DC187191 DOI: 10.

2118/187191-PA Date: 19-November-19 Stage: Page: 345 Total Pages: 11

Consistent Model for Injection and Falloff

Pressure Match of Diagnostic Fracture
Injection Tests (DFITs)
G. Liu and C. Ehlig-Economides, University of Houston

Summary
The diagnostic fracture injection test (DFIT), a fracture-injection/falloff test, is a reliable tool for quantifying the formation of closure
stress, leakoff coefficient, formation permeability, and pressure. The current analytical DFIT model (used before and after closure)
enables one to match the pressure falloff of abnormal leakoff behaviors, and quantifies more formation parameters than a traditional
DFIT model. However, this model design addresses only the falloff data after shut-in; thus, analysts have expressed concerns that the
net pressure implied by the falloff is inconsistent with the injection pressure behavior. Therefore, this paper provides a model capable
of matching both injection and falloff pressure behaviors.
The pressure-falloff model is capable of quantifying essential pressure values including, in order of occurrence, instantaneous shut-
in pressure (ISIP), minimum fracture-propagation pressure, one or more closure-stress values, formation minimum principal stress, and
pore pressure. The early pressure response represents the dissipation of three kinds of friction—wellbore, perforation, and near-
wellbore friction. Each of them is quantified, and, together, they comprise the difference between the pressure at the end of injection
and the ISIP. Presence of tip extension enables the quantification of the minimum fracture-propagation pressure. The minimum princi-
pal stress is consistent with the final closure stress. Subtracting the closure stress and friction pressure losses from the recorded or calcu-
lated bottomhole pressure (BHP) provides the fracture net pressure. The model match for injection pressure behavior incorporates the
same pressures and consistent values for 2D fracture geometry and leakoff coefficient.
The global match confirms not only the estimation of formation and fracture properties from the pressure-falloff analysis, but also the
friction losses along the wellbore, through the perforations, and in the near-wellbore tortuosity during and after injection. In particular, by
matching both injection and falloff, the model incorporates friction pressure losses that can explain apparent excessive net pressure.
The match with both injection and falloff pressure variation addresses concerns that the net pressure implied by the falloff model
match cannot be consistent with the observed injection behavior. When the pressure difference between the final DFIT pick for closure
stress and the pressure at the end of injection is large, the reason might be tip extension and/or large friction pressure losses, the latter
of which can be addressed by the main treatment design.

Introduction
For a DFIT, a small amount of fluid is pumped into the formation to create a fracture in the target layer. Then, the well is shut in to
allow the injected fluid to leak off into the formation (Feng and Gray 2016). Previous DFIT models, Nolte (1979, 1986), Mayerhofer
et al. (1995), and Craig and Blasingame (2005, 2006), did not account for abnormal leakoff behaviors, as commonly observed in the
falloff data and flow friction losses. Therefore, these models could not be used to match the injection data. Liu and Ehlig-Economides
(2015, 2019) described available models for the DFIT analysis. These models were designed to match the entire falloff, including
abnormal leakoff behaviors such as perforation and near-wellbore friction, tip extension, pressure-dependent leakoff area, height reces-
sion, variable compliance, and multiple closure events. The model for the falloff data also matched the fracture propagation during
pumping. The focus of this paper was to match the injection pressure during pumping with the model that matches the falloff behavior.
Three 2D fracture models widely used in hydraulic-fracturing analysis and design are the PKN (Perkins and Kern 1961; Nordgren
1972), the KGD (Zheltov 1955; Geertsma and de Klerk 1969), and the GDK radial (Geertsma and de Klerk 1969). All three models assume
that the formation rock is continuous and homogeneous, and that it follows the linear elastic relationship between the stress expressed as
net pressure and the strain expressed as half of the fracture width. Fracture width is a key parameter because it is the bridge connecting pres-
sure change and material balance. Mack and Warpinski (2000) explained that the net pressure at the tip of all three models is assumed to be
zero. This implies that the fracture shape is elliptical along the length. Therefore, fracture net pressure and fracture width vary, with maxi-
mum values at the wellbore and zeros at the tip. Nolte (1979, 1986) provided a factor, b, to describe the ratio between the average fracture
net pressure and the maximum fracture net pressure at the wellbore. For a Newtonian fluid, the b-factors are 2=3 for the injection period of
all the 2D fracture models, and 4=5, 0.9, and 3p2/32 for the PKN, KGD, and GDK radial models, respectively, after shut-in.
Besides the nonuniform pressure distribution along the fracture, all three models assume constant-rate injection. However, field data
frequently show that the surface injection rate is not constant during the test. Some operators deliberately vary the injection rate before
the end of pumping with a step-rate test, also known as a step-down test, designed to measure friction pressure loss under different
injection rates (Barree et al. 2014; Naidu et al. 2015). Barree et al. (2014) suggested that each rate step last for 10 to 15 seconds.
The commonly used 2D fracture models also assume that fluid leakoff is negligible during pumping. The PKN model used by
Nordgren (1972) obtained analytical solutions for two limiting cases—one with a large leakoff rate and the other with zero. Typically,
the analytical PKN solution with zero leakoff is preferred, especially in tight formations. For the other two models, the KGD and GDK
radial fracture models, fluid leakoff was neglected. At the end, the fracture surface area changed or, correspondingly, the fracture propa-
gation rate was approximately expressed in a power law function given by Nolte (1986) as:
 a
Af t
¼ ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð1Þ
Af 0 tp

Copyright V
C 2019 Society of Petroleum Engineers

This paper (SPE 187191) was accepted for presentation at the SPE Annual Technical Conference and Exhibition, San Antonio, Texas, USA, 9–11 October 2017, and revised for publication.
Original manuscript received for review 21 June 2017. Revised manuscript received for review 31 January 2019. Paper peer approved 7 February 2019.

December 2019 SPE Drilling & Completion 345

ID: jaganm Time: 16:38 I Path: S:/DC##/Vol00000/190010/Comp/APPFile/SA-DC##190010

 DC187191 DOI: 10.2118/187191-PA Date: 19-November-19 Stage: Page: 346 Total Pages: 11

where Af is the surface area of the one-sided fracture at any time t during injection; Af 0 is the Af at shut-in, or at t ¼ tp , and tp is the
injection time; and a is the area exponent. According to Nolte et al. (1993), the area exponent is close to unity for low leakoff, and
1/ 2 for high leakoff. In particular, for a Newtonian fluid with no leakoff, the upper-bound area exponents for the PKN, KGD, and GDK

radial fracture models are 4=5, 2=3, and 8=9, respectively (Nolte 1986). Although Eq. 1 is derived from 2D fracture models that originally
assumed zero leakoff, Nolte (1986) relaxed the requirement and added fluid leakoff during pumping with the G-function. Furthermore,
Nolte et al. (1993) suggested calculating the area exponent in the nonzero leakoff situation as

a ¼ 0:5 þ gða1  0:5Þ: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð2Þ

The parameter g in Eq. 2 is the fluid efficiency, and a1 is the upper-bound area exponent—4=5, 2=3, and 8=9 for the PKN, KGD, and GDK
radial fracture models, respectively.
In this paper, we constructed a consistent model that matched the injection profile using the parameters quantified from the falloff
data. We identified inconsistencies between the injection and falloff modeling and explained how to resolve them. Finally, the model
was applied on a field case to demonstrate that a single model was capable of matching both the injection and falloff data and to illus-
trate the clear distinctions between net pressure and friction pressure losses, both of which provided information for the subsequent
main-treatment design.

