Professional Documents
Culture Documents
B.J. Carter
, J. Desroches
, A.R. Ingraffea
L
h
|
\
|
.
n / (2+
n )
(
(
(1)
w =
2 /(2 + n )
L
h
n /(2 + n )
c
1
( n ) c
2
( n )
L
|
\
|
.
2+ 3 n
4+2 n
(
(
(2)
where
L
h
= V
K
E
|
\
|
.
1/ n
(3)
is a characteristic length particular to hydraulic fracturing and
p
h
= E
cos((1 ))
sin()
|
\
|
.
1+ n
2 n +1
n
2
(2 + n )
|
\
|
.
n
(
(
1 (2 + n )
(4)
is a characteristic pressure with ( n ) = 2 (2 + n ) . V is the crack tip speed which is equal to
the fluid velocity at the tip. L is the fracture half length and is the position from the crack tip.
3
is the far-field minimum stress. E = E (1
2
) is the effective Young's modulus where E
is the Young's modulus and
is Poisson's ratio. K is the consistency index and n is the
power-law exponent. c
1
and c
2
are constants (SCR, 1994) that are evaluated best numerically;
for a Newtonian fluid, they are c
1
7.21 and c
2
3.17. Note that the flow rate at the tip q
b
is a
function of the speed and the crack opening, q
b
= V w = f (V) .
From these equations, the fluid pressure at the crack tip is found to be singular; for a Newtonian
fluid, n = 1 and the order of the singularity is 1/3. The stress at the crack tip has the same order
of singularity. The order of the singularity depends on the fluid properties only and, within the
assumptions made here, is always weaker than the 1/2 obtained from linear elastic fracture
mechanics.
A similar solution was developed for the permeable case and can be found in Lenoach (1995). It
involves a supplementary length scale because of the leak-off process and yields a solution which
exhibits yet another singularity, also weaker than that of linear elastic fracture mechanics.
4.2.2 Extension of the LEHF Solution to 3D
The behavior of a hydraulic fracture in the vicinity of its propagating front is easily described by
the LEHF solution, provided two additional assumptions are made: the crack front is considered
to be locally under plane strain conditions, and the local fluid flow parallel to the crack front is
negligible. We shall restrict ourselves to the case of a Newtonian fluid, but the process can easily
be extended to power-law fluids. The width w
in the vicinity of the fracture front is then
described by the LEHF solution along any normal to the crack front:
w = (2) 3
7 6
( )
V
E
|
\
|
.
1 3
2 3
= V
1 3
(5)
where = (2) 3
7 6
( )
E
|
\
|
.
1 3
2 3
and is the curvilinear distance, measured on the fracture
surface, between any point and the fracture front. is the fluid viscosity.
The equation describing the mass conservation within the bulk of the fracture can then be
transformed in order to accommodate the introduction of the LEHF solution. The fracture is
considered as a surface in 3D space, F. Consider a subdomain of the fracture, characterized
by a boundary , a pressure field p , an aperture field w and fluid flux field q . t is the time.
If one considers any closed domain A included in (Figure 7), fluid enters and leaves A
through its boundary A. Fluid is stocked by an increase of the fracture width w with time t .
Considering also a source term at a point O included in A (injection or sink of value Q(t ) ),
one then can write for an incompressible fluid:
q n dA
A
} +
w
t
dA
A
} = Q(t)
at O
(6)
After transformation of the contour integral into a surface integral, one gets
(
w
t
A
}
+ divq Q(t)(O))dA = 0 (7)
where is the Dirac operator. This equation must hold whatever the chosen domain A, which
yields the mass conservation equation for any point inside the considered domain :
w
t
+ divq = Q(t)(O) (8)
One can write a variational equation for (8) by considering an auxiliary pressure field p:
p
w
t
}
d + p divq d
}
= Q(t )P(O) (9)
Using s as the curvilinear abscissa along the boundary and n
s
as the outward normal to at
point s (Figure 7), one can apply the divergence theorem to the second integral of equation (9),
to get
p
}
divq d = q grad(p)
}
d + p q
n
s
( )
}
ds (10)
The lubrication approximation for a Newtonian fluid of viscosity gives
q =
w
3
12
gradp (11)
Setting q
n
s
= q
b
(Figure 7) and plugging the lubrication equation into equation (9), it becomes
p
w
t
}
d+ (
w
3
12
}
gradp) (gradp) d + p
}
q
b
d = Q(t)p(O) (12)
As illustrated in Figure 7, it is convenient to choose so as to describe the bulk of the fracture,
whereas F- corresponds to the domain of the fracture in the vicinity of the crack front that is
described by the LEHF solution. As now is the boundary between and the front region of
the fracture where the asymptotic solution holds, then
q
b
= V w (13)
where V is the velocity of the crack and w is now a known function of V and the distance
from the crack front. Therefore, q
b
= V
4 3
, which gives the final fluid flow equation to be
solved outside the vicinity of the crack front:
p
w
t
}
d+ (
w
3
12
}
(gradp gradp) d+ p
}
V
4/ 3
d = Q(t)p(O) (14)
The first two integrals correspond to the usual hydraulic fracture solution. The third integral is a
contour integral that incorporates the asymptotic solution for the fluid flow. The right hand side
accommodates the boundary condition of a source term at any point and can be extended for the
case of a line source.
5.0 Finite Element Implementation of Fluid Flow Coupled with the Structural Response
The finite element method is used to solve the fluid flow equation coupled with the structural
response. The special form of the fluid flow equation developed in section 4 allows one to
separate the fracture surface in two regions: the bulk of the fracture where width and pressure
are to be determined and regular shell finite elements are used, and the near vicinity of the crack
front where special tip elements adjacent to the crack front are used. Thanks to the form of
equation (14), the special tip elements only represent the LEHF solution, and nodal width and
pressure are never computed, except at the boundary between the regular elements and the tip
elements.
5.1 Finite Element Formulation of Fluid Flow Equation
Considering a set of n shape functions N over the domain , the pressure can be approximated
in the finite element sense by p = N
i i =1
i =n
p
i
where p
i
is the pressure at node i (or in matrix form
p = N
T
p ). The small increment in pressure p is approximated in a similar way
p = N
i i =1
i =n
p
i
. These two expressions, when introduced into (14), lead to
N
i
p
i
i=1
i =n
|
\
|
.
}
w
t
d +
w
3
12
}
grad N
j
p
j
j =1
j =n
|
\
|
.
| grad N
i
p
i
i =1
i= n
|
\
|
.
d
+ N
i
p
i
i =1
i =n
|
\
|
.
}
V
4 / 3
d = Q(t) N
i
(O)Q(t)
i =1
i =n
|
\
|
.
