SPE-199728-MS

Developing Methodology for DFIT Design and Pressure Interpretation by

Coupled Reservoir Geomechanics Flow Simulation

Mohamed Mohamed, Colorado School of Mines; Tirth Thaker, Computer Modeling Group; Mazher Ibrahim, Apache
Corp.; Erdal Ozkan, Colorado School of Mines

Copyright 2020, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Hydraulic Fracturing Technology Conference and Exhibition held in The Woodlands, Texas, USA, 4-6 February
2020.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
A numerical, coupled reservoir flow and geomechanics model has been built using a black-oil simulator
and evaluated for diagnostic fracture injection test (DFITs) in unconventional reservoirs. DFITs have
evolved into a commonly used technique to generate direct estimates of some key unconventional
reservoir characteristics. Challenges, however, arise due to the complexity of the unconventional reservoir
characteristics, long shut-in times to reach the closure pressure, and the long test durations to obtain
realistic reservoir parameters from pseudoradial flow. One way to improve the analysis of minifrac tests
in unconventional reservoirs is to use a coupled geomechanics and flow simulator, which is capable of
representing the interactions between the reservoir fractures and minifrac developed during minifrac tests.
The numerical model used in this work is based on coupling the geomechanical rock properties with
the reservoir flow model. The geomechanical model simulates the growth and subsequent closure of the
hydraulic and secondary fractures by modeling the change in the reservoir stresses. Furthermore, the model
simulates the minifrac pressure response before fracture-closure as well as during the after-closure falloff
period. The pressure response during the falloff period is then analyzed to evaluate the reservoir properties
using Nolte pre-closure and after-closure analysis techniques.
The coupled geomechanics and flow simulation of minifrac provides the capability of modeling the
fracture breakdown and fracture closure and estimating the fracture dimensions. The traditional fracture
design tools provide similar information focusing only on the geomechanics of the rock and ignoring the
effect of the reservoir flow. An additional advantage of the simulator used in this work is to provide estimates
of fracture stiffness, fracture closure permeability, residual permeability, and permeability of the matrix.
The simulator allows pressure matching and successfully simulates the reservoir flow for extended shut-
in periods. The simulator allows designing DFITs with different injection volumes, rates, and shut-in periods
and thus helps provide the optimum parameters for the test. The after-shut-in flow regimes generated by
the simulation model agree with the DFIT theory.
2 SPE-199728-MS

Introduction
DFIT is a short injection/falloff test performed without proppant before a main fracture stimulation
treatment. During a DFIT, 10-25 bbls of water (usually KCL water) is pumped at a constant rate of 2-10 bpm
(Mohamed et al. 2019). The intent is to break the formation to create a short fracture during the injection
period and then to observe the closure of the fracture system during the ensuing falloff period.
The test must be run long enough to reach pseudoradial flow to determine reservoir pressure as well as
reservoir permeability. The duration of the shut-in period to reach pseudoradial flow is dependent upon the
formation permeability and the pumping time. Other completion properties can also be obtained, including
instantaneous shut-in pressure (Mohamed et al. 2019), fracture gradient, net extension pressure, fluid leakoff
mechanism, closure time, closure pressure (minimum horizontal stress), an approximation of maximum
horizontal stress, anisotropy, fluid efficiency, effective permeability, transmissibility, and pore pressure.
DFIT creates complex interactions between effective stress and joint deformation. The induced
deformations can take the form of shear, dilation, normal closure, and opening of fractures. Deformation due
to fracturing can cause as much as three orders of magnitude change in the rock conductivity at moderate
compressive stress levels.
The model used in this work simulates stress- and fracture size-dependent coupling of shear stress,
displacement, dilation, conductivity, normal stress, closure, and conductivity. The model couples the fluid
flow and induced-stress in three-dimensions. Most simulation models discussed in the literature focus more
on the flow in the reservoir; however, they do not typically include the underlying geomechanical effects
on fracture creation and the corresponding stress effects on the subject formation and their impact on the
formation properties. These models have pressure-dependent permeability and porosity function at best.
Few numerical flow models incorporate full geomechanical capabilities, as presented by Dean et al. (2006).
One advantage of flow simulators that incorporate full geomechanics is that they can be easily turned around
for flow simulation and have all the related geomechanics effects captured during the flowback stage of the
hydraulic-fracturing operation and during the subsequent production stage.
In this work, we have used CMG IMEX, which is a numerical reservoir flow simulation tool that is
designed to handle the flow in the reservoir in a rigorous manner with a coupled geomechanics feature.
The simulator can model fracture initiation, propagation, closure, and falloff behavior of a typical DFIT.
These models are discussed in detail and have been applied for geomechanical evaluation of a saline aquifer
during CO2 sequestration by Tran et al. (2005) and Tran et al. (2009).
One of the advantages of the numerical simulator used in this study is the grid-scale specification of
reservoir and fracture properties. This simulator is a fully coupled 3D model that is robust in determining the
fracture dimension and its propagation in the vertical and linear-horizontal directions. The model provides
the ability to estimate the time to reach pseudoradial flow during pressure falloff prior to the actual field
DFIT. This allows better design, planning, and execution of the DFIT.
This paper is organized as follows; first part is the description of the coupled reservoir geomechanics and
flow simulator and the input parameters. The second part of the paper presents simulations of minifrac tests
for different reservoir properties to validate the capability of the model of honoring the reservoir properties.
Finally, we present recommendations for DFIT designs based on our numerical model findings.

