You are on page 1of 10

J166132 DOI: 10.

2118/166132-PA Date: 4-June-15 Stage: Page: 508 Total Pages: 10

A New Method To Interpret Fracturing


Pressure—Application to Frac Pack
Elias Pirayesh, Mohamed Y. Soliman, Mehdi Rafiee, and Ali Jamali, Texas Tech University

Summary ture (e.g., proppant screenout) (see Fig. 1). The other two modes
Real-time analysis of fracturing data is an invaluable tool for are (3) a mode when pressure drops rapidly, which is usually the
determining whether a fracturing job is progressing as planned. sign of uncontained fracture-height growth, and (4) an elongated
Since early days, understanding of fracturing pressure was em- flat pressure (i.e., critical pressure) for which there is no definitive
phasized and practiced by the industry. The most well-known explanation. On the basis of the succeeding pressure trend, several
fracturing-pressure-analysis tool is the Nolte-Smith technique. To interpretations are possible for the latter mode, which include
predict the geometry of a hydraulically induced fracture, the rapid height growth, increasing fracture compliance, and opening
Nolte-Smith technique analyzes the pressure response of a forma- of fissures.
tion during pumping. Extensive application of this technique has In addition to the basic assumptions noted by Nolte and Smith
proved reliable to interpret fracturing events. However, the com- (1981), this technique has two additional implied assumptions. The
pression of data imposed by logarithmic scale may make the first assumption is that the injection rate is constant. The second
detection of some events difficult. In addition, the Nolte-Smith assumption is that the fracture propagation is continuous (smooth
technique necessitates prior accurate knowledge of formation-clo- function of time). Furthermore, to ensure a correct interpretation of
sure pressure. fracturing events, the Nolte-Smith technique necessitates precise
In this paper, we present a real-time fracturing diagnostic. This knowledge of formation-closure pressure. This requires conducting
method, which is based on a modification of the Nolte-Smith tech- prefracturing tests, such as minifrac tests, that are not routinely per-
nique, has proved reliable in the interpretation of fracturing formed in every fracturing job. This issue is furthered with the
behavior while a fracturing job is being carried out. In addition, it increasing application of multistage multicluster-fracturing
eliminates the shortcomings of the original technique, meaning schemes in which the subsequent fracturing stages experience
that while making the interpretation of fracturing pressure faster, higher instantaneous shut-in pressures and thus higher closure
the new technique does not require prior knowledge of formation stresses (Soliman et al. 2008; Mayerhofer et al. 2011).
in-situ stresses. This technique was reached by a new innovative
moving-reference-point concept assembled with the power-law Pressure Derivative. Ayoub et al. (1992) introduced the pressure
fracture-propagation theory. Application of the new technique in derivative in fracturing-pressure analysis. The pressure derivative
the analysis of a variety of field cases, including several frac-pack introduced in well testing is known to enhance pattern-recognition
and regular fracturing treatments, proved successful. capabilities in well testing (Ayoub et al. 1988). Application of it
to fracturing treatments produces similar results.
Taking the derivative of the power-law fracture-propagation
Introduction
theory (see Eq. 1) yields
With the fracture-propagation model developed by Perkins and
Kern (1961) and refined by Nordgren (1972), one may write the dP
fracturing pressure at the wellbore as a power function of time as ¼ e c t e1 : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð3Þ
dt
Pnet / te ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð1Þ Multiplying by t gives
dP
where 18  e  15. A large exponent is an indication of low leakoff t ¼ e c te : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .ð4Þ
rate. In other words, more fluid is maintained inside the fracture dt
and contributes to fracture propagation. The bounds given in Eq. A log-log plot of t(dP/dt) vs. time will yield the same slope e as in
1 are based on a Newtonian fluid, which was generalized by Nolte the net-pressure plot (see Fig. 1); however, it does not require
(1979) in the following form in Eq. 2: knowledge of formation-closure stress. In addition, because of its
1 1 enhanced sensitivity, the pressure-derivative technique magnifies
e ; . . . . . . . . . . . . . . . . . . . . . . . . ð2Þ and detects fracturing events earlier in time.
4n þ 4 2n þ 3 The pressure-derivative technique retains the assumptions of
where n is the flow-behavior index of a power-law fluid. the Nolte-Smith technique; however, as is discussed later, some
of these assumptions, such as continuous fracture propagation, do
not always hold. In addition, this technique is prone to the effect
Nolte-Smith Technique. With dimensional analysis, Nolte and of starting-time error (Ayoub et al. 1992).
Smith (1981) reached the conclusion that there are four modes of
fracture propagation. Beginning with the start of the fracturing
treatment, each of those modes is defined by a specific slope on a New Approach: The Moving-Reference-Point
plot of log of Pnet vs. log of time. The two basic modes are (1) a Technique
mode in which a small positive slope on the log-log plot is The approach that we have developed builds on the work of Nolte
observed, indicating that the fracture is propagating normally, and and Smith by coupling the fracture-propagation theory with basic
(2) a mode in which a unit slope on the log-log plot is observed, testing technology. The Nolte-Smith analysis assumes that the
implying that a significant flow restriction has formed in the frac- fracture continuously and smoothly propagates with time. Some
of the recent field observations through microseismic monitoring,
especially in fractured shale formations, imply that a fracture may
Copyright V
C 2015 Society of Petroleum Engineers
grow intermittently. This sporadic fracture growth implies that a
This paper (SPE 166132) was accepted for presentation at the SPE Annual Technical fracture might go through periods of ballooning followed by peri-
Conference and Exhibition, New Orleans, 30 September–2 October 2013, and revised for
publication. Original manuscript received for review 15 October 2013. Revised manuscript ods of growth. Identifying the periods of ballooning and growth
received for review 3 March 2014. Paper peer approved 11 August 2014. will help in diagnosing problems and identifying potential sandout