(15)
Requiring stationarity of (15) leads to a set of n differential equations. Introducing the finite
element approximation for the width w
together with the time discretization, the i -th equation
can be written as:
N
i
}
N
j
w
j
(t
n+1
) w
j
(t
n
)
t
n+1
t
n j =1
j =n
|
\
|
.
|
d
+
1
12
N
k
w
k
k=1
k =n
|
\
|
.
}
3
gradN
i
grad N
j
p
j
j =1
j =n
|
\
|
.
|
d
+ N
i
V
4 3
}
d = Q(t)N
i
(O)
(16)
where w
j
is the value of the width w at node j .
For simplicity, the i -th shape function is associated with the node at which it takes a value of 1.
It is worth noting that a shape function N
i
is equal to zero when the point considered for the
integration is not part of an element containing the associated i -th node. Therefore, the integrals
in (16) can be reduced to integrals over a single element
i
.
N
i
i
}
N
j
w
j
(t
n+1
) w
j
(t
n
)
t
n+1
t
n j =1
j = n
|
\
|
.
|
d
+
1
12
N
k
w
k
k =1
k= n
|
\
|
.
i
}
3
gradN
i
grad N
j
p
j
j =1
j = n
|
\
|
.
|
d
+ N
i
V
4 3
i
}
d = Q(t)N
i
(O)
(17)
where
i
is the portion of the boundary which intersects the element
i
. Two cases can be
considered, depending on the location of the element
i
with respect to the boundary .
5.2 Element not Adjacent to
If
i
is void (the element is not adjacent to the boundary), the computation of the various
integrals over
i
is straightforward and the integral over
i
is zero. The i -th equation can be
reordered in the following way:
Q(t )N
i
(O) = N
i
N
j
w
j
(t
n+1
) w
j
(t
n
)
t
n+1
t
n j =1
j =n
|
\
|
.
|
i
}
d
+ p
j
j =1
j =n
1
12
N
k
w
k
k=1
k =n
|
\
|
.
3
gradN
i
gradN
j
i
}
|
\
|
.
| d
(18)
5.3 Element Adjacent to
If
i
is not void (the element is adjacent to the boundary), the second and third term of equation
(17) are to be modified. The pressure p
i
at each node belonging to the boundary
i
satisfies
p
i
=
P
i
E
2 3
3
i
|
\
|
.
|
1 3
V
i
1 3
(19)
Note that this equation simply means that the derivative of the pressure normal to the boundary is
known. Assuming that there are n nodes in the considered element, m of which are not on the
boundary, the second term I
i
of equation (17) can be transformed into:
I
i
=
1
12
i
}
N
k
w
k
k
|
\
|
.
3
gradN
i
grad N
j
p
j
j =1
n
|
\
|
.
| d
=
1
12
i
}
N
k
w
k
k
|
\
|
.
3
gradN
i
grad N
j
p
j
+ N
j
(
P
j
j
V
j
1 3
)
j =m+1
n
j =1
m
|
\
|
.
| d
(20)
where
j
= E
2 3
3
j
|
\
|
.
|
1 3
.
Finally,
I
i
=
1
12
i
}
N
k
w
k
k
|
\
|
.
3
gradN
i
grad N
j
p
j
+ N
j
P
j
j
V
j
1 3
j =m+1
n
)
j =m+1
n
j =1
m
|
\
|
.
|
d (21)
For convenience,
P
j
can be lumped with p
j
(note that p
j
is not the true pressure at the nodes
sitting on
i
), leading to a more compact notation
I
i
=
1
12
i
}
N
k
w
k
k
|
\
|
.
3
gradN
i
grad N
j
p
j
j=1
n
|
\
|
.
| d
1
12
i
}
N
k
w
k
k
|
\
|
.
3
gradN
i
grad N
j
B
j
V
j
1/ 3
j =m+1
n
|
\
|
.
| d
(22)
Assuming l nodes along
i
, one can also define a finite element approximation for the flow q
b
of the fluid normal to
i
. For the case of a Newtonian fluid:
q
b
= M
k
k =1
k =l
k
V
k
4/ 3
(23)
where
V
k
is the speed at node k ,
k
is the evaluation of at node k , and M is a set of shape
functions. Then, the third term of (17) becomes:
N
i
i
}
q
b
V
4 /3
d = N
i
M
k
k=1
k=l
k
V
k
4/ 3
|
\
|
.
i
}
d (24)
where
k
is the curvilinear distance from the fracture tip to node k .
If the various nodal speeds are known, the i -th equation can now be reordered as follows:
N
i
N
j
w
j
(t
n+1
) w
j
(t
n
)
t
n+1
t
n j =1
j = n
|
\
|
.
|
i
}
d + p
j
1
12
i
}
j =1
j = n
N
k
w
k
k =1
k= n
|
\
|
.
3
grad N
i
grad N
j
d
1
12
i
}
N
k
w
k
k
|
\
|
.
3
grad N
i
grad N
j
B
j
V
j
1/ 3
j = m+1
n
|
\
|
.
|
d + N
i
M
k
k
V
k
4 / 3
k =1
l
|
\
|
.
i
}
d
= Q(t )N
i
(O)
(25)
Equations (18) and (25) were implemented in FRANC3D in order to model fluid flow in the
fractures while accounting for the near crack tip behavior.
5.4 Coupling with the Structural Response of the Rock
To adequately consider the coupling between elasticity and fluid flow, both the structural and the
fluid flow equations are solved at the same time for both the nodal widths and fluid pressures.
Linear elasticity provides the following linear relation between the fracture apertures w and the
fluid pressure p
w = w
p
+ w
o
w = G p + w
o
(26)
where w
o
is the contribution from the external stresses and G is the influence function relating
the fracture opening due to the internal fluid pressure ( w
p
) to the fluid pressure in the fracture.
This relation can be expressed as a set of equations using the flexibility matrix K computed by
BES (see previous section). For the nodes which are located neither on the crack front nor on ,
the i -th equation of this set can be written as:
K
i , j
j
p
j
w
i
+ w
o i
= 0 (27)
where w
o i
is the nodal width induced by the external stresses. For nodes located on the crack
front, the width is set to zero. The fluid pressure is set to zero also in order to conform with the
assumed existence of a fluid lag; therefore this relation disappears at the crack front. For the
pressure at the nodes located on , there is an extra term to be taken into account and the
equation becomes:
K
i , j
j
p
j
w
i
= K
i ,k
B
k
V
k
1/ 3
k
nodes
w
o i
(28)
When solving for both nodal width and pressure, a strong non-linearity arises in the set of fluid
flow equations due to the cubic power of the crack opening displacement w
3
. This is handled by
considering that, for every term in which w
3
arises, the value of w will be fixed and equal to
either the value of the width at the previous iteration or, for the first iteration of any time stage,
the value of the width at the previous time stage.