Description of the Coupled Reservoir Geomechanics and Flow Simulator

A black-oil unconventional simulator (IMEX module) has been used in this study for reservoir flow
simulation. The simulator adopts a Logarithmically Spaced, Locally Refined, Dual-Porosity (LS-LR-DP)
reservoir model to adequately represent the complex natural fracture and mini fracture behavior. The dual
Porosity model is used to incorporate the effect of randomly distributed secondary fractures intersecting the
mini fracture, and a logarithmically spaced grid is used to create refinement around the fracture where it is
crucial to capture the pressure transients with the least amount of grid blocks. Evenly spaced gridding was
SPE-199728-MS 3

not used to minimize the runtime. Flow is allowed to occur from matrix to matrix, fracture to fracture, and
matrix to fracture, and the porosity of the matrix and fracture media varies from grid block to grid block.
Fig. 1 illustrates the allowed flow connections in a dual-porosity system. The matrix to fracture flow is
modeled by using a Gilman and Kazemi (1983) matrix-fracture transfer term defined by:

(1)

where, L represents the fracture spacings in the x, y, and z directions (ft), and k is the matrix permeability
(mD).

Figure 1—Schematic of flow connections in the dual-porosity system.

In the numerical simulator, the geomechanics and fluid flow calculations are coupled through the pressure
and porosity function using a two-way coupling method. Initially, the simulator calculates the pressures over
time and passes the pressure values to the geomechanical module that updates the formation deformation in
response to the updated pressure values. The geomechanical model computes the stress and strain profiles
and then passes the new deformation parameters back to the fluid flow simulator for the next time interval in
a closed loop. Fig. 2 shows a flowchart that illustrates the coupled calculations between the geomechanics
module and the fluid flow simulator. The porosity is calculated in the fluid flow model as a function of the
pressure, where the mass is conserved between the time steps. The geomechanical deformation response
to the pressure changes is through changing parameters in the porosity function, where the porosity is a
function of pressure, temperature, and total mean stress, as shown in the following equations.
(2)
where,

(3)

and

(4)

(5)

(6)

(7)
where Vpore is the pore volume (ft3), cb is the bulk compressibility (psi-1), Vb is the bulk volume (ft3), α is
the Biot's coefficient, γ is the volumetric thermal expansion ( ), and σm is the mean total stress (psi). The
previous set of equations for the deformation response are computed for each grid block.
4 SPE-199728-MS

Figure 2—Flowchart of the coupled geomechanics and fluid flow simulator.

Barton-Bandis stress-permeability relationship model (Barton et al. 1985; Tran et al. 2009) was used
for this simulation model. Barton-Bandis model describes the relationship between the permeability of the
fracture system and fracture opening. In this model, the secondary fracture system is described in the grid
system using the dual-porosity formulation, where the secondary fracture represents the natural fractures
in the system and is coupled with a porous rock matrix on a one-to-one basis such that the stresses are
specified in every grid block.
The stress-permeability relationship of the Barton-Bandis model is described in Fig. 3. In Fig. 3, the y-
axis is the fracture permeability, x-axis is the normal fracture effective stress ( ), kccf is fracture closure
permeability, khf is the maximum fracture permeability, and krcf is the residual value of fracture closure
permeability.

Figure 3—Modified Barton-Bandis stress-permeability relationship model (Left).

Illustration of the load applied on a rock specimen and the generation of fractures (Right).

According to the Barton-Bandis model, initially, there is no fracture in the matrix rock. The effective
normal stress on the fracture increases as the pressure increases in the grid block. Fracture propagation
starts immediately when the stress increases past the failure envelope of the rock, allowing the fluid to
flow through the fracture system as well as the surrounding matrix blocks. The rock material is assumed to
be brittle such that when rock cracks, the value of fracture permeability increases from zero to maximum
SPE-199728-MS 5

permeability khf. The fracture aperture continues to increase until the pressure within the rock is reduced.
The decrease in pressure causes normal fracture effective stress to increase, and the fracture aperture to
decrease. Due to the asperities of the fracture surface, the fracture aperture cannot be closed completely, and
the closing-fracture permeability does not follow the opening-fracture-permeability path. The subsequent
stress path is shown in Fig. 3 and the fracture permeability follows the curve FG when the value of fracture
normal effective stress changes its direction from negative to positive, allowing the fluids to flow through
the crack. The residual value of fracture closure permeability indicates that the aperture cannot be closed
completely. If the fracture is mainly caused by tensile stress, the threshold normal fracture effective stress
is negative or zero. If the fracture is mainly due to shear stress and is parallel to the crack as shown in Fig.
3, the threshold value may be positive as the failure may obey the Mohr-Coulomb criteria, rather than the
tensile criteria. The threshold value for the rock used in this study was obtained from laboratory tests under
different loading conditions.
The geomechanical parameters (stress and strain) calculated from the Barton Bandis model are only
coupled to the matrix blocks; however, the model allows the calculation of the fracture permeability from
the normal fracture effective stress. The fracture closure permeability krcf (mD) is calculated by the following
equation:

(8)

Where kccf is the fracture permeability at zero stress (mD), (e) is defined as the current fracture aperture
(ft), e0 is the initial fracture aperture (ft); that is,
(9)
and Vj is the stress to fracture stiffness ratio and is calculated as follows:

(10)

In Eq. 10, Vm is the minimum fracture aperture correlated to closure permeability (ft).

(11)

Rock Mechanical Testing and Core Analysis for Geomechanical Properties

Core samples were analyzed for geomechanical properties of Mohr failure envelope, Young's modulus,
Poisson's ratio, and failure strength. Altogether, 3 cores of Well A, within the depth range of 4957.40 ft.
to 5081.45 ft, were used.
To characterize the mechanical properties of the rock sample, the triaxial compressive tests were
performed at a range of confining pressures. The triaxial compressive tests are commonly used to simulate
in-situ stress conditions of the reservoirs and provide compressive strength and static values of Young's
modulus and Poisson's ratio. Moreover, Mohr-Coulomb analyses, determined from the triaxial compressive
strength collected data for cohesion (initial shear strength) and internal friction angle. The compressive
strength from the triaxial compressive tests is used to determine the critical shear stress and predict the
conditions of failure during fracturing. The summary of the mechanical properties obtained from rock testing
is shown in Table 1.
6 SPE-199728-MS

Table 1—Summary of rock mechanical properties obtained from lab testing.

