0.5
we have:
A
-max
max
p A
Pf
= Pf
+ JZ^~ A P
P

. _n.,
(5.12)

From the in-situ stress measurements the horizontal rock stress in cap and
c/b
base rock, S
, is known. Vertical fracture containment at reservoir
H
pressure p is guaranteed if

PTX < sZ/b

(5.13)

From (5.11) and (5.13) we can determine the maximum reservoir pressure,
p
, for which the maximum propagation pressure satisfies (5.13). Putting
-max _ S_c/b . , c ,, .
, , .
- .
c
p.
u
in (5.11) and solving for
p gives:
r

ti

-max

_L

=p

1cA

i T^TT

, c/b

(S

H "

max,

.c

(5 14)

which gives with c = 0 . 5 :

-max
P

^
= P

p
A

/ f ,c/b
(S

"

max,
f 5

( 5

'15)

..

- 221 -

The significance of (5.14) or (5.15) is that for any reservoir pressure


below p

vertical fracture containment is guaranteed. This means, for


c/b
instance, that if the initial stress contrast between S. and S
is not
H .
large enough to contain the fracture so that therefore p, > S
, a lower
r
H
reservoir pressure can be determined from (5.14) at which vertical

propagation into cap and base rock will not occur. On the other hand, for
large initial stress contrasts, (5.14) may be used to determine the rise in
reservoir pressure that is allowed before fracturing into cap and base rock
occurs.
If the vertical fracture growth was monitored during the injectivity
test then the maximum injection pressure may have been established for which
the fracture propagated into cap and base rock but not far enough to
establish communication with other reservoirs. In that case this pressure
c/b
can be used in (5.13) rather than S
.
H

5.6.2 Determination of maximum injection rate


For reservoir pressure p, with p < p
corresponding maximum injection rate, q

, we can calculate the


, for which the fracture

propagates in the pseudo-radial mode or,, in other words, for which the sweep
efficiency is unaffected. From Section 5.4, Eq. (5.2) we have:

= 0.61 2ffh - (pf-p)

where the propagation pressure p

(5.16)
is defined in (5.6).

To be on the safe side the expected minimum propagation pressure may be


used in (5.16). This minimum may be calculated by determining the minimum
propagation pressure at reservoir pressure p. as predicted by the analytical
model and using (5.10) to go back to reservoir pressure p.

5.6.3 Fractured well pattern and reduction in the number of wells

From the observed production behaviour of a typical well at reservoir


pressure p., a typical productivity index (PI) can be determined. It is
defined by: PI = q/(p

-p.) with q the production rate and p


wp r i

wp

the

- 222 -

corresponding bottomhole pressure in the production well. From the PI the


maximum production rate q

at reservoir pressure p can be obtained.

Depending on the ratio of production wells to fractured injection


wells, R, at reservoir pressure p, with R given by

_
-max
R =3
%

(5.17)

the appropriate well pattern at reservoir pressure p, with p < p

, can be

determined. Subsequently the reduction in the number of wells compared to


injection at non-fracturing conditions can be calculated. Finally, the
fractured well pattern and the reservoir pressure below p

that give the

maximum reduction in the number of wells can be selected.

A detailed_example
Let us return to our example in Section 5.2 in more detail. It was
assumed that at the prevailing reservoir pressure, p., the maximum injection
rate, q

, into a non-fractured injector equals the maximum production

rate, q , for a producer. Let us furthermore assume that the injectivity


index (II) of a typical non-fractured injector equals the PI of a typical
producer. The II is defined as: II = q./(P -~P) with q. the injection rate
and p . the corresponding bottomhole pressure in the injection well. This
means that at reservoir pressure p. a pattern of five-spots is the optimal
non-fractured well pattern and furthermore the total number of wells
required to provide a certain field offtake Q, while balancing injection and
production, is at a minimum. Any other reservoir pressure requires more
(non-fractured) wells. Let us assume that an improvement in injectivity of a
factor of three can be obtained by injecting under fracturing conditions
without impairing sweep efficiency, i.e.,
max
3

= 3

(5.18)

-max
q
According to Section 5.4 this is typical for the situation where the
unfractured well had a skin of S = -2.
In Appendix 5-A we have calculated the ratio of producers to fractured
injectors, R, as a function of the change in reservoir pressure Ap, with

- 223 -

Ap = p-p.. We have also calculated the reduction in the number of wells that
can be obtained by injecting under fracturing conditions at reservoir
pressure p compared to the number of non-fractured wells required at
reservoir pressure p..
We introduce the following definitions:
N

= total number of non-fractured wells at reservoir pressure p.,


required for a certain field offtake Q.

= total number of wells, required for a certain field offtake Q, while


injecting under fracturing conditions at reservoir pressure p.
= ratio of required number of producers at reservoir pressure p to the
P

F.

required number of producers at reservoir pressure p..


= ratio of required number of fractured injectors at reservoir pressure
p to the required number of producers at reservoir pressure p..

Ap

= dimensionless change in reservoir pressure defined as:

APn =
D

with p f

= ,

r^
(Pf

in

AP

(5.19)

( P i _ P w )

-Pi)

the minimum pressure required for fracture propagation at

reservoir pressure p. and p


1

the minimum bottomhole pressure in the


W

producers.
From the appendix we have:
F

= (1 + A p D ) - 1

F. = (3 - 1.98 A p D ) _ 1

(5.20)
(5.21)

Using (5.20) and (5.21) the ratio of producers to fractured injectors and
the reduction in the number of wells can be obtained from:
F
R = -fi

(5.22)

N - N
- = 1 - 0.5 (F. + F )
N
i
p

(5.23)

- 224 -

The minimum propagation pressure p f

/ enters the calculations because it

was used to determine the maximum allowable injection rate under fracturing
conditions in accordance with Section 5.6.2. In translating p f

to

different reservoir pressures Eq. (5.10) was used in Section 5.6.1 with
c = 0.5 and A = 0.5 as typical values.
P
It should be noted that the reduction in the number of wells is
determined by comparison with the optimal non-fractured case. It may not
always be possible to operate the reservoir at the corresponding optimal
reservoir pressure in which case the reduction in the number of wells as
calculated in (5.23) is too pessimistic.
The ratio of producers to fractured injectors and the reduction in the
number of wells as a function of the dimensionless change in reservoir
pressure have been plotted in Pig. 5.2 . The dotted curves represent
Eq. (5.22) and (5.23). For certain values of Ap
non-integer values. For instance, at Ap

the ratio R takes on

= 0.1, R equals 2.5. This means

that the injection rate into a fractured injector balances the production
rate of 2.5 producers. This injection rate is the maximum, as calculated
from Eq. (5.16), section 5.6.2, that guarantees pseudo-radial flow. Higher
injection rates would influence the sweep efficiency, unfortunately no
regular well pattern can be constructed for which R = 2.5. Since higher
injection rates are not allowed, the only solution to obtain a regular well
pattern at this particular reservoir pressure, is to lower the injection
rate to twice the production rate per well. The penalty, of course, is that
more injection wells have to be drilled to maintain balanced conditions and
therefore ,the reduction in the number of wells becomes smaller.
This effect is expressed by the solid curves in Fig. 5.2. The reduction
in the number of wells for these cases was calculated from

~ N = 1 - 0.5 F (1 + 1/R)
N

with F

(5.24)

from (5.20). R was constrained to integer values for which a regular

well pattern can be constructed. The corresponding well patterns are shown
in Fig. 5.3.