One can then reorder the i -th equation of the set of fluid flow equations as:
w
j
t
j =1
j =n
N
i
N
j
i
}
d + p
j
1
12
i
}
j =1
j =n
N
k
w
k
k=1
k=n
|
\
|
.
3
grad N
i
grad N
j
d =
w
j
(t
n
)
t
j =1
j =n
N
i
N
j
i
}
d N
i
M
k
k
V
k
4 / 3
k =1
l
|
\
|
.
i
}
d + Q(t)N
i
(O) +
1
12
i
}
N
k
w
k
k
|
\
|
.
3
grad N
i
grad N
j
B
j
V
j
1/3
j =m+1
n
|
\
|
.
|
d
(29)
This results in a set of n fluid flow equations and n structural equations with n unknown nodal
widths and n unknown nodal fluid pressures. The solution process is described in more details in
section 7.
6.0 Extensions to the Model and Formulation
Current trends in research and in field applications are pushing development of the simulator
towards being able to model indirect vertical fracturing (with fracture containment, pinching,
closure and fluid recession as corollaries), sand control (with screen-out as a corollary) and steam
injection. It could also be used as a tool to study the need for special completions (with the
ability to handle multiple fractures from cased and inclined wellbores as a corollary). Some of
the features that are needed to model these problems include: 1) partial propagation of a fracture
front across material interfaces, 2) simultaneous solution of crack surface contact tractions and
fluid flow pressures during fracturing of layered formations with differing stresses and
stiffnesses; 3) branching, intersecting, and merging cracks (for example, consider T-shaped
fractures that can form at material interfaces), 4) proppant transport, and 5) bi-material interface
slip and pressurization. As a step towards those developments, BES has been modified recently
to run on parallel computers such as the IBM SP2 and the SG Power Challenge (Shah et al.,
1997), allowing larger more complex problems to be modeled with increased accuracy and
efficiency.
Unlike the subjects mentioned above, each a research topic in itself, a fairly simple extension to
the simulator can be made in order to model hydraulic fracture propagation in the presence of a
porous and permeable rock mass where fluid is able to leak out of the fracture into the
surrounding rock.
6.1 Introduction of Leak-off
Carter's 1-D formulation (Carter, 1957), which uses a single lumped leak-off coefficient C
l
, is
the standard model in oil field applications for taking into account fluid leaking away from the
fracture into the formation. It is a simple model of fluid diffusion through the fracture walls.
The underlying assumptions are that the outward flow is normal to the direction of fracture
propagation and that variations in the difference between the fluid pressure in the fracture and the
pore pressure remain small.
The Carter leak-off model takes the form:
u
l
=
2C
l
(x)
(30)
where u
l
is the fluid velocity of the leaking fluid, and is the exposure time at point x. Note
that the factor 2 in front of C
l
corresponds to fluid leaking through the two faces of the fracture,
where C
l
is an empirical constant related to the porosity and permeability of the rock.
This introduces an extra term in the mass balance equation which now becomes:
w
t
+ divq +
2C
l
(x)
= Q(t) (O) (31)
This impacts the finite element formulation of the fluid flow equation described in section 5. An
extra term has to be added on the right hand sideof (29). For the i -th equation, this
supplementary term is:
N
i
i
}
2 N
j
C
lj
j
t
n
+ t N
j
j
j
|
\
|
.
| d (32)
where C
lj
is the value of the leak-off coefficient at node j and
j
the time at which the fluid
reached node j .
The presence of leak-off also modifies the behavior in the vicinity of the tip. The pressure and
width tip fields are now governed by a new series of equations (see Lenoach, 1995):
w = 6. 6687
C
l
E
|
\
|
.
1 4
V
1 8
5 8
+ 2. 2545
C
l
E
|
\
|
.
|
1 2
V
3 4
3 4
(33)
p() = p
o
0. 43 E
3/ 4
C
l
( )
1/ 4
V
1/ 8
3/ 8
0. 42 E
1/ 2
C
l
|
\
|
.
|
1/ 2
V
3 / 4
1/ 4
(34)
with similar notations as in section 4. p
o
is a constant dependent upon fracture geometry.
This changes the expression for the flow q
b
= V w = f (V) across the boundary between
regular elements and tip elements. Introducing the variational formulation for the fluid flow
equation, the right hand side term corresponding to the flow across the boundary still is:
p (q
} n
s
)ds = p
} q
b
ds (35)
Introducing the new expression for the width as given by (33) and the finite element
discretization, this becomes for the i th equation:
N
i
V
9 /8
+ V
7/ 4
( )
i
}
d =
N
i
N
j
j
j
|
\
|
.
|
N
k
V
k
k
|
\
|
.
9/ 8
+ N
j
j
j
|
\
|
.
|
N
k
V
k
k
|
\
|
.
7/ 4
(
(
i
}
d
(36)
i
= 6. 6687
C
l
E
|
\
|
.
1 4
i
5 8
i
= 2. 255
C
l
E
|
\
|
.
|
1 2
i
3 4
(37)
where E and C
l
are evaluated at point i also.
It is clear that tends to zero whereas
tends to infinity when the leak-off coefficient tends to
zero. In short, the permeable solution does not tend in a clear manner towards the impermeable
solution when the leak-off coefficient tends to zero. It is, however, clear physically and verified
numerically that the width field corresponding to a very small amount of leak-off is extremely
similar to that described by the impermeable solution (equation 5).
A criterion, therefore, is needed to switch from the permeable to the impermeable solution when
the amount of leak-off is small. The permeable solution actually relies on a parameter to be
small (Lenoach, 1995), where is given by
=
E
|
\
|
.
1/ 3
V
4C
l
|
\
|
.
|
2 /3
(38)
As the exponent in is different for the two solutions (impermeable and permeable), no absolute
criterion can easily be established. It was determined empirically that as soon as was greater
than 1/20, one could safely switch from the permeable solution to the impermeable solution. In
order to be able to mix the two solutions, two nodal switch functions G(x
i
) and H(x
i
) were
introduced: so that G = 0 and H =1 if (x
i
) < 1/ 20 (permeable case), and G = 1 and H = 0
otherwise (impermeable case).
A mixed sink term is then
N
i
H
j
N
j
j
j
|
\
|
.
| H
k
N
k
V
k
k
|
\
|
.
9/ 8
+ H
j
N
j
j
j
|
\
|
.
| H
k
N
k
V
k
k
|
\
|
.
7/ 4
|
\
i
}
+ G
j
N
j
j
j
|
\
|
.
|
G
k
N
k
V
k
k
|
\
|
.
4/ 3
|
.
| d
(39)
i
= 6. 6687
C
l
E
|
\
|
.
1 4
i
5 8
i
= 7. 206
E
|
\
|
.
