Vous êtes sur la page 1sur 18

SPE 127742PP

SPE 102801

ADVANCED HORIZONTAL WELL MODEL


Imaging Seismic Deformation Induced by Hydraulic Fracture Complexity
Maxwell, C.K.SPE,
Waltman,
N.R. Warpinski,A.A.
M.J. Mayerhofer,
andS.V.
N. Boroumand,
Technologies.
D.S. Kuznetsov, SPE, A.N.S.C.
Cheremisin,
Schlumberger;
Chesnokov,
Golovin, Pinnacle
Institute
of Hydrodynamics
injection and a pre-existing fracture network2-6. Even between
neighbouring wells, the geometry of the stimulated fracture
Exhibition held in San Antonio, Texas, U.S.A., 2427 September 2006.
network shows a high degree of variability due to localized
2
This paper was prepared for presentation
atwas
the
SPEforNorth
Africa
Conference
Exhibition
held
in Cairo,
1417
February
2010.
This paper
selected
presentation
by an Technical
SPE Program Committee
followingand
review
of
differences
in Egypt,
the fracture
network
. The
Barnett has very low
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
intrinsic matrix permeability, and the permeability
This paper was selected for presentation
by byantheSPE
program
committee
following
review ofreflect
information
contained in an abstract submitted by the author(s). Contents
correction
author(s).
The material,
as presented,
does not necessarily
any
associated
with the
fracture
stimulationreflect
resultsany
in
Society of Petroleum Engineers, its officers, or members. Papers presented at
of the paper have not been reviewed position
by theof the
Society
of Petroleum Engineers and are subject to correction byenhancement
the author(s).
The material
does
not necessarily
SPE meetings are subject to publication review by Editorial Committees of the Society of
permeable fracture networks sufficient for economic gas
Engineers.
reproduction,
distribution,Electronic
or storage of any
part of this paper distribution, or storage of any part of this paper without the written
position of the Society of PetroleumPetroleum
Engineers,
its Electronic
officers,
or members.
reproduction,
commercial purposes without the written consent of the Society of Petroleum Engineers is
recovery in the field. Previous studies have shown a
consent of the Society of Petroleumfor
Engineers
is prohibited.
reproduce
in more
print
prohibited.
Permission
to reproduce in Permission
print is restricted to
to an
abstract of not
thanis restricted to an abstract of not more than 300 words; illustrations
correlation between the volume of the reservoir that is
300 words;
illustrations
may not be
copied. The abstract of
mustSPE
contain
conspicuous
may not be copied. The abstract must
contain
conspicuous
acknowledgment
copyright.
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
stimulated, as measured by the volume of the reservoir that
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
emits microseisms during the stimulation, and the production
ultimately realized from the well2,4,6. The correlation is
Abstract
attributed
to larger fracture networks being stimulated in wells
Abstract
Microseismic mapping is extensively utilized in the Barnett
where a large microseismically active volume of the reservoir
Shale, to map hydraulic fracture complexity associated with
has been
realized, resulting
in more permeable
interactions
of the stimulation
with pre-existing
fractures.
A new semi-analytical model
developed
allows making
fast modeling
of the
heterogeneous
reservoirs,
drained fracture
by a
pathways connected to the well and hence a higher potential
Previous studies have indicated a fair correlation between the
single horizontal well withwell
multiple
transverse
hydraulic
fractures.
It
is
assumed
that:
for gas flow to the well. Recently, many operators in the
performance and extent of the seismically active volume.
Barnett have attempted horizontal completions, which have
However, in addition to this measure of the extent of the
allowed large volumes of the reservoir to be stimulated with
the flow in the reservoir
obeys
Darcy
law;
stimulated
fracture
network,
the characteristics of this fracture
large fracture networks. Many of these completions use
network
is
also
expected
to
impact
the
well
performance.
In
the fluid contains only
one phase;
perforated, cemented liners and the microseismic images allow
particular, the fracture spacing is believed to be important
for along
indentification
of improved
perforation staging to
factor controlling
gas flow.
In this
paper, wedirected
the reservoir permeability
tensor the
is potential
a function
of the
variable,
the horizontal
wellbore;
maximize the stimulated reservoir volume (SRV)4.
utilize the density of the total seismic moment release (a
Copyright 2006, Society of Petroleum Engineers

Copyright 2010, Society of Petroleum


ThisEngineers
paper was prepared for presentation at the 2006 SPE Annual Technical Conference and

the well trajectory isrobust


notmeasure
a straight
line; strength) as an indication of
of the microseism
the
seismic
deformation
the wellbore has finite conductivity. that may correlate to the fracture

Many of the Barnett stimulations are water fracs, where high


volumes of water are injected at high rate7. One possible
density. The study uses a set of microseismic maps of
mechanism for the success of waterfracs is that increased fluid
hydraulic fracture stimulations, including cases where the
The proposed model isstimulated
fast inreservoir
terms volume
of themeasured
execution
the timepressure
required
to fractures
build ainduced
new shear
case.failure, resulting in
in natural
by thetime
extentand
of the
dilation associated with mismatched surfaces on
seismically
active
region poorly correlated
with the well
In the first section of this
article
a description
of the solution
method isfracture
given.
opposite sides of the fracture. Within this conceptual
performance. Incorporating the seismic moment density to
The second part is dedicated
to validating the proposed model. The framework,
recognized
finite-difference
hydrodynamic
the microseismic
events correspond
to the actual
assess the fracture density with the network extent, an
fracture are
movement.
The earlier investigations
SRV
simulator (FDS) was used
as reference
software,
though was
theobserved.
model and FDS
not competitors
becauseofofthetheir
improved
correlation with
the well performance
measured the total volume of the microseismically active
differing applicability. For this reason the execution times of the FDS and
the
Advanced
Horizontal
Wellvolume
(AHW)
region.
However,
this measure
of the stimulated
does
Introduction
model are not compared with
one another.
More significant
is that
thistakemodel
one the
canproperties
generate
a fracture
new case
in
into account
of the
network,
Microseismic
mapping of hydraulic
fracture stimulations
has using not
has
also been
indicated
to impact
well performance6.
a common technique
to reservoir
map the fracture
growth
and and which
fifteen minutes even whenbecome
the properties
of
the
are
complex
when
there
are
multiple
fractures.
1-5
Furthermore, the permeability enhancement of the fracture
. Microseismic images provide details of the
The primary focus of geometry
the AHW
model is a single well productivity index
and candidate
for
may be analysis,
related to deformation
associatedselection
with fracturing.
fracture azimuth, height, length, and complexity resulting from
Beyond the basic hypocentral locations of the microseisms
stimulation.
interaction with pre-existing fratures. The resulting images can
used to calculate the SRV, additional seismic signal
be used to calibrate numerical simulations of the fracture
characteristics allow investigation of the source of the
growth,
allowing
more
confident
modeling
of
other
Introduction
mechanical deformation resulting in the microseisms. In
stimulations in the field, and a better identification of the
particular, the seismic moment8, a robust measure of the
stimulated region that may ultimately be drained by the well.
strength of an investment
earthquake or microearthquake,
be used to
Fracturing horizontal wells has gained wide acceptance as a viable option to maximize
returns (i.e.can
increased
quantify the seismic deformation9.
Arguably, the Barnett Shale is the field that has had the most