Young's Modulus (psi) 1.1 - 4.1 e6

Poisson's Ratio 0.22-0.24

Cohesion for Mohr-Coulomb (psi) 527

Angle of internal friction for Mohr-Coulomb (degrees) 53.59

Coefficient of internal friction 1.36

Bulk density (g/cm ) 3 2.53

Biot's Constant at Confining Pressure of 770 psi 0.72

Biot's Constant at Confining Pressure of 1,200 psi 0.68

Total Porosity (%) 5

DFIT Modeling using Dilation-Compaction Process (Low Fidelity) Approach

In this section, we provide a test case where stresses are ignored, and only compaction/dilation curves
are used to model the pressure-dependent degradation of permeability and porosity. A table of ϕ and/or k
multipliers as a function of pressure is used for describing the stress changes with the formation response
represented as branch paths and possibly hysteresis. Each path is described by a table of transmissibility
multiplier or permeability versus stress value, and the various paths are linked at common stress values.
Each multiplier is applied only to fluid-flow and not to any geomechanical properties. The porosity,
permeability and transmissibility multipliers at a given pressure are determined from table lookup with
linear interpolation.
Fig. 4 represents the dilation model used in the test case. When the pressure in a rock is lower than its
yield point pressure (3,206 psi in this example), the rock behaves elastically or reversibly. As the pressure
in the rock is further increased, the rock starts behaving plastically. When the pressure in the rock suddenly
decreases, the rock behavior follows one of the unloading/dilation paths in Fig. 4 based on the pressure
the block had achieved at the end of the dilation period. As the pressure decreases, the rock will continue
following the unloading path until it reaches the minimum pressure.

Figure 4—Dilation model used in test case 1.

SPE-199728-MS 7

Initial oil, gas, and water saturations were not provided within the core dataset, accordingly published
literature values were used from wells in nearby counties as initial model inputs. Organic-rich shales
deposited in lakes and marine environments are typically assumed to yield mainly water-wet characteristics;
however, wettability can change to weakly water-wet and strongly oil-wet as a result of the organic content
and mineralogy. Water was assumed to be the wetting phase for this study. Using history matching from
production data, the relative permeability curves were generated and imported into the model. The total
liquid saturation is defined as the oil saturation plus the connate water saturation for a water-wet case. The
curves generated from these values are given in Fig. 5, and Fig. 6. Table 2 provides a summary of the
reservoir properties and model description.

Figure 5—Relative permeability as a function of water saturation used in this study.

Figure 6—Relative permeability as a function of total liquid saturation used in this study.
8 SPE-199728-MS

Table 2—Summary of model inputs.

Top Depth (ft) 5168

Thickness (ft) (Overburden/Formation/Underburden) 5/5/5

Fracture Injection Depth (ft) 5,208

Initial Reservoir Pressure (Psia) 3,206

Reservoir Temperature (F) 200

Permeability (md) (I, J, K) 0.0085/0.0085/0.00085

Fracture Compressibility (1/psi) 2E-6

Formation Compressibility (1/psi) 2E-6

Injection Fluid Water

Injection Rate (bpm) 5

Injection Duration (min) 6

The pressure response is captured during injection as well as during the falloff period after the end of
injection. Fig. 7 shows the pressure history obtained from the simulation. The procedure involves pumping
25 bbl. of water; once the initial breakdown occurs, the injection rate is held constant for a short time span.
The pressure breakdown was around 11,000 psi in our example, which is comparable to the breakdown
pressure observed from the field data.

Figure 7—Rate and pressure (results) from the DFIT simulation for the test case.

The pressure profile after the end of the injection is then used for the Nolte pre-closure, G-function
analysis, and after closure analysis. The following plots are used to analyze the before closure analysis
(BCA) and after closure analysis (ACA) of the fall-off pressure data to identify the fracture parameters for
the anomalous field DFIT cases discussed below:

• Log-log of pressure and derivative, introduced by Bourdet et al. (1989)

• G-Function Plot

• plot
SPE-199728-MS 9

The log-log plot can be used to establish the flow regimes for both the BCA and ACA analyses. Typically,
the log-log plot is used for the pressure build-up analysis test. However, Cinco et al. (1978) and Bachman et
al. (2012) demonstrated it to be a valid analysis technique when applied to a DFIT dataset. The pressure data
and its derivative are plotted on log-log scale, which identifies the flow regimes along with the characteristic
slopes. Fig. 8 shows the log-log Bourdet derivative plot (Bourdet et al. 1989). The diagnostic plot indicates
the following flow regimes:

• Fracture flow from the ¼ slope;

• Unit slope defining the before closure region;

• Formation Linear flow from the ½ slope; and

• Zero slope is defined as a pseudoradial flow regime.

Figure 8—Log-log Diagnostic plot for pressure response for the test case.

The unit slope can be interpreted by fracture closing (Craig and Blasingame 2006). The fracture closure
is picked at the deviation from the unit slope. After identifying the flow regime and fracture closure, a G-
function plot is used to confirm the closure pressure. From the G-function plot in Fig. 9, we can observe
a pressure drop during the first hours of shut-in caused by tip extension or near-wellbore tortuosity. The
fracture closure is picked at the start of the inflection and was confirmed from the log-log diagnostics at
the end of the unit slope. The fracture closure for this case was found as 5,200 psi, and the closure time
is 25.5 hrs.
10 SPE-199728-MS

Figure 9—Nolte G-function plot for closure pressure determination for the test case.