1 3
i
2 3
i
= 2. 255
C
l
E
|
\
|
.
|
1 2
i
3 4
(40)
This approach was implemented and was found to function well. The only drawback it presents
is the necessity to gradually increase the leak-off coefficient from zero to its desired value during
the solution process for the first time stage in order to preserve reasonably quick convergence of
the algorithm.
7.0 Solution Procedure
Hydraulic fracture propagation is a non-linear, time-dependent, moving boundary problem that
involves simultaneous satisfaction of solid deformation, fluid flow and fracture mechanics.
Following the discretization in both time and space, the solution consists of a series of
"snapshots" that correspond to unique instances in time and crack shape. Two approaches can be
followed to obtain the next term in such a series: one can either fix the time step and look for the
corresponding geometry, or fix the geometry and look for the corresponding time. Although the
first approach is more intuitive, the latter scheme was chosen as it minimizes the amount of
computation. Note that the crack initiation process is not modeled; rather a model with an initial
starting crack is the beginning point for the simulation. The mechanics of fracture initiation and
breakdown pressure are beyond the scope of this paper.
7.1 Computing the Solution for a Particular Model
The special case of the initial solution for the starting crack will be detailed in section 7.2. For
subsequent time stages (actually fracture geometry stages), it is assumed that the solution at the
previous time stage t
n
is known. It consists of crack geometry
*
(t
n
) , crack volume Vol(t
n
) , a
set of nodal widths w
j
(t
n
) , a set of nodal pressures p
j
(t
n
) , and a set of nodal crack front speeds
V
j
(t
n
) . Assuming the geometry of the current model was computed from the previous solution,
the process which leads to the solution of the current model is complicated and has been
separated into three subsections for the purpose of clarity.
7.1.1. Meshing and Updating Information
As described in section 3, crack propagation involves a change in the model geometry. The mesh
that is attached to the crack surface is merely a mathematical representation of the true geometry.
For a given stage of fracture propagation, the previous,
*
(t
n
) , and current,
*
(t
n +1
), fracture
geometries are known. The current fracture surface contains the previous surface plus some new
surface representing the propagation. A mesh is attached to the geometry surface; in general, the
mesh on the portion of the current fracture surface that was the previous fracture surface can
remain fixed or can be modified (to reduce the number of elements for example). The set of
nodal widths and pressures from the previous solution must be mapped onto the mesh of the
current fracture. This is a trivial procedure if the mesh has remained fixed. If the mesh has been
modified, the nodal widths and pressures for the new mesh must be interpolated from the
previous mesh. Efficient algorithms to find the location of the node on the previous mesh and to
interpolate the response have been written for this purpose (see Potyondy, 1993), but will not be
discussed here.
7.1.2 Computation of the Flexibility Matrix
From the fracture geometry
*
(t
n +1
), the elastic properties of the model and the corresponding
boundary conditions, BES is used to compute the flexibility matrix as described in section 4.2.
Using the flexibility matrix, one can convert the nodal pressures inside the fracture into nodal
fracture widths. This reflects both the influence of the geometry of the fracture, the geometry of
the model, the elastic properties of the model, and the boundary conditions on the model. The
computation of the elastic flexibility matrix is by far the most time consuming part of the
analysis and guided the earlier choice of fixing the geometry instead of the time.
7.1.3 Iterative Solution of the Fluid Flow
The results from the previous time stage, t
n
, are used as starting values for the solution process.
Given an initial set of values for crack opening displacement, fluid pressure and crack front
speed, the iterative scheme proceeds at two levels. First, the set of nonlinear equations
describing the fluid flow inside the fracture coupled to the structural response of the model is
solved in an iterative manner. Once convergence is achieved based on the crack opening
displacements, the global mass balance and the crack front speeds are considered in the following
way to determine the absolute time corresponding to the current geometry. The total volume of
fluid injected should be equal to the volume of the fracture minus the volume of fluid that has
leaked away into the formation. Also, the fluid speed at each point of the crack front should
equal the crack propagation speed, i.e. the crack advance divided by the time step. Note that,
within the framework used here, these two relationships both express satisfaction of global mass
balance. If these two relationships are not satisfied, the time step is adjusted and the fluid flow
equation is solved iteratively again. This process continues until the solution has converged on
both the nodal values and the time, therefore satisfying both elasticity and fluid flow.
7.1.4 Fracture Propagation
Before a crack can be propagated, the total model displacements must be determined. This is
done by combining the displacements due to the far-field applied loading with the displacements
produced by the fluid pressurization in the fracture.
The process of propagating a crack in FRANC3D has been described elsewhere (Martha et al.,
1993; Carter et al., 1997). The only difference in this case is that the crack is propagated using
an ad-hoc propagation criterion for hydraulic fracturing (Figure 8). It is worth describing this
criterion in more detail as the authors believe that more research is needed in this area.
The local extension of the fracture propagation along the crack front is proportional to the speed
of the fluid at that point. It is scaled by a maximum propagation length L
0
(input by the user) so
that the point where the maximum speed is attained is propagated by L
0
. Indeed, a reasonable
value for L
0
will depend on the model and crack geometry as well as the desired accuracy but,
because of the linear interpolation in time, should not be greater than 10% of the fracture
penetration into the formation.
The direction of fracture propagation at each point is determined according to the maximum
circumferential stress criterion. The criterion is evaluated at discrete points along the front in a
plane normal to the crack front tangent (Figure 8). The angle is computed using the mode I
and II stress intensity factors K
1
and K
2
. Quasi-static propagation of the fracture requires that
the mode I stress intensity factor be equal to the fracture toughness of the rock K
1c
at each point.
The assumption underlying the special solution used here at the fracture tip is that, although not
explicitly taken into account, a fluid lag would exist and automatically adjusts its length to satisfy
this condition (SCR, 1993). Therefore, (x) is taken to satisfy:
K
1c
sin( (x)) + K
2
(x)(3cos( (x)) 1) = 0 (41)
Once the crack has been propagated, the model can be remeshed for the next stage of analysis.
7.2 Special Case for the First Time Stage
As described in the previous section, the solution process requires the results of a previous step.
This is achieved for the first model through approximate analytical solutions. These solutions
are used to produce the complete description of a "dummy" crack, smaller than the initial crack,
which is then considered as the previous model by the iterative solver. This provides the initial
values to be used for the real fracture geometry, and the solution scheme then proceeds as
described in section 7.1.
Two solutions have been developed covering most initial crack geometries in hydraulic
fracturing, one for a slot crack and one for a penny-shaped crack (or radial crack). Following the
approach of the SCR Geomechanics Group (SCR, 1994), the main assumption behind these
solutions is that the pressure profile near the tip is governed by a singular field and that the
resulting stress intensity factor is zero. This allows one to obtain a pressure profile from which
all other quantities are derived. These solutions, therefore, produce time, crack opening
displacements, crack front speeds and fluid pressure inside the crack as a function of fracture
extension, elastic parameters, and pumping parameters.