Models To Connect Injection With Falloff Modeling

As two successive processes, the injection and falloff should be consistent in the following parameters: fracture geometry at shut-in, for-
mation leakoff coefficient, natural-fracture surface area if its effect is present, closure stress, and rate-dependent friction. These param-
eters were quantified by means of the falloff analysis, and were used as input parameters in the model to fit the injection pressure.
Although the injection and falloff occurred in the same fracture and formation setting, the pressure response and fracture behavior
during these two periods was not exactly the same. In this study, we investigated the gap between pressure-falloff modeling and
fracture-propagation modeling, and then built a general model that worked for both.

Wellbore Storage and Variable Bottomhole Injection Rate. In modeling fracture initiation and propagation, the injection rate refers
to the bottomhole (BH) rate as it enters the created fracture, not the recorded rate at the wellhead. Frequently, the DFIT injection rate
varies. Some operators include a step-rate test, also called a step-down test, to estimate the relationship between the friction pressure
loss and the injection rate (Barree et al. 2014). In such cases, the wellbore-storage (WBS) effect distorts changes in the BH injection
rate when compared with the step-rate changes at the surface. Estimation of the BH rate and the total pumping volume to create a
hydraulic fracture requires consideration of the WBS effect (Cramer and Nguyen 2013; Barree et al. 2014; Ehlig-Economides and Liu
2017). In this subsection, we demonstrate how to calculate the BH rate by considering the WBS, and then explain its impact on the fall-
off analysis and hydraulic-fracture modeling.
The volume change of the fluid storage in the wellbore (DVfluid) with pressure was written as

DVfluid ¼ cw Vw Dpw ðtÞ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð3Þ

where cw is the water compressibility, Vw is the total wellbore volume at the start of the injection, and Dpw ðtÞ is the treating-
pressure change.
As suggested by Barree et al. (2014), besides fluid volume change, the volume change of the injection string was also considered,
and its diameter change with the pressure was described as

Dpðd  dÞ2
Dd ¼ ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð4Þ
2dE
where Dp is the internal pressure change, d is the string outer diameter (OD) and Dd is its change, d is the string wall thickness, and
E is the elastic modulus of steel. Then, the volume change with pressure caused by string expansion, DVstring , was approximately
derived as
d
DVstring  Vw Dpw ðtÞ: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð5Þ
dE
The total volume change caused by WBS (DVWBS ) was then determined as
 
d
DVWBS ¼ DVstring þ DVfluid ¼ þ cw Vw Dpw ðtÞ: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð6Þ
dE

 in the American Petroleum Institute (API) specifications (API SPEC 5CT 2011) with an OD between 5 and 8.5 in.,
For a casinglisted
d
the ratio of to cw lies between 12 and 31%, by assuming E ¼ 3107 psi and cw ¼ 3106 psi1. This means the string volume
dE
change is small in comparison to the fluid volume change because of the large magnitude of the steel modulus, E. These two factors
were combined in Eq. 7 as one single effective WBS compressibility (cWBS ) that was 11–24% larger than the water compressibility for
API casings with ODs ranging from 5 to 8.5 in.
d
cWBS ¼ þ cw : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð7Þ
dE
We suggest calculating the BH rate by considering the compressibility of fluid and injection string with Eq. 8:
X X
qBH;i Dti ¼ qsurf;i Dti  cWBS Vw ½pw ðti Þ  pw;start ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð8Þ

where qsurf;i and qBH;i are surface and BH rates during a time interval of Dti , pw ðti Þ is the treating pressure, and pw;start is the pressure
level at the start of injection, or pw;start ¼ pw ðt ¼ 0Þ.

346 December 2019 SPE Drilling & Completion

ID: jaganm Time: 16:38 I Path: S:/DC##/Vol00000/190010/Comp/APPFile/SA-DC##190010

 DC187191 DOI: 10.2118/187191-PA Date: 19-November-19 Stage: Page: 347 Total Pages: 11

If there is no pre-existing fracture, all the initially injected fluid pressurizes the wellbore with minimal flowing out of the wellbore
until the formation breaks down. During this period, the pressure rises linearly with the injection volume, and the injection-rate time
remains constant. This feature was used to estimate the effective WBS compressibility defined in Eq. 7.
By considering the WBS effect, we concluded that there are two possible ways to obtain a valid relationship between the friction
loss and the injection rate from a step-rate test. The first is to maintain each rate step for an adequate time period to observe a stable
pressure level so that the BH rate is similar to the rate at the surface. (The time period might be longer than 15 seconds.) The second
approach is to correct the BH rate with Eq. 8 so that the friction loss can be estimated using the corrected BH-rate values. Liu and
Ehlig-Economides (2019) quantified friction losses and the ISIP by matching pressure-falloff data points after shut-in. When a step-
down test is performed, it should confirm the falloff model. Therefore, matching the injection and falloff pressures could eliminate the
need for a step-down test.

Fracture Propagation with Variable BH Injection Rate. Two important parameters for the DFIT analysis are the injection time and
the total injection volume. Injection volume is the basis of the material balance that must be strictly obeyed for the DFIT analysis (Liu
and Ehlig-Economides 2018). Variation in the injection rate has little explicit impact on the pressure-falloff behavior, and it is usually
assumed to be constant.
Yet, the injection rate has a significant impact on the pressure response during pumping. First, friction losses are rate-dependent and,
at high rate, friction losses often account for much of the total pressure increases during pumping. As a result, the pressure and the rate
behavior are highly correlated. Second, increasing the pump rate increases the BH rate as it enters the created fracture, which increases
the net pressure in the fracture and should accelerate fracture propagation. However, the commonly used fracture-propagation model
described in Eq. 1, which was derived by assuming a constant injection rate, implies that the fracture-propagation rate is independent of
the injection rate and is dependent only upon the time. To honor the rate effect and material balance, we modified Eq. 1 by replacing
time on the right-hand side with the injection volume, giving
X !a
Af ;i qBH;i Dti
¼ ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð9Þ
Af 0 VBH;inj

where Af ;i represents the time-dependent fracture area during injection for the time interval Dti . The BH flow rate, qBH;i (in Eq. 9), was
calculated by solving

ðqsurf;i  qBH;i ÞDti ¼ cWBS Vw Dpsurf;i ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð10Þ

where qsurf;i and psurf;i are the surface-recorded injection rate and pressure for the time interval Dti , respectively. The total effective
injection volume into the fracture before the shut-in, VBH;inj (in Eq. 9), was calculated as

VBH;inj ¼ Vsurf;inj  cWBS Vw ðpws  pw;start Þ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð11Þ

where Vsurf;inj is the total volume injected at surface and pws is the BH pressure at shut-in, which is not the same as the ISIP. The relation
between the pressure before the start of injection and the shut-in pressure depends upon operating procedures. For example, after plug-
ging and perforating, the wellbore pressure might be low before the start of injection. In such cases, there might be a sharp linear rise in
pressure associated with pressurizing the wellbore. The total fluid volume flowing into the fracture at shut-in is less than the surface
injection volume because part of it is used to pressurize the wellbore. In contrast, we have seen field data in which the hydraulic pres-
sure required to open a sliding sleeve exceeded the shut-in pressure, creating the possibility that the total fluid flowing into the fracture
at shut-in might be greater than the reported surface volume because of the fluid expansion. Such operating procedures must be consid-
ered in the DFIT design to ensure that the volume injected into the created fracture is according to the design.