* Note that when p is small compared to p. as, for instance in an


optimal gaslift design, Ap in Eq. (5.19) is approximately equal to the
change in reservoir pressure as a percentage of p..

225

DIMENSIONLESS CHANGE IN RESERVOIR PRESSURE


0-4

-0-2

-04

-0-2

00

0-0

0-2

0-2

0-4

0-6

0-4

0-6

0-8

0-8

DIMENSIONLESS CHANGE IN RESERVOIR PRESSURE

Fig.5-2 RATIO OF PRODUCERS TO INJECTORS AND REDUCTION IN THE


NUMBER OF WELLS AS A FUNCTION OF RESERVOIR PRESSURE

- 226 -

INVERTED

INVERTED 5-SPOT
R = 1

7-SPOT

R =2

INVERTED

INVERTED 13-SPOT

9-SPOT

R =5

R =3

= PRODUCER
/

= FRACTURED INJECTOR
R = RATIO OF PRODUCERS TO FRACTURED INJECTORS
IN A MULTIPLE PATTERN SYSTEM

Fig.53 POSSIBLE WELL - PATTERNS FOR WATERFLOODING


UNDER FRACTURING CONDITIONS

- 227 -

Fig. 5.2 clearly shows that waterflooding under fracturing conditions,


as far as no impairment of sweep efficiency is concerned, offers an enormous
scope for the reduction in the number of wells. The reduction in the number
of wells for non-integer R reaches a maximum of 42% at Ap

= 0.45. This

curve has a maximum since according to Bq. (5.20) and (5.21) fewer producers
and more injectors are required as the reservoir pressure increases and vice
versa.
*

If a strictly regular well pattern

is adhered to the maximum reduction

in the number of wells is 40%. This reduction can be obtained with a


fractured 5-spot or a fractured 7-spot with the

latter being operated at a

lower level of the reservoir pressure.


As discussed in Section 5.6.1 vertical fracture containment is only
ensured for reservoir pressures p such that p < p

.In the terminology of

Fig. 5.2 this means that we must define a maximum dimensionless change in
reservoir pressure Ap

. The only cases that satisfy both the conditions of

vertical containment and unimpaired sweep efficiency are located in the part
of the graph for which Ap

< Ap

We take again c = 0.5 and A

= 0.5 as typical values. Let us

furthermore assume that the thermo-elastic stress reduction, Lo

, and the

critical stress intensity factor, KTr/ are small, so that we have from
Eq. (5.8), section 5.6.1:

max

=p

mm

2S. - A p.
Hi
pi

T^T~
p
= 1.33 S. - 0.33 p.

(5.25)

Hi

From Eq. (5.15) and (5.25) we have:


-max

max

A
AP

->c/b

iZi

3S

4S

Hi * P i

=
* " ^
1.33(8,. - P.)
Pf
" Pt
Hi
*l
c/b
with S
the horizontal rock stress in cap and base rock.

(5

'26)

* In practice, deviations from a strictly regular pattern are unavoidable


due to the presence of already existing wells, reservoir boundaries or
possible faults. Of course, such deviations occur both for fractured and
non-fractured well patterns.

- 228 -

Since we had assumed that the II and PI are equal and that the
prevailing reservoir pressure, p., equals the optimum pressure for injecting
under non-fracturing conditions we have (see Appendix 5-A, Eq. (5-A-2)):
min ,
P<= + P.
p

with p

(5.27)

the minimum bottomhole pressure in a typical producer.

Assuming a gradient for S . of 0.17 bar/m and assuming that the wells
are produced by gaslift so that the bottomhole pressure has a typical
gradient of 0.016 bar/m, we find for p., from (5.27) and (5.25) a gradient
of 0.10 bar/m, i.e. the reservoir pressure, p., is practically hydrostatic.
For a typical range in gradients for the stress in cap and base rock, we
,. , max
find for AAp
:
S/b/depth
(bar/m)

s c/b /s
H

Apmax

Hi

0.170

1.00

- 0.78

0.175

1.03

- 0.61

0.180

1.06

- 0.44

0.185

1.09

- 0.28

0.190

1.12

- 0.11

0.195

1.15

0.06

0.200

1.18

0.22

Hi
=0.17 bar/m
depth
P

i
= 0.10 bar/m
depth

It can be inferred from the table above and from Fig. 5.2 that if,
initially, the horizontal rock stress in the reservoir and cap and base rock
are equal the lowering of the reservoir pressure to a level that guarantees
vertical containment does not result in a reduction in the number of wells.
c/b
However, when there is a moderate stress contrast, with S
/S . = 1.09, the
H
Hi
required lowering of the reservoir pressure still results in a reduction of
the number of wells of some 16%. The most suitable well pattern in this case
c/b
is the inverted 13-spot. With S_ /S. = 1.14 we find from Eq. (5.26) that
H
HI
Ap D
= 0. Therefore, with a stress contrast of 14% the prevailing reservoir
pressure is the optimal pressure for a fractured 9-spot giving a maximum
reduction in the number of wells of 33%. For larger contrasts with
c/b ,
S
/S
> 1.18 the reservoir pressure can be increased to obtain the
ii

HI

- 229 -

maximum reduction in the number of wells of 40%. The most suitable well
pattern in this case is the inverted 7-spot because it requires a lower
reservoir pressure than the 5-spot.
We see from the table above that Ap

is extremely sensitive to the

actual stress contrast. This emphasizes the importance of accurate stress


data for the optimal design of waterflooding under fracturing conditions.
Comgarison with_Hagoort's_work
Hagoort, in his thesis

, compared the conductivities for fractured and

non-fractured five-spots as a function of L/d with L the stable fracture


half-length during steady-state and d the well spacing. The conductivity of
a pattern is defined as
C =

p .- p
wi
wp

(5.28)

with q the steady-state injection rate into the pattern and p .-p
the
wi wp
difference between the bottomhole pressures in the injector and the
producer.
As was shown in Chapter 3, section 3.5.1, for a dimensionless injection
rate of q

= 0.61, which is the largest possible injection rate that leaves

sweep efficiency unimpaired, a stable situation will be reached in the


5-spot with L/d = 0.21. According to Hagoort's results this gives an
improvement in conductivity compared to the non-fractured five-spot of
C/C = 1.66 with C the fractured five-spot conductivity. This gives for the
reduction in the number of wells at a certain field offtake:

~ N = 1 - C/C = 0.40

(5.29)

which agrees with the results of Fig. 5.2 for a fractured 5-spot at the
maximum injection rate that leaves sweep efficiency unimpaired.
Hagoort's method of only comparing conductivities for different ratios
of L/d has the disadvantage that the corresponding difference in reservoir
pressure is not explicitly considered. When the reservoir pressure is
explicitly calculated as in Pig. 5.2, it becomes apparent that a fractured
5-spot is not a realistic option since for an attractive improvement in

- 230 -

conductivity a relatively high reservoir pressure is required which would


result in most cases in a loss of vertical fracture containment.
Hagoort also considered a doubly fractured 5-spot in which both the
injector, and producer are fractured. Of course, for L/d = 0.21 for both
injector and producer a doubly fractured 5-spot is possible at the same
reservoir pressure as the non-fractured 5-spot (equal fractured II and PI).
With the same conditions as considered in Fig. 5.2 only a third of the
injectors and a third of the producers are required as compared to
non-fracturing, giving a reduction in the total number of wells of 67%.
Therefore, the fact that mainly 9-spots and 7-spots appear as viable options
in Fig. 5.2, strongly reflects the fact that only the injector is fractured.