508 June 2015 SPE Journal

ID: jaganm Time: 15:51 I Path: S:/3B2/J###/Vol00000/140069/APPFile/SA-J###140069


J166132 DOI: 10.2118/166132-PA Date: 4-June-15 Stage: Page: 509 Total Pages: 10

2000

Net Pressure
(psi)

200
0.1 1 10 100
Time (min)

Fig. 1—The Nolte-Smith analysis of a frac-pack treatment.

very early. In the new approach, the reference time is not the time Principle. Assuming that the reference time is ti , we note that
of the start of injection but rather changes. This change in refer- Eq. 6 is the general power-law equation of fracture growth. In the
ence points makes the identification of potential error quicker. Nolte-Smith technique, the reference point is ðti ; Pi Þ ¼ ð0;
The Nolte-Smith technique depends on the power-law equa- Pclosure Þ. In this analysis, this reference time is the start of a
tion for propagation of a hydraulic fracture, given as Eq. 5: growth period. As shown next, one can derive Eq. 7 from Eq. 6:

logðP  Pclosure Þ / e logðtÞ: . . . . . . . . . . . . . . . . . . ð5Þ P  Pi ¼ cðt  ti Þe ; . . . . . . . . . . . . . . . . . . . . . . . . . .ð6Þ


dP
¼ e c ðt  ti Þe1 : . . . . . . . . . . . . . . . . . . . . . . . . . ð7Þ
The Nolte-Smith technique assumes that the fracture passes dt
through various phases and each phase is continuous. Thus, it
assumes that the log-log plot of net pressure vs. time should yield Combining Eqs. 6 and 7 yields
a straight line with slopee. The value of the slope e depends on dP
the fracturing-fluid flow-behavior index n. ðt  ti Þ ¼ eðP  Pi Þ: . . . . . . . . . . . . . . . . . . . . . . ð8Þ
dt
It was observed that the assumption of continuous propagation,
although helpful, may not always be accurate. As noted previ- If the fracture is propagating, then the exponent e will generally
ously, fracture propagation may consist of periods of ballooning have a value in the range determined by Eq. 2. If the fracture is
followed by periods of intermittent growth. These periods of bal- ballooning, the exponent will be unity, similar to what one would
looning and growth may alternate through the injection period. observe in any storage situation. In the case of fracture balloon-
Identification of periods of intermittent growth may be very ing, Eq. 8 becomes
helpful and is described in the following analysis. Jeffrey et al. dP
(2009) present a 2D numerical simulator that is built around this ðt  ti Þ ¼ ðP  Pi Þ: . . . . . . . . . . . . . . . . . . . . . . . ð9Þ
dt
idea in which a fracture may balloon when it intersects a natural
fracture. Fig. 2 shows a mined fracture that was created in a Also, Eqs. 6 and 7 will take the forms
coalbed-methane formation. The figure shows that a fracture fol-
lowed a plane of weakness (cleat) that, in turn, intersected the cre- P  Pi ¼ cðt  ti Þ; . . . . . . . . . . . . . . . . . . . . . . . . . ð10Þ
ated hydraulic fracture. In such a shallow formation, however, the
stress level is very low, and the stress anisotropy between the two and
horizontal stresses may be very low or nonexistent and may yield dP
behavior inconsistent with deep-formation fracturing. ¼ c: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð11Þ
dt

a b
96 mm diameter
borehole

Fig. 2—Offsets in hydraulic fracture exposed by mining; propped fracture trace is shown in roof rock above mined coal seam
[Jeffrey et al. (2009)].