well productivity). This is especially true in the case of tight gas formations. Compared to an openhole horizontal
fracs mapped over the last several years. Microseismic
well, this type of completion
provides a greater area of oil penetration and
a better
flow
efficiency.
It also microseismic
provides
In this
paper, we
examine
several published
mapping in the Barnett has repeatedly demonstrated extreme
projects in a
thevertical
Barnett forfracture,
correlation between
production
more even pressure drop fracture
in the complexity
reservoirresulting
compared
to a vertical
whichtheenhances
from interaction
betweenwell
the intersecting
productivity.
The value of horizontal multiple fractured wells can be maximized when the parameters with the greatest effect
on productivity are understood. As a step towards achieving maximum value from horizontal multifractured wells,
a semi-analytical Advanced Horizontal Well model (AHW) has been developed that allows the user to quickly assess
the impact of various well, fractures or reservoir properties on well performance.
The procedure described has many restrictions in use. The domain (reservoir) must be of special shape (ellipse,
rectangular), the external and internal boundary conditions are usually constant values of pressure or flow rate. The
properties of the wellbore are simplified, its deviation from the straight line is often omitted.
Taking into account all restrictions above, numerical methods for solving reservoir problems are in the vanguard
now. Nevertheless, numerical methods are powerless against high gradient solutions, which occur in singular points
of the domain such as source (or sink), or in the neighborhood of faults. One problem caused by high gradients is
accuracy and stability of numerical methods. Another problem is time resources and machine memory needed for
accurate calculation.
Common features of the existing analytical models are dictated by the methods of solution. These are: constant
reservoir properties with respect to spatial variables (x, y, z), type of external boundary conditions (constant pressure
or no-flow). Also, the pressure drop in the wellbore is neglected.
Of course, all of this simplifies the solution, but it also simplifies the description of the fluid flow in the reservoir.

SPE127742PP

In the review below only models with multiple transverse fractures are considered. The orientation of the fractures
to the wellbore is arbitrary.
The inner boundary condition (at the wellbore) is known flow rate q(t). This is due to fact that the majority of
results are supposed to be used in welltest analysis.
The well is usually positioned at the geometry center of the drainage region (reservoir), and the well is treated as
a straight line, no deviation.
A short overview of most relevant existing models is done below.
An analytical solution for diffusivity equation was performed[1] . The reservoir is homogenous, external boundary
conditions are either of Dirichlet or Neumann type. Boundary condition at the wellbore is flow rate q(t). The fluid is
slightly compressible. The conductivity of the wellbore is infinite. There can exist multiple transverse fractures along
the wellbore. The conductivity of fractures is finite, non-Darcy flow is considered inside the fracture. A solution for
multiple wells with fractures in an isolated reservoir is obtained.
Another semi-analytical model of horizontal well with multiple fractures in a brick-shaped drainage volume was
developed[2] . A 2D Fourier transform is used to obtain an analytical solution. At the same time pressure is calculated
using single-phase numerical simulator. Those two values are used to calculate the effective wellbore radius. General
solutions are developed for horizontal wells with multiple 2D and 3D fractures. The reservoir is homogenous with
constant porosity and constant permeabilities kx , ky , kz . The reservoir has impermeable top and bottom; infinite
acting reservoir or box-shaped reservoir is considered. Multiple fractures with infinite or finite conductivity can be
arbitrarily oriented to the wellbore. The inner boundary condition (at the wellbore) is total flow rate q(t).
Fully analytical model based on the boundary element method combined with Laplace transform is used to perform
a fracture flow model and a reservoir flow model[3] . Based on the solution built for vertical wells in a closed reservoir,
a solution for horizontal wells intersected by multiple hydraulic fractures is developed. The fluid is supposed to be
slightly compressible, the reservoir anisotropy is taken into account. The wellbore has infinite conductivity, and a flow
rate q(t) of the well is known. Early- and long-time approximations are derived. It can be used in welltest analysis
for evaluation of reservoir properties.
An interesting combination of the numerical fracture model with an analytical reservoir model can be found[4] .
The reservoir is anisotropic (kx 6= ky 6= kz ) and infinite acting. The solution is obtained for one well with one fracture,
but the number of fractures can be extended by the method of superposition. The fracture has varying properties
(e.g. conductivity), and its direction might not be straight. The boundary condition at the wellbore is known flow
rate q(t). It is shown that the approximation of complex fracture geometry with pseudo-skin give non-satisfactory
errors at the early stages of production.
The article[5] compares different methods of estimating the horizontal well productivity. These are the well-known
Joshi method, its augmented version, a new analytical solution presented in paper, and a numerical simulation. The
constant-pressure outer boundary ellipse is used in the model. The numerical simulation showed a deviation with an
analytical one. The cause of this deviation was identified, and an augmentation to Joshis solution was done. Also a
new more appropriate analytical solution provided results much closer to the simulation.
Based on the information above, and business needs, a new semi-analytical method for 3-D simulation of reservoir
flow towards the horizontal well was developed. Our model is a combination of a discrete approach in one direction,
and an analytical solution in the other two directions plus time.
The following requirements for the model were made:

Permeabilities are step functions along the direction of the horizontal wellbore;
Wellbore has finite conductivity;
Trajectory of the horizontal wellbore is not a straight line;
There can exist multiple transverse hydraulic fractures. Each fracture has its own size and conductivity; all
fractures can be positioned at any point of the wellbore.

The Background

The differential equation for pressure distribution in the reservoir is a consequence of the Darcy law
k
w
~ = p

(1)

and continuity equation


()
+ div(w)
~ = F .
t

(2)

System (1)(2) of the differential equations for unknown velocity w


~ and pressure p is closed if functions = (p, ~x),
= (p, ~x), = (p, ~x), k = k(p, ~x) defining porosity , liquid density , dynamic liquid viscosity , and permeability

SPE127742PP

coefficient k as functions of pressure p(t, ~x) and coordinates ~x = (x, y, z), and the right-hand side F (t, x, y, p) of the
equation (2) are specified.
Assuming that

Dynamic viscosity is constant;


Permeability tensor k is diagonal and depends on x;
Liquid is almost incompressible: (p) = 0 (1 + l (p p0 ));
Porous media is almost incompressible: (p) = 0 + p (p p0 ).

we arrive at the parabolic equation:




k
p

= div
p + F.
t

(3)

Here F = 1
0 F . The horizontal wellbore is introduced by means of the additional term q(t, x) y yw (x), z yw (x) .
It models a well with the trajectory
W = {(x, yw (x), zw (x)) |0 6 x 6 X},

(4)

and time-dependent flow rate q(t, x). The reservoir permeability is a function of the variable, directed along the
wellbore: kx = kx (x), ky = ky (x), kz = kz (x). The equation (3) reduces to



kz (x) 2 P
kx (x) P
ky (x) 2 P
p
+
+ q(t, x) y yw (x), z yw (x) + F.
(5)
=
+

2
2
t
x
x
y
z
Piezoconductivity equation (5) with suitable initial and boundary conditions determines the pressure distribution
in the reservoir and fluid inflow to the wellbore.
2

Heterogeneous Reservoir Without Fracture

This section describes the basic algorithm to calculate the pressure distribution in the reservoir.
2.1

Formulation of the Problem The equation (5) is considered in the domain


= {(x, y, z) R3 | x (0, X), y (0, Y ), z (0, H)}.