Even though for this test case, there were no fractures modeled in the system, and only a dilation-
compaction process is used, the G-function plot shows the behavior of the pressure-dependent leakoff from
dilated fractures/fissures. This esteems the need for a fully coupled reservoir and geomechanical approach
to the diagnostic fracture injection test problem, and secondary fracture response to pressure falloff should
not be ignored.
The log-log plot of pressure minus assumed reservoir pressure, versus the square of the linear flow time
function, can be used to identify the after-closure flow regimes. The after-closure analysis depends on an
accurate closure pick. The pressure difference curve is dependent on the value of reservoir pore pressure
used, but the pressure derivative is insensitive to the pressure estimate. Fig. 10a shows the ∆p vs. the square
of the linear flow time function, where the linear flow and pseudoradial flow regimes can be identified,
respectively, by a ½-slope and unit-slope behavior of the pressure derivative. During the pseudoradial flow
period, both pressure difference and derivative curves will lie on the same unit slope line if the pressure
estimate used for the pressure difference function is correct. In this test case, the pressure estimate was
3,217 psi. Once the pseudoradial flow regime has been identified, the cartesian plot of pressure versus
the pseudoradial flow time function can be constructed. As shown in Fig. 10b, a straight line through the
appropriate data in the pseudoradial flow period can be drawn. The intercept gives the pore pressure, where,
in this test case, the pressure was 3,217 psi, which is reasonably close to the input reservoir pressure as
3,206 psi. The slope is used to calculate the transmissibility from which the permeability can be determined
from the following equation, knowing the net pay and reservoir fluid viscosity:

(12)
SPE-199728-MS 11

Figure 10—After Closure Analysis- Square Linear Flow Time Function

(left) and Radial Flow Time Function Plot (right) for the test case.

Assuming a fluid viscosity of 0.313 cp and knowing fracture height of 25 ft, the effective reservoir
permeability can be calculated as 0.003 mD. This is sufficiently close to the input permeability of 0.008
md. The fracture half-length was also calculated as 23 ft.
Fig. 11a shows the pressure distribution at the initiation of the fracture at the beginning of the injection,
Fig. 11b shows the breakdown point of the rock, and Fig. 11c pressure distribution at the end of the injection.

Figure 11—Pressure distribution at the a) beginning of injection, b) rock breakdown c) end of injection.

Fracture Modeling using Coupled Geomechanics

The Barton-Bandis geomechanical model (Barton et al. 1985) was used to simulate three cases. The first
case is a homogenous formation with no natural fractures within the matrix. This is to simulate a normal
leakoff behavior and fracture closure due to matrix leakoff. The second case includes natural fractures. The
leakoff, in this case, occurs faster than expected for a normal bi-wing fracture due to the interception of
natural fracture with the hydraulic fracture. The last case is to simulate a fracture height recession case
where the fracture propagates through impermeable layers above or below the fractured zone during the
injection. All three cases were created and simulated to prove the validity of the Barton-Bandis approach.
No history matching was performed for these cases; rather, results were used to analyze flow behavior and
validate model capability to simulate the pressure response during mini-frac operation.

Case 1: Homogenous Formation Example

The reservoir fluid properties, mechanical properties, as well as the injection schedule for this example were
adopted from the compaction-dilation simulation presented in the previous section. A total of 20 bbls of
water was injected at a constant rate of 3.5 bpm. The minimum horizontal stress in the x-direction, maximum
horizontal stress in the y-direction, and the overburden stress in the z-direction are set at 6,248 psi, 6,498
psi, and 9,181 psi, respectively, at the top of the formation. In general, the stresses can be specified at each
12 SPE-199728-MS

individual grid block in the numerical simulator allowing more rigorous accommodation of stress variations
in the horizontal and vertical directions. In this test case, we kept the overburden stress and the minimum
horizontal stress constant across the layers. The Barton-Bandis parameters used in all validation cases are
shown in Table 3.

Table 3—The Barton-Bandis parameters used for Test Cases 1-3.

Initial Fracture Aperture (ft) 2e-05

Fracture Stiffness (psi/ft) 7e5

Fracture Opening Stress (psi) 4,200

Hydraulic Fracture Permeability (mD) 1,000

Fracture Closure Permeability (mD) 100

Fracture Residual Permeability (mD) 15

Although this case is a dual-porosity scenario, that is, natural fractures are supposedly present, the
transmissibility between the matrix and fracture media is assumed to be zero, and the permeabilities of the
fracture and matrix systems are assumed to be equal at 0.007 mD. These adjustments amount to simulating
a homogenous formation with no natural fractures.
The pressure profile after the end of injection is used for the Nolte pre-closure, G-function, and after-
closure analyses. The log-log plot can be used to establish the flow regimes for both the BCA and ACA
techniques. The pressure data and its derivative are plotted on a log-log scale to identify the flow regimes
along with their characteristic slopes. Fig. 12 shows the log-log Bourdet derivative plot. The diagnostic plot
indicates the following flow regimes:

• The 3/2-slope signature defining the before closure region;

• Formation Linear flow indicated by the ½-slope trend; and

• The zero-slope behavior signifying the existence of pseudoradial flow regime.

Figure 12—Log-Log Diagnostic plot for pressure response for normal leakoff case.
SPE-199728-MS 13

The fracture closure is picked at the deviation from the 3/2-slope behavior. After identifying flow regimes
and closure, a G-function plot was used to confirm the closure pressure. The G-function plot shown in Fig.
13 yields a homogenous leakoff signature, where the closure pressure is also confirmed by the deviation
from the straight line behavior passing through the origin. The fracture closure pressure and closure time
are obtained as 6,411 psi and 11.46 hrs, respectively, in Fig. 13. The closure pressure obtained from the G-
function plot is comparable to the input minimum horizontal stress of 6,248 psi.

Figure 13—Nolte G-function plot for closure pressure determination for normal leakoff case.