Fracture Internal Pressure Distribution. To account for the nonuniform distribution of the fracture net pressure along the fracture,
Nolte (1979, 1986) provided the b-factor with different, but constant, values during the injection and after the shut-in, respectively. It is
a convenient and simple method to relate the maximum fracture net pressure at the wellbore to the average fracture net pressure, or to
relate the calibrated BH pressure to the fracture volume. The b-factor was then used in their falloff analysis (Nolte 1979, 1986). How-
ever, that approach was not sufficient to match the field data because two constant b-factor values, before and after shut-in, cannot
describe the varying BH flow rate and the “long-lasting after flow” after shut-in. To construct a consistent model that included both
injection and falloff, we interpolated for the varying injection rate between the two constant b-factors as
q
bðqÞ ¼ ðbp  bs Þ þ bs ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð12Þ
qBH;inj

where bp is the b-factor during pumping (equal to 2=3 for all fracture models), bs is the b-factor after shut-in (equal to 4=5 for the PKN,
0.9 for the KGD, and 3p2/32 for the GDK radial fracture model) (Nolte 1979, 1986), q is the BH flow rate (calculated with Eq. 8), and
qBH;inj is the average BH injection rate (obtained by dividing the total BH injection volume defined in Eq. 11 by the injection time).

Leakoff During Fracture Propagation. Although Nolte et al. (1993) used Eq. 2 to calculate the area exponent for fluid loss, the equa-
tion had little bearing on falloff analysis other than calculating the G-function of time to generate the G-function diagnostic plot.
Marongiu-Porcu et al. (2014) showed that, in reality, the G-function is not highly sensitive to the value of the area exponent. However,
the area exponent calculated with Eq. 2 is required for injection modeling, and the fluid efficiency quantified from the falloff before-
closure (BC) analysis determined the estimation of this value.

Pressure Calculation During Injection. In this subsection, we used the model parameters from the falloff match to calculate the pres-
sure at each time point during the injection. The input data were as follows: (1) pump rate and time recorded at the surface; (2) wellbore
volume, effective WBS compressibility calculated from the injection curve (cWBS defined in Eq. 7); and (3) parameters quantified from

December 2019 SPE Drilling & Completion 347

ID: jaganm Time: 16:38 I Path: S:/DC##/Vol00000/190010/Comp/APPFile/SA-DC##190010

 DC187191 DOI: 10.2118/187191-PA Date: 19-November-19 Stage: Page: 348 Total Pages: 11

the BC analysis, including surface areas of the hydraulic fracture and secondary fractures at shut-in, friction constants in the friction-
rate function (Liu and Ehlig-Economides 2019), leakoff coefficient, and fluid efficiency.
The first step required the area exponent, estimated from Eq. 2, as the initial guess to calculate the fracture area (Af,i) at each time
point using Eq. 9. The G-function model (Nolte 1979) was used to calculate the leakoff volume (Vl,i) through each area. Using Eq. 10
to calculate the BH rate (qBH;i ) for each time point, we quantified the friction losses in the perforation and the near-wellbore tortuosity,
respectively. As for the BC modeling, the pressure response during hydraulic-fracture propagation was governed by the material-
balance function given as

qBH;i Dti ¼ n½bðqBH;i Þ cf ;i ðpnet;i  pc ÞAf ;i  bðqBH;i1 Þ cf ;i1 ðpnet;i1  pc ÞAf ;i1  þ ðVl;i  Vl;i1 Þ ; . . . . . . . . . . . . . . . . . . . ð13Þ

where n is the unit conversion coefficient from cubic feet to barrels, the value cf ;i is the fracture compliance at the ith time point calcu-
lated from the fracture geometry (Nolte 1986), and pc is the final fracture closure stress from the injection falloff model. Rearranging
Eq. 13 calculated the value of pnet;i , the net pressure at the wellbore for the time interval Dti . Finally, summing the net pressure and the
friction losses provided the BH pressure. We then iterated on the area exponent value to achieve a match between the BH-pressure
values computed from the recorded surface pressure data and the BH pressure computed from the model.

Case Study
Liu and Ehlig-Economides (2015, 2016, 2019) conducted the BC analysis of a field case with multiple apparent closures. In this paper,
we reinterpreted the BC data and the after-closure (AC) data, and then generated a single consistent model to match both the injection
and falloff pressure data. Approximately 31.45 bbl of freshwater was injected into the well in 7.13 minutes, with an average rate of
4.41 bbl/min, as shown in Fig. 1. It should be pointed out that the rate profile is missing. To ensure that the injection rate was constant
during injection at a given value, we divided the total injected volume by the injection time. However, rates of a few early points were
calculated from the pressure curve because the injected fluid pressurized the wellbore during that time. The primary input parameters
are listed in Table 1.

12,000 5

Wellhead Injection Rate (bbl/min)

10,000
4

8,000
BHP (psi)

6,000
2
4,000 BHP
Wellhead injection rate
1
2,000

0 0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Δt (hours)

Fig. 1—Injection profile.

Parameters Values

Injection time (minutes) 7.13

Surface injection volume (bbl) 31.45

Fracture model GDK radial

6
Formation plane-strain modulus (psi) 6×10

Wellbore length (ft) 15,236

Wellbore volume (bbl) 338

True vertical depth (ft) 9,265

BH temperature (°F) 234

Formation thickness (ft) 360

pw,start (psi) 3,670

Formation-fluid viscosity (cp) 0.027

Porosity 0.058
–4
Total compressibility (1/psi) 1.06×10

Table 1—Input parameters.