5.7 FULL-SCALE IMPLEMENTATION

In the previous section it was shown how the reservoir pressure can be
determined for which the reduction in the number of wells is optimal while
the conditions of unimpaired sweep efficiency and vertical fracture
containment are satisfied. If there is only a moderate contrast between the
horizontal rock stress in the reservoir and that in cap and base rock the
reservoir pressure may have to be lowered. Whether this is a realistic
option depends of course very much on the bubble-point pressure of the oil.
If the reservoir is initially at or above bubble-point pressure, some
pressure depletion may in fact be advantageous to the oil recovery since, in
parts of the reservoir where the pressure is low enough, trapped gas and
residual oil may coexist after the water front has passed, resulting in a
lower residual oil saturation

. Too large a pressure depletion however may

result in the creation of a secondary gascap which could provide a


preferential flow path for the water and/or gas and could result in
accelerated water or gas breakthrough. In other words, whether a change in
reservoir pressure is permitted depends very much on its effect on oil
recovery.
Let us consider how the process of waterflooding under fracturing
conditions at a lower reservoir pressure could be implemented in practice.
Initially only the producers have to be drilled since some pressure
depletion is desired. This has the advantage that the investment in the
injection wells and facilities can be delayed, resulting in improved
economics. When water injection starts, the downhole injection pressure has

- 231 -

c/b
to be constrained to a maximum S
, the horizontal rock stress in cap and
H
base rock. This maximum may be slightly higher if sufficient containment at
this higher pressure was inferred from the injectivity test. Suppose that
initially the reservoir pressure is lowered to a level p with p larger than
p
, the reservoir pressure that permits fracture growth at the fracture
c/b
propagation pressure S
. Since the injection pressure is constrained to a
n
maximum that is lower than the fracture propagation pressure at reservoir
pressure p, the injectivity will be reduced. This results in a situation of
temporary underinjection until the reservoir pressure reaches p

and the

fracture can propagate. Therefore with a constrained injection pressure, the


reservoir pressure will adjust itself to the level that is required to keep
the fracture open. With the fracture open, the injection rate may exceed the
maximum rate for which sweep efficiency is unimpaired. This would result in
temporary overinjection and therefore in a rise in reservoir pressure which
would eventually reduce the injectivity again. Nevertheless, the fracture
may grow too long in this period and, depending on its healing ability,
could permanently impair sweep efficiency.
It is therefore important to constrain both injection pressure and
injection rate to their optimum values. As injection continues, cooling of
the reservoir may result in a lowering of the reservoir rock stress and
therefore in a propagation pressure that is lower than the maximum allowable
c/b
injection pressure S
. Owing to conductive cooling a reduction in rock
H
stress will also occur in the parts of cap and base rock immediately
adjacent to the reservoir. Eventually, depending On its thickness, the whole
of cap and base rock may have been cooled giving a uniform lowering of rock
stress. In the light of vertical fracture containment, therefore, it may be
important to lower the maximum allowable injection pressure in
correspondence with the observed decrease in propagation pressure.A correct determination of the maximum injection pressure that keeps
the fracture vertically contained may be of critical importance to the
project. In this case it is important to get an indication how the rock
stress in cap and base rock varies over the field. During drilling the sonic
log should be run in selected new injectors to check for a possible change
in contrast of elastic moduli between reservoir and bounding layers. Such a
change could indicate a different stress contrast and may necessitate an
adjustment of the maximum allowable injection pressure.

- 232 -

If in the casing sequence a casing shoe is set into the cap rock, a
leak-off test may be performed, after drilling out the shoe, to measure the
formation strength. If this is done across the field, additional information
on lateral consistency of formation strength is obtained.

5.8 SPECIAL APPLICATIONS

Line drive
If there is an indication that the direction of minimum in-situ stress
is the same across the field, this circumstance may be exploited to set up a
line drive pattern. Alternate rows of injectors and producers with the rows
parallel to the fracture orientation could result in a very high sweep
efficiency.
Such an indication of rectilinear horizontal stress trajectories could
come from measurements of microfrac orientation together with, for instance,
a clear trend in the orientation of a fault system.
Of course, for completely aligned fractures the constraint q

< 0.6 can

be dropped.

High dimensionless injection rates


For larger dimensionless injection rates deviations from pseudo-radial
flow occur during propagation. However, for very favourable mobility ratios
the sweep efficiency may still be rather good.
These cases are best evaluated with a numerical fracture/reservoir
12
simulator. Recently , a prototype of such a simulator, including poro- and
thermo-elastic effects on propagation pressure following the method outlined
in Chapter 3, section 3.5.3, was developed.
This simulator is a very powerful tool in studying more general
conditions than can be dealt with in analytical models. It is especially
useful for making operational decisions such as, for instance, the
conversion of producers with premature water breakthrough into fractured
injectors. Care should be taken, however, that thermo-elastic effects are
not underestimated and poro-elastic effects overestimated by too large grid
blocks. A specific grid lay-out should be validated by comparison with the
analytical model.

- 233 -

Poor-quality injection water


For injection water of poor quality consideration may be given to
fracturing the injector rather than to filtering the water to maintain
injectivity.
The continuous plugging of new areas of the fracture face would, of
course, result in continuing fracture growth. The effect on sweep efficiency
can again be checked with the simulator of Ref. 12. It incorporates a model
for describing permeability deterioration as a function of water quality and
pore volumes injected.
It is of interest to note that the pressure fall-off technique of
Chapter 4, together with deconvolution as described in Ref. 6, can be used
to calculate the skin at the fracture face. Several such calculations in
time can then be used to validate the plugging model.

5.9 CONCLUSIONS
1) Waterflooding under fracturing conditions has a large potential for a
reduction in the number of wells.
2) Waterflooding under fracturing conditions can, in many cases, be designed
in such a way that sweep efficiency is unimpaired and vertical fracture
containment is ensured. Design formulae are presented for calculating the
optimum reservoir pressure, injection rate and well pattern.
3) The case of a reservoir at hydrostatic pressure, with a typical gradient
for the horizontal reservoir rock stress of 0.17 bar/m is investigated in
detail for unit mobility ratio. With the stress in cap and base rock
exceeding that in the reservoir by 14% or more, the most attractive
options are a fractured 7-spot and a fractured 9-spot with a maximum
reduction in the number of wells of 40% and 33% respectively.

4) For more moderate stress contrasts the reservoir pressure has to be


lowered to ensure vertical fracture containment. With the horizontal rock
stress in cap and base rock exceeding that in the reservoir by only 9% a
fractured 13-spot at a lower reservoir pressure results in a 16%
reduction in the number of wells.

- 234 -

5) The fact that 7-spots and 9-spots are the most attractive options
reflects the fact that only the injectors are fractured. If both
injectors and producers are fractured the maximum reduction in the number
of wells is 67% with the fractures in the producers of the same length as
the stable fracture length at the injectors. For a stress contrast of 14%
or more a doubly fractured 5-spot would be the preferable pattern in this
case.

6) In-situ stress measurements should be performed in the reservoir and cap


and base rock. The stress data are essential for an optimal design of the
process.

7) An injectivity test should be carried out to establish a trend in


fracture propagation pressure. Regular fall-off testing can be used to
determine the change in fracture dimensions with time. A match with the
analytical propagation model results in values for the thermo- and poroelastic constants. From the model the maximum fracture propagation
pressure can be determined that will occur during fracture propagation
under balanced conditions. This quantity is important for determining the
maximum reservoir pressure for which fracturing occurs under the
condition of vertical containment.