(see Fig. 1)
Viscosity and storage coefficient = ct are constants. The term of equation (5) with Delta-function is
responsible for the fluid inflow q(t, x) to the wellbore.
Sections with
horizontal well

Sections with fracture

y
z

Sections without horizontal well

Fig. 1. Schematic view of the reservoir

The bottom and the top of the reservoir are impermeable:



P
= 0.
z

(6)

z=0,H

view of reservoir
At the remaining part of the outer boundary Schematic
either pressure
with
horizontal
well

P = pe (t)

or zero flow condition



P
=0
n

(7)

(8)

SPE127742PP

is specified. Here n is the outer normal vector to the boundary . At the initial time moment the pressure distribution
is uniform:
P |t=0 = pe (0).

(9)

Pressure along the wellbore (4) is distributed according to the Hagen-Poiseuille law:


P
R4

(x)
= q(t, x), where (x) =
x
x
8

(10)

Functions R(x), (x) are radius and empirical conductivity of the wellbore. The inner boundary condition is specified
either as the given pressure pw (t) at the start point of the horizontal wellbore
P = pw (t),

(x = xl , y = yw (xl ), z = zw (xl ))

(11)

or as the given total flow rate Q(t) of the well


Z
q(t, x) dx = Q(t).

(12)
Sections with
horizontal well

Sections with fracture

At the end point of the wellbore the flow rate equals zero:
y

P
(x)
= 0,
x

(x = xr , y = yw (xr ), z = zw (xr )).

(13)

The mathematical statement of the problem is the xfollowing:


ithorizontal
is required
to satisfy equation (5) in the domain
Sections without
well
with given

initial condition (9);


outer boundary condition (6), and either (7) or (8);
inner boundary condition either (11), or (12);
closure boundary condition (10), (13).

Schematic view of reservoir


with horizontal well

The reservoir pressure and either flow rate or bottomhole flowing pressure should be determined.
Replacing pressure P with the function p(t, x, y, z) = P (t, x, y, z)pe (t) in the case of condition (7), or p(t, x, y, z) =
P (t, x, y, z) pe (0) in the case of condition (8), we arrive to the similar problem, but with zero outer boundary
conditions.
2.2

Method of Solution The idea of solving the formulated problem is following:


Discretization along Ox.
The Ox-axis is splitted into N subintervals by the set of points x1 , . . . , xN , where xi < xi+1 (see Fig. 2).
hi
x1/2 =0 xNl -1/2 =xl

x i-1

xi

x i+1

xN+1/2
=xr
r

x N+1/2 =X

Fig. 2. Discretization along Ox

All functions are assumed to be piecewise constant with respect to x, equal to their mean values over intervals
[xi1/2 , xi+1/2 ]. Then all equations and boundary conditions are to be rewritten using this finite-difference
approach.
Laplace transform with respect to time.
The system of equations obtained at the previous stage is reduced using the Laplace transformation with respect
to time. Thus the initial 3-dimensional evolutionary problem becomes only 2-dimensional over variables (y, z).
Fourier decomposition.
To solve the remaining problem, all functions are to be decomposed in Fourier series over sines and cosines. The
choice of basic functions makes it possible to satisfy the outer boundary conditions: either the specified pressure
or the constant inflow.
The linear algebraic system of equations for Fourier coefficients is 3-diagonal with diagonal predominance, and
the method of its solution is sweep procedure.
Closing relations for this linear system is a pressure continuity condition within the wellbore side surface.

SPE127742PP

Inverse Fourier and Laplace transformations for solution images.


According to the workflow above, the method is semi-analytical: it uses the finite-difference approach along the
wellbore, and the analytical solution in transverse directions.
3

Heterogeneous Reservoir With Multiple Hydraulic Fractures

The algorithm described in the previous section can be adopted to the case with multiple transverse hydraulic fractures
along the horizontal wellbore. It is assumed that each layer along the Ox axis (except for the first and the last one)
may contain a hydraulic fracture (HF), located at the center point x = xi of the corresponding layer (see Fig. 3). We
assume that the height of each HF equals the reservoir thickness Z. In the horizontal direction along the Oy-axis each
fracture has its individual size.
Hydraulic fracture (HF)

wellbore

xi-1

x i-1/2

xi

x i+1/2

Fig. 3. Hydraulic fracture in i-th layer

Hydraulic fractures are taken into account by adding terms

Nr
P

k (t, y)(x xk ) to the right hand side of the

k=Nl

main equation (5), which are responsible for the inflow to the wellbore from the fractures.
Fluid inflow to the fracture in the i-th layer is determined by the discharge intensity i (t, y). For simplicity and
due to the thickness of the reservoir in the z-direction we assume that functions i do not depend on z. In the case
when there is no fracture in the i-th section, the corresponding function i 0.
The flow inside the fracture is supposed to be laminar, and the same approach as with the finite conductivity well
is valid here: pressure drop in HF satisfies the equation


CF

p
y


= (t, y).

(14)

Here the empirical function CF determines HF conductivity. The following boundary conditions for equation (14) are
taken:


p
CF
= 0, p|yi = pi W
(15)

0
y yi ,yi
l

i
i
i
Here
yl and yr are the left and right limiting values of y along HF in the i-th layer; y0 is the position of the wellbore;
i
p W is the wellbore pressure in i-th layer.
The method of solution of modified equations describing the flow in the reservoir with a horizontal well and
multiple transverse hydraulic fractures remains the same as in section 2. Nevertheless, one must add closure relations
i . The following conditions will be used:
to determine unknown flow rates qi and intensities

The reservoir pressure at the wellbore coincide with the wellbore pressure;
The reservoir pressure at HF coincides with the HF pressure.
At least 2 approaches can be used to determine the reservoir pressure at the HF:
Pressure at HF coincides with the average layer pressure: pf = pi .
i1 (4
pi pi1 ) + i (4
pi pi+1 )
, where i is the
Pressure at HF is calculated by linear approximation: pf =
3(i1 + i )
length of the i-th interval.