348 December 2019 SPE Drilling & Completion

ID: jaganm Time: 16:38 I Path: S:/DC##/Vol00000/190010/Comp/APPFile/SA-DC##190010

 DC187191 DOI: 10.2118/187191-PA Date: 19-November-19 Stage: Page: 349 Total Pages: 11

The fluid was pumped into the formation through a casing with an inner diameter of 4.78 in. The measured depth of the perforation
was 15,236 ft. From this information, the wellbore volume was estimated at 338 bbl, which was greater than the injected-fluid volume
and sufficiently large to warrant that the WBS effect cannot be neglected. Indeed, the WBS effect is apparent from the straight line in
Fig. 1, marked by the solid green line during early injection. Assuming that the entire injected volume was stored in the wellbore, Eq. 6
estimated the WBS compressibility, defined by Eq. 7, at 2.92106 psi, which corresponded to the water and pipe compressibility at
the test temperature and pressure condition, and the observed straight-line behavior also confirmed that the flow rate was constant.
The diagnostic plots were graphed in the Bourdet derivative in a log-log plot (Bourdet et al. 1983) and the composite G-function
plot. Mohamed et al. (2011) provided the definition of the superposition time function used to create the Bourdet derivative plot. Nolte
(1979) derived the G-function of time to account for the fluid leakoff during injection. (Note, the G-function plot is designed to show
the fracture-closure trend, from which closure stress can be picked, and can also reveal potential abnormal leakoff behaviors, such as
pressure-dependent leakoff, tip extension, height recession, or transverse storage.) Besides the fracture-closure trend and abnormal leak-
off behaviors, the Bourdet derivative log-log plot also presented the early WBS effect and the AC pseudolinear and pseudoradial flows.
These data showed a sequence of both BC and AC behaviors that determined which features a model for this behavior must include.
Fig. 2 shows the analysis presented previously by Liu and Ehlig-Economides (2015, 2016, 2019). The left graph in Fig. 2 is
the Bourdet derivative in a log-log plot with markings that indicate slope trends of interest, and the right graph shows the composite
G-function plot that displays the exact same closure trends. The three dashed lines in the bottom right corner of the Bourdet derivative
plot are guidelines and represent the three slopes used in the plot: The brown line indicates the 1/ 2 slope, the green line indicates the unit
slope, and the red line indicates the 3=2 slope. Briefly, the first behavior, indicated by the green unit slope (marked as Circle 1) in the
Bourdet log-log plot, was identified as the WBS with friction dissipation. The next derivative transition (Circle 2) related to the tip
extension after shut-in, coupled with the near-wellbore friction dissipation. The end of the tip extension was denoted as the start of the
first closure trend, at approximately 0.05 hours after shut-in. The minimum fracture-propagation pressure, denoted at the end of the tip
extension, was 6,640 psi. Then, the unit slope, ending near Dt¼tp (Circle 3), was identified as the first constant area closure trend. The
following four 3=2-slopes (Circles 4 through 7) were successive additional closure trends, producing five successive closures in total.
Mohamed et al. (2011) showed a similar multiple-closure field case from a different formation.

1×1012 10,000 1,000

Δm(p) BHP
m(p)′ Gdp/dG

Dp/dG and Gdp/dG (psi)

dp/dG
Δm(p) and m(p)′ (psi2/cp)

1×1011 End of constant area (eca) 7 8 9 8,000 800

Fracture closure End of constant area (eca)
Hydraulic-fracture closure 6 Fracture closure
Hydraulic-fracture closure
BHP (psi)

1×1010 5 6,000 600

tp
4
1×109 4,000 400
2

1×108 3 2,000 200

1

1×107 0 0
0.0001 0.001 0.01 0.1 1 10 100 1,000 0 5 10 15 20 25 30
Δt (hours) G(ΔtD)

Fig. 2—Bourdet derivative plot (left) and composite G-function plot (right).

The two unit-slope trends that appeared in this example represent two different behaviors, the first being WBS and the second being
a constant leakoff area closure trend ending at approximately the pumping time tp, after shut-in. Works by Bachman et al. (2012) and
Marongiu-Porcu et al. (2014) showed that closure trends before the time tp had a unit slope and had a 3=2 slope after 10tp.
Liu and Ehlig-Economides (2015, 2016) previously quantified a different leakoff coefficient from each apparent closure trend. We
assumed all the leakoff trends related to the same formation and, therefore, were governed by one constant leakoff coefficient. Different
leakoff rates, indicated by the unit and 3=2 slopes in the Bourdet derivative, or extrapolated Gdp/dG slopes in the G-function plot, were
caused by different leakoff areas. The leakoff area decreased from the highest value it associated with the first closure trend to the
lowest value with the last final closure.
A possible interpretation for the closure trends is that each represented a different secondary fracture set, and a possible explanation
for the distinct closing pressures might relate to the angle of the fracture planes in a given fracture set. Table 2 lists the interpretation
parameter values for the final model match. Fig. 3 shows the leakoff area for each closure trend associated with the pressure at the end
of the constant area leakoff. By subtraction, each closure trend had an associated leakoff area. The red curve in Fig. 3 shows that the
area associated with the first closure was extremely large. The area closed in 0.47 hours after shut-in, according to the Bourdet deriva-
tive plot in Fig. 2.
Fracture surface areas, shown by the red line in Fig. 3, were calculated by assuming a constant leakoff, because all leakoff occurs
into the same formation. An alternative explanation for some of the closure events might be hydraulic-fracture closures in layers sur-
rounding the tested layer. The tested layer was thin and had a likely fracture barrier below, but no effective fracture barrier to the over-
lying layers. One of the overlying formations also had a DFIT test that showed a minimum stress estimated at approximately 5,700 psi
and a leakoff coefficient up to 6.8 times the value estimated from the final closure for this test. The minimum stress in the overlying
zone was similar to the first closure stress value in this DFIT. Therefore, the first closure might be related to the overlying zone, and the
higher leakoff rate could be explained by a higher leakoff coefficient through a smaller apparent leakoff area. The dashed blue curve in
Fig. 3 shows a leakoff area corresponding to the possibility that the first constant area leakoff occurred in the overlying layer with a
higher leakoff coefficient. The two trends overlaid each other after the end of the first closure event.
This possibility underscores the importance of the closure stress magnitude associated with each leakoff event. Stress estimates cor-
responding to known magnitudes, such as the vertical stress or the minimum stress levels observed in layers near the tested layer, sug-
gest potential layering behavior. Constant area leakoff ending at a known value for overburden stress might relate to vertical leakoff
along a horizontal plane between adjacent beds. Constant area leakoff ending at a known minimum stress level for a particular forma-
tion suggests that the vertical fracture height grew into that formation.

December 2019 SPE Drilling & Completion 349

ID: jaganm Time: 16:38 I Path: S:/DC##/Vol00000/190010/Comp/APPFile/SA-DC##190010

 DC187191 DOI: 10.2118/187191-PA Date: 19-November-19 Stage: Page: 350 Total Pages: 11

Fracture and Pressure Behaviors in Chronological Order

WBS
Injection and and Tip Linear Extrapolated
Falloff Feature Injection Friction Extension Closure 1 Closure 2 Closure 3 Closure 4 Closure 5 Flow Radial Flow

c WBS (psi) 2.9×10

–6

VBH,inj (bbl) 25.6

α 0.67

Vl (bbl) 9.6

Δpfric,perf &wb (psi) 370

Δpfric,tort (psi) 2,002

Rf,initial (ft) 38.9

pp, min (psi) 6,640

Rf, final (ft) 39.1

peca (psi) 6,119 5,474 5,358 5,121

pc (psi) 5,588 5,427 5,291 4,972 4,784

Total area of
one-sided fracture 124,956 19,056 12,839 10,811 4,791
2
(ft )

A f (ft )
2
105,900 6,217 2,028 6,020 4,791

C L ( ft / t ) 5.8×10
–5

ŋ 0.625

k (nd) 200

pi (psi) 4,065

Table 2—Interpretation results.