8) The value for the in-situ stress in cap and base rock should serve as a
maximum for the downhole injection pressure. Lateral uniformity of this
stress should be checked with sonic logging and, if possible, with
leak-off tests.

9. If the injection rate and injection pressure are constrained to their


appropriate maximum values, fracture growth will occur in a completely
controlled manner.

- 235 -

LIST OF SYMBOLS

poro-elastic constant
P

conductivity

coefficient of proportionality in poro-elastic stress change

distance from injector to nearest producer

reservoir height

II

injectivity index

ratio of the number of producers at reservoir pressure p to the

F.

number of producers at reservoir pressure p.


ratio of the number of fractured injectors at reservoir pressure
p to the number of producers at reservoir pressure p.

permeability

critical stress intensity factor

fracture half-length

total number of non-fractured wells at reservoir pressure p.

total number of producers at reservoir pressure p.


p

l
total number of producers and fractured injectors at reservoir
pressure p

PI

productivity index

p.

prevailing reservoir pressure

arbitrary reservoir pressure

maximum reservoir pressure at which vertical fracture


containment is ensured

minimum bottomhole pressure in producer

pf

fracture propagation pressure at reservoir pressure p.

p,

fracture propagation pressure at reservoir pressure p

A p D = (P-P.)/(P.-PW)
Ap = p - P i
q
max
q

injection rate
. . .

,. .

maximum injection rate under fracturing conditions with


unimpaired sweep efficiency at reservoir pressure p.

maximum injection rate into a non-fractured well at reservoir


pressure p.

maximum injection rate under fracturing conditions with


unimpaired sweep efficiency at reservoir pressure p

- 236 -

g
q
rw
R

maximum production rate at reservoir pressure p.


maximum production rate at reservoir pressure p
wellbore radius
penetration depth of pressure transients

ratio of the number of producers to the number of fractured


injectors at reservoir pressure p

SIT
.
Hi

initial horizontal reservoir rock stress at reservoir rpressure


p

horizontal reservoir rock stress at reservoir pressure p

SHc/b

horizontal rock stress in cap and base rock

Greek
X
Aa
Aa

fluid mobility
poro-elastic stress change at fracture face
thermo-elastic stress change at fracture face

- 237 -

REFERENCES
1) Breckels, I.M. & van Eekelen, H.A.M., Relationship between horizontal
stress and depth in sedimentary basins.
JPT (September 1982), p. 2191.
2) Abou-Sayed, A.A., Brechter, C.E. & Clifton, R.J., In-situ stress
measurements by hydrofracturing: A fracture mechanics approach.
J. Geoph. Res. (June 1978), p. 2851.
3) Daneshy, A.A., Slusher, G.L., Chisholm, P.T. & Magee, D.A., In-situ
stress measurements during drilling.
JPT (August 1986), p. 891.
4) Griffin, K.W., Induced fracture orientation determination in the Kuparuk
reservoir.
SPE 14261, 1985.
5)

Schlumberger Brochure, M-090033, April 1986.

6) Newberry, B., Nelson, R., Cannon, D. & Ahmed, U., Prediction of vertical
hydraulic fracture migration using compressional and shear wave
slowness.
SPE 13095, 1985.
7) Koning, E.J.L. & Niko, H., Application of a special fall-off test method
in a fractured North Sea water injector.
SPE 16392, 1986.
8) Dobkins, T.A., Improved methods to determine hydraulic fracture height.
JPT (April 1981), p. 719.
9) Hsu, K., Brie, A. & Plumb, R.h.i

A new method for fracture

identification using array sonic tools.


SPE 14397, 1985.
10) Hagoort, J., Waterflood-induced hydraulic fracturing.
Ph.D. Thesis, Delft Technical University, 1981.
11) Craig, F.P., The reservoir engineering aspects of waterflooding.
SPE Monograph Vol. 3, 1973, p. 41.
12) Dikken, B.J. & Niko, H., Waterflood-induced fractures: A simulation
study of their propagation and effects on waterflood sweep efficiency.
KSEPL, Publication 798, 1987.

- 238 -

APPENDIX 5-A
DETERMINATION OF THE RATIO OF PRODUCERS TO FRACTURED INJECTORS
AND THE REDUCTION IN THE NUMBER OF WELLS

We make the following assumptions:

1) We consider unit mobility ratio and injectors and producers with


identical typical properties so that:
II = PI

(5-A-l)

Let p' denote the reservoir pressure at which the maximum injection rate
into a non-fractured injector equals the maximum production rate from a
producer. It follows from (5-A-l) that:
p. =

where p

(p

min

+ pw)/2

(5-A-2)

is the minimum bottomhole pressure in a producer and p'

is

the minimum fracture propagation pressure at reservoir pressure p'. p'


may include a thermo-elastic lowering of rock stress owing to cooling.
p'

is taken as the maximum injection pressure at which injection takes

place without fracturing, p'

can be obtained from its counterpart p.

at the prevailing reservoir pressure p. using Eq. (5.10) in Section


5.6.1. p' can be determined by simultaneously solving Eqs. (5.10) and
(5-A-2). From its definition it follows that at reservoir pressure p' a
5-spot is the optimal non-fractured well pattern and that the number of
wells required for a certain field offtake is a minimum at this reservoir
pressure. Any other reservoir pressure requires more wells.

2) For simplicity we assume that the prevailing reservoir pressure, p., is


equal to the optimum reservoir pressure, p', for waterflooding under
non-fracturing conditions, i.e.:

pi

= p'

(5-A-3)

- 239 -

The maximum injection rate into a non-fractured injector and the maximum
production rate are given by, respectively:
~max _
min
, __
q
= (Pf
~ PL) - II

, _ . ..
(5-A-4)

q = (p. - p ) . PI
p
l
w

(5-A-5)

From ((5-A-l) to (5-A-3) we have:


~max
q
= q

,c > e
(5-A-6)

Since the optimum non-fractured well pattern is a 5-spot we have for the
total number of wells, N, at reservoir pressure p.

N = 2N
P
in which N

(5-A-7)

is the total number of producers at reservoir pressure p. .

3) The maximum injection rate for injecting under fracturing conditions is


given by:
q m a X =0.61 27rhX(pin - p.)

(5-A-8)

with X the mobility of the injection fluid.


Injection under fracturing conditions is assumed to give an improvement
in injectivity of a factor of three:
max
CL_
~max
q

. ., _ ,.
0.61 rtV =
II

According to Section 5.4, Eq. (5.3) this corresponds typically to an


unfractured well with skin S = -2.
., _,
. .
. .
-min
.
min
4) The minimum propagation pressure, p f , at reservoir pressure p and p
are related through Eq. (5.10) in Section 5.6.1. Taking c = 0.5 and
A = 0.5 as typical values we have:
P

- 240 -

min
mm
. __ .
p.
= p. + 0.33 Ap

_
(5-A-10)

with Ap = p - p..
With the above assumptions we can now calculate the required quantities*.
We have for the maximum production rate at reservoir pressure p:

q p = (P - Pw) PI

(5-A-ll)

q = q + Ap . PI
P
P

(5-A-12)

so that

with q from (5-A-5).