Fig. 14a shows the ∆p vs. the square of the linear flow time function, where the linear and pseudoradial
flow regimes can be identified by a ½-slope and unit-slope of the pressure derivative, respectively. During
the pseudoradial flow period, both curves will lie on the same unit slope line if the pressure estimate used
for the pressure difference function is correct. In this test case, the pressure estimate was 3,100 psi. Once
the pseudoradial flow regime has been identified, a Cartesian plot of pressure versus pseudoradial flow time
function can be constructed, as shown in Fig. 14b. The intercept of the straight line through the appropriate
data in the pseudoradial flow period gives the pore pressure. From Fig. 14b, the pore pressure was estimated
as 3,106 psi, which is close to the input reservoir pressure of 3,206 psi. The slope is used to calculate
the transmissibility from which the permeability can be determined, knowing the net pay and reservoir
fluid viscosity and using Eq. 12. Assuming fluid viscosity of 0.313 cp and knowing fracture height of
25 ft. the effective reservoir permeability can be calculated as 0.004 mD, which is in agreement with the
input permeability of 0.007 md. The start and end of linear flow were found to be 14.5 hrs and 22.5 hrs,
respectively. The beginning of the pseudoradial flow was found to be 48.22 hrs.
14 SPE-199728-MS

Figure 14—After Closure Analysis- Square Linear Flow Time Function (left)
and pseudoradial Flow Time Function Plot (right) for normal leakoff case.

Case 2: Multiple Closures Examples

In this example, multiple closures caused by fracture complexity due to natural fractures is simulated.
Often time, this leakoff behavior is referred to as pressure dependent leakoff. For the dual-porosity model,
the permeability and porosity of natural fractures are assumed to be different from that of the matrix
(Table 4), where permeability is given in mD. Fracture spacing is used to calculate the matrix-to-fracture
transmissibility in each grid. Case A represents a scenario of sparse natural fractures with an opening stress
of 4,248 psi.; thus, the spacing is higher than Case B. Case B is the case of a denser fracture network (less
fracture spacing) as well as a lower fracture opening stress of 4,200 psi. Other Barton-Bandis parameters
are presented in Table 3.

Table 4—Fracture and Matrix Properties for Case A and Case B.

Matrix Fracture Fracture Spacing

PERM I PERM J PERM K Porosity PERM I PERM J PERM K Porosity I-Dir J-Dir D-Dir

Case A 0.0007 0.0007 0.0007 0.05 0.007 0.007 0.007 0.01 1 1 10000
Case B 0.0007 0.0007 0.0007 0.05 0.007 0.007 0.007 0.01 0.5 0.5 10000

Reservoir fluid properties, mechanical properties as well as the injection schedule was kept equal to
the case discussed above. The minimum horizontal stress in the x-direction, maximum horizontal stress in
the y-direction, and the overburden stress in the z-direction are set at 6,248 psi, 6,498 psi, and 9,181 psi,
respectively, at the top of the formation. The overburden stress and the minimum horizontal stress were
held constant across the layers in this example.

Multiple Closures Case A

The pressure profile after the end of the injection is used for the Nolte pre-closure, G-function, and after-
closure analyses. The log-log plot can be used to establish the flow regimes for both the BCA and ACA
techniques. The data and its derivative are plotted on the log-log scale that identifies flow regimes along with
their characteristic slopes. Fig. 15 shows the log-log Bourdet derivative plot. The diagnostic plot indicates
the following flow regimes:

• ¼-slope defining the before-closure, early, bi-linear flow, in the natural and hydraulic fractures.

• ½-slope for fracture linear flow, where fluid flows along the fracture, thus increasing fracture width.
SPE-199728-MS 15

• Early unit-slope before closure for composite flow.

• 3/2-slope defining the before closure region.

• After closure formation Linear flow from the ½ slope, where fluid flows into the formation in paths
normal to the fracture plane.
• The zero slope defined as the pseudoradial flow regime, where fluid flows radially into the
formation from the wellbore.

Figure 15—Log-log Diagnostic plot for pressure response for Multiple Closures Case A (left).
Physical illustration of composite flow due to secondary fractures intersecting main frac (right)

In this case, the pressure gradient in the finite-conductivity fractures is comparable to the formation
pressure gradient in natural fractures and thus, the bi-linear flow regime is observed. The bi-linear flow is
identified by a ¼-slope trend on the Bourdet log-log plot. According to the analytical model presented by
Cinco-Ley and Meng (1988), if mbf is the slope of the corresponding straight line on a Cartesian plot of
Δpwf vs. , then

(13)

where,

(14)

At the point of transition from the unit-slope behavior to 3/2-slope behavior before fracture closure, the
pressure equals to 6,482 psi, which is equal to the input maximum horizontal stress, and the minimum
horizontal stress (main fracture closure) is estimated at the end of the 3/2-slope period as 5,655 psi.
This is certainly an important observation as in many field cases, a unit-slope behavior is observed on the
Bourdet derivative plot. This is an indication of the composite flow behavior before closure. The before-
closure composite flow is attributed to enhanced leak-off due to the opening of existing natural fractures and
connecting to the hydraulic fractures, resulting in a larger total fracture surface area. The composite flow
regime occurs when the matrix to fracture permeability ratio is greater than or equal to 10. Warpinski (1991)
interpreted this behavior in a pre-closure setting as a pressure dependent leakoff. The physical illustration
for this scenario is on the right-hand side of Fig. 15.
16 SPE-199728-MS

On the G-function plot shown in Fig. 16, an early peak is observed on the derivative plot, which represents
the closing of natural fractures at 7,282 psi. The main closure pressure is observed at 5,578 psi. The natural-
fracture closure pressure obtained from the G-function plot is significantly higher than the pressure obtained
from the log-log plot.

Figure 16—Nolte G-function plot for closure pressure determination for Multiple Closures Case A.

The linear and pseudoradial flow regimes can be identified by the ½-slope and unit-slope behaviors of
the pressure derivative, respectively, on the plot of ∆p vs. the square of the linear flow time function shown
in Fig. 17a. During the pseudoradial flow period, both pressure and derivative curves will collapse on the
same unit slope line if the pressure estimate used for the pressure difference is correct. Fig. 17b presents the
cartesian plot of pressure vs. the pseudoradial flow time function. A straight line through the appropriate
data in the pseudoradial flow period is constructed and the vertical intercept gives a pore pressure estimate
of 2,910 psi, which is slightly less than the input pressure. The slope is used to estimate the transmissibility
and permeability by Equation 12. The start of linear flow and end of linear were found to be 0.8 hrs and 1.2
hrs, respectively. The beginning of the pseudoradial flow was found to be 1.5 hrs.