140,000
128,005 Constant leakoff coefficient for all fractures
Total Surface Area of One-Sided Fracture (ft2)

Larger leakoff coefficient for the first fracture closure

120,000

100,000

80,000

60,000

40,000
34,942
18,928
20,000
12,711 10,886
4,866
0
6,700 6,500 6,300 6,100 5,900 5,700 5,500 5,300 5,100 4,900 4,700
BHP (psi)

Fig. 3—Decline of total surface area of one-sided fracture with pressure during falloff.

350 December 2019 SPE Drilling & Completion

ID: jaganm Time: 16:38 I Path: S:/DC##/Vol00000/190010/Comp/APPFile/SA-DC##190010

 DC187191 DOI: 10.2118/187191-PA Date: 19-November-19 Stage: Page: 351 Total Pages: 11

The apparent 1/ 2 slope (circle 8 in Fig. 2) in the Bourdet log-log plot observed after the final closure is commonly interpreted as the
formation pseudolinear flow (Mohamed et al. 2011; Bachman et al. 2012). From this behavior, we estimated the formation permeability
to be approximately at 200 nd, using the method introduced by Liu et al. (2016), and the formation pressure at 4,065 psi. In Fig. 2, the
derivative trend seems to bend toward flat at the very end and is marked (Circle 9) as a possible pseudoradial-flow regime. However,
the permeability estimated using this trend was 1,200 nd and differed from the permeability estimated from the pseudolinear trend
(Circle 8) using the fracture half-length quantified from the BC analysis.
The permeability could be estimated from either the apparent pseudolinear trend or the apparent pseudoradial trend, but not both.
However, in this case, the AC data could not be matched using the permeability estimated from the apparent pseudoradial flow, as
shown by the dashed curve in Fig. 4. Further, Xue and Ehlig-Economides (2013) explained the relationship between permeability
and formation-pressure estimates. The permeability estimate from the apparent pseudoradial trend produced a formation pressure of
4,472 psi. Because this formation pressure was inconsistent with the independent external data, we reported that the permeability and
formation pressure were consistent with the match, shown as a solid curve in Fig. 4. The match with injection and BC falloff data was
not impacted by the permeability estimate.
Figs. 4 and 5 show matches for both falloff and injection models achieved through slight adjustments in the friction and area-
exponent estimates with only the falloff data. From our experience, the error between the quantified parameter from the falloff data and
those used to match the pumping and falloff pressure curve is less than 10% in most cases. Several interesting points merit discussion.
First, the total fluid volume flowing into the formation before shut-in was 25.6 bbl, which was almost 6 bbl less than the surface injected
volume because of the WBS. Further, 9.6 bbl out of the injected 25.6 bbl leaked off into the formation during the injection; therefore,
the fluid efficiency during injection was relatively low (62.5%), even though the well was drilled in shale.

1×1012
m(p)′, data Δm(p), data 10,000 1,400
Δm(p), model m(p)′, model BHP, data
Δm(p) and m(p)′ (psi2/cp)

1×1011 Δm(p) with high m(p)′ with high BHP, model 1,200
permeability permeability 8,000
Gdp/dG, data
Gdp/dG, model 1,000
tp

Gdp/dG (psi)
1×1010

BHP (psi)
6,000
800

1×109 4,000 600

400
1×108 2,000
200

1×107 0 0
1×10–4 1×10–3 1×10–2 1×10–1 1×100 1×101 1×102 1×103 1×104 1×105 0 10 20 30 40 50 60 70 80
Δt (hours) G(ΔtD)

Fig. 4—Model match for falloff pressure behavior.

12,000

BHP, data
10,000
BHP, model

8,000
BHP (psi)

6,000

4,000

2,000

0
0 0.2 0.4 0.6 0.8 1
Real Time (hours)

Fig. 5—Model match for injection and falloff pressure.

Fig. 6a presents the comparison between the surface injection rate and the BH flow rate at the formation face. At a very early injec-
tion time, the injected fluid only pressurizes the wellbore with no fracture initiation. After breakdown, the BH rate rises quickly to the
wellhead-rate level. The surface rate drops to zero immediately after shut-in, while the BH rate continues flowing at a low rate for some
time, as a result of the afterflow effect. The rate-dependent b-factor, b(q), defined in Eq. 12, varies between bp ¼ 2=3 and bs ¼ 3p2/32.
Fig. 6b shows the fracture internal pressure (red line); BH-pressure data (orange dots); friction loss occurring in the wellbore, the perfo-
ration, and the near-wellbore tortuosity (blue line); fracture net pressure at the wellbore (purple line), which is the maximum net pres-
sure along the fracture because of the fracture’s internal pressure gradient; and the average net pressure (green line).
In this example, the difference between the pressure at the end of pumping and the minimum closure stress was 4,953 psi and must
not be confused with the fracture net pressure, which was 2,032 psi, as indicated by the green curve in Fig. 6b at the end of pumping.
The excess pressure difference seen during pumping was caused by the sum of wellbore and perforation, tortuosity, and internal fracture
pressure losses of 370 psi, 2,002 psi, and the difference between the fracture internal maximum (purple line in Fig. 6b) and average
(green line in Fig. 6b) pressures during pumping, estimated by the model as 530 psi.

December 2019 SPE Drilling & Completion 351

ID: jaganm Time: 16:38 I Path: S:/DC##/Vol00000/190010/Comp/APPFile/SA-DC##190010

 DC187191 DOI: 10.2118/187191-PA Date: 19-November-19 Stage: Page: 352 Total Pages: 11

5 1 12,000
BHP
Fracture internal pressure
4 0.8 10,000 Friction (WB and perforation, tortuosity)
Flow Rate (bbl/min) Net pressure at wellbore

Pressure (psi)
8,000 Average net pressure
3 0.6

β(q)
6,000
2 0.4
Surface injection rate 4,000
BH flow rate
1 0.2
β(q) 2,000

0 0 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0 0.05 0.1 0.15 0.2 0.25 0.3
Real Time (hours) Real Time (hours)
(a) (b)

Fig. 6—(a) Surface injection rate, BH flow rate, and the rate-dependent b-factor; (b) BHP and fracture internal pressure, friction, net
pressure at the wellbore, and average net pressure during injection and falloff.