8.0 Verification and Validation
One of the most important aspects in the development of any numerical simulator is the
verification of the approach as well as the computer software. To verify this 3D simulator, one
must analyze problems for which the solution is known. In this case, the fully 3D simulator is
used to model simple 2D fractures: a KGD and a radial hydraulic fracture. The results from the
3D simulator are compared with those of a previously verified 2D simulator. Furthermore, the
3D simulator is used to model some experiments that were performed at Delft University which
provides additional validation and shows the capability of the simulator to model fully 3D
hydraulic fractures in the near-wellbore region.
8.1 Modeling Hydraulic Fractures of Simple Geometries
FRANC3D has been compared with a hydraulic fracturing model (Desroches and Thiercelin,
1993) which can only model simple geometries such as KGD and radial, but considers full
coupling of fluid flow, elastic deformation and fracture propagation. It includes a complete
description of the fracture tip, in particular the effect of fracture toughness and the possibility for
the fracturing fluid not to reach the tip of the fracture (fluid lag). This model, referred to as
Loramec, has been validated using laboratory experiments (de Pater et al., 1996). Although the
formulation of these two models is radically different, the underlying physical assumptions are
similar so they should yield similar results for a similar fracture geometry.
In order to simulate a plane strain geometry with FRANC3D, a small slot crack was introduced
in a cube (Figure 9). The fracture was propagated with an evenly distributed line source. Data for
comparison with Loramec were taken in the middle of the crack. The relevant parameters were:
Young's modulus of 24.1 GPa, Poisson's ratio of 0.02, fracture toughness of 0.25 MPam, fluid
viscosity of 130000 cp, flow rate/height of 5.92x10
-9
m
3
/s and closure stress of 9.7 MPa.
Figure 10a shows the fracture extension versus time for Loramec and FRANC3D together with a
profile of pressure (Figure 10b) and width (Figure 10c) from the fracture mouth to the fracture
tip. Excellent agreement between FRANC3D and Loramec is observed. Note that the
FRANC3D model used only six linear elements along the fracture extension in addition to the
last tip element. This demonstrates how well the tip element captures the essence of the solution
allowing one to use a coarse mesh away from the tip.
In order to simulate an axisymmetric (radial) geometry, a penny-shaped crack was introduced in
the center of a cube (Figure 11). The size of the crack was two orders of magnitude smaller than
that of the cube to simulate an infinite medium. The fracture was propagated using a point
source at the center of the crack. Data for comparison with Loramec were taken along a crack
radius. The relevant parameters were similar to those for the KGD geometry except for the
injection rate of 2.96x10
-9
m
3
/s. These parameters correspond to those of one of the laboratory
experiments that was used to validate the Loramec model. Figure 12a shows the fracture radius
versus time for Loramec and FRANC3D, together with a profile of pressure (Figure 12b) and
width (Figure 12c) from the fracture center to the fracture tip. There is excellent agreement
between the two solutions for this model as well.
8.2 Modeling Experimental Data
To study the 3D geometry of hydraulic fractures, a matrix of hydraulic fracturing experiments
using a true triaxial loading frame was carried out on physical models at the University of Delft
(Weijers, 1995). The simplest configuration, an open-hole model was chosen for comparison
with FRANC3D here so that the extra complexity due to the perforations and casing could be
avoided. Simulations of the other test configurations are on-going (Shah et al., 1997), but cannot
be described here due time and space limitations.
The geometry of the chosen configuration, COH10, is presented in Figure 13. A cylindrical
wellbore of radius 1 cm was drilled in a 30 cm edge cube. The wellbore was drilled parallel to
the direction of the minimum stress, perpendicular to the preferred fracturing plane. The
experimental parameters were: Young's modulus E=35 GPa, Poisson's ratio =0.15, fracture
toughness K
1c
=0.6 MPa m, fluid viscosity =100000 cp, injection rate Q=0.088 cc/min,
1,2,3
=23, 12.1 and 9.7 MPa respectively. Splitting of the blocks after the experiment showed
that a single fracture initiated parallel to the wellbore and slowly turned to propagate
perpendicularly to the wellbore. The pressure in the wellbore was recorded during the
experiment. Two displacement transducers DIS2 and DIS3 were set 90 degrees apart in the
middle of the wellbore to measure the deformation of the wellbore. DIS2 measures the
deformation in the direction of the maximum principal stress. The output of FRANC3D was
compared against both this data and the radius of reorientation of the fractures which was
measured after dismantling the blocks.
Although categorized as open-hole, there was still a borehole assembly in this experiment,
which consisted of two 10 cm long packers glued at the two extremities of the wellbore, leaving
an open-hole section of 10cm at the center of the wellbore. To simulate the wellbore assembly
that was used during the experiment, the cylinder representing the wellbore was split into three
regions; the middle section corresponds to the open-hole section whereas the two other
correspond to the packer arrangement. As the packer assembly was glued to the wellbore, it was
considered as a rigid inclusion and was modeled by a zero normal displacement boundary
condition on the packer zone. In the open-hole section, two initial slot fractures were added, 180
degrees apart. Their location was taken as indicated by the experiments; the fracture initiation
process was not modeled.
Figure 14 shows the comparison between the data obtained from the experiment and the output
of FRANC3D. The comparison is made in terms of wellbore pressure (Figure 14a) and
displacements of the wellbore wall as measured by the two LVDT sensors DIS2 (Figure 14b)
and DIS3 (Figure 14c). Apart from numerical noise, the pressure predicted by FRANC3D is
comparable to that measured in the laboratory and the two horizontal displacements are well
predicted. As DIS3 is an indirect measurement of the fracture aperture, this indicates that the
fracture aperture at the wellbore is well predicted by the model. The only discrepancy between
the experiment and the output of the model is the radius of reorientation of the fracture; the
experiment showed that the fracture was slow to turn (about 10 cm for the radius of reorientation
Weijers, 1995), whereas the simulated fracture turned much more quickly towards the preferred
fracturing plane (a radius of reorientation of about 2cm was computed). It is proposed that this is
due to some influence of the packer assembly that is not taken into account by our modeling (de
Pater, personal communication, 1996).
9.0 Conclusions
It is necessary to have a fully 3D simulator to correctly model some of the problems regularly
encountered in the field during hydraulic fracturing, such as the near wellbore fractures of a
deviated well. If a robust and accurate simulator is available, improvements in oil/gas recovery
are possible as well as savings in time and money during completion and re-injection.