SPE127742PP

Validation

A reliable finite-difference simulator FDS was used to validate this new model, which can have application not only
for slightly compressible fluid flow, but also for gas flows. The usual approach to switch from liquid to gas is a use of
pseudo-functions, e.g. pseudo-pressure and pseudo-time[7, 8, 9] .
Standard SPE tests for horizontal well simulators are not applicable here, because they are developed for multiphase flow instead of 1-phase flow considered in our model[10] .
Tests have started from simple reservoir models, which complexity was increased with each step. All types of tests
are listed in the tree below.
Homogenous Reservoir
No fractures
Fractured
Heterogenous (along Ox) reservoir
Deviated horizontal well
4.1

Models of the Reservoirs

Numerical Hydrodynamic Model. Productivity layer is supposed to have the form of rectangular parallelepiped with dimensions 1100 100 10 m. The dimensions were chosen based on typical reservoirs for which the
model will be used and based on the widely accepted Seventh SPE comparative solution project for horizontal well
simulation[10] . Cell dimensions of the Cartesian grid in FDS are selected according to the case tested. Several simulations were done on different grids to be convinced that the result is not sensitive to the grid selection. Simulation
of the vertical hydraulic fracture in Oyzplane was done on a refined grid with several buffer blocks in front of and
behind the fracture (see Fig. 7 for explanation).
Permeability of grid cells is chosen depending upon the case considered (type of reservoir, fractured or not).
Porosity and total compressibility ct are constant values in all tests. Only the 1-phase fluid flow is considered.
A numerical aquifer is used in FDS to simulate cases with a constant outer boundary pressure condition.
Fractures in FDS are modeled via a local permeability increase of the cells in the refined grid: the grid in the
Oxdirection was successively decreased from 100 m to 1 cm. Permeability of the cells containing fracture is chosen
so that the equivalent fracture conductivity kf wf , where kf is the real fracture permeability, and wf is the fracture
width, has the same value both in the model and the reservoir.
4.1.1

4.1.2

4.1.3

AHW Model. The model is determined by the following parameters:

Dimensions of the reservoir;


Fluid viscosity ;
Total compressibility ct
Start and end point of the horizontal well, its deviation survey;
Number of layers of heterogeneity along the horizontal well;
Permeability of each layer in x, y and z directions;
Hydraulic fractures, its properties;
Wellbore radius and wellbore conductivity;
Boundary conditions both at the wellbore and at the outer boundary;
Time period for simulation.
Parameters to Compare. The AHW model makes it possible to track the following reservoir and production

parameters:

Pressure as a function of time at the starting point of the wellbore;


Flow rates from each layer as a function of time;
Total flow rate of the wellbore;
Pressure as a function of y at any fixed time t = const, and any section x = const.
Validation Methods and Acceptance Criteria Seventh SPE Comparative Solution Project

[10]

reports comparisons results of simulation runs performed by 14 organizations involving production from a horizontal well in a
rectangular reservoir.
The results from the Seventh SPE comparison study showed that maximum deviation (among 14 simulators) in
cumulative horizontal well production for most cases is about 10%. It is impossible to find a base case here, because

4.1.4

SPE127742PP

there is no exact solution for horizontal well inflow in rectangular reservoir. Therefore, SPE community reports mean
value and standard deviation for each case. For most cases 5%, and we take this value as an acceptance criterion
for the AHW model validation. Expected results between the developed model and FDS are also 10%, but application
of FD simulators for horizontal well inflow modeling is limited (e.g., reservoirs with extremely low permeability).
Thus, the decision was made to compare the most simple cases first and then to increase complexity of the cases to
validate all functionalities of the new model.
This approach makes it possible to identify the validity area of the model as well as to determine its weak points.
Stability of the Models Before running the validation tests, both models in FDS and AHW were put on trials
with different grid block dimensions. FDS gives a stable result with an inessential distinction when:
4.1.5

the number of grid blocks > 3 in both Ox, Oy directions;


the number of grid blocks > 1 in the Oz direction.
Note that the permeability of the reservoir can be a function of only variable x (see Section 2).
The AHW model showed good convergence on different grids1 . Typical plots are given on Fig. 4. Here the following
notation is used: 7(5) Sections mean that the reservoir was split into 7 sections, and 5 of them in the center contain
a well.
The solution method of the model requires at least 5 sections in the Ox direction with mandatory 3 mid-sections
containing a horizontal well.

30

15
14,9
Flow rate, m3/day

Flow rate, m3/day

25
15(13) Sections
20

7(5) Sections
7(3) Sections

15
10

14,8
14,7
14,6

15(13) Sections

14,5

7(5) Sections

14,4

7(3) Sections

14,3
14,2

14,1
14

0
0

50

100
Time, days

150

25,0

200

25,2

25,4

25,6

25,8

26,0

Time, days

Fig. 4. Flow rate curves for AHW. Varying discretization along Ox (left); Zoomed plots (right).

4,5

4.2 4 Homogeneous Reservoirs


3,5
Relative error, %

4.2.1

4,5

No Fractures Tests start from simple homogeneous reservoirs with a straight (not deviated) horizontal well

Relative error, %

4
positioned at the symmetry axis (along Ox) of the rectangular reservoir.
All physical parameters of the rock and the
fluid2,5remain the same in all these runs. Conductivity of the well3,5in our model was set manually while FDS calculates
7(3) Sections
pressure
drop at the wellbore due to friction according to engineering
formulas used in pipeline flow calculations.
2
3
7(5) Sections
Such
parameters as porosity, compressibility, and viscosity 2,5
are just multiplicative constants for one-phase flows;
1,5
7(3)regime
Sections in the wellbore.
permeability influences the duration of different flow regimes in the reservoir and the flow
1
5
1 2
Thus, the storage coefficient ct in all tests is 4.05 10 bar , the permeability in all directions
is 0.1 mD, the fluid
7(5) Sections
1,5
0,5
viscosity
= 0.88 cP.
1
0
Case
#1 (see Appendix for details) shows good correspondence
between FDS and AHW (Fig. 5). The overall
disagreement
for
cumulative
production
at
end
of
simulation
period
is
even less than 1%. Pressure and rate gradients
0
50
100
150
200
0,5
are always high at early times,
hence,
it
is
very
hard
to
obtain
good
matching
there. FDS is very time step sensitive
Time, days
0
within the first 1020 days of production, and it negatively affects
the
match
between100AHW and
FDS. Thus,
the
0
50
150
200
disagreement at early time period had no influence on validation.
Time, days

1 The

term grid for semi-analytical AHW model means how detailed is the discretization along the Ox axis.

SPE127742PP

200
10,0

FDS

8,0

AHW

7,0

Flow Rate

6,0

Cumulative

Error, %

Flow rate, m3/day

9,0

150

100

5,0
4,0
3,0

50

2,0
1,0

0,0

50

100
Time, days

150

200

50

100
Time, days

150

200

Fig. 5. Flow rates, Case #1 (left); Relative Error, Case #1 (right).

10000

200

Eclipse
AHW

Flow raate, m3/day

Cumulative, m
m3

8000

6000

4000

2000

FDS

150

AHW
100

50
0
0

50

100 0
Time, days 0

150

200

100

200

300
Time, days

400

500

600

Fig. 6. Flow rates. Case #2.