June 2015 SPE Journal 509

ID: jaganm Time: 15:51 I Path: S:/3B2/J###/Vol00000/140069/APPFile/SA-J###140069


J166132 DOI: 10.2118/166132-PA Date: 4-June-15 Stage: Page: 510 Total Pages: 10

Start

Pick the current point as reference point


(tref , Pref)

Pick the next point


(t, P ) No

Yes
δP

Find e(t) = δt
P – Pref ε < threshold?
t – tref
P – Pref
And c(t) =
(t – tref)e(t)

t
1
– =
Find e(t)
t – tref
∫ e(t) dt – –
tref ε (t) = P – c(t)(t–t ref)
e(t)
t
1
∫ c(t) dt
– =
And c(t)
t – tref tref

Fig. 3—Data-analysis flow chart.

Numerical Procedure. Analysis of fracturing pressure with the and


moving-reference-point technique is performed according to the
data-analysis flow chart presented in Fig. 3. BHPest ¼ cðtÞ ðt  tref ÞeðtÞ : . . . . . . . . . . . . . . . . . . . ð14Þ
To begin the analysis, an initial reference point, such as
ðtref ; Pref Þinitial , which meets the following criteria, is picked: If the difference between the BHPest: and the observed BHP [i.e.,
I. For intact formations, the first reference point must be P(t)] exceeds a predetermined threshold, then the next point in
picked after the formation breakdown has occurred. time is chosen as the new reference point.
II. For formations with existing flaws, such as small cracks
resulting from minifrac tests, the first reference point can be Application to Homogeneous Formations. If the formation is
picked at any time after the existing crack has been reopened. The homogeneous and isotropic, it is expected that the fracture growth
sandstone formation of Example 1, which was previously subject would be continuous and the slope of the net pressure with time
to minifrac and step-rate tests, falls into this category. would follow the theory developed by Nolte and Smith (1981).
To obtain meaningful results, it is recommended that the first This may be the case in hard sandstone rocks; however, this is not
few points of fracturing-pressure data be omitted from analysis expected to happen in many reservoirs and definitely not in heter-
because such data usually contain severe fluctuations and are usu- ogeneous and shale formations. Application of the moving-refer-
ally affected by formation breakdown/fracture reopening. After ence-point technique in the fracturing-pressure analysis of two
the initial reference point is picked, analysis continues by select- frac-pack examples is provided in subsequent sections.
ing ðt; PÞ pairs and then by calculating eðtÞ with Eq. 8. Values of
eðtÞ are then plotted vs. time and used to interpret fracture behav- Application to Heterogeneous Formations. If the formation
ior. In every timestep, eðtÞ and cðtÞ are averaged over ½tref ; t with contains various heterogeneities, natural fractures, and planes of
Eqs. 12 and 13, respectively; eðtÞ and cðtÞ are subsequently used weakness, the fracture growth is likely to consist of periods of
to find estimated bottomhole pressureðBHPest: Þ (see Eq. 14): propagation and ballooning. In the described formations, the frac-
ture usually propagates following the established theory until the
ðt fracture tip hits a region of heterogeneity. After the fracture
1 reaches the region of heterogeneity, it may start ballooning, caus-
eðtÞ ¼ eðtÞ dt; . . . . . . . . . . . . . . . . . . . . . ð12Þ
t  tref ing an increase in pressure. During this ballooning period, it is
tref
expected that the fracture volume increases and includes the
ðt increase in the volume of the natural fracture. The ballooning
1 effect is similar to the tip-screenout effect discussed by Nolte and
cðtÞ ¼ cðtÞ dt; . . . . . . . . . . . . . . . . . . . . . ð13Þ
t  tref Smith. One can calculate the volume of the fracture and natural
tref fracture with Eq. 15:

510 June 2015 SPE Journal

ID: jaganm Time: 15:51 I Path: S:/3B2/J###/Vol00000/140069/APPFile/SA-J###140069


J166132 DOI: 10.2118/166132-PA Date: 4-June-15 Stage: Page: 511 Total Pages: 10

Simulator 1 Simulator 2 Simulator 1 Simulator 2


Simulator 3 Observed Simulator 3 Observed

900 6000
Net Pressure (psi)

