You are on page 1of 41

Fracture Simulators

A Practical Comparison

March 25, 1996


Models for Comparison

• FracCADE
• StimPlan
• MFRAC
• FRACPRO
• FRACANAL

March 25, 1996


FracCADE "P3D" Model
Description from Documentation
"Placement is a P3D numerical hydraulic fracture simulator
that provides the capability to model fracture growth into
layers above and below the payzone along with fracture
extension and recession. By modeling extension and
recession, tip screenouts can be simulated. The simulator
can continue after proppant bridging or fluid dehydration.
A Lateral Coupling option is available which represents
the gradual evolution of a fracture from a short KGD-type
fracture to a longer PKN-type fracture."

FracCADE 4.1 Help March 25, 1996


StimPlan "3D" Model
Description from Documentation
"The 3D model is a true, discretized, finite difference
simulator with each "cell" or portion of the fracture treated
independently. Thus, for example, the fracture shape is not
apriori assumed to be elliptical. This is seen in the example
below which shows a fracture breaking through a zone,
and starting downward growth - but only late in the job."
StimPlan Technical Background 1992

Not actual drawing, hand reproduction

March 25, 1996


MFRAC "3-Dimensional" Model
Description from Documentation
"This is a variable height model with both lateral and
vertical fracture growth. For large length to height ratios
the model approaches the PKN model. When no stress,
toughness, or moduli contrast is entered, the model reduces
to the Vertical Radial type Geometry. This model produces
the most realistic geometries and is applicable for all
length to height ratios. The model assumes a bounded
geometry at the leading edge(perimeter). In order to use
this model the reservoir and adjacent layers must be
described fully in terms of rock properties, stress and
leakoff."
MFRAC-II Users Guide Jan. 1994
March 25, 1996
FRACPRO "FRACPRO 3D MODEL"
Description from Documentation
"The fracture module is three-dimensional, in that spatial variations in
reservoir stress, modulus, pressure, and flow distribution are taken
into account. However, it does not need to calculate the variations at
specific points within the fracture. Instead, the effects are integrated
into functional coefficients of the governing differential equations,
greatly simplifying the calculation of the fracture dimensions. The
module can therefore run many times faster than real time, as
required for history matching on-site. The coefficients necessary to
calculate the spatial variations are calculated from a full three
dimensional model and checked against experimental and field test
data. FRACPRO handles up to three modulus zones (payzone, upper,
and lower), up to fifty stress zones, and up to fifty permeable
(leakoff) zones."
March 25, 1996
FRACPRO for Windows Help
FRACANAL II MODEL
Description from Documentation
• "dynamic height growth calculated for finite difference
grids as the fracture propagates(as opposed to the usual
equilibrium height modeling)
• 3 or 5 stress layers plus barriers to growth permitted to
model fracture containment.
• variable height leading edge
• fracture tip permitted to move out of pay zone into
potentially weaker zones
• model can compute temporary containment by thin high
stress layers(which cannot be determined by equilibrium
height models)
• Tip effects and lateral containment (e.g., due to faults or
reservoir boundaries" FRACANAL USERS GUIDE JULY 1994
March 25, 1996
Things all Models have in common
• All based on Linear-elastic rock mechanics
– Assumes that the formation behaves as a "competent"
rock sample would under laboratory conditions. The
definition of a "competent rock" is seldom agreed upon,
however the generality; the higher the youngs modulus-
the more competent the rock, is seldom disputed.
– All rocks will have some degree of "non-linear-elastic"
behavior , this can be accounted for in fracture models
by changing the "dilatency" or "Fracture Toughness" of
the rock which will introduce "non-linear-elastic"
behavior to the model.
March 25, 1996
Things that vary between models
• Model Type
• Degree of "Non-linear-elastic" behavior that is
modeled
• Height Growth Modeling
• Fluid Leakoff Modeling
• Proppant Transport Modeling