8000

4.2.2

Cumulative, m3

Case #2 simulates a depletion drive type of production. In addition, the time of the forecast increased up
to 500 days (Fig. 6). A minor disagreement Eclipse
between FDS and AHW cumulative production curves appeared on
day 200 of the production, after
some
oscillations
AHW in AHW flow rates have occurred due to the transverse Laplace
6000
transform procedure. This disappointing singularity of the semi-analytical approach can be eliminated by using
another procedure, e.g. the Stephest algorithm instead of transformation with Legendre polynomials. Later on the
mismatch in flow rates (oscillations around zero) has generated a divergence of 6.5% in the cumulative production at
4000
late time stages.
Multiple Transverse Fractures

2000
Case #3 (see Appendix) is intended
to make sure that AHW and FDS give similar results on the refined grid (Fig. 7),
prepared to simulate vertical transverse fractures (Fig. 8).
0

fracture in one cell of dx =0.01m


5*100m

600
1*25m

500
2*10m

2*1m

400
1*2.5m

Time, days

1*0.25m

2*0.1m

10*0.01m

300
2*0.1m

1*0.25m

200
2*1m

2*10m

1*2.5m

100
1*25m

5*100m

5*100m - 5 ggrid blocks with dx =100m

Fig. 7. Progressive grid refinement for fracture simulation

It is observed that both simulators perform equally on the refined grid without fractures, which shows good stability
of the numerical methods used.

SPE127742PP

10,0

10,0

9,0

9,0
8,0

7,0

Flow Rate

6,0

Cumulative

Error, %

Error, %

8,0

5,0
4,0

7,0

Flow Rate

6,0

Cumulative

5,0
4,0

3,0

3,0

2,0

2,0

1,0

1,0

0,0

0,0

50

100
Time, days

150

200

50

100
Time, days

150

200

Fig. 8. Relative Error. Case #3 (left); Case #4 (right).

Hydraulic fractures with significantly different conductivities were simulated in cases #47. Case #5 showed
minor oscillations of the flow rate (Fig. 9) due to very close placement of the fracture tip to the outer boundary of
the reservoir, and it resulted in the pressure gradient of approximately 100 bars within 5 meters of porous media. To
reduce the relative error in this case (7.5% between the blue and the red lines), an increase of the number of basic
functions is needed. Of course, the time of simulation grows much faster than the rate of convergency: the number of
basic functions in both directions (y, z) was increased by factor 1.5, but the accuracy increased by only 2.4% (greenred
lines, Fig. 9).
250

Flow
w rate, m3/day

200

FDS
AHW short series

150

AHW extended series


100

50

0
0

50

100
Time, days

150

200

Fig. 9. Flow rates. Case #5.


16000

10,0
9,0

Eclipse

8,0

12000

AHW short series 8,0

6,0
5,0
4,0

10000

Flow Rate

9,0

8000
6000

5,0
4,0

4000

3,0

2,0

2000

2,0
1,0

0,0
0

50

100
Time, days

Cumulative

6,0

3,0

1,0

Flow Rate

AHW externed series


7,0

Cumulative

Error, %

Cumulative, m33

7,0
Error, %

10,0

14000

150

50

200

100

0,0

Time, days

150

50

200

100
Time, days

150

200

Fig. 10. Relative Error. Case #6 (left); Case #7 (right).

Cases #4, #8 with real fracture conductivity (10003000 mD m) show good agreement (approximately 2%)
with the numerical simulator (Fig. 8 (right), Fig. 11 (left)), while other cases with extremely low fracture conductivity
(Fig. 10) gave a relative error of 34% in the flow rate, and resulted in the total error of 24% in cumulative production.

10

SPE127742PP

The AHW model has an option to simulate non-symmetry vertical fractures, whenthe extreme case of which is
when only one wing of the fracture exists. Such experiment (Fig. 11, right) was launched using a grid for symmetry
fractures, and the result is quite good: disagreement between AHW and FDS is small, and the diagram for pressure
distribution at the section with fracture shows how the solution is sensitive to fracture parameters (Fig. 12).
10,0

10,0

9,0

9,0
8,0

7,0

Flow Rate

7,0

Flow Rate

6,0

Cumulative

6,0

Cumulative

Error, %

Error, %

8,0

5,0
4,0

5,0
4,0

3,0

3,0

2,0

2,0

1,0

1,0

0,0

0,0

50

100
Time, days

150

200

50

100
Time, days

150

200

Fig. 11. Relative Error. Case #8 (left); Case #9 (right).

200

Presssure, bar

150

T=0,6

100

T=3,5
T=9,1
50

T=21,3
T=148

0
0

10

20

30

40

50

60

70

80

90

100

Y, meters

Fig. 12. Pressure distribution along the section with non-symmetry fracture. Case #9.

Before making multiple fractures, the stability of calculations was checked on the refined grid. As in Case #3, each
fracture in FDS was simulated via a local increase of permeability so that the equivalent permeability (in mD m) both
in AHW and FDS was similar. Thus, Fig. 13 contains a production forecast on the fine grid with original reservoir
properties; plots on Fig. 14 are the product of 3 fractures simulated at X0 = 550m, X0 = 650m, and X0 = 750m.
Again, some oscillations in production curves occur, which has further influence on cumulative production.

Flow
w rate, m3/day

200

FDS

150

AHW
100

50

0
0

50

100
Time, days

150

Fig. 13. Flow rates. Case #10.


12000
10000

Eclipse

e, m33

AHW
8000

200

SPE127742PP

11

350
10,0
9,0
8,0

FDS

250

7,0

AHW
200

Error, %

Flow
w rate, m3/day

300

150
100

6,0
5,0
4,0
3,0
2,0

50

Flow Rate

1,0

Cumulative

0,0

50

100
Time, days

150

200

50

100
Time, days

150

200

Fig. 14. Flow rates, Case #11 (left); Relative Error, Case #11 (right).

Due to this the relative error of flow rate increased up to 9.2% at the steady state regime, though the error of
14000
the cumulative production curve decreases to 6% with further negative gradient (Fig. 14). Such oscillations may
12000
sometimes
appear,Eclipse
but they are connected only with the numerical algorithm for the transverse Laplace transform
AHW
procedure (see also
Fig. 6).
Straight Well Shifted to the Boundary All previous tests were conducted with a well positioned at the
symmetry
axis of the reservoir. Here (Appendix, cases #1214) the well is shifted towards one or two boundaries of
8000
the reservoir, no fractures. Extreme cases are considered, when the well is only 0.5 m off the external boundary.
6000
150

4000

Flow
w rate, m3/day

2000
0
0

50

FDS
100

AHW

100

150

200

Time, days
50

0
0

50

100
Time, days

150

200

Fig. 15. Flow rates. Case #12.