P
From the definition of F in Section 5.6.3 in the text and from
P
(5-A-12) we have:
q,
M
p
Ap PI -1
F = -= = (1 + -*
)
p
q

(5-A-13)

From assumptions 1 and 2 we have:


PL_Ap_
q
P

II_A
~max
q

Ap.
a _Ap.
min
p. - p
Pf
Pt
i
w

(5-A-14)

We introduce the following definition for the dimensionless change in


reservoir pressure:
Ap n = r ^
= ^ ~
D
mm
p. - p
Pf
" Pt
i
w

(5-A-15)

F = (1 + Ap) _ 1
r
p
D

(5-A-16)

so that:

We have for the maximum injection rate under fracturing conditions at


reservoir pressure p,

- 241 -

q m a X = 0.61 27ThX (pin - p")

(5-A-17)

From (5-A-8) and (5-A-10) we have:


-max _ gmax _

(5-A-18)

0 > g l 2ffhX 0 # 6 g A p

so that from (5-A-4) and (5-A-9):


-max
3

= 3 - 1.98 Ap
~max
D
q

(5-A-19)

From the definition of F. in Section 5.6.3 we have using (5-A-6) and


(5-A-19):
q
jmax
F. = - = a
= (3 - 1.98 Ap,J
l
-max
-max
D
q
q

(5-A-20)

The ratio of producers to fractured injectors as a function of the


dimensionless change in reservoir pressure is now given by:
F

3-1.98 Ap

<5"A-21>

"^Tnr1

At Ap

*D

= 0 we have R = 3 in accordance with (5-A-6) and (5-A-9).

The reduction in the number of wells is given by, using (5-A-7):

N - N
,
c N
- = 1 - 0.5
N

,
.
= 1 - 0.5 F

+ F )
1

P
= 1 - 0.5{(3 - 1.98Ap ) _ 1 + (1 + Ap D ) -1 }

(5-A-22)

- 243 -

WATERFLOODING UNDER FRACTURING CONDITIONS

SUMMARY
When an oilfield is exploited by simply producing oil and gas from a
number of wells, the reservoir pressure, in many cases, drops rather quickly
and so does the production rate. Therefore, water is often injected to
maintain the reservoir pressure. The injection wells are located in such a
way that as much oil as possible is driven to the producers before they
experience water breakthrough.
Recovering oil by water injection is called waterflooding. The
efficiency with which the water sweeps the oil to the producers without
bypassing it is called the sweep efficiency.
A major saving of the exploitation costs of an oilfield can be obtained
by a reduction in the number of wells. The required number of injectors can
be reduced by increasing their injection capacity. An effective way of doing
this is by rupturing the permeable reservoir rock surrounding the wellbore.
The communication of a well with a fracture in the rock results in a
considerable enlargement of the surface area of injection. As a consequence,
with the same pressure gradient, a dramatic increase in injection rate can
be obtained.
Rupturing of the reservoir rock can be effected by injecting at a
pressure that is higher than the rock stress and the tensile strength that
keep the rock particles together. Continued injection at this pressure
causes the fracture to propagate into the reservoir. Water injection in this
manner is called waterflooding under fracturing conditions.
As opposed to the advantage of a reduction in the number of wells,
there is the disadvantage of possible excessive fracture growth that may
adversely affect the sweep efficiency of water injection. A fracture that
comes too close to a production well will cause premature water
breakthrough. Moreover, if the fracture is vertical, excessive vertical
growth may establish communication with other reservoirs, resulting in loss
of injection water or, even worse, loss of oil.

- 244 -

The objectives of this thesis are to:


a) investigate the mechanisms of fracture initiation and propagation under
the influence of continued water injection,
b) evaluate the effect of fracture growth on sweep efficiency,
c) improve the methods for determining fracture dimensions,
d) give rules for designing the process of waterflooding under fracturing
conditions in such a way that sweep efficiency is unimpaired and vertical
fracture growth is limited.

This thesis consists of five self-contained chapters. In Chapter 1 a


general introduction is given, together with a survey of the work that was
done in the past and of the new elements contributed by this thesis. In
Chapter 2 a complete, analytical solution is given for the stress field in
the reservoir rock surrounding an unfractured well. Three-dimensional theory
of poro- and thermo-elasticity has been used to calculate the effect of
changes in pressure and temperature on the rock stress. It is shown that the
commonly used reduction of the stress calculations to two horizontaldimensions may lead to a serious underestimation of the magnitude of the
stress changes.
The calculated stress field in the rock is used to determine the'
injection pressure at which fractures can be initiated.
In an example it is shown that injection of large quantities of cold
water into a reservoir of good permeability can lead to a considerable
reduction in rock stress, which may result in thermal fracturing.
Chapter 3 gives a description of fracture growth and the effect on
sweep efficiency. An analytical model is presented with a complete
two-dimensional description, in the plane of the reservoir, of the fluid
leak-off from the fracture. It is shown that previously developed models
with a one-dimensional description lead, in most cases, to a large
overestimation of the fracture length.
The situation is considered in which every injection well is surrounded
by a number of production wells in a regular pattern. When in a pattern
element the injection rate equals the production rate then, after a certain
period of time, the pressure distribution will not change anymore and the
fracture will stop growing.

- 245 -

Special attention is focused on the mode of fracture propagation in


which the pressure transients travel radially-symmetrically in the plane of
the reservoir. It is shown that when the injection rate is small enough to
establish propagation with this pseudo-radial flow, the sweep efficiency of
water injection is not affected. Three-dimensional theory of poro- and
thermo-elasticity has been used for an analytical calculation of the effect
of pressure and temperature changes on the stress field around a propagating
fracture with pseudo-radial flow. The effect of a change in the stress field
on the fracture propagation velocity is illustrated with two examples.
Further, a numerical method is given for the calculation of pressure- and
temperature-dependent changes in reservoir rock stress that can easily be
incorporated into numerical fracture growth simulation models such as
developed in the past.
Chapter 4 discusses a method for determining the dimensions of an
induced fracture from a pressure fall-off test. The technique consists of
measuring the decreasing fluid pressure downhole during an interruption of
the injection. From the behaviour of the falling pressure with time,
information can be obtained on the size of the fracture. It is shown that
the falling pressure with time is characterised by four distinct periods,
each of which gives independent information on the fracture size.
Furthermore, the in-situ rock stress can be found by determining the
fluid pressure at which the fracture has closed. This method of determining
fracture size is important for the measurement of fracture growth in
practice and for gauging theoretical fracture propagation models.
Finally, Chapter 5 gives a practical programme for designing the
process of waterflooding under fracturing conditions in a given field in
such a way that sweep efficiency is unimpaired and vertical fracture
containment is ensured. It consists of:
- Measurements of the in-situ rock stress in the reservoir and bounding
formations to investigate vertical fracture containment. The rock stress
in the bounding formations must be higher than in the reservoir to limit
vertical propagation.
- An injectivity test together with regular fall-off testing to determine
the parameters in the theoretical model. When a correct description of the
injection pressure and the fracture size with time has been obtained the
maximum fracture propagation pressure can be determined that will occur in
an actual pattern of injection and production wells with the total

- 246 -

injection balancing the total production. This quantity should be lower


than the horizontal rock stress in cap and base rock in order to prevent
vertical fracture propagation. If it is too high, the reservoir pressure
must be lowered such that the corresponding decrease in the horizontal
reservoir rock stress results in a maximum fracture propagation pressure
that prevents vertical propagation.
- Design calculations to establish a) the maximum reservoir pressure at
which vertical containment is ensured, b) the maximum injection rate at
which sweep efficiency is unimpaired, c) the optimal well pattern and
reservoir pressure, d) the reduction in the number of wells compared to
injection under non-fracturing conditions.
It is shown in a typical example that with the horizontal rock stress
in cap and base rock exceeding that in the reservoir by 14% or more the
reduction in the number of wells by fracturing the injection wells ranges
from 33% to a maximum of 40%. If the production wells are also fractured
with standard fracture stimulation techniques and if these fractures have
the same length as the stabilised fracture at the injection well the
reduction in the number of wells can reach a maximum of 67%.
- A practical scheme for the full-scale implementation of waterflooding
under fracturing conditions in the field. It is shown that if the
injection pressure and injection rate are constrained to appropriate
maximum values fracture growth will occur in a completely controlled
manner.
Sonic logging and, if possible, leak-off tests should be performed
during drilling to detect variations in the horizontal rock stress in cap
and base rock across the field. The maximum allowable injection pressure
should be adjusted accordingly.