Figure 17—After Closure Analysis- Square Linear Flow Time Function (left)
and Radial Flow Time Function Plot (right) for Multiple Closures Case A.
SPE-199728-MS 17

Multiple Closures Case B

In the presence of dense natural fractures with closer spacing, the log-log plot (Fig. 18) for Case B does
not show a 3/2-slope behavior before closure; instead, a unit-slope behavior is observed on the Bourdet-
derivative, followed by a linear flow period and a pseudoradial flow period after two log cycles. The
observed unit-slope behavior correlates well with the expectation of a composite flow behavior before
closure, which is dominant in this case due to the denser natural fractures within the matrix. Main fracture
closure is masked on the log-log plot. Nonetheless, the closure of secondary fractures can be determined at
the end of the unit slope as 6,671 psi. The main fracture closure is somewhere between the end of the unit
slope and before the beginning of the linear flow regime.

Figure 18—Log-log Diagnostic plot for pressure response for Multiple Closures Case B.

Next, we use the G-function plot (Fig. 19) to corroborate the closure pick on the derivative plot. The
G-function plot shows a large hump above the straight line. Subsequent to the hump, G-function shows
another smaller hump. The end of the second hump represents the fissure opening/closure pressure. The
closure stress from the first and the second humps are 6,663 psi and 6,238 psi, respectively. The second
closure stress is in agreement with the input minimum horizontal stress, and the first closure is in agreement
with the input maximum horizontal stress. The first hump represents the closure of natural fractures, which
precedes that of the main hydraulic fracture, resulting in an early closure signature. The magnitude of the
first hump is significantly greater than the one observed in Case A due to the denser fracture spacing in
Case B. Therefore, we conclude that the coupled reservoir geomechanics and flow model is sensitive to the
density of the secondary (natural) fractures.
18 SPE-199728-MS

Figure 19—Nolte G-function plot for closure pressure determination for Multiple Closures Case B.

Fig. 20a shows the ∆p vs. the square of the linear flow time function, where the linear flow and
pseudoradial flow regimes can be identified. Fig. 20b presents the cartesian plot of pressure versus the
pseudoradial flow time function. A straight line through the appropriate data in the pseudoradial flow period
is constructed. The intercept gives a pore pressure estimate of 3,315 psi, which is in agreement with the input
pressure. The slope is used to calculate the transmissibility and permeability by Equation 12. Assuming a
fluid viscosity of 0.313 cp and knowing the fracture height of 25 ft, the effective reservoir permeability can
be calculated as 0.0011 mD. The start and end of linear flow period were found to be 1.4 hrs and 3.5 hrs,
respectively. The beginning of the pseudoradial flow was at 84.5 hrs.

Figure 20—After Closure Analysis- Square Linear Flow Time Function (left)
and Radial Flow Time Function Plot (right) for Multiple Closures Case B.

Case 3: Height Recession Example

The height recession concept is observed when the fracture extends out of the target zone into overlying
and/or underlying layers where the leakoff rate is relatively small due to lower permeability and higher
local formation stress than that in the target zone. Fluid in the fracture beyond the target zone leaks into
formation with lower permeability at a slower rate than the decline rate of fracture face in the target layer,
and thus, height recession is observed.
SPE-199728-MS 19

The height recession concept is simulated with a single fracture, that propagates into overlying and
underlying layers out of the target zone. As shown in Fig. 21, the horizontal stress in the upper and lower
bounding intervals are increased, and the permeability is decreased.

Figure 21—Sideview with target region (red) at permeability higher one order of
magnitude and stress that is 300 psi lower than the underlying and overlying layers.

The simulation evaluates a three-layer scenario. The upper and lower bounding intervals have different
stress and matrix permeability values, with the minimum horizontal stress being 300 psi greater and the
permeability being one order of magnitude lower as compared to the target interval. Fig 22 shows the
distribution of fracture permeability and pressure at the beginning of injection, breakdown, and end of
injection. The fracture predominantly propagates in the target zone (low-stress layer) in the horizontal
directional, with some height growth in the underlying and overlying layers (high-stress layers). The fracture
is expected to propagate slightly more in the upward direction than downward because of the increase in
stress with depth.

Figure 22—JK Plane of pressure propagation (upper row) and fracture permeability
(lower row) at a) beginning of injection, b) rock breakdown c) end of injection

At shut-in, the volume of the fluid stored in the fracture is large; nonetheless, the leakoff rate is small
and, hence, the rate of pressure decline is small. As shown in Fig. 23, compared to the target layer, the
fracture permeability decreases faster at the tip of the fracture that is invading the higher-stress layers. As
fracture continues closing, the fracture tips (at the high-stress layers and low permeability zones) are forced
to close first. This results in a height recession and requires that a larger volume of fluid leak-off to reach
fracture closure than that expected for a single, planar, constant-height fracture. As the fracture empties, the
rate of leakoff relative to the remaining, stored fluid accelerates, and the pressure declines more rapidly.
Nonetheless, the fracture empties from the overlying and underlying layers, and the rate of leakoff in the
20 SPE-199728-MS

target zone increase. Fig. 23 displays the fracture permeability as it increases at the target layer while fracture
closes in the high-stress overlying and underlying layers. Fig. 24 shows a physical illustration of the height
recession.

Figure 23—IK Plane Distribution of fracture permeability distribution at different

time points starting from the end of injection for Height Recession Scenario.

Figure 24—Physical illustration of Height Recession Scenario

The logarithmic derivative in the log-log diagnostic plot (Fig. 25) shows a steep upward trend, and the
closure time and pressure are picked when the tangent to the derivative has a slope of 3/2. In this case, the
closure time is picked at 14.3 hrs, and the fracture closure pressure is estimated as 6,255 psi, which is in
excellent agreement with the input stress parameters.

