Académique Documents
Professionnel Documents
Culture Documents
A three dimensional
hydraulic fracture
analysis code
Version 1
User's Guide and Menu Reference
April 15, 1999
1 INTRODUCTION
This document is a user's guide for the hydraulic fracture simulator HYFRANC3D, a
FRacture ANalysis Code for 3 Dimensional simulation of hydraulic fracturing. It
describes how to use the program and some of the concepts behind the program for
performing hydraulic fracture analysis. This assumes that the user is already familiar with
FRANC3D, the basis of HYFRANC3D. Information for FRANC3D can be obtained
from the other FRANC3D reference documentation (www.cfg.cornell.edu/software/).
The theory behind the coupled fluid flow - fracture propagation is available in several
publications.
2 MENU REFERENCE
This section describes the additional menus that are added to FRANC3D for hydraulic
fracture. The first is Write HyFSys File, which is available from the Read/Write
Analysis Files (Figure 1).
The maximum number of iterations controls how long the solution is allowed to
iterate. This number is based on the global mass balance iterations. Within each
of these iterations, we iterate on cod as well. This maximum number can be
increased on the terminal window if the solution appears to be converging and the
maximum is exceeded.
The remaining two parameters control the solution in the case of leakoff. We
solve for leakoff using the one-dimensional empirical leakoff coefficient. To
improve the solution procedure, we ramp up the leakoff starting from zero leakoff
and ending at the proper solution. The user controls the number of steps and the
rate at which the leakoff is ramped-up.
Solve Current Model
Solve the set of coupled fluid flow and structural equations. The iterative process
is displayed on the terminal window as the solution proceeds. The user can
monitor the convergence (or lack of convergence). If the solution process
converges within the specified number of iterations, the iterations terminate and
control is passed back to the user. At this stage, the user can examine the results,
save the solved model, and/or propagate the crack. If the solution does not
converge, the user can increase the allowable number of iterations and/or modify
the parameters in the dialog box that control the iterative solution procedure.
Display Response Info
A submenu is presented (Figure 11).
crack surfaces. A new window is then presented. This window is similar to the
Deformation/Contour window in FRANC3D. However, only the crack surface(s)
are displayed by default. The deformed surfaces and color contours can only be
shown for the crack surfaces.
Line plots
Presents a new window for displaying data in x-y form. The user is prompted to
pick lines on the crack surface. Multiple lines can be given by picking the first
point and last point of the lines until the user selects FINISH. A new window is
presented showing the line(s) on the model surface and the user can then select the
response value to be plotted. The normalized length of the lines is plotted on the
x-axis while the data is plotted on the y-axis. This window is similar to the Line
Plot window in FRANC3D.
Point response
Displays response data for a surface point. The user is prompted to select a point
on the crack surface. The data is printed on the terminal window.
Node response
Displays response data for a node on the crack. The user is prompted to select a
node on the crack surface. The data is printed on the terminal window.
Write and Read .besout File
A submenu is presented (Figure 12).
the fluid flow - structure equations in the subsequent model. The advance, fluid
or crack front speed, and time step must be consistent.
read solved model
Reads a previously saved solved model to use as the initial solution for the current
model. A file selector box is presented; select the appropriate .hyd file. A solved
model must read into HYFRANC3D before solving for the second and subsequent
models where the crack has been propagated.
Apply Constant Pressure
A constant pressure is applied to the crack surfaces. This feature allows the user
to quickly check to see if the model and starting fluid pressure are reasonable. An
initial constant pressure is applied to all nodes of the crack. The resulting crack
opening displacements are then computed. The results can be viewed using the
features described above.
Delete All Models
Delete the models and free the associated memory. This feature is only needed if
the user is examining a number of different models with HYFRANC3D in a
single session.
6. Select Build Model From File and then select the .hyf file from the list. The .flex file
is read automatically and the binary .infl file created. The .infl file combines all the .flex
matrices into a single matrix.