Cumulative, m33

In case #12 the well is drawn12000


up to the impermeable upper boundary Z = const (top of the reservoir). An overall
error in flow rate is approximately 3% (Fig. 15),Eclipse
which shows that AHW and the numerical model of FDS are still
10000
in satisfactory agreement. The disagreement
between AHW and FDS is between 2% and 4% depending on the well
AHW
deviation from the center line of the reservoir.
8000
10,0
9,0
6000
8,0

4000
7,0
Error, %

Cumulative, m33

10000

4.2.3

6,0

Flow Rate

2000

5,0

Cumulative

4,0

3,0

50

2,0

100

150

200

150

200

Time, days

1,0
0,0
0

50

100
Time, days

Fig. 16. Relative error. Case #13.

12

SPE127742PP

FDS and AHW coincide better (only 0.7% of relative error) when the well is moved close to the open boundary,
i.e. Y = const (case #13, Fig. 16).
Such difference in two previous cases is caused by the fact that a smaller number of basic functions is used in the
Ozdirection in comparison with the Oydirection (100 against 200). The increase in the series length in Ozdirection
leads to a significant increase of time for the execution with an increase in accuracy by only several fractions. Thus,
the golden mean is preferable in this case: both satisfactory accuracy and running time of the program.
Case #14 is the superposition of two previous cases. The number of harmonics in the Fourier series was optimized,
and the total relative error dropped to 0.6%. This was done just to show that the adjustment between the models is
possible, but prior to forecast execution one must set an appropriate mistake according to common sense and select
the series length. This will remain as an option of AHW, similar to FDS time step choice.
Cases with Extreme Permeability Values Three cases with reservoir permeability of 0.01 mD, 10 mD and
10 D were studied (see table below). Extremely low permeability (case #15, Fig. 17, Appendix) leads to inefficient
disagreement between AHW and FDS at early-time flow rates. The flow rates on steady-state regime are nearly equal:
0.34% of relative error.
4.2.4

150

30
25
FDS

Flow
w rate, m3/day

Flow
w rate, m3/day

FDS
100

AHW

50

20

AHW

15
10
5
0

0
0

50

100
Time, days

150

200

100

200
Time, days

300

400

Fig. 17. Flow rates. Case #14 (left); Flow rates. Case #15 (right).

automatically
tracks the type of the fluid flow in the well: as Reynolds number
exceeds 4000, the friction factor is
AHW
14000
AHW
calculated using another formula, while the AHW model still
2500 uses the Poiseuille equation (10). Case #16 (Fig. 18)
12000 this disagreement between FDS and AHW. After the conductivity of the well in AHW was manually decreased,
shows
the
results of simulations nearly coincided (green curves on2000
the Fig. 18).
10000
Cumulative, m33

8000
5000
60004500

10,0

1000 9,0

40004000
Flow
w rate, m3/day

1500

FDS

20003500

500

50

100

2000

8,0
7,0

AHW

03000
2500
0

AHW
150 low conduct.
200

Time, days

Error, %

Cumulative, m33

3500
18000
An increase in reservoir permeability up to 10 mD results in higher well productivity. The flow in the wellbore
switches
to the turbulent
type, and the pressure gradient 3000
along the horizontal
16000
Eclipse segment of the well increases. FDS
Eclipse

0 6,0
100

200

3,0

1000

2,0

500

1,0

300 Cumulative 400

Time, days

4,0

1500

Flow Rate

0
5,0

0,0

50

100
Time, days

150

200

50

100
Time, days

150

200

Fig. 18. Flow rates. Case #16 (left); Relative error. Case #16 (right).

1000000

Cumulative, m33

900000
Cases #17, 18 are
similar except the wellbore friction option. Once increased reservoir permeability up to 10 D
Eclipse
resulted
800000 in a sufficiently turbulent flow in the wellbore, and the disagreement between two simulators became visibly
AHW
700000
AHW low conduct.
conduct
600000
500000
400000
300000

SPE127742PP

13

high (Fig. 19). After the wellbore friction option was disabled, the agreement between simulators was restored
(Fig. 20).
180000
160000
Flow
w rate, m3/day

140000
120000
100000

FDS

80000

AHW

60000
40000
20000
0
0

50

100
Time, days

150

200

Fig. 19. Flow rates. Case #17.

Cumulative,
m33
Error,
%

35000000
10,0
30000000
9,0
25000000
8,0

Eclipse
AHW
Flow Rate

7,0
20000000

Cumulative

6,0

15000000
5,0
10000000
4,0
3,0
5000000
2,0
0
1,0

0,0

50

50

100

150

Time, days
100
Time, days

150

200
200

Fig. 20. Relative error. Case #18.

4.3

Heterogeneous Reservoirs

Constant Permeability The reservoirs simulated here have constant, but different permeability in x, y, and z
directions. It means that kx , ky , kz are constant values, though they are not equal.
The results of FDS runs with different number of cells in each coordinate axis direction were analyzed before
comparison with the AHW model. During these adjustment runs the optimal number of cells was selected: the
minimal number of cells in each direction, when its further increment had a negligible effect on the result.
4.3.1

500

30

FDS

25

AHW

450
Flow
w rate, m3/day

Flow
w rate, m3/day

35

20
15
10

400

FDS

350

AHW

300
250
200
150
100

50
0

0
0

50

150

200

70000

3500

Eclipse

3000

AHW

60000

Eclipse
AHW

ve, m33

50000
2500

50

Fig. 21. Flow rates. Case #19 (left); Case #20 (right).

4000

ive, m33

100
Time, days

40000

100
Time, days

150

200

14

SPE127742PP

Cases #1922 (see Appendix) are typical with respect to the variety of tested ones within this section. The overall
trend is as follows: when permeability in Oy direction dominates permeability in both Ox and Oz directions, AHW
model production curves lay slightly below the FDS production curves (Fig. 21Fig. 22).
Then, if kx values are dominant, the rates calculated by the AHW model exceed those obtained via FDS (Fig. 22).
The behavior of production curves described above does not depend on the number of cells in FDS or on the
number of layers and series lengths in AHW.

Flow
w rate, m3/day

200
FDS

150

AHW
100

50

0
0

50

100
Time, days

150

200

Fig. 22. Flow rates. Case #22.


12000

Varying Permeability The heterogeneity of the reservoir in the AHW model can be magnified in comparison

4.3.2

Cumulative, m33

with section 4.3.1, meaning that10000


the componentsEclipse
of the permeability tensor kx , ky , and kz can be step functions of x
AHW
variable (along the main direction of the wellbore).
8000
New model was tested on reservoirs
with the following properties (see table on page 17).
10,0
9,0

6000

10,0
9,0

4000

8,0

8,0

2000

7,0

Cumulative

6,0

5,0

50

4,0

100

Time, days

5,0

2,0

2,0

1,0

1,0

0,0

0,0
100
Time, days

150

200

150

200

4,0
3,0

50

Cumulative

6,0

3,0

Flow Rate

7,0
Error, %

Error, %

Flow Rate

50

100
Time, days

150

200

Fig. 23. Relative error. Case #23 (left); Case #24 (right).