A simulator that is capable of modeling multiple, fully 3D, non-planar hydraulic fractures has
been built. It maintains a consistent geometric representation of the fracture geometry
throughout the analysis and properly couples the fluid flow in the fracture with the fracture
deformation. It is verified using previous solutions for simple 2D crack geometries as well as 3D
experimental results. Enhancements and additions are in progress to allow more complex
problems to be modeled and to increase the speed, efficiency, and accuracy of the simulator.
An additional advantage of the software is that it is easily adapted to other problems of fracture
in geomechanics, such as pressure grouting of cracks in dams (Ingraffea et al., 1995), cohesive
crack propagation (Carter et al., 1993), and compression induced fracture from inclined flaws in
brittle materials (Germanovich et al., 1996).
10.0 Acknowledgments
The authors wish to acknowledge Schlumberger for permission to publish this work as well as
for their intellectual and financial support during the development of the simulator. The
additional support of the Cornell Fracture Group and the Cornell National Supercomputing
Facility is also hereby acknowledged.
11.0 References
Abass, H.H., Hedayati, S. and Meadows, D.L. (1996) Nonplanar fracture propagation from a
horizontal wellbore: experimental study, SPEPE, August, p. 133-137.
Advani, S.H., Lee, T.S. and Lee, J.K., (1990) Three dimensional modeling of hydraulic fractures
in layered media: Finite element formulations, J. Energy Res. Tech., 112, p. 1-18.
Bahat, D. (1991) Tecton-fractography. Springer-Verlag Publishers.
Barree, R.D. (1983) A practical numerical simulator for three-dimensional hydraulic fracture
propagation in heterogeneous media, SPE Paper 12273, SPE Symp. Reservoir Simulation,
San Francisco, Nov. 15-18.
Baumgart, B.G. (1975) A polyhedron representation for computer vision, AFIPS Proc., 44, p.
589-596.
Behrmann, L.A. and Elbel, J.L. (1991) Effect of perforations on fracture initiation, JPT, May, p.
608-615.
Carter, B.J., Ingraffea, A.R. and Bittencourt, T.N. (1993) Topology-Controlled Modeling of
Linear and Non-Linear 3D-Crack Propagation in Geomaterials, in "Fracture of Brittle
Disordered Materials: Concrete, Rock and Ceramics," Proc. IUTAM Conf., Brisbane,
Australia, E&FN Spon Publishers, p. 301-318.
Carter, B.J., Wawrzynek, P.A., Ingraffea, A.R. and Morales, H. (1994) Effect of casing on
hydraulic fracture from horizontal wellbores, 1st North American Rock Mechanics
Symposium, Austin, TX, June, p. 185-192.
Carter, B.J., Chen, C-S., Ingraffea, A.R., and Wawrzynek, P.A. (1997) Recent advances in 3D
computational fracture mechanics, invited Lecture for the Ninth International Conference on
Fracture, Sydney, April, 1997.
Carter, R.D. (1957) Derivation of the general equation for estimating the extent of the fractured
area, Appendix to "Optimum Fluid Characteristics for Fracture Extension", by Howard, G.C.
and Fast, C.R., Drilling and Production Practices, API, p. 261-270.
Clark, J. B., (1949) A hydraulic process for increasing the productivity of wells, Trans. AIME
186, p. 1.
Clifton, R. J., and Abou-Sayed, A.S., (1979) On the computation of the three-dimensional
geometry of hydraulic fractures, SPE Paper 7943.
Desroches, J. and Thiercelin, M., (1993) Modelling the propagation and closure of micro-
hydraulic fractures, Int. J. of Rock Mech. & Geomech. Abstr., 30, p. 1231-1234.
Emermann, S.H., Turcotte, D.L. and Spence, D.A., (1986) Transport of magma and
hydrothermal solutions by laminar and turbulent fluid fracture, Physics of the Earth and
Planetary Interiors, 41, p. 249-259.
Gardner, D.C., (1992) High fracturing pressures for shales and which tip effects may be
responsible, Proc. 67th SPE Annual Tech. Conf. and Exhib., Washington, DC, p. 879-893.
Geertsma, J., (1989) Two-dimensional fracture propagation models, in "Recent Advances in
Hydraulic Fracturing", Monograph Series, 12, SPE, Richardson, TX, p. 81-94.
Geertsma, J. and de Klerk, F. (1969) A rapid method of predicting width and extent of
hydraulically induced fractures, JPT, p.1571-1581.
Germanovich, L.N., Carter, B.J., Dyskin, A.V., Ingraffea, A.R. and Lee, K.K., (1996) Mechanics
of 3D crack growth in compression, in "Tools and Techniques in Rock Mechanics", 2nd North
American Rock Mechanics Symposium, NARMS'96, Montreal, Canada, June, p. 1151-1160.
Gu, H. and Leung, K.H. (1993) 3D numerical simulation of hydraulic fracture closure with
application to minifracture analysis, JPT, March, p. 206-211.
Hallam, S.D. and Last, N.C. (1991) Geometry of hydraulic fractures from modestly deviated
wellbores, JPT, June, p. 742-748.
Hoffmann, C.M. (1989) Geometric and Solid Modeling: An Introduction. Morgan Kaufmann
Publishers, Inc., San Mateo, California.
Howard, G. C., and Fast, C. R., (1970) Hydraulic Fracturing, Monograph Series, 2, SPE,
Richardson, TX.
Ingraffea, A.R., Carter, B.J. and Wawrzynek, P.A., (1995) Application of computational fracture
mechanics to repair of large concrete structures, in Fracture Mechanics of Concrete
Structures, 3, Edited by Folker Wittmann, Proc. 2nd Int. Conf. Fracture Mech. of Concrete
Structures (FRAMCOS2), Zurich, Switzerland, AEDIFICATIO Publishers.
Jeffrey, R.G., (1989) The combined effects of fluid lag and fracture toughness on hydraulic
fracture propagation, Proc. Joint Rocky Mountains Regional Meeting and Low Permeability
Reservoir Symposium, Denver, CO, p. 269-276.
Johnson, E. and Cleary, M.P. (1991) Implications of recent laboratory experimental results for
hydraulic fractures, Proc. Rocky Mountain Regional Meeting and Low-Permeability
Reservoirs Symp., Denver, CO, p. 413-428.
Khristianovic, S.A. and Zheltov, Y.P. (1955) Formation of vertical fractures by means of highly
viscous fluid. Proc. 4th World Petroleum Congress, Rome, Paper 3, p. 579-586.
Lam, K.Y., Cleary, M.P. and Barr, D.T. (1986) A complete three dimensional simulator for
analysis and design of hydraulic fracturing, SPE Paper 15266.
Lenoach, B. (1995) Hydraulic fracture modelling based on analytical near-tip solutions, in
"Computer Methods and Advances in Geomechanics", Edited by: Siriwardane& Zaman,
Balkema Publishers, p. 1597-1602.