7. For this first model (the initial crack), a dialog box is presented that allows you to
adjust the solution parameters. If you don't know what the parameters should be, just
select Accept. For subsequent propagation steps model, see step 13.
8. Select Solve Current Model. The iterative solution process is started; monitor the
progress on the terminal window. The key is to try to get both the mass balance and the
speed error terms to converge. The mass balance error should converge toward 0.0. The
speed error should converge to zero, but might go from positive to negative. If this
happens, a message is printed to the terminal window and you are asked if you want to
iterate further. The solution actually gets worse if you continue to iterate even if the mass
balance error continues to decrease. Type in "n" to stop the iterations. If the solution
does not converge, see the Solution Procedure section below.
9. You can Display the Response Info to make sure the solution is reasonable.
10. Write and Read the .beset file next. This combines the fluid pressure with the
influence matrices along with the far field loading induced displacements and tractions to
produce the global equilibrium displacements and tractions. These values are then
written to a .beset file. This file can then be read in again, which is the same as Read
BES File from the Read/Write Analysis Files menu. You need to read in the .beset file
to advance the crack.
11. You can now Advance the Crack Front based on the stress intensity factors and the
crack front or fluid speed. The stress intensity factors are determined using displacement
correlation. The mode I stress intensity factor is then scaled down to equal Kic. (Note
that this scaling is debatable.) The direction of propagation is based on the mode I and II
stress intensity factors. The crack extension is based on the speed. The user supplies a
maximum extension. The extension at points along the crack front is scaled based on the
relative speeds. The new crack front points are then fitted with a polynomial and the
crack is then propagated. You must perform these steps manually in the given order.
12. The solved model must be saved after propagating even if you already saved it after
the solution converged; this is because the subsequent iterative process for the next crack
step requires the amount of extension along the crack front. Select the Write/Read .hyd
File menu button, select write curent model and save the .hyd file.
13. If you are dealing with the second (or later models), you must read the solved model
of the previous step before building the current model. This provides the solution of the
previous time stage necessary to solve for the current time stage (mass balance). You
should select read solved model and read in the .hyd file before going back to Step 6.
4 Solution Procedure
This section describes in more detail the actual solution procedure and methods for
obtaining convergence for cases where the convergence is difficult to achieve.
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.
The essential boundary condition consists of fluid injection sources, either point or line
sources. The natural boundary condition is the speed at the crack front, actually the fluid
speed at the boundary between the near-crack front and the bulk of the fracture.
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.
4.1 Computing the Solution for a Particular Model
The special case of the initial solution for the starting crack will be detailed in section 4.2.
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 (tn ) ,
crack volume Vol(tn ) , a set of nodal cod widths w j (t n ) , a set of nodal pressures p j (t n ) ,
and a set of nodal crack front speeds V (t ) . Assuming the geometry of the current
j
model was computed from the previous solution, the process that leads to the solution of
the current model has been separated into subsections.
The mesh on the crack surface is constrained by the fact that a boundary between the
near-front region of the crack and the bulk of the fracture surface must exist. The nearfront region is governed by the analytical linear elastic hydraulic fracture (LEHF)
equations (SCR Geomechanics Group, 1994). It is necessary to keep track of the element
edges that form this boundary along with the orthogonal distance between the crack front
and the boundary. The near-front analytical solution implicitly provides for a fluid lag
and a singular pressure distribution without having to perform prohibitive discretization
of the crack front region.
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. These procedures are implemented in
HYFRANC3D and are invoked automatically when reading in the propagated crack
models.
Given 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. (The
flexibility matrix is a set of resulting displacements and tractions for the body assuming a
unit traction load on a node of the crack surface.) Using the flexibility matrix, one can
convert the nodal pressures inside the fracture into nodal fracture widths. This reflects
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.
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.
main assumption behind these solutions is that the pressure profile near the crack front 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.