- 247 -

WATERSTWING BIJ SPLIJTINGSCONPITIES

SAMENVATTING
Wordt een olieveld ontgonnen door eenvoudigweg olie en gas te
produceren uit een aantal putten dan zal dat veelal leiden tot daling van de
reservoirdruk en, dientengevolge, van de produktiesnelheid. Daarom wordt
vaak water geinjecteerd om de reservoirdruk zoveel mogelijk te handhaven. De
injectieputten worden zodanig geplaatst dat zoveel mogelijk olie naar de
produktieputten wordt gestuwd voordat water hierin doorbreekt. Oliewinning
door waterinjectie wordt waterstuwing genoemd. De fractie van het reservoirvolume dat uiteindelijk door het genjecteerde water bestreken wordt, heet
het veegvermogen.
Een van de grootste besparingen op de ontginningskosten van een
olieveld kan bereikt worden met een vermindering van het aantal putten. Het
vereiste aantal injectieputten kan verminderd worden door een vergroting van
hun injectiecapaciteit. Een effectieve manier om dit te bereiken is door het
splijten van het permeabele reservoirgesteente om de put. De verbinding van
een put met een scheur in het gesteente zorgt voor een belangrijke
vergroting van het injectieoppervlak. Zodoende kan met dezelfde drukgradient
een aanzienlijke verhoging van de injectiesnelheid worden verkregen.
Splijting van het reservoirgesteente kan bereikt worden door te
injecteren met een druk die hoger is dan de gesteentespanning en de
rekkracht die de gesteentedeeltjes bij elkaar houden. Onder voortdurende
injectie bij deze druk zal de scheur zich uitbreiden in het reservoir. Men
spreekt dan van waterstuwing bij splijtingscondities.
Tegenover het voordeel van een kleiner aantal putten staat het nadeel
dat overmatige scheurgroei het stromingspatroon en daardoor het veegvermogen
van waterinjectie ongunstig kan beinvloeden. Komt een scheur te dicht bij
een produktieput dan zal dit tot een vroege waterdoorbraak leiden. Bovendien
kan, in het geval van een vertikale scheur, overmatige vertikale groei een
verbinding met andere reservoirs tot stand brengen. Dit kan resulteren in
verlies van injectiewater of erger nog, van olie.
De doelstellingen van dit proefschrift zijn:
a) het onderzoek van de mechanismen van scheurvorming en scheuruitbreiding
onder de invloed van voortdurende water injectie,
b) het vaststellen van de invloed van scheuruitbreiding op het veegvermogen,

- 248 -

c) het verbeteren van de methoden om de scheurafmetingen te bepalen,


d) het geven van een een voorschrift om tot een ontwerp te komen van
waterstuwing bij splijtingscondities voor een gegeven veld waarin het
veegvermogen onaangetast blijft en de vertikale scheuruitbreiding beperkt
is.
Dit proefschrift bestaat uit vijf op zichzelf staande hoofdstukken. In
Hoofdstuk 1 wordt een algemene inleiding gegeven samen met een overzicht van
het werk dat in het verleden is gedaan en de nieuwe elementen die met dit
proefschrift worden toegevoegd. In Hoofdstuk 2 wordt een volledige,
analytische oplossing gegeven voor het spanningsveld in het reservoir
gesteente om een ongescheurde put. Drie-dimensionale poro- en thermoelasticiteitstheorie zijn gebruikt om het effect van druk- en temperatuurs
veranderingen op de gesteentespanning te berekenen. Aangetoond wordt dat de
vaak gebruikte reductie van de spanningsberekeningen tot het horizontale
vlak tot een ernstige onderschatting van de spanningsveranderingen kan
leiden. Het berekende spanningsveld in het gesteente wordt gebruikt om de
injectiedruk te bepalen waarbij scheuren'ontstaan.
Aan de hand van een voorbeeld wordt getoond dat injectie van grote
hoeveelheden koud water in een goed doorlatend reservoirgesteente kan leiden
tot een aanzienlijke verlaging van de gesteentespanning hetgeen kan
resulteren in thermische scheurvorming.
Hoofdstuk 3 geeft een beschrijving van scheurgroei en het effect op
veegvermogen. Een analytisch model wordt gepresenteerd dat de vloeistof
stroming uit de scheur beschrijft als twee-dimensionaal in het vlak van het
reservoir. Aangetoond wordt dat voorheen ontwikkelde modellen met een
een-dimensionale beschrijving in de meeste gevallen tot een sterke
overschatting van de scheurlengte leiden.
De situatie wordt beschouwd waarin elke injectieput omringd wordt door
een aantal produktieputten in een regelmatig patroon. Wanneer in een element
van zo'n patroon de injectiesnelheid gelijk is aan de produktiesnelheid
verandert het drukprofiel na een zekere tijd niet meer en stopt de scheur
met groeien. Speciale aandacht wordt gevestigd op de wijze van scheur
uitbreiding, waarbij de drukveranderingen zich radiaal-symmetrisch in het
reservoir voortplanten voordat deze stabiele situatie wordt bereikt.

- 249 -

Aangetoond wordt dat wanneer de injectiesnelheid laag genoeg is om scheuruitbreiding met deze zogenaamde pseudo-radiale stroming te bewerkstelligen,
het veegvermogen van water inject ie niet wordt beinvloed.
Drie-dimensionale poro- en thermo-elasticiteitstheorie zijn gebruikt
voor een analytische berekening van het effect van druk- en temperatuurs
veranderingen op het spanningsveld rond een zich uitbreidende scheur met
pseudo-radiale stroming. Het effect van de met het spanningsveld
veranderende snelheid van de scheuruitbreiding wordt gellustreerd aan de
hand van twee voorbeelden. Verder wordt een numerieke methode behandeld voor
het berekenen van druk- en temperatuursafhankelijke spanningsveranderingen
die eenvoudig ingebouwd kan worden in numerieke scheurgroei-simulatiemodellen zoals die in het verleden zijn ontwikkeld.
Hoofdstuk 4 behandelt een methode om de afmetingen van een aangebrachte
scheur te bepalen met een zogenaamde drukdalingstest. De techniek bestaat
uit het meten van de druk onderin de put tijdens een onderbreking van de
injectie. Aangetoond wordt dat in het drukverloop vier perioden zijn te
onderscheiden die elk onafhankelijke gegevens opleveren over de
scheurgrootte. Bovendien kan de in-situ gesteentespanning bepaald worden
door vast te stellen bij welke vloeistofdruk de scheur zich sloot. Deze
methode om scheurgrootte te bepalen is van belang voor het meten van de
scheurgroei in de praktijk en voor het ijken van theoretische scheurgroei
modellen.
Hoofdstuk 5, tenslotte, geeft een praktisch programma om tot een
ontwerp te komen van waterstuwing bij splijtingscondities voor een gegeven
veld waarin het veegvermogen onaangetast blijft en de vertikale
scheuruitbreiding beperkt is. Het programma bestaat uit:
- Meting van de in-situ gesteentespanning in het reservoir en aangrenzende
lagen om de mogelijkheid van vertikale scheuruitbreiding te onderzoeken.
De gesteentespanning in de aangrenzende lagen moet hoger zijn dan in het
reservoir om de vertikale groei te beperken.
- Een injectietest met regelmatige tussentijdse drukdalingstesten met het '
doel de parameters van het theoretische model te bepalen. Wanneer een
correcte beschrijving van de gemeten injectiedruk en scheurafmetingen is
verkregen kan de maximum scheuruitbreidingsdruk worden bepaald die zal
optreden in een patroon van injectie- en produktieputten waarin de
injectiesnelheid gelijk is aan de totale produktiesnelheid. Deze grootheid
moet kleiner zijn dan de horizontale gesteentespanning in de aangrenzende