March 25, 1996


Model Types
• Cell-based (FracCADE, StimPlan, FRACANAL)
– integrated across a cell
» typically assume constant or linear across each cell
– fracture contains several (5-25) cells
– more rigorous
• Lumped (MFRAC, FRACPRO)
– integrated across entire fracture
– results generally look smoother
» averaged across fracture
» profiles always elliptical

March 25, 1996


Degree of "Non-linear-elastic"
behavior that is modeled
The more "non-linear-elasticity" that is assumed:
• The more Net Pressure
• The more Fracture Width
• Greater tendency for height growth
• The shorter the fracture length
• The higher the fluid efficiency(less fracture area)
• The less pressure drop due to friction down the
length of the fracture

March 25, 1996


Height Growth Modeling
FracCADE
Two Options:
1. Equilibrium solution, This solution assumes that
the fluid flow is generally horizontal. This
assumption implies that the vertical pressure drop
in the fracture is only due to the hydrostatic head
in the fluid.
2. Non- Equilibrium solution, The non-equilibrium
model calculates the movement of the top and
bottom fracture tips, accounting rigorously for all
the stress layers in the system including reductions
in stress with distance from the payzone.
March 25, 1996
Height Growth Modeling
StimPlan

Equilibrium height criterion, with additions


which limit the rate of height growth to
prevent "runaway height growth"

March 25, 1996


Height Growth Modeling
MFRAC

It is not clear how height growth is modeled


in MFRAC, however, the results tend to
agree well with "Cell Based" models such
as StimPlan

March 25, 1996


Height Growth Modeling
FRACPRO

It is not clear how height growth is


modeled in FRACPRO, however, it appears
that FRACPRO uses a time constant to
control height growth.

March 25, 1996


Height Growth Modeling
FRACANAL
• "dynamic height growth calculated for finite
difference grids as the fracture
propagates(as opposed to the usual
equilibrium height modeling)
• model can compute temporary containment
by thin high stress layers(which cannot be
determined by equilibrium height models)

March 25, 1996


Leakoff Modeling
FracCADE
Two options:

1. Cw defined, Based on Cw entered in the Fluid Loss form and from


the formation and fluid properties (compressibility, permeability,
and so on) the dominant leakoff mechanism(wall building or
reservoir controlled) is determined and used by the simulator. The
leakoff coefficient changes during the job depending on the pressure
difference between the formation and the fracture. Cw can be
defined vs. permeability in which case leakoff is handled based on
area of exposed zone or zones at each step of the simulation.

2. Ct defined, Ct entered in the Fluid Loss form is applied to the entire


area intersected by the fracture regardless of permeability, pressure
and so on. March 25, 1996
Leakoff Modeling
StimPlan
Two options:
1.Single layer, single leakoff coefficient is used for entire
leakoff area.
2. Multiple layer, different leakoff coefficients are defined for
different layers.
Mechanism for both options:
A "spurt time" input is allowed to better characterize spurt
loss in high permeability formations(enter a spurt time of 0
for comparison to other models). There is a utility on the
Loss data screen that will calculate a Ct based on Cw,
viscosity of the reservoir fluid, the total compressibility of
the reservoir, and the pressure difference between the
fracture and the reservoir.
March 25, 1996
Leakoff Modeling
MFRAC
Three options:
1. Constant Fluid loss model, Ct value is entered and is used
by the fracture model for entire fracture area.
2. Harmonic Fluid loss model, Cw is input for each zone.
Based on the reservoir pressure, total compressibility,
permeability, porosity, reservoir fluid viscosity, filtrate
viscosity, and spurt loss a Ct is calculated for each layer.
That Ct is used by the fracture model.
3. Dynamic Fluid loss model, same as Harmonic but with
with different preferential weighting of the three
components which make up the total fluid loss coefficient.

March 25, 1996


Leakoff Modeling
FRACPRO
Fluid loss is modeled (for each of 50 or less layers) as one-
dimensional flow perpendicular to the fracture face, following Darcy-
law behavior, including spurt loss, filter cake build up on the fracture
face, and a compressible reservoir- fluid region. The rise in confining
stress due to poro-elastic effects is included if the backstress option is
selected.