P(r ) = c + 3.72 Ph
1 3
L1
R
13
Ph
L1
13
E and L1 =
Pw = c + 1.89E
23
13
R
To keep the initial solution simple, consider a width profile w(r) resulting from a
constant pressure profile of value Pw
w( r) = wo 1
r
R
wo =
8 Pw R
=Q t
3
Assuming a constant flow rate Q , we can eliminate the speed in favor of the flow rate to
get
v = 0.0962 Q
34
14
R5 4
This provides us with the speed at the near-crack front boundary and an initial pressure
distribution on the crack surface. Using the flexibility (influence) matrix, we can
compute the equilibrium crack opening displacement. The dummy crack is smaller
than the actual crack. We can compute the pseudo-crack advance to arrive at the actual
crack front geometry. All data necessary to start the iterative solution procedure for the
actual crack geometry are thus provided.
E 2 v
P ( ) = c +
3
6 2 1
1
1
L3 x 3
where c is the stress acting normal to the fracture surface, L is the fracture length, is
the distance from the fracture tip, v is the fracture speed. At the wellbore ( = L ), this
gives a pressure Pw equal to
Pw = c + 1.179 E
L
Assuming a constant flow rate Q , one can derive an expression for fracture length as a
function of time
L(t ) = 0.656Q0
v = 0.354Q0
The rest of the solution follows that for the radial crack.
4 Tutorial Example
This section provides a step by step tutorial guide for modeling a radial type hydraulic
fracture in 3D from an uncased wellbore. The modeling procedures to create the initial
model, place the initial crack, and apply static boundary conditions will not be described.
If the user is unfamiliar with these topics, he/she should follow through the FRANC3D
tutorial and learn how to use FRANC3D first.
The initial crack modeling including the crack is shown in Figure 18. We are using
symmetry conditions and modeling a half-radial crack from an uncased wellbore. The
material properties consist of Youngs modulus = 7.5e6 psi and Poissons ratio=0.25.
The fracture toughness is 700 psi. The far field loading consists of uniform pressures of
9090 psi parallel to the wellbore, 10100 psi perpendicular to the crack surface, and 12820
psi parallel to the crack. The wellbore pressure is 10500 psi. The initial crack radius is
10 inches. The closure stress is 10100 psi. The fluid viscosity is 0.0001 cP. The flow
rate along the crack mouth is 0.2 bpm; this is applied to the edges of the crack mouth.
This data is written to the .hyf file.
solution for this model requires about half an hour on a DEC ALPHA workstation. (This
mesh is much coarser than would normally be used for an actual simulation).
Once BES has finished, we read the .fys file into HYFRANC3D. We proceed to the
Hydraulic Fracture menu and select Build Model From File. We select the proper .hyf
for this model (it should have been written previously). HYFRANC3D reads the .hyf file
and the corresponding .flex file. Once finished, the solution parameter dialog box is
displayed. To begin, just select Accept on this dialog and proceed to Solve Current
Model. The iteration steps are displayed on the terminal window and should be similar to
the following information. (Only the final set of iterations starting from the time step
prior to negative speed error is shown.)
Decreasing starting time factor 0.0188639 to 0.017149.
Time step set to: 0.039861; Speed at one node: 50.1743
COD Iteration #1
COD Error (target = 0.001 ): 0.0137515
COD Iteration #2
COD Error (target = 0.001 ): 0.000891194
Mass balance error: 0.367349.
Crack/fluid speed error: 0.10807.
Starting iterations from last solution.
Time step set to: 0.0478333; Speed at one node: 41.8119
COD Iteration #1
COD Error (target = 0.001 ): 0.0253502
COD Iteration #2
COD Error (target = 0.001 ): 0.00128419
COD Iteration #3
COD Error (target = 0.001 ): 2.94335e-05
Global Iteration #1
Mass Balance Error (target = 0.01 ): 0.331611
Crack/Fluid Speed Error : -0.196452
Speed error is -0.196452 and mass balance error is 0.331611.