Some final points regarding the implications of this model match are in order. We prefer the interpretation of the first closure as
height growth into an overlying layer with known closure stress, thus offering critical insight into the treatment fracture design. While
fracturing fluid (slickwater) can enter the overlying layer, it might be difficult for proppant to do so and could lead to premature screen
out. Even if proppant can enter the overlying layer, if it gets there through a shear displacement at the interlayer interface, vertical
stress could close off access to the proppant placed above the interface. Avoiding either of these adverse outcomes could be accom-
plished by pumping below the closure stress of the overlying layer. Indeed, some of the subsequent treatment stages showed premature
screenout, and the well performance was poorer than that of wells drilled and fractured in the overlying formation.

Discussion
The following are remarks regarding key points addressed in this paper.

Net Pressure Magnitude. Some analysts have expressed concerns about the magnitude of pressure decline from the shut-in pressure
to the fracture-closure stress. The pressure-decline magnitude includes friction losses and the fracture net pressure. Friction losses dissi-
pate quickly after shut-in (within 5 minutes in most cases), and the BH pressure without friction loss is the ISIP. Menouar et al. (2018)
showed an approach to estimate wellbore and perforation friction, near-wellbore-tortuosity friction, and the ISIP. The difference
between the ISIP and the fracture closure stress is the fracture net pressure, which represents the energy needed to create the fracture
system during pumping. In Fig. 6b, we showed that accurate accounting of friction pressure losses determines fracture net-pressure esti-
mates with magnitudes in an expected range.

Implication of Successful Matching with the Injection Pressure Response. Analysis of pressure falloff following injection is analo-
gous to analysis of pressure buildup following production. In the latter case, current analysis will always use the model that matches the
pressure-buildup data to also match pressures recorded during production before the buildup if these data exist (Pang et al. 2016).
A buildup model that does not history match pressures recorded during production is lacking critical features or represents an incorrect
model for the data. For DFITs, injection pressure data normally exist. The capability to match both injection and falloff data with the
same model strongly improves confidence in the model and the insights it offers regarding the created fracture and
formation characteristics.

Insights Gained by the Analysis, Including a Match with Injection Pressure. Liu and Ehlig-Economides (2019) demonstrated the
model capability to quantify BC parameters. Those quantified parameters include (1) friction losses through wellbore, perforation, and
near-wellbore tortuosity; (2) ISIP; (3) tip extension distance after shut-in and the minimum fracture-propagation pressure; (4) fracture
area associated with pressure-dependent leakoff and/or transferred storage; (5) leakoff coefficient and stress magnitudes related to clo-
sure of layers and/or secondary fractures; and (6) minimum stress of the tested formation. All these parameters have a bearing on
hydraulic-fracture design (Liu and Ehlig-Economides 2018).

Validation of the Minimum Propagation Pressure at the End of Tip Extension After Shut-In. The minimum propagation pressure
picked at the end of the tip extension after shut-in represents the minimum pressure needed to create a static fracture in the formation (Liu
and Ehlig-Economides 2019). The difference between the minimum propagation pressure and the formation closure stress might be com-
parable with laboratory measurements of the rock tensile failure strength. Or, the fracture intensity factor at the tip reaches the fracture
toughness when fracture pressure drops to the minimum propagation pressure (Griffith 1921). Fracture propagation will stop when the
pressure drops below the minimum propagation pressure. The pressure should be less than the stable injection pressure during injection,
even after the friction is deducted, because extra momentum or pressure is required to maintain fracture growth at a certain growth rate.

Uncertainty and Validation of the Model. The DFIT estimates the formation and fracture parameters critical for fracture-treatment
design, with the added contribution of formation-permeability and -pressure estimates from the AC data that cannot be estimated from
conventional buildup or wireline formation tests. Experience from pressure-transient-test analysis has demonstrated the value of show-
ing a model match for data on a log-log Bourdet diagnostic plot because this presentation, particularly the derivative, exposes mis-
matches with data that relate to incorrect model selection and/or model-parameter estimation errors. (The second point in this
discussion underscored the added value in matching both injection and falloff data.)
Model matching data on the Bourdet log-log diagnostic plot relies on several parameters, some of which are sensitive to multiple
flow regime behaviors that are coupled by the model representing the entire series of flow regimes. This observation justifies confidence

352 December 2019 SPE Drilling & Completion

ID: jaganm Time: 16:38 I Path: S:/DC##/Vol00000/190010/Comp/APPFile/SA-DC##190010

 DC187191 DOI: 10.2118/187191-PA Date: 19-November-19 Stage: Page: 353 Total Pages: 11

in the model and reduces uncertainty in the parameter estimates that might otherwise have been made with a set of isolated trends, each
represented on a separate plot for a separate behavior. Still, ultimately, parameters estimated with a model match are subject to the
uncertainty of model input parameters such as Young’s modulus and Poisson’s ratio, formation height and porosity, and formation-
fluid properties.
To reduce the uncertainty, it is always helpful to have independent external measurements besides the DFIT data, such as (1) a shear
sonic log to show stress profile, (2) a microseismic mapping to show the potential fracture geometry, (3) a core analysis and image log
to show the existence of natural fractures and their orientation, and (4) DFIT results from overlying or underlying layers to show the
closure magnitude and leakoff coefficient. For instance, the stress profile and a DFIT performed in the overlying layer validated a falloff
model accounting for fracturing into that layer.
All models for pressure-transient data are subject to the potential nonuniqueness arising when multiple models match data equally
well. Until now, there were no other models capable of matching the features that our models can match. When external measurements
indicate a need to alter a model that already matches the falloff data, the analytical models usually enable an alternative match that
restores consistency with all known data and information.
A word of caution—this paper was proposed to build a consistent model connecting the pressure response during surface pumping
described by the fracture-propagation model with the pressure falloff after shut-in presented by the DFIT model. However, both models
were simplified, and only future use will determine in which downhole situations they can be applied, and when alternative approaches
are indicated. Further work, combined with other independent measurements, is required to demonstrate the general applicability and
field value of the fracture-propagation model and the DFIT model.

Conclusions
The paper presented a model capable of matching both the injection data and the falloff data from a DFIT. Understanding the inconsis-
tency between classic 2D fracture models and models used for the falloff analysis revealed the need to construct a consistent model that
was based on the falloff analysis and could match the entire injection and falloff pressure behavior. The following are key conclusions:
1. The BH should be considered when estimating rate-dependent friction losses from a step-rate test.
2. We developed a rate-dependent b-function to connect the fracture net pressure at the wellbore with the average fracture net pressure
during the injection and the falloff.
3. The leakoff volume cannot be neglected during the injection, even for a tight formation, and the match with injection data must con-
sider the resulting fluid volume loss.
4. The net pressure in the fracture is likely to be much smaller than the observed pressure drop after shut-in. Matching both the injec-
tion and falloff data with the same model validates the model and demonstrates that actual net pressure can be much less than a
superficial estimate on the basis of the large pressure drop after shut-in.
5. Step-rate tests can lead to an incorrect friction loss estimation, and we recommend excluding the step-rate test from a DFIT.
6. Other independent measurements or tests besides the DFIT, such as microseismic mapping, core analysis, and sonic and image logs,
can provide validation and reduce uncertainty in the DFIT model parameter estimates.