Lutz, E.E (1991) Numerical Methods for Hypersingular and Near-Singular Boundary Integrals
in Fracture Mechanics, Ph.D. Thesis, Cornell University.
Mntyl, M. (1988) An Introduction to Solid Modeling. Computer Science Press, Rockville,
Maryland.
Martha, L.F. (1989) A Topological and Geometrical Modeling Approach to Numerical
Discretization and Arbitrary Fracture Simulation in Three-Dimensions, Ph.D. Thesis, Cornell
University.
Martha, L.F., Wawrzynek, P.A. and Ingraffea, A.R. (1993) Arbitrary crack representation using
solid modeling, Engineering with Computers, 9, p. 63-82.
Medlin, W.L. and Fitch, J.L, (1983) Abnormal treating pressures in MHF treatments, Proc. SPE
Annual Technical Conference and Exhibition, San Fransisco, CA.
Mendelsohn, D. A., (1984a) A review of hydraulic fracture modeling - I: General concepts, 2D
models, motivation for 3D modeling, J. Energy Res. Tech., 106, p. 369-376.
Mendelsohn, D. A., (1984b) A review of hydraulic fracture modeling - II: 3D modeling and
vertical growth in layered rock, J. Energy Res. Tech., 106, p. 543-553.
Morales, R.H. (1989) Microcomputer analysis of hydraulic fracture behavior with a pseudo-
three dimensional simulator, SPEPE, Feb., p. 69-74.
Morita, N., Whitfill, D.L. and Wahl, H.A. (1988) Stress-intensity factor and fracture cross-
sectional shape predictions from a three-dimensional model for hydraulically induced
fractures, JPT, Oct., p. 1329-1342.
Mortenson, M.E. (1985) Geometric Modeling. John Wiley & Sons, New York.
Nordgren, R.P. (1972) Propagation of a vertical hydraulic fracture. SPEJ, p. 306-314, Aug.
Ong, S.H. and Roegiers, J-C. (1996) Fracture initiation from inclined wellbores in anisotropic
formations, JPT, July, p. 612-619.
Palmer, I.D. and Veatch R.W., (1990) Abnormally high fracturing pressures in step rate tests,
SPE Production Engineering, p. 315-323.
Papanastasiou, P. and Thiercelin, M. (1993) Influence of inelastic rock behaviour in hydraulic
fracturing, Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 30, p. 1241-1247.
de Pater, C.J., Weijers, L., Savic, M., Wolf, K.A.A, van den Hoek, P.J. and Barr, D.T. (1994)
Experimental study of non-linear effects in hydraulic fracture propagation, SPE Production
and Facilities, 9, p. 239-246.
de Pater, C.J., Desroches, J., Groenenboom, J. and Weijers, L. (1996) Physical and numerical
modeling of hydraulic fracture closure, SPE Production and Facilities, May, p. 122-127.
de Pater (1996) Personal Communication.
Perkins, T.K. and Kern, L.R. (1961) Widths of hydraulic fractures. J. Petrol. Technol., p. 937-
949, Sept.
Potyondy, D.O. (1993) A software framework for simulating curvilinear crack growth in
pressurized thin shells. PhD Thesis, Cornell University, Ithaca, NY.
Potyondy, D.O., Wawrzynek, P.A. and Ingraffea, A.R. (1995) An algorithm to generate
quadrilateral or triangular element surface meshes in arbitrary domains with applications to
crack propagation, Int. J. Numer. Meth. in Engng., 38, p. 2677-2701.
SCR Geomechanics Group, (1993) On the modelling of near tip processes in hydraulic fractures,
Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 30, p. 1127-1134.
SCR Geomechanics Group, (1994) The crack tip region in hydraulic fracturing, Proc. Royal
Society, Mathematical and Physical Sciences, Series A, 447, p. 39-48.
Settari, A. and Cleary, M.P. (1986) Development and testing of a pseudo-three-dimensional
model of hydraulic fracture geometry, SPE, Trans. AIME, 281, p.449-466.
Shah, K., Carter, B.J. and Ingraffea, A.R. (1997) Simulation of hydrofracturing in a parallel
computing environment, 36th U.S. Rock Mech. Symp., Columbia Univ., NY, Special Volume
of the Int. J. Rock Mech. Min. Sci. & Geomech. Abstr.
Shlyapobersky, J., Wong, G.K. and Walhaug, W.W., (1988) Overpressure calibrated design of
hydraulic fracture stimulations, Proc. SPE Annual Tech. Conf. and Exhib., Houston, TX, p.
133-148.
Shlyapobersky, J., Walhaug, W.W., Sheffield, R.E. and Huckabee, P.T. (1988) Field
determination of fracturing parameters for overpressure calibrated design of hydraulic
fracturing, Proc. 63rd Annual Tech. Conf. and Exhib., Houston, TX, SPE Paper 18195.
Spence, D.A. and Sharp, P., (1985) Self-similar solutions for elastodynaic cavity flow,
Proceedings of the Royal Society of London A, 400, p. 289-313.
Soliman, M.Y., Hunt, J.L. and Azari, M. (1996) Fracturing horizontal wells in gas reservoirs,
SPE Paper 35260, Mid-Continent Gas Symposium, Amarillo, TX.
Sousa, J.L.S., Carter, B.J., Ingraffea, A. R., (1993) Numerical simulation of 3D hydraulic
fracture using Newtonian and power-law fluids, Int. J. Rock Mech. Min. Sci. & Geomech.
Abstr., 30, p. 1265-1271.
Touboul, E., Ben-Naceur, K. and Thiercelin, M. (1986) Variational methods in the simulation of
three-dimensional fracture propagation, 27th US Rock Mech. Symp., p. 659-668.
Vandamme, L., Jeffrey, R.G., Curran, J.H., (1988) Pressure distribution in three-dimensional
hydraulic fractures, SPE Production Engineering, 3, p. 181-186.
van den Hoek, P.J., van den Berg, J.T.M. and Shlyapobersky, J. (1993) Theoretical and
experimental investigation of rock dilatancy near the tip of a propagating hydraulic fracture,
Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 30, p. 1261-1264.
Warpinski, N.R., (1985) Measurement of width and pressure in a propagating hydraulic
fracture, SPEJ, Feb., p. 46-54.
Warpinski, N.R., Moschovidis, Z.A., Parker, C.D. and Abou-Sayed, I.S. (1994) Comparison
study of hydraulic fracturing modelsTest case: GRI staged filed experiment No. 3, SPE
Production & Facilities, Feb., p. 7-16.
Wawrzynek, P.A. (1991) Discrete modeling of crack propagation: theoretical aspects and
implementation issues in two and three dimensions, Ph.D. Thesis, Cornell University.
Wawrzynek, P.A. and Ingraffea, A.R. (1987a) Interactive finite element analysis of fracture
processes: an integrated approach, Theor & Appl Fract Mech, 8, p. 137-150.