Figure 25—Log-log Diagnostic plot for pressure response for Height Recession Scenario (left).
SPE-199728-MS 21

The G-function plot shown in Fig. 26 shows the derivative signature of a belly below the straight line
through the origin and tangent to the semi-log derivative of pw vs. G-time at the point of fracture closure.
The fracture closure occurs at the G-time corresponding to 11 hrs and the corresponding closure pressure is
6,627 psi, which is 400 psi higher than the results obtained from the log-log diagnostic analysis and from
the input stress parameters.

Figure 26—Nolte G-function plot for closure pressure determination for Height Recession Scenario.

Fig. 27a shows the ∆p vs. the square of the linear flow time function, where the linear flow and
pseudoradial flow regime are identified. Fig. 27b presents the cartesian plot of the pressure vs. the radial
flow time function. The vertical intercept of the straight line through the appropriate data in the pseudoradial
flow period gives a pore pressure estimate of 3,300 psi, which is in excellent agreement with the input
reservoir pressure. Using the slope of the straight line, fluid viscosity of 0.313 cp, and the fracture height
of 25 ft, the effective reservoir permeability can be calculated as 0.004 mD from Equation 12. The start and
end of linear flow are estimated as 14 hrs and 20 hrs, respectively and the beginning of the pseudoradial
flow is found to be 26.5 hrs.

Figure 27—After Closure Analysis- Square Linear Flow Time Function (left)
and Radial Flow Time Function Plot (right) for Height Recession Scenario.
22 SPE-199728-MS

Effect of Injection Volume on Observed Flow Regimes

Two crucial design parameters for diagnostic fracture injection tests are the volume of injection and injection
time (Barree 1998 and Barree et al. 2009). Therefore, in this section, we test the effect of the injected volume
and injection time on the closure time as well as the time to reach pseudoradial flow regime for a given
matrix permeability. We used the inputs form Case 1 presented above at a fixed permeability of 0.007 mD
for matrix and fracture. We considered the injection times of 2.4 minutes, 4.8 minutes, and 7.1 minutes,
and the corresponding injection volumes of 20 bbls, 40 bbls, and 60 bbls. Thus, the time to closure is not a
function of the injection volume and insensitive to rate as long as a high enough rate is achieved to reach a
stable fracture propagation and the time to closure is a function of pumping time, not the volume injected.
Fig. 28 presents the log-log diagnostic plot of the logarithmic derivative for the three injected volumes.
Overall, the derivative trends for the three injection volumes are comparable. A ¼-slope behavior is
observed, which indicates the existence of a before closure, early-time, bi-linear flow regime between
the reservoir and hydraulic fractures. This is followed by a 3/2-slope period defining the before closure
region. The end of the 3/2-slope period coincides in all three cases, which indicates the fracture closure at
approximately 11 hours after the shut-in time. After closure, formation linear flow indicated by ½ slope
forms, where fluid flows into the formation in paths normal to the fracture plane. The time to reach linear
flow regime is approximately equal for all cases; nonetheless, the period of the linear flow is dependent
on the injection volumes, and hence, the linear flow period increases as the injection volumes increases.
Accordingly, the pressure transient is observed to reach pseudoradial flow regime at different times that are
dependent on the injection volumes. The time to reach pseudoradial flow regime is approximately equal to
30 hours, 33 hours, and 36 hours for the corresponding injection volumes of 20 bbls, 40 bbls, and 60 bbls.

Figure 28—Log-log Diagnostic plot for three injection volumes.

Nolte (1979) provided approximate equations for the g-function for the two asymptotic values of the
fracture growth exponent α

(15)

where Nolte dimensionless shut-in time is defined as

(16)
SPE-199728-MS 23

where te is the injection time. Fig. 29 indicates that as the injection time increases, the dimensionless shut-
in time decreases, and g-function time decreases for the same time point. Analysis of pseudoradial flow
regimes yields an effective reservoir permeability of 0.005 mD for the three injection volumes considered.

Figure 29—G-function plot for three injection volumes.

Effect of System Permeability on DFIT Design

Barree et al. (2015) used an empirical correlation and concluded that the time for the fracture closure with
stable geometry is controlled by the reservoir permeability. In moderately high permeability formations, the
time required for pressure transient to propagate within the zone of interest is short, and thus, the extended
pumping time maybe not be a serious concern.
Fig. 30 and Fig. 31 present the log-log diagnostic plot and the G-function time plot, respectively. Overall
the derivative trends on the diagnostic plot for all cases are relatively close. From the derivative shapes (Fig.
30), as expected, the early time behavior for all cases is the same since the injected volume is the same and
constant stress parameters are used; hence, the created fracture geometry is identical for all cases.

Figure 30—Log-log Diagnostic plot at different system permeability.

24 SPE-199728-MS

Figure 31—G-function plot at different system permeability.

We introduce the macroscopic permeability ratio, K, which is defined as

(17)

At the transition period, the fluid transfer between matrix and fractures is dominant as the matrix
permeability decreases the macroscopic permeability ratio, κ, increases, and thus a bigger dip is observed
(Chen et al. 1990; Bourdet 2002). At late times, all curves merge into the homogeneous reservoir solution,
and a 3/2-slope behavior is dominant. Fracture closure is then picked at the separation of the 3/2-slope
behavior from the derivative curves.
For higher permeability formations, closure can happen too quickly, and pseudoradial flow can develop
quickly after closure. As shown in Table 5, times for fracture closure and pseudoradial flow increase
significantly when matrix permeability decreases. For a matrix permeability of 0.7 mD and pumping time of
21 minutes, the fracture will reach closure in approximately 3 hrs of shut-in time, and the pressure transient
will reach pseudoradial flow after 13 hrs. At 0.07 mD matrix permeability fracture closure developed after
approximately 5 hrs of shut-in time and the pressure transient will reach pseudoradial flow after 16 hrs. At
0.007 mD matrix permeability, fracture closure occurs after 10 hours of the shut-in time and the pressure
transient will reach pseudoradial flow after approximately 30 hrs. While at 0.0007 mD matrix permeability,
fracture closure occurs after 12 hrs and pseudoradial flow after 33 hrs. At ultra-low permeability of 0.00007
mD (70 nD), fracture closure occurs after 13 hours of the shut-in time, and the pressure transient will reach
pseudoradial flow after approximately 38 hrs. At ultra-low matrix permeability of 7nD, fracture closure
occurs after 22 hrs, and the pseudoradial flow develops after 40 hr following a long linear flow period.