Net Pressure (psi)


600 4000

300 2000

0 0
–10 10 30 50 0 20 40 60

Time (min) Time (min)

Fig. 4—In this example, commercial fracturing simulators over- Fig. 5—In this example, two out of three fracturing simulators
estimated fracturing pressure. failed to anticipate a major event.

dP 0:041665
¼ : . . . . . . . . . . . . . . . . . . . . . . . . . . ð15Þ dP 2qi E0
dt Cff Vff ¼ : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð17Þ
dt p h2 L
Eq. 15 is easily derived from the basic well-testing equation for
the storage period. One may also use the equation developed by As Nolte and Smith (1981) have suggested, one may use
Nolte and Smith, which is based on the compliance of the fracture: Eq. 17 to estimate the distance to a restriction and conse-
quently to determine whether the restriction is caused by tip
dP 2ðqi  ql ÞE0 screenout or near-wellbore restriction. One may use Eq. 16 in
¼ ; . . . . . . . . . . . . . . . . . . . . . . . . . ð16Þ the same fashion to determine the distance to an obstruction.
dt p h2 L
Calculating the volume of the fracture with Eq. 15 at different
where E0 ¼ 1vE
2 is plane-strain Young’s modulus. If the leakoff rate is times during fracturing may be taken as a measure of fracture
negligible compared with the injection rate, one can simplify Eq. 16 as complexity.
Shale

4800 6400 0.0 2.4 0.000 0.060 3920 4120 0.10 0.30

Closure pressure Young’s modulus Fluid-loss coefficient Reservoir pressure Porosity


ft
(psi) (106psi) ( ) (psi) (fraction)
min

Fig. 6—Example 1: Geology.

June 2015 SPE Journal 511

ID: jaganm Time: 15:51 I Path: S:/3B2/J###/Vol00000/140069/APPFile/SA-J###140069


J166132 DOI: 10.2118/166132-PA Date: 4-June-15 Stage: Page: 512 Total Pages: 10

1000 n' K'

Viscosity (cp) at 170 1/sec


0.75 0.2
750

0.5
500
n' 0.1 K'
0.25
250

0 0 0
0 0.5 1 1.5 0 0.5 1 1.5
Time (hours) Time (hours)
at formation temperature (200°F) at formation temperature (200°F)

Fig. 7—At a constant shear rate, the viscosity of the fracturing Fig. 8—Variations of power-law-fluid parameters k and ń as a
fluid declines with time. function of time.

Examples volume of 37,000 gal of slurry, the first treatment is almost twice
Application of the moving-reference-point technique is illustrated the size of the second one, which injected 21,000 gal of slurry. In
through two frac-pack examples. Frac packs are normally high- addition, the first treatment was performed in a relatively thin
rate treatments designed to create infinite-conductivty fractures to sandstone, which was only 40 ft in height, whereas the second one
bypass the skin damage in high-permeability formations. Because was pumped into 215 ft of perforated interval spanning through
of high rates of injection, fracture packing can start and can lead several layers of high-permeability sandstone, shale, and silty
to pressures much beyond the allowable levels in a matter of sand. Results of fracture-analysis studies with a 3D fracturing
minutes or sometimes even in a fraction of a minute. Therefore, simulator indicate that the first and second treatments create
quick identification of the onset of fracture packing is of utmost fractures with length/height ratios of approximately 4.65 and
importance to prevent both intolerable pressure levels and over- 0.75, respectively.
flush of proppants. Each example presents analysis results in the form of an eðtÞ
Currently, fracturing-diagnostic techniques include numerical vs. time plot. One can use the four fracturing modes introduced
simulation and the Nolte-Smith technique; however, fracturing by Nolte and Smith (1981) with this plot to monitor fracture
simulators are not fully capable of replicating fracturing behavior behavior during pumping. As was discussed previously, values of
1 1
in real time. Figs. 4 and 5 compare the net pressures of two frac- eðtÞ in the range of 4nþ4  e  2nþ3 indicate that the created frac-
turing treatments with those predicted by three commercial frac- ture is propagating under the assumptions of Perkins and Kern
turing simulators. In the first case (Fig. 4), all the simulators (1981), which are confined height, constant fracture compliance,
overestimated the average net pressure, but this is not nearly as and unrestricted extension. The eðtÞ  1 usually means that frac-
serious as failure to predict screenouts. Fig. 5 shows the net pres- ture propagation has decreased significantly, and, instead, fluid
sure observed during a frac pack that reached extremely high val- storage is taking place in the form of increasing Pnet and increasing
ues, whereas two out of three fracturing simulators did not foresee average fracture width. In addition, a rapid pressure drop [i.e.,
a screenout. One may note that the nature of the log-log plot of eðtÞ << 1] is the sign of rapid height growth. No certain explana-
the original technique by Nolte and Smith tends to compress data. tion exists for a constant-fracturing-pressure trend [i.e., eðtÞ  0],
The technique presented in this paper corrects this problem. and on the basis of the succeeding pressure behavior, several inter-
The frac-pack examples illustrate how one can use the mov- pretations are possible, including rapid height growth, increasing
ing-reference-point technique in different geologies to obtain a fracture compliance, and opening of fissures. Because of the uncer-
more-detailed understanding of fracture behavior, as well as to tainties associated with this period and the potential undesirable
accelerate identification of fracturing problems. The two frac consequences, Nolte and Smith (1981) named this constant pressure
packs are vastly different from each other. With an injected “critical pressure” and provided guidelines for treating it.