Do you want to continue iterating (y/n) :n
Global mass balance convergence achieved - system solved
The speed error changes from 0.10807 to -0.196452 in one set of iterations. The mass
balance error is 0.331611. This is clearly not a very good solution. We can improve on
this by adjusting the relaxation factor in the dialog box. Select Set Solve Parameters to
view the dialog box. Set the global time step relaxation factor to 0.001, set the
maximum number of iterations to 500, and then select Accept. Select Solve Current
Model to obtain the equilibrium solution again. This time, the solution procedure takes a
lot longer to converge. After 349 global iterations, the speed error is reduced to a value
of -0.00016398. The mass balance error is 0.35311. It is not possible to achieve a good
mass balance in this case. The starting solution is not good enough; the mass balance
error should decrease significantly for the subsequent steps of propagation because we
start from better starting solutions.
The next step is to Write/Read the .besout file. You can display the response information
first if desired. HYFRANC3D reads the .flex file again, combines the base solution due
to the far field loading along with the fluid pressure and influence matrices to produce the
equilibrium displacements and tractions which are stored in the .besout file. The .besout
file is read as soon at it is written so that we can proceed to propagate the crack.
Select the Advance Crack Front menu button and then select stress intensity factors. Pick
the crack front and the Stress Intensity Factors dialog is displayed. Select Accept. You
can keep or destroy the plots that are displayed. The next step is to calculate the new
crack front points. Select show new front points from the submenu. In the dialog boxes
(Figure 15a,b) set the extension to 2 and select Accept. You are prompted to enter a
tolerance on the terminal window for partial propagation (this is temporary). Enter 0.001
and press the Return key. The new points for the new crack front are displayed. Fit a
polynomial to these points by selecting the proper submenu entry; a single third order
polynomial is appropriate. Finally, propagate the crack. The crack should be propagated
for you. It will have to be meshed, but we will do that after saving the solution. Select
Write/Read .hyd Files and select write current model. Save the data using the default
file name.
Now, subdivide the edges and mesh the surfaces. You can use a similar mesh to that for
step 1. Save the HYFRANC3D model (.fys) file, save the BES input (.bes) file, and save
the .hyf file. Note that you do not have to reset any far-field boundary conditions. The
previous solution (.hyd file for step 1) provides most of the necessary fluid flow boundary
condition data and fluid properties for step 2. The flow rate boundary conditions must be
reapplied to the crack mouth edges, however.
Perform the BES analysis using the same command as for step 1, but change the file
name to that for step 2. The solution requires a little longer than step 1 because of the
additional crack surface elements. Once the .flex file is generated, proceed with reading
the model back into HYFRANC3D. Then proceed to the Hydraulic Fracture menu.
This time, we need to read in the .hyd file for the previous step before reading the current
model .hyf file. Select Write/Read .hyd Files and select read solved model. Select the
.hyd file for step 1.
Now, select Build Model From File and select the .hyf file for step 2. The .flex file is
read in along with the .hyf file and a .infl file is written. Once the model has been read in,
proceed to solve it using the default parameters. From this initial solution, decide on the
relaxation factor and maximum number of iterations and resolve. In this case, a
relaxation factor or 0.01 and maximum number of iterations of 500 should suffice. The
final converged solution after 174 global iterations has a speed error of -0.00088 and a
mass balance error of 0.0802. This is a vast improvement from the initial solution. The
error could be reduced somewhat by reducing the relaxation factor and performing more
iterations. For this tutorial, the solution is adequate. The mass balance error should
continue to decrease with each step if the amount of propagation is not too large.
The user is encouraged to continue the simulation on his/her own. The response
information can be examined at each step or at the end of the simulations. The user can
read in all the solved models in succession and then display animated sequences of
images showing the evolving crack and the fluid pressures or crack opening
displacements at each step of propagation. For instance, Figure 19 shows the pressure
contours on the crack surface for step 2.
Figure 19. Fluid pressure contours on the deformed crack surfaces for step 2.
5.0 Bibliography
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. 1518.
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. 13291342.
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. 239246.
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 threedimensional 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.