Wawrzynek, P.A. and Ingraffea, A.R. (1987b) An edge-based data structure for two-dimensional
finite element analysis, Engin. with Comp., 3, p. 13-20.
Weijers, L. and de Pater, C.J. (1992) Fracture reorientation in model tests, SPE Paper 23790,
Formation Damage Control Conference, Lafayette.
Weijers, L. (1995) The near-wellbore geometry of hydraulic fractures initiated from horizontal
and deviated wells, Ph.D. Dissertation, Delft University of Technology.
Weiler, K. (1985) Edge-based data structures for solid modeling in curved-surface
environments, IEEE Comp. Graph & Appl., 5, p. 21-40.
Weiler, K. (1986) Topological Structures for Geometric Modeling, Ph.D. Dissertation,
Rensselaer Polytechnic Institute, Troy, NY, Univ. Microfilms Intl., Ann Arbor, MI.
Yew, C.H. and Li, Y. (1988) Fracturing of a deviated well, SPEPE, Nov, p. 429-437.
U
C
R
i
A
i
E
i
F
i
,
i
i+1
i=i+1
S
M D
Figure 1. Incremental crack growth simulations; i denotes the increment of crack growth.
Stress/Flow/Thermal Analysis
Model Representation
Discretize
FRANC3D
Analysis Input
Update
Model
Analysis Output
Stress/Flow/Thermal Analysis
Model Representation Model Representation
Discretize Discretize
FRANC3D
Analysis Input Analysis Input
Update
Model
Update
Model
Analysis Output Analysis Output
Figure 2. FRANC3D encompasses most of the conceptual crack growth simulation model.
x
y
z
topology geometry tied to
vertex
edge
face
cubic b-spline
curve
Cartesian
coordinates
x
y
z
topology geometry tied to
vertex
edge
face
x
y
z
topology geometry tied to
vertex
edge
face
cubic b-spline
curve
Cartesian
coordinates
Figure 3. Relationship between topology and geometry.
face
edge
pair of mate
faceuses
pair of mate
edgeuses
pair of radial
edgeuses
face
edge
pair of mate
faceuses
pair of mate
edgeuses
pair of radial
edgeuses
face
edge
pair of mate
faceuses
pair of mate
edgeuses
pair of radial
edgeuses
face
edge
pair of mate
faceuses
pair of mate
edgeuses
pair of radial
edgeuses
Figure 4. The radial-edge database relies on the radial ordering of edge uses about each edge. A
face has two face uses and the edge has a use with respect to each face use (after Weiler, 1986).
main
side
face
mate
side
face
front
edge
open boundary edge
tip vertex
front
front
branch
edges
branch
vertices
tip vertex
main
mate
main
side
face
mate
side
face
front
edge
open boundary edge
tip vertex
front
front
branch
edges
branch
vertices
tip vertex
main
mate
main
side
face
mate
side
face
front
edge
open boundary edge
tip vertex
front
front
branch
edges
main
side
face
mate
side
face
front
edge
open boundary edge
tip vertex
front
front
branch
edges
branch
vertices
tip vertex
main
mate
Figure 5. The topology of a typical branching crack.
existing fracture
surface and mesh
new
fracture
surface
Figure 6. Fracture propagation is modeled by adding new crack surface geometry. The mesh
which is attached to the geometry can remain the same for the existing surface.
F
A
A
O
n
s
Figure 7. Fracture geometry for fluid flow analysis (showing F, , , A and
A
).
t
n
b
u
v
r
t
n
b
u
v
r
Figure 8. Propagation of a hydraulic fracture.
Figure 9. Modeling a plane strain fracture with FRANC3D.
1000 1500 2000 2500
0.08
0.09
0.1
0.11
0.12
0.13
0.14
pumping time (s)
f
r
a
c
t
u
r
e
h
a
l
f
l
e
n
g
t
h
(
m
)
FRANC3D/Loramec: one big slot crack, same time
__: Loramec
o: FRANC3D
0 0.02 0.04 0.06 0.08 0.1 0.12
0
2
4
6
8
10
12
14
position of the nodes (m)
t
o
t
a
l
f
l
u
i
d
p
r
e
s
s
u
r
e
(
M
P
a
)
FRANC3D/Loramec: one big slot crack, same length
*: Loramec
o: FRANC3D
0 0.02 0.04 0.06 0.08 0.1 0.12
0
1
2
3
4
5
6
7
x 10
5
pumping time (s)
f
r
a
c
t
u
r
e
a
p
e
r
t
u
r
e
(
m
)
FRANC3D/Loramec: one big slot crack, same length
*: Loramec
o: FRANC3D
Figure 10. Comparison between FRANC3D and Loramec: plane strain case.
Figure 11. Modeling an axisymmetric fracture with FRANC3D.
o: FRANC3D
_: Loramec
0 50 100 150 200 250
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
Pumping time (s)
F
r
a
c
t
u
r
e
r
a
d
i
u
s
(
m
)
FRANC3D/Loramec: radial, same time
*: Loramec
o: FRANC3D
0 0.01 0.02 0.03 0.04 0.05
0
5
10
15
20
25
30
35
p
r
e
s
s
u
r
e
(
M
P
a
)
position of the nodes (m)
FRANC3D/Loramec: radial, same length
*: Loramec
o: FRANC3D
0 0.01 0.02 0.03 0.04 0.05
0
1
2
3
4
5
6
7
8
x 10
5
a
p
e
r
t
u
r
e
(
m
)
position of the nodes (m)
FRANC3D/Loramec: radial, same length
Figure 12. Comparison between FRANC3D and Loramec: axisymmetric case.
Figure 13: Geometry of experiment COH10 (from Weijers, 1995).
2000 2200 2400 2600 2800 3000 3200 3400 3600 3800 4000
0
5
10
15
20
25
30
35
40
Experimental time (s0
F
l
u
i
d
p
r
e
s
s
u
r
e
(
M
P
a
)
PRESCOH10
__: experiment
o: FRANC3D
2000 2200 2400 2600 2800 3000 3200 3400 3600 3800 4000
0
5
10
15
20
25
30
35
40
45
50
DIS2COH10
Experimental time (s0
d
i
s
p
l
a
c
e
m
e
n
t
(
m
i
c
r
o
n
s
)
__: experiment
o: FRANC3D
2000 2200 2400 2600 2800 3000 3200 3400 3600 3800 4000
0
5
10
15
20
25
30
35
40
45
50
DIS3COH10
Experimental time (s0
d
i
s
p
l
a
c
e
m
e
n
t
(
m
i
c
r
o
n
s
)
__: experiment
o: FRANC3D
Figure 14: Comparison between FRANC3D and experiment COH10.