Cases #2326 (see Appendix) cover several extreme scenarios of the reservoir structure. It is known, that numerical
algorithms may have difficulties (instabilities) when functions in the design domain are highly discrete. To overcome
such problems, special methods of calculations are required.
Listed cases were executed using several grids with different mesh sizes (for AHW the term mesh of the grid
means the number of layers along Ox bounded by planes x = const). The final time of simulation T = 200 days is
sufficient to reach a steady-state production period.
Typical relative errors are shown on plots (Fig. 23,Fig. 24). The agreement in flow rates and cumulative production
is excellent except that AHW rates have oscillations at some time segments.
4.4 Deviated Wellbore Survey One more option is available in AHW model: the wellbore can be positioned
at any place of the reservoir, and its deviation survey can be not a straight line, but a step function determined at
every layer separately. Thus, the well becomes a broken line with infinite conductivity connections at the boundaries
of layers with x = const.

SPE127742PP

15

10,0

10,0

9,0

9,0

8,0

Flow Rate

7,0

Cumulative

6,0

Error, %

Error, %

8,0

Flow Rate

7,0

5,0
4,0

5,0
4,0

3,0

3,0

2,0

2,0

1,0

1,0

0,0

Cumulative

6,0

0,0

50

100
Time, days

150

200

50

100
Time, days

150

200

Fig. 24. Relative error. Case #25 (left); Case #26 (right).

The reservoir under consideration is still rectangular shaped with original dimensions: 1100 m along Ox, 100 m
along Oy, and 10 m along the Oz axis. The permeability is constant: 0.1 mD in all directions; the fluid and the wellbore
properties are as described in section 4.1. The well starts at x0 = 100 m and finishes at the point x1 = 1000 m. The
fluid in the wellbore flows to the left, e.g. towards the starting point x0 .
In case #27 (see Appendix) the well changes its direction in the Oxz plane. The vertical coordinate of the well
center Zc 6= const and it jumps from 0.5 m (distance to the bottom of the reservoir) to Zc = 5.5 , which is approximately
the center of the reservoir.
Next, in case #28 the wellbore lays in the Oyz plane, and switches between 16.7 and 83.3 meters stand-off the
side reservoir boundary y = 0.
Case #29 is the superposition of two previous cases.
10,0

10,0

9,0

9,0

8,0

Flow Rate

7,0

Cumulative

6,0

Error, %

Error, %

8,0

Flow Rate

7,0

5,0
4,0

Cumulative

6,0
5,0
4,0

3,0

3,0

2,0

2,0

1,0

1,0
0,0

0,0
0

50

100
Time, days

150

200

50

100
Time, days

150

200

Fig. 25. Relative error. Case #27 (left); Case #29 (right).

From the plots above one can see how wellbore placement influences the production rates during transient and
steady-state regimes.
Relative error between AHW and FDS is less than 1% in all considered cases.
5

Area of Applicability of the Advanced Horizontal Well Model

Based on more than 50 comparative runs of AHW and FDS, one can determine the boundaries of new model applicability.
5.1

Good Performance
Homogenous and heterogeneous reservoirs with non-constant permeability along a horizontal well;
Reservoirs with multiple hydraulic fractures, each with its own physical and geometry properties;
Reservoirs with deviated horizontal wells, minimum recommended distance between well and outer boundary of
the reservoir is 0.5 meters (2 feet);
Laminar flow in the wellbore.

16

5.2

SPE127742PP

Poor Performance
Reservoirs with high permeability and long horizontal wells lead to turbulent flows inside the wellbore. Then
pressure distribution equation (10) needs additional adjustment via special type of function, or the model
should be equipped with the turbulent flow regime in the wellbore (Fig. 19);
Long time simulation on depletion drive type reservoirs due to minor oscillations of the flow rate curve near
zero values may lead to unacceptable mismatch (Fig. 6). This issue, as it was mentioned on page 8, is only a
problem of numerical algorithm.

Model Summary

The Advanced Horizontal Well Model is developed to meet the requirements listed in Introduction on page 2. According
to the specific of the task, we used finite-difference approximation in the Ox direction combined with Fourier series
in both Oy and Oz directions. Also, as governing equations are linear with respect to time t, the Laplace transform
technique was applied.
The whole reservoir was divided into sections by planes, parallel to the Oyz-plane. The model makes it possible
to describe cross-flows between these sections, and moreover, the cross-flows behind the fractures (as tips of fractures
do not reach the boundary of the reservoir).
The pressure drop along the wellbore is taken into account, though the current realization assumes the laminar
type of flow inside the wellbore only.
Brief comments on some functionalities of the model:
Varying Permeability
Components of the permeability tensor depend upon variable x, directed along the horizontal wellbore. This
functionality makes it possible to simulate reservoirs with pronounced heterogeneity in one horizontal direction.
Comparison tests with FDS have showed good agreement, even when the permeability of adjacent sections
differed by factor 10 (see section 4.3.2).
Hydraulic Fractures
Transverse hydraulic fractures can be placed within any section that contains a well. All fractures can have
non-symmetry wings with different conductivities.
Tests have showed that even in extreme conditions AHW gives results very close to FDS. There are no limitations
in fracture conductivity, but there is a limitation in fracture length: the tip of the fracture should not be close
to the outer boundary (see Fig. 9 and the description).
Finite Wellbore Conductivity
The model and the numerical algorithm used are designed to take into account finite wellbore conductivity.
One of the possible solutions is to use HagenPoiseuille law (10), which is good only for laminar flows. As the
Reynolds number grows, the flow regime transforms from laminar to turbulent. FDS automatically tracks the
flow type, while the AHW model has no such option (see Fig. 19). As a result, there may be some unacceptable
discrepancies between FDS and AHW at high flow rates.
Deviated Well
The wellbore in AHW can have curved survey. Coordinates of the well center in every section containing well
are specified independently from other sections. Thus, the wellbore becomes a broken line, the vertical segments
of which have infinite conductivity.
The most complicated cases are when the well is positioned close to outer boundary, i.e. when the pressure
has high gradient. Cases #1214 and #2729 show that even if the wellbore is only 0.5 m stand-off from the
boundary, the disagreement between AHW and FDS is only 23%.
7

Conclusion

This paper illustrates the use of a prediction tool for improved modelling of oil production from horizontal or deviated
wells with induced fractures. This was achieved by implementing original semi-analytical solutions into the model,
to provide a useful diagnostic and prognostic tool for the oil industry. The model robustness and speed makes this
programme for well production optimization of a horizontal well with induced fractures easy to use.
The approach has been validated against industry recognized finite-difference simulator.
Overall, an obvious benefit to the industry from this model is a possibility to make fast production prediction for
long horizontal wells with multiple transverse hydraulic fractures in heterogeneous reservoirs.
Acknowledgements
The authors express their gratitude to Schlumberger for the problem statement and support.