Nomenclature
Af ¼ surface area of one-sided fracture, L2, ft2
Af 0 ¼ surface area of one-sided fracture at shut-in, L2, ft2
cf ¼ fracture compliance, L2t2/m, ft/psi
cw ¼ water compressibility, Lt2/m, 1/ psi
2
cWBS ¼ effective WBS compressibility,
pffi pffi Lt /m, 1/ psi
CL ¼ leakoff coefficient, L/ t, ft/ t
d ¼ outer string diameter, L, in.
G ¼ G-function, dimensionless
E ¼ elastic modulus of steel, m/Lt2, psi
ISIP ¼ instantaneous shut-in pressure, m/Lt2, psi
k ¼ permeability, L2, md
m( p) ¼ pseudopressure, m/Lt3, psi2/cp
m( p)0 ¼ pseudopressure derivative, m/Lt3, psi2/cp
pc ¼ closure stress, m/Lt2, psi
pfric ¼ friction loss, m/Lt2, psi
pi ¼ formation initial pressure, m/Lt2, psi
pnet ¼ fracture net pressure, m/Lt2, psi
pp,min ¼ minimum fracture-propagation pressure, m/Lt2, psi
psurf ¼ surface-recorded pressure during injection, m/Lt2, psi
pw ¼ bottomhole pressure (BHP), m/Lt2, psi
pw,start ¼ bottomhole pressure at the start of injection, m/Lt2, psi
pws ¼ bottomhole pressure at shut-in, m/Lt2, psi
q ¼ flow rate, L3/t, bbl/min
qsurf ¼ surface-recorded injection rate, L3/t, bbl/min
qBH ¼ bottomhole flow rate into the fracture, L3/t, bbl/min
Rf ¼ fracture radius in the GDK radial fracture model, L, ft
Rf,final ¼ fracture radius at the end of tip extension, L, ft
Rf,initial ¼ fracture radius at shut-in, L, ft
t ¼ time, t, minutes
tp ¼ pumping time, t, minutes
V ¼ volume, L3, bbl
Vl ¼ leakoff volume, L3, bbl
Vw ¼ injection string or wellbore volume, L3, bbl
VBH,inj ¼ bottomhole flow volume into the fracture, L3, bbl

December 2019 SPE Drilling & Completion 353

ID: jaganm Time: 16:38 I Path: S:/DC##/Vol00000/190010/Comp/APPFile/SA-DC##190010

 DC187191 DOI: 10.2118/187191-PA Date: 19-November-19 Stage: Page: 354 Total Pages: 11

Vsurf,inj ¼ surface injection volume into the wellbore, L3, bbl

xf,initial ¼ fracture half-length at shut-in, L, ft
xf,final ¼ fracture half-length after tip extension, L, ft
a ¼ area exponent, dimensionless
b ¼ the b-factor, dimensionless
D ¼ difference, dimensionless
i ¼ the ith time point
d ¼ string wall thickness, L, in.
g ¼ fluid efficiency, fraction

Subscripts
c ¼ closure
D ¼ dimensionless
eca ¼ end of constant area
f ¼ fracture
fric ¼ friction
inj ¼ injection
max ¼ maximum
p ¼ pumping
perf ¼ perforation
s ¼ shut-in
tort ¼ near-wellbore fracture tortuosity
w, WB ¼ wellbore
WBS ¼ wellbore storage

References
API SPEC 5CT, Specification for Casing and Tubing, ninth edition. 2011. Washington, DC: API.
Bachman, R. C., Walters, D. A., Hawkes, R. et al. 2012. Reappraisal of the G Time Concept in Mini-Frac Analysis. Presented at the SPE Annual Techni-
cal Conference and Exhibition, San Antonio, Texas, 8–10 October. SPE-160169-MS. https://doi.org/10.2118/160169-MS.
Barree, R. D., Miskimins, J. L., and Gilbert, J. V. 2014. Diagnostic Fracture Injection Tests: Common Mistakes, Misfires, and Misdiagnoses. Presented
at the SPE Western North American and Rocky Mountain Joint Meeting, Denver, Colorado, 17–18 April. SPE-169539-MS. https://doi.org/10.2118/
169539-MS.
Bourdet, D., Whittle, T. M., Douglas, A. A. et al. 1983. A New Set of Type Curves Simplifies Well Test Analysis. World Oil 196 (6): 95–106.
Craig, D. P. and Blasingame, T. A. 2005. A New Refracture-Candidate Diagnostic Test Determines Reservoir Properties and Identifies Existing Conduc-
tive or Damaged Fractures. Presented at the SPE Annual Technical Conference and Exhibition, Dallas, 9–12 October. SPE-96785-MS. https://
doi.org/10.2118/96785-MS.
Craig, D. P. and Blasingame, T. A. 2006. Application of a New Fracture-Injection/Falloff Model Accounting for Propagating, Dilated, and
Closing Hydraulic Fractures. Presented at the SPE Gas Technology Symposium, Calgary, 15–17 May. SPE-100578-MS. https://doi.org/10.2118/
100578-MS.
Cramer, D. D. and Nguyen, D. H. 2013. Diagnostic Fracture Injection Testing Tactics in Unconventional Reservoirs. Presented at the SPE Hydraulic
Fracturing Technology Conference, The Woodlands, Texas, 4–6 February. SPE-163863-MS. https://doi.org/10.2118/163863-MS.
Ehlig-Economides, C. A. and Liu, G. 2017. Comparison Among Fracture Calibration Test Analysis Models. Presented at the SPE Hydraulic Fracturing
Technology Conference and Exhibition, The Woodlands, Texas, 24–26 January. SPE-184866-MS. https://doi.org/10.2118/184866-MS.
Feng, Y. and Gray, K. E. 2016. A Comparison Study of Extended Leak-Off Tests in Permeable and Impermeable Formations. Presented at the 50th US
Rock Mechanics/Geomechanics Symposium. Houston, 26–29 June. ARMA-2016-033.
Geertsma, J. and De Klerk, F. 1969. A Rapid Method of Predicting Width and Extent of Hydraulically Induced Fractures. J Pet Technol 21 (12):
1571–1581. SPE-2458-PA. https://doi.org/10.2118/2458-PA.
Griffith, A. A. 1921. The Phenomena of Rupture and Flow in Solids. Philos Trans R Soc A 221 (582–593): 163–198. https://doi.org/10.1098/
rsta.1921.0006.
Liu, G. and Ehlig-Economides, C. 2015. Comprehensive Global Model for Before-Closure Analysis of an Injection Falloff Fracture Calibration Test.
Presented at the SPE Annual Technical Conference and Exhibition, Houston, 28–30 September. SPE-174906-MS. https://doi.org/10.2118/
174906-MS.
Liu, G. and Ehlig-Economides, C. 2016. Interpretation Methodology for Fracture Calibration Test Before-Closure Analysis of Normal and Abnormal
Leakoff Mechanisms. Presented at the SPE Hydraulic Fracturing Technology Conference, The Woodlands, Texas, 9–11 February. SPE-179176-MS.
https://doi.org/10.2118/179176-MS.
Liu, G. and Ehlig-Economides, C. 2018. Practical Considerations for Diagnostic Fracture Injection Test (DFIT) Analysis. J Pet Sci Eng 171: 1133–1140.
https://doi.org/10.1016/j.petrol.2018.08.035.
Liu, G. and Ehlig-Economides, C. 2019. Comprehensive Before-Closure Model and Analysis for Fracture Calibration Injection Falloff Test. J Pet Sci
Eng 172: 911–933. https://doi.org/10.1016/j.petrol.2018.08.082.
Liu, G., Ehlig-Economides, C., and Sun, J. 2016. Comprehensive Global Fracture Calibration Model. Presented at the SPE Asia Pacific Hydraulic Frac-
turing Conference, Beijing, China, 24–26 August. SPE-181856-MS. https://doi.org/10.2118/181856-MS.
Mack, M. G. and Warpinski, N. R. 2000. Mechanics of Hydraulic Fracturing. In Reservoir Stimulation, third edition, ed. M. J. Economides and K.G.
Nolte, Chap. 6, 1–49. Chichester, England: John Wiley & Sons Ltd.
Marongiu-Porcu, M., Retnanto, A., Economides, M. J. et al. 2014. Comprehensive Fracture Calibration Test Design. Presented at the Hydraulic Fractur-
ing Technology Conference, The Woodlands, Texas, 4–6 February. SPE-168634-MS. https://doi.org/10.2118/168634-MS.
Mayerhofer, M. J., Ehlig-Economides, C. A., and Economides, M. J. 1995. Pressure Transient Analysis of Fracture Calibration Tests. J Pet Technol
47 (3): 229–234. SPE-26527-PA. https://doi.org/10.2118/26527-PA.
Menouar, N., Liu, G., and Ehlig-Economides, C. 2018. A Quick Look Approach for Determining Instantaneous Shut-In Pressure ISIP and Friction
Losses from Hydraulic Fracture Treatment Falloff Data. Presented at the SPE International Hydraulic Fracturing Technology Conference and Exhibi-
tion, Muscat, Oman, 16–18 October. SPE-191465-18IHFT-MS. https://doi.org/10.2118/191465-18IHFT-MS.