More info on Notes page.

March 25, 1996


Leakoff Modeling
FRACANAL II

Based on user input Cw and reservoir properties, leakoff


is calculated for each layer defined. Those leakoff
properties are used by the simulator along with variation
of pressure down the length of the fracture to calculate the
leakoff of the fracture geometry.

March 25, 1996


Proppant Transport Modeling
Two Options: FracCADE
1. One Dimensional Flow, proppant stage fronts are
modeled as vertical lines (concentration constant
from top to bottom of fracture at any given point).
Proppant Settling is modeled down the length of the
fracture.

2. Two Dimensional Flow, Proppant Transport


modeled based on fluid rheology, fracture width,
and velocity of the fracturing fluid. (under
development) March 25, 1996
Proppant Transport Modeling
StimPlan
Two Options:
1. 1D Proppant Transport model

2. 2D Proppant Transport model (under development)


• Rigorous fluid flow/proppant transport model
• Convection and single particle settling
• Variation of fluid and proppant properties with
fracture length and width

March 25, 1996


Proppant Transport Modeling
MFRAC
Proppant Transport based on proppant
settling only. Mechanisms which may occur
other than settling are addressed by
allowing user input of settling rates for each
proppant concentration.

March 25, 1996


Proppant Transport Modeling
FRACPRO
Three + Options:
1. Proppant Settling, Selecting Proppant Settling allows
settling of proppant in the fracture based primarily on fluid
viscosity and particle diameter (i.e., Stokes Law).
2. Proppant Convection, Process whereby heavier treatment
stages (e.g., proppant stages) displace rapidly downward
from the perforations to the bottom of the fracture. Those
stages nearest the perforations may then be replaced by the
pad, or by low-concentration proppant stages.
3. No convection or settling

+. Proppant Convection Coefficient is allowed to control


speed of Convection.
March 25, 1996
Proppant Transport Modeling
FRACANAL II
Two Options:
1. 1D Proppant Transport model

2. 2D Proppant Transport model, Proppant


Transport modeled based on fluid leakoff,
fluid rheology, fracture width shape, and
velocity of the fracturing fluid. (under
development)

March 25, 1996


Frequently asked Questions
• Why do some models predict screenout and others
do not with the same input data?
• Why do some models predict different results after
screenout occurs?

March 25, 1996


Why do some models predict
screenout and others do not with
the same input data?

• Virtually impossible to have same input


• User configured options for Model
• Differences in Leakoff modeling
• Differences in predicted Geometry,
therefore more or less permeable area
exposed to the fracture.
March 25, 1996
Differences in input to Model
Fracturing Programs have extensive user interface's
which isolate the user from the actual input to the
fracture model. It is virtually impossible to input the
same data into a single model twice, much less put
the same data into two different models.

March 25, 1996


User configured options for simulation
Most Fracture models have user configurable
options that may drastically effect results. These
are intended to allow the user to "fine tune " the
model to each region, field, well, or rock. It is very
difficult to have all of these options set to behave
similarly from one fracture model to another.

March 25, 1996


Differences in Leakoff modeling
As you can see from the description of the
leakoff modeling for each of the simulators
leakoff modeling is similar but not exact.
There should not be significant differences
in results simply because of these minor
differences, however, if each simulator
predicts a different fracture shape and
applies the same leakoff modeling to
different areas of varying permeability the
results can be very different.
March 25, 1996
Differences in predicted Geometry

10 md

1 md
0.1 md
0 md

1/6 of fracture
area in high
permeability
layer.

March 25, 1996


Differences in predicted Geometry

10 md

1 md
0.1 md
0 md

1/3 of fracture
area in high
permeability
layer.

March 25, 1996


Why do some models predict different
results after screenout/bridging occurs?