- 250 -

lagen om vertikale scheuruitbreiding te voorkomen. In het geval van een te


hoge waarde moet de reservoirdruk verlaagd worden zodat de overeenkomstige
afname in de horizontale reservoir-gesteentespanning een maximum
scheuruitbreidingsdruk geeft die vertikale uitbreiding verhindert.
- Ontwerpberekeningen voor het vaststellen van a) maximum reservoirdruk
waarbij vertikale scheuruitbreiding beperkt blijft, b) maximum
injectiesnelheid waarbij het veegvermogen onaangetast blijft, c) het
optimale putpatroon en de overeenkomstige reservoirdruk, d) de
vermindering van het aantal putten ten opzichte van het geval waarin
geinjecteerd wordt bij niet-splijtingscondities.
In een karakteristiek voorbeeld wordt aangetoond dat wanneer de
horizontale gesteentespanning in de aangrenzende lagen die in het
reservoir overtreft met 14% of meer, de vermindering in het aantal putten
reikt van 33% tot een maximum van 40%. Wanneer de produktieputten ook
gescheurd worden met standaard scheur-stimulatie technieken en wanneer
deze scheuren dezelfde afmeting hebben als de gestabiliseerde scheur bij
de injectieput dan kan de vermindering in het aantal putten een maximum
van 67% bereiken.
- Een praktisch schema voor het op grote schaal implementeren van
waterstuwing bij splijtingscondities in het veld. Aangetoond wordt dat
wanneer de injectiedruk en de injectiesnelheid worden beperkt tot de in
het ontwerp bepaalde maxima, de scheur zich op een volkomen gecontroleerde
wijze zal uitbreiden.
Met acoustische metingen en, indien mogelijk, met een formatiesterktetest moet worden bepaald in hoeverre de horizontale gesteentespanning in
de aangrenzende lagen varieert over het veld. Aan de hand hiervan kan de
maximum toelaatbare injectiedruk eventueel worden aangepast.

- 251 -

ACKNOWLEDGEMENTS
Most of this thesis is based on the work I did during my last two years
at Koninklijke/Shell Exploratie en Produktie Laboratorium (KSEPL).
I completed this thesis during subsequent leaves while already working for
Petroleum Development Oman (PDO).
I should like to thank the Management of KSEPL for permission to
publish this work and for providing me with the opportunity to produce this
thesis.
I am especially indebted to Helmut Niko of KSEPL, with whom I spent
many enjoyable and stimulating hours discussing the subject. Throughout
those two years Helmut was an enthusiastic critic and supporter. Helmut has
made several contributions to this work and wrote the computer programme for
the numerical inversion in Chapter 4.
I am grateful to Joost v.d. Burgh, Jan Geertsma and Rik Drenth, with
whom I had numerous instructive discussions.
I am also indebted to my successor at KSEPL, Ben Dikken, who read my
reports very critically. Ben has been most helpful in providing some of the
figures in this thesis.
I am grateful to Jacques Hagoort for critically reading the manuscript
and especially for his discussion of the last chapter.
I should like to thank my promotor Prof. Jan de Haan for his pleasant
and stimulating coaching and for his patience during the many times that the
communication from Oman fell silent.
Finally, I appreciate the great effort provided by the typingpool of
KSEPL, especially by Jopie Aarden, in completing the tremendous task of
typing those long strings of formulae.

STELLINGEN

behorende bij het proefschrift "Waterflooding under Fracturing


Conditions", E.J.L. Koning, September 1988.

1. De vereenvoudigde, een-dimensionale modellering van vloeistofuitstroming


die algemeen in scheurstimulatie ontwerpprogramma's wordt toegepast (zie
bijv. Howard h Fast) kan tot een optimistische voorspelling van de
scheurlengte leiden. Het verdient dan ook aanbeveling deze beschrijving
uit te breiden naar twee dimensies.

Howard, G.G. i Fast, C.R., Hydraulic Fracturing,


SPE Monograph, Volume 2, 1970.

2. Sommige oplossingen voor problemen in de poro- en thermo-elasticiteitstheorie met gemengde randvoorwaarden (zie bijv. Smith en Hagoort) kunnen
sterk vereenvoudigd worden door een met de Goodier verplaatsings
potentiaal gegenereerde oplossing te superponeren op een in de homogene
elasticiteitstheorie reeds bekende oplossing.

Smith, M.B., Stimulation design for short, precise


hydraulic fractures - MHF, SPE 10313, 1981.
Hagoort, J., Waterflood-induced hydraulic
fracturing, Hfdst. 4, proefschrift, TH Delft, 1981.

3. De twee-dimensionale covariante Poisson vergelijking voor Goodier's


verplaatsingspotentiaal genereert alleen op vlakke twee-dimensionale
ruimtes een particuliere oplossing voor de thermo-elasticiteitsvergelijkingen. Hetzelfde geldt ook in drie dimensies.

4. De door Nolte ontwikkelde methode om de drukdaling na een minischeurstimulatie te analyseren berust op enkele niet zonder meer
toelaatbare veronderstellingen. Het verdient dan ook aanbeveling het
geldigheidsgebied van deze veel gebruikte methode te bepalen.

Nolte, K.G., Determination of fracture parameters


from fracturing pressure decline, SPE 8341, 1979.

- 3 -

- 2 -

5. De door Ramey en Agarwal gegeven oplossing voor de bodemdruk tijdens

9. Bij het testen van een geavanceerde oliewinningsmethode op veld-schaal

productie met een veranderende boorgatvuil ing (wellbore storage) is

is succes sterk afhankelijk van nauwe samenwerking tussen de

onjuist.

verschillende betrokken disciplines, met name tussen die in de research


en die in het veld.
Ramey, H.J., Jr & Agarwal, R.G., Annulus unloading
rates as influenced by wellbore storage and skin

Koning, E.J.L., Mentzer, E., t, Heemskerk, J.,

effect, SPEJ (Oct. 1972) 453.

Evaluation of a pilot polymer flood in the


Marmul field, Oman, SPE 18092, 1988.

6. De bewering van Oake dat gedurende veranderingen in de reservoirdruk de


totale verticale reservoirgesteentespanning onveranderd blijft is alleen

10. In reservoirs met een matig hoge olieviscositeit kan polymeerstuwing

juist indien de laterale uitgestrektheid van de drukverandering groot is

onder splijtingscondities een grote besparing in het aantal putten

ten opzichte van de reservoirhoogte.

opleveren.