Table 5—Time needed to reach fracture closure, linear and pseudoradial flow at
different matrix permeability, and fixed hydraulic fracture permeability = 1000 mD.

Matrix Perm Time to Frac Time to End linear Time to

(mD) Closure (hrs) Linear (hrs) (hrs) Radial (hrs)

0.000007 22 15 23.3 40
0.00007 13 14 22 38
0.0007 12 13.8 21 33
0.007 10 13.5 19 30
0.07 5 6.5 11 16
0.7 3 5.8 8 13
SPE-199728-MS 25

Conclusions
In this work, we presented a fully coupled reservoir geomechanics and flow model that can successfully
simulate the fluid flow and matrix deformation in the presence of secondary fractures and complex stresses.
The following observations and conclusions are warranted based on the discussions of this paper:

• The coupled geomechanics and flow model is sensitive to the intensity of natural fractures.

• The model results agree with the minifrac theory and the model could produce the flow regimes
after shut-in considered in standard type curves.
• The time to closure is not a function of the injection volume and it is insensitive to the injection rate.
However, the time needed for the pressure transient to reach pseudoradial flow regime increases
as the injection volume increase.
• It is strongly recommended not to use an injection volume in excess of 60 bbls in a reservoir with
nanodarcy range permeability as it might delay the time needed to reach pseudoradial flow and
hampers the ability to accurately calculate reservoir properties.
• The coupled geomechanical and flow model is capable of computing fracture dimensions, allowing
us to numerically simulate DFITs in complex scenarios which is not possible with for most
commercial (analytical) packages.
• The coupled geomechanical and flow model allows extracting time depended fracture properties.

• Currently, the model is in a single-phase model. However, it can be extended to simulate multiphase
flow.

Acknowledgments
We very much appreciate the endorsement and the support of this work by UREP (Unconventional Reservoir
Engineering Project) at Colorado School of Mines. We would like to thank Apache Corporation and
Computer Modeling Group, for their aid in conducting this study.

References
Bachman, Robert Clayton, Walters, Dale A, Hawkes, Robert et al 2012. Reappraisal of the G time concept in mini-frac
analysis (includes associated discussion). Proc., SPE Annual Technical Conference and Exhibition.
Barree, RD. 1998. Applications of pre-frac injection/falloff tests in fissured reservoirs-field examples. Proc., SPE Rocky
Mountain Regional/Low-Permeability Reservoirs Symposium.
Barree, Robert D, Barree, Valencia L, and Craig, David. 2009. Holistic fracture diagnostics: consistent interpretation of
prefrac injection tests using multiple analysis methods. SPE Production & Operations 24 (03): 396-406.
Barton, Nick, Bandis, S, and Bakhtar, K. 1985. Strength, deformation and conductivity coupling of rock joints. Proc.,
International journal of rock mechanics and mining sciences & geomechanics abstracts 3, 121-140.
Bourdet, Dominique. 2002. Well test analysis: the use of advanced interpretation models, Vol. 3: Elsevier.
Bourdet, Dominique, Ayoub, JA, and Pirard, YM. 1989. Use of pressure derivative in well test interpretation. SPE
Formation Evaluation 4 (02): 293-302.
Chen, Her-Yuan, Poston, SW, and Raghavan, R. 1990. The well response in a naturally fractured reservoir: Arbitrary
fracture connectivity and unsteady fluid transfer. Proc., SPE Annual Technical Conference and Exhibition.
Cinco-Ley, H and Meng, H-Z. 1988. Pressure transient analysis of wells with finite conductivity vertical fractures in
double porosity reservoirs. Proc., SPE Annual Technical Conference and Exhibition.
Cinco, Ley, Samaniego, V, and Dominguez, A. 1978. Transient pressure behavior for a well with a finite-conductivity
vertical fracture. Society of Petroleum Engineers Journal 18 (04): 253-264.
Craig, David P and Blasingame, Thomas Alwin. 2006. Application of a new fracture-injection/falloff model accounting
for propagating, dilated, and closing hydraulic fractures. Proc., SPE Gas Technology Symposium.
Dean, Rick H, Gai, Xiuli, Stone, Charles M et al 2006. A comparison of techniques for coupling porous flow and
geomechanics. Spe Journal 11 (01): 132-140.
26 SPE-199728-MS

Ibrahim Mohamed, Mohamed, Ibrahim, Ahmed Farid, Ibrahim, Mazher et al 2019. Determination of ISIP of Non-Ideal
Behavior During Diagnostic Fracture Injection Tests. Proc., SPE Annual Technical Conference and Exhibition.
Mohamed, Mohamed Ibrahim, Salah, Mohamed, Coskuner, Yakup et al 2019. Investigation of Non-Ideal Diagnostic
Fracture Injection Tests Behavior in Unconventional Reservoirs. Proc., SPE Hydraulic Fracturing Technology
Conference and Exhibition.
Tran, D, Nghiem, L, and Buchanan, L. 2005. Improved iterative coupling of geomechanics with reservoir simulation.
Proc., SPE Reservoir Simulation Symposium.
Tran, David, Nghiem, Long, and Buchanan, Lloyd. 2009. Aspects of coupling between petroleum reservoir flow and
geomechanics. Proc., 43rd US Rock Mechanics Symposium & 4th US-Canada Rock Mechanics Symposium.
Warpinski, Norman R. 1991. Hydraulic fracturing in tight, fissured media. Journal of Petroleum Technology 43 (02):
146-209.