• User configured options


• Coupling of Fracture Model to
Proppant Transport Model
• Fluid Flow through proppant pack

March 25, 1996


What happens after screenout/bridging?
FracCADE
The FracCADE P3D model arrests fracture growth after a
screenout (height and length). Once screenout has
occurred the portion of the fracture which has an
immobile proppant pack is no longer simulated(no
leakoff). The fracture height and length dimensions are
frozen and the simulation continues. The only fracture
growth that occurs after this point is in fracture width.

March 25, 1996


What happens after screenout/bridging?
StimPlan
As the volume fraction of proppant in a particular fluid stage
approaches 0.6 due to fluid loss concentrating the proppant,
the slurry viscosity will approach infinity. This viscosity
increase, if the slurry is filling a significant volume of the
fracture will force fracture propagation to stop, and the
solution will enter "tip screen-out mode". Execution
continues, allowing a prediction of post screenout pressure
behavior and proppant distribution. Leakoff is allowed
through the walls of the created fracture (as always) and
through the immobile proppant pack at the tip of the fracture
or wherever the dehydration of the slurry dictated that the
simulator switch to " tip screen-out mode".
March 25, 1996
What happens after screenout/bridging?
MFRAC
Two Options:
1. Default option, stop fracture growth (height and length)
when bridging occurs, the only fracture growth that
occurs after this point is in fracture width.
2. Uncouple proppant modeling from fracture modeling
option, If this option is selected the model ignores the
bridging and continues the simulation based on all other
input data.

March 25, 1996


What happens after screenout/bridging?
FRACPRO
The default is the quadratic backfill model. Each stage
has an upper and lower quadrant, if there is proppant
bridging in one of these quadrants fracture growth will
stop in that quadrant at that point(growth can continue in
the other quadrant). If there is bridging in an upper and
lower quadrant of any stage, fracture growth will stop in
both quadrants and only the fracture width will grow with
subsequent pumping.

The quadratic backfill model can be altered by changing


the quadratic backfill coefficient which will change the
rate of pressure increase after bridging has occurred.
March 25, 1996
What happens after screenout/bridging?
FRACANAL II
After bridging occurs in the fracture the simulation is
stopped. There is a "Soft Tip Screen Out" option which
allows the fracture to continue to grow after bridging. If
this option is selected the model ignores the bridging and
continues the simulation based on all other criterion
specified in the input file.

March 25, 1996


Results from Comparison
THREE-LAYER RESULTS AT END OF PUMP
Length Height Pressure Max. Width Efficiency
Model (ft) (ft) (psi) (in) (%)
RES-FP 1,744 544 1,227 0.9 80
Marathon 1,360 442 1,387 1.04 96
Texaco-FP 1,938 435 1,132 0.72 68
Meyer-2 2,692 360 1,109 0.72 74.3
Advani 2,089 357 1,113 0.66 43
FracCADE 2,568 338 1,045 0.71 76.7
SAH 3,408 318 1,009 0.65 77
Arco-Stimplan 3,598 306 992 0.57 67
Meyer-1 3,549 291 987 0.58 70.3
NSI-Stimplan 3,750 283 903 0.56 66

Data in Tables, with exception of FracCADE, from SPE #25890 presented 1993

March 25, 1996


Results from Comparison
FIVE-LAYER RESULTS AT END OF PUMP
Length Height Pressure Max. Width Efficiency
Model (ft) (ft) (psi) (in) (%)
RES-FP 1,754 501 1,119 0.83 82
Marathon 1,224 476 1,250 1.03 97
Advani 1,594 438 1,129 0.81 58.1
Texaco-FP 2,011 428 1,008 0.68 69
SAH 2,905 394 960 0.72 80.1
Arco-Stimplan 3,399 394 944 0.64 68
FracCADE 2,332 384 934 0.94 79
NSI-Stimplan 3,709 361 852 0.63 66
Meyer-1 2,962 328 669 0.5 70.5
Meyer-2 2,407 327 768 0.6 74.8

Data in Tables, with exception of FracCADE, from SPE #25890 presented 1993
March 25, 1996

You might also like