Dake, L.P., Fundamentals of Reservoir Engineering,


Elsevier Scient. Publ. Comp., 1976, p. 4.

11. Tijdens polymeerstuwing in sterk heterogene reservoirs kan het


stapsgewijs reduceren van de polymeerviscositeit volgens de methode van
Claridge een verslechtering van de olieopbrengst opleveren ten opzichte

7. Bij het bepalen van de in-situ gesteentespanning door het maken van

van voortdurende injectie met de oorspronkelijke viscositeit. Een zinvol

micro-scheuren kan het nodig zijn poro-elastische effecten op de scheur-

ontwerp van dit reductieproces is daarom alleen mogelijk indien een goed

sluitingsdruk in rekening te brengen. De in dit proefschrift gegeven

model van de reservoirheterogeniteiten beschikbaar is.

oplossing voor het poro-elastische spanningsveld kan hiervoor worden


gebruikt.

Claridge, E.L., Control of viscous fingering in


enhanced oil recovery processes: effect of
Dit proefschrift, Hoofdstuk 3, p. 127.

8. Deconvolutie van drukdalingsdata voor een zich sluitende scheur of voor


een scheur met boorgat- en/of scheurvulling (wellbore/fracture storage)

heterogeneities. SPE 7662, 1978.

12. Gezien de stormachtige ontwikkeling in de digitalisering van muzikale


informatie, waarbij voornamelijk gebruik gemaakt wordt van toetsenborden

heeft het voordeel boven andere methoden dat het expliciet de skin aan

in combinatie met computertechnieken, is het muziekstudenten aan te

het scheuroppervlak kan opleveren.

bevelen piano als hoofdvak en informatica als bijvak te kiezen.

Koning, E.J.L. t Niko, H., Application of a special


falloff test in a fractured North-Sea injector,

13. Het in Oman ingestelde verbod op een huwelijk tussen Omani's en


buitenlanders is in strijd met de geest van de Islam.

SPE 16392, 1986.


14. Het toenemende aantal jeep-achtige voertuigen onder stadsbewoners
suggereert een markt voor de terreinfiets.

- 2 -

5. De door Ramey en Agarwal gegeven oplossing voor de bodemdruk tijdens

- 3 -

9. Bij het testen van een geavanceerde oliewinningsmethode op veld-schaal

productie met een veranderende boorgatvulling (wellbore storage) is

is succes sterk afhankelijk van nauwe samenwerking tussen de

onjuist.

verschillende betrokken disciplines, met name tussen die in de research


en die in het veld.
Ramey, H.J., Jr 4 Agarwal, R.G., Annulus unloading
rates as influenced by wellbore storage and skin

Koning, E.J.L., Mentzer, E., & Heemskerk, J.,

effect, SPEJ (Oct. 1972) 453.

Evaluation of a pilot polymer flood in the


Marmul field, Oman, SPE 18092, 1988.

6. De bewering van Dake dat gedurende veranderingen in de reservoirdruk de


totale verticale reservoirgesteentespanning onveranderd blijft is alleen

10. In reservoirs met een matig hoge olieviscositeit kan polymeerstuwing

juist indien de laterale uitgestrektheid van de drukverandering groot is

onder splijtingscondities een grote besparing in het aantal putten

ten opzichte van de reservoirhoogte.

opleveren.

Dake, L.P., Fundamentals of Reservoir Engineering,


Elsevier Scient. Publ. Comp., 1978, p. 4.

11. Tijdens polymeerstuwing in sterk heterogene reservoirs kan het


stapsgewijs reduceren van de polyoeerviscositeit volgens de methode van
Claridge een verslechtering van de olieopbrengst opleveren ten opzichte

7. Bij het bepalen van de in-situ gesteentespanning door het maken van

van voortdurende injectie met de oorspronkelijke viscositeit. Een zinvol

micro-scheuren kan het nodig zijn poro-elastische effecten op de scheur-

ontwerp van dit reductieproces is daarom alleen mogelijk indien een goed

sluitingsdruk in rekening te brengen. De in dit proefschrift gegeven

model van de reservoirheterogeniteiten beschikbaar is.

oplossing voor het poro-elastische spanningsveld kan hiervoor worden


gebruikt.

Claridge, E.L., Control of viscous fingering in


enhanced oil recovery processes: effect of
Dit proefschrift. Hoofdstuk 3, p. 127.

8. Deconvolutie van drukdalingsdata voor een zich sluitende scheur of voor


een scheur met boorgat- en/of scheurvulling (wellbore/fracture storage)

heterogeneities. SPE 7662, 1978.

12. Gezien de stormachtige ontwikkeling in de digitalisering van muzikale


informatie, waarbij voornamelijk gebruik gemaakt wordt van toetsenborden

heeft het voordeel boven andere methoden dat het expliciet de skin aan

in combinatie met computertechnieken, is het muziekstudenten aan te

het scheuroppervlak kan opleveren.

bevelen piano als hoofdvak en informatica als bijvak te kiezen.

Koning, E.J.L. f, Niko, H., Application of a special


falloff test in a fractured North-Sea injector,

13. Het in Oman ingestelde verbod op een huwelijk tussen Omani's en


buitenlanders is in strijd met de geest van de Islam.

SPE 16392, 1986.


14. Het toenemende aantal jeep-achtige voertuigen onder stadsbewoners
suggereert een markt voor de terreinfiets.

STELLINGEN

behorende bij het proefschrift "Waterflooding under Fracturing


Conditions", E.J.L. Koning, September 19B8.

1. De vereenvoudigde, een-dimensionale modellering van vloeistofuitstroming


die algemeen in scheurstimulatie ontwerpprogramma's wordt toegepast (zie
bijv. Howard & Fast) kan tot een optimistische voorspelling van de
scheurlengte leiden. Het verdient dan ook aanbeveling deze beschrijving
uit te breiden naar twee dimensies.

Howard, G.G. & Fast, C.R., Hydraulic Fracturing,


SPE Monograph, Volume 2, 1970.

2. Sommige oplossingen voor problemen in de poro- en thermo-elasticiteitstheorie met gemengde randvoorwaarden (zie bijv. Smith en Hagoort) kunnen
sterk vereenvoudigd worden door een met de Goodier verplaatsings
potentiaal gegenereerde oplossing te superponeren op een in de homogene
elasticiteitstheorie reeds bekende oplossing.

Smith, H.B., Stimulation design for short, precise


hydraulic fractures - KHF, SPE 10313, 1981.
Hagoort, J., Waterflood-induced hydraulic
fracturing, Hfdst. 4, proefschrift, TH Delft, 1981.

3. De twee-dimensionale covariante Poisson vergelijking voor Goodier's


verplaatsingspotentiaal genereert alleen op vlakke twee-dimensionale
ruimtes een particuliere oplossing voor de thermo-elasticiteitsvergelijkingen. Hetzelfde geldt ook in drie dimensies.

4. De door Nolte ontwikkelde methode om de drukdaling na een minischeurstimulatie te analyseren berust op enkele niet zonder meer
toelaatbare veronderstellingen. Het verdient dan ook aanbeveling het
geldigheidsgebied van deze veel gebruikte methode te bepalen.

Nolte, K.G., Determination of fracture parameters


from fracturing pressure decline, SPE 8341, 1979.

Potrebbero piacerti anche