354 December 2019 SPE Drilling & Completion

ID: jaganm Time: 16:38 I Path: S:/DC##/Vol00000/190010/Comp/APPFile/SA-DC##190010

 DC187191 DOI: 10.2118/187191-PA Date: 19-November-19 Stage: Page: 355 Total Pages: 11

Mohamed, I. M., Azmy, R. M., Sayed, M. A. I. et al. 2011. Evaluation of After-Closure Analysis Techniques for Tight and Shale Gas Formations. Pre-
sented at the North American Unconventional Gas Conference and Exhibition, The Woodlands, Texas, 24–26 January. SPE-140136-MS. https://
doi.org/10.2118/140136-MS.
Naidu, R. N., Guevara, E. A., Twynam, A. J. et al. 2015. Understanding Unusual Diagnostic Fracture Injection Test Results in Tight Gas Fields—
A Holistic Approach to Resolving the Data. Presented at the SPE Middle East Unconventional Resources Conference and Exhibition Muscat, Oman,
26–28 January. SPE-172956-MS. https://doi.org/10.2118/172956-MS.
Nolte, K. G. 1979. Determination of Fracture Parameters from Fracturing Pressure Decline. Presented at the SPE Annual Technical Conference and
Exhibition, Las Vegas, Nevada, 23–26 September. SPE-8341-MS. https://doi.org/10.2118/8341-MS.
Nolte, K. G. 1986. A General Analysis of Fracturing Pressure Decline with Application to Three Models. SPE Form Eval 1 (6): 571–583. SPE-12941-PA.
https://doi.org/10.2118/12941-PA.
Nolte, K. G., Mack, M. G., and Lie, W. L. 1993. A Systematic Method for Applying Fracturing Pressure Decline: Part I. Presented at the SPE Rocky
Mountain Regional Low Permeability Reservoirs Symposium, Denver, 26–28 April. SPE-25845-MS. https://doi.org/10.2118/25845-MS.
Nordgren, R. P. 1972. Propagation of a Vertical Hydraulic Fracture. SPE J. 12 (4): 306–314. SPE-3009-PA. https://doi.org/10.2118/3009-PA.
Pang, W., Du, J., Zhang, T. et al. 2016. Actual and Optimal Hydraulic-Fracture Design in a Tight Gas Reservoir. SPE Prod & Oper 31 (1): 60–68. SPE-
168613-PA. https://doi.org/10.2118/168613-PA.
Perkins, T. K. and Kern, L. R. 1961. Widths of Hydraulic Fractures. J Pet Technol 13 (9): 937–949. SPE-89-PA. https://doi.org/10.2118/89-PA.
Xue, H. and Ehlig-Economides, C. 2013. Permeability Estimation from Fracture Calibration Test Analysis in Shale and Tight Gas. Presented at the
Unconventional Resources Technology Conference, Denver, 12–14 August. URTEC-1569587-MS. https://doi.org/10.1190/URTEC2013-023.
Zheltov, A. K. 1955. Formation of Vertical Fractures by Means of Highly Viscous Liquid. Presented at the 4th World Petroleum Congress, Rome,
6–15 June. WPC-6132.

SI Metric Conversion Factors

bbl  1.5898 E01 ¼ m3
cp  1.0* E03 ¼ Pas
psi  6.8947 Eþ03 ¼ Pa
nd  9.8692 E22 ¼ m2
ft  3.084* E01 ¼ m
in.  2.54* E02 ¼ m
ft2  9.29 E02 ¼ m2
*
Conversion factor is exact.

Guoqing Liu is a post-doctoral-degree research fellow in the Petroleum Engineering Department at the University of Houston. His
research interests are hydraulic-fracturing modeling, pressure-transient analysis, and the DFIT. Liu has authored or coauthored
more than 10 technical papers. He earned a BS degree in mechanical engineering from Qingdao University of Science and
Technology and an MS degree in petroleum engineering from China University of Petroleum, and he holds a PhD degree in
petroleum engineering from Texas A&M University.
Christine Ehlig-Economides is a professor and holds the Hugh Roy and Lillie Cranz Cullen Distinguished University Chair at the Uni-
versity of Houston. Her research interests include production and reservoir engineering of conventional and unconventional res-
ervoirs. Ehlig-Economides has authored or coauthored more than 120 technical papers and holds two patents. She holds a PhD
degree in petroleum engineering from Stanford University, an MS degree in chemical engineering from the University of Kansas,
and a BA degree in math-science from Rice University. Ehlig-Economides is an Honorary Member of SPE and a member of the
National Academy of Engineering.

December 2019 SPE Drilling & Completion 355

ID: jaganm Time: 16:38 I Path: S:/DC##/Vol00000/190010/Comp/APPFile/SA-DC##190010