SPE127742PP

17

Appendix. Cases
#
X
1
2
3
4
5
6
7
8
9
10
11

Reservoir, m
Y
Z

1100
1100
1100
1100
1100
1100
1100
1100
1100
1100
1100

100
100
101
101
101
101
101
101
101
101
101

10
10
11
11
11
11
11
11
11
11
11

FDS Cells
Y

11
11
482
48
48
48
48
48
48
1504
150

3
3
3
3
3
3
3
3
3
3
3

1
1
1
1
1
1
1
1
1
1
1

Boundary Cond.
Well
Outer
BHP
BHP
BHP
BHP
BHP
BHP
BHP
BHP
BHP
BHP
BHP

Fracture
Y (m)/N

CP
NF
CP
CP
CP
CP
CP
CP
CP
CP
CP

Conductivity,
mD m

N
N
N
40m
90m
40m
40m
40m
20m3
N
3 40m5

1000
1000
10
1
3000
3000
3000

Well standoff6
Y = const
Z = const
12
13
14

1100
1100
1100

15
16
17
18

101
101
101

1100
1100
1100
1100

100
100
100
100

11
11
11

11
11
11

10
10
10
10

101
101
101

11
11
11
11

3
3
3
3

100
11
100

1
1
1
1

BHP
BHP
BHP

CP
CP
CP

BHP
BHP
BHP
BHP

50.5
0.5
0.5

0.5
5.5
0.5

Permeability,
mD

Wellbore
friction (Y/N)

0.01
10
10 000
10 000

Y
Y
Y
N

CP
CP
CP
CP

Permeability, mD
kx
ky
kz
19
20
21
22

1100
1100
1100
1100

100
100
100
100

10
10
10
10

11
11
11
11

3
3
3
3

1
10
10
10

BHP
BHP
BHP
BHP

CP
CP
CP
CP

0.1
1
0.1
1

0.1
1
1
0.1

0.01
0.1
0.1
0.1

Y
Y
Y
Y

BHP: Bottomhole pressure control;


CP: Constant pressure;
NF: No flow.

0100
ky

kx
23
24
25
26
#
23
24
25
26

kz

0.01
1
1
1

0.01
0.01
1
1
1
0.1
0.1
0.5
600700
0.07
0.07
0.07
1
1
1
1
1
0.1
1
0.1
0.5

100200
ky

kx
0.02
0.1
0.1
0.1

0.02
0.02
0.1
0.1
0.1
1
1
5
700800
0.08
0.08
0.08
0.1
0.1
0.1
0.1
0.1
1
0.1
1
5

0100
Yc
Zc

100200
Yc
Zc

27
28
29
#

N
N
N
N
N
N
800900
Yc
Zc
50
5.5
83.3
5.5
83.3
5.5

50
0.5
16.7
5.5
16.7
0.5
9001000
Yc
Zc
50
0.5
16.7
5.5
16.7
0.5

27
28
29

kz

kx

Permeability along Ox, m in mD


200300
300400
ky
kz
kx
ky

0.03
1
1
1

0.03
0.03
1
1
1
0.1
0.1
0.5
800900
0.09
0.09
0.09
1
1
1
1
1
0.1
1
0.1
0.5

0.04
0.1
0.1
0.1
0.1
0.1
0.1
0.1

0.04
0.1
0.1
1
9001000
0.1
0.1
0.1
1

kz

kx

0.04
0.1
1
5

0.05
1
1
1

0.1
0.1
1
5

0.05
1
1
0.1
10001100
0.11
0.11
1
1
1
1
1
0.1

Wellbore Coordinates in the Layer (Yc , Zc )7 , m


200300
300400
400500
500600
Yc
Zc
Yc
Zc
Yc
Zc
Yc
Zc
50
5.5
83.3
5.5
83.3
5.5
1000-1100
Yc
Zc
N
N
N
N
N
N

50
16.7
16.7

0.5
5.5
0.5

50
83.3
83.3

5.5
5.5
5.5

50
16.7
16.7

0.5
5.5
0.5

N: No well in current layer.

2 Progressively

refined towards the layer with fracture Fig. 7


fracture: only one wing l = 20m exist.
4 Progressively refined grid analogous to Fig. 7 for 3 separate fractures in different Ox-sections
5 3 fractures, each 40m of length (20m half length)
6 In meters from the corresponding boundary
7 The origin of the Cartesian coordinate system is placed at the bottom corner of the reservoir
3 Non-symmetry

400500
ky

kz

kx

500600
ky

kz

0.05
1
0.1
0.5

0.06
0.1
0.1
0.1

0.06
0.1
0.1
1

0.06
0.1
1
5

0.11
1
0.1
0.5

600700
Yc
Zc
50
83.3
83.3

5.5
5.5
5.5

700800
Yc
Zc
50
16.7
16.7

0.5
5.5
0.5

18

SPE127742PP

Nomenclature
=
ct =
CF =
=
i =
=
=
=
=
=
k=
ki =
=
n=
P, p =
p =
q=
Q=
=
R=
t=
=
~x =
w
~ =
W =

storage coefficient
isothermal total compressibility
fracture conductivity
Dirac delta-function
length of i-th interval over Ox axis
porosity
discharge intensity of the fracture
wellbore conductivity
reservoir outer boundary
empirical specific wellbore conductivity
permeability tensor
diagonal elements of k, i = x, y, z
fluid viscosity
outer normal vector to
fluid pressure
Laplace image of p
specific flow rate
total flow rate
fluid density
wellbore radius
time
reservoir domain
Cartesian coordinates, ~x = (x, y, z)
fluid velocity vector
wellbore manifold

References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.

Brusswell, G., Banerjee, R., Thambynayagam, R.K.M., Spath, J.: Generalized Analytical Solution for Reservoir
Problems With Multiple Wells and Boundary Conditions, SPE 99288
Wan, J., Aziz, K. Semi-Analytical Well Model of Horizontal Wells With Multiple Hydraulic Fractures // SPE
81190
Zernar, A., Bettam, Y. Interpretation of Multiple Hydraulically Fractured Horizontal Wells in Closed Systems //
SPE 84888
Al-Kobaisi, M., Ozkan, E., Kazemi, H. A Hybrid Numerical/Analytical Model of a Finite-Conductivity Vertical
Fracture Intercepted by a Horizopntal Well // SPE 92040
Economides, M., Deimbacher, F., Brand, C., Heinemann, Z. Comprehensive Simulation of Horizontal Well
Performance // SPE Formation Evaluation, December 1991
Krylov, V.I., Skoblia, N.S. Handbook on numerical inversion of the Laplace trnsformation Izdatelstvo Nauka
i Tekhnika, Minsk, 1968.
AlHussainy, Ramey, H.J., Jr., Crawford, P.B. The Flow of Real Gases Through Porous Media // JPT, May,
1966
Agarwal, R,G. Real Gas Pseudotime A New Function for Pressure Buildup Analysis of MHF Gas Wells //
SPE 8279
Chen, H.Y., Poston, S.W. Application of a Pseudotime Function to Permit Better Decline Curve Analysis //
SPE 17051PA
Nghiem, L., Collins, D., Sharma, R. Seventh SPE Comparative Solution Project: Modeling of Horizontal Wells
in Reservoir Simulation // SPE 21221

Vous aimerez peut-être aussi