Bottomhole Pressure Slurry Rate Proppant Concentration

7000 50

40

6000 Slurry Rate


30
(bbl/min)
BHP
(psi)
20 Proppant Concentration
5000 (lb/gal)

10

4000 0
0 10 20 30 40 50
Time (min)

Fig. 9—Example 1: Treatment schedule.

512 June 2015 SPE Journal

ID: jaganm Time: 15:51 I Path: S:/3B2/J###/Vol00000/140069/APPFile/SA-J###140069


J166132 DOI: 10.2118/166132-PA Date: 4-June-15 Stage: Page: 513 Total Pages: 10

2,000 Period I: ( 0 < t < 20 minutes) The slope of this period fits in
with Mode I, which indicates that fracture propagation is likely to
be the predominant event in this period.
Period II: ( t  20 minutes) With a slope greater than unity,
this period fits the definition of Nolte-Smith Mode III, a potential
Net Pressure explanation for which is the creation of a substantial flow barrier
(psi) in the fracture.
In summary, fracturing Modes I and III were identified on the
Nolte-Smith plot of Example 1 that helped determine the onset of
fracture-packing at about t ¼ 25 minutes.
Analysis With Moving-Reference-Point Technique. The plot
of eðtÞ vs. time is shown in Fig. 12. According to this figure,
the moving-reference-point technique confirms the conclusions
200
achieved by means of the Nolte-Smith technique, meaning that a
0.1 1 10 100
period of overall fracture propagation (i.e., Mode I) (on the e-vs.-
Time (min)
time plots, this range is highlighted in green) is followed by a
period during which fracturing pressure increases as a linear func-
Fig. 10—Example 1: Analysis with the Nolte-Smith technique.
tion of time (i.e., Mode III) (on the e-vs.-time plots, this range is
highlighted in red). Fig. 12 also shows that, from t ¼ 0 minutes to
t ¼ 20 minutes, the induced fracture goes through periods of bal-
Example 1: High-Permeability Oilwell Frac Pack. A frac pack looning and growth. Marked by signs, periods of balloon-
performed in a high-permeability sandstone formation (see Fig. 6) ing may be recognized by the two peaks that almost reach into the
pumped 37,000 gal of a 25 lbm/1,000 gal seawater-based fracturing red zone. This is in accord with microseismic observations that
fluid (see Figs. 7 and 8) and 54,000 lbm of a 12/18 lightweight show that rather than growing continuously, fractures actually
synthetic proppant. Even though a constant slurry-injection rate of tend to grow intermittently.
25 bbl/min was maintained throughout the treatment, proppant As previously discussed, quick detection of the beginning of
injection started at t ¼ 18:4 minutes and continued until the end fracture packing and screenout is very important in fracturing
of the treatment when the fracture could be packed no more (see treatments, especially in frac packs. For the example in hand, the
Fig. 9). response to fluid storage starts by as early as t ¼ 21 minutes
Mini-frac test results show that the sandstone formation, which (marked by an orange circle on Fig. 12), where the curvature of
is highlighted in yellow in Fig. 6, has an average closure stress of the plot changes from positive to negative. By t ¼ 23 minutes, the
5,022 psi, 200 to 250 psi lower than the surrounding shale barrier. plot falls well within the red zone, which is a better indicator of
This, along with a modulous of elasticity (i.e., E) on the order of fluid storage (marked by a red circle on Fig. 12).
the moduli of the surrounding shale and a relatively small fracture Real-time application of the Nolte-Smith technique is such
toughness (i.e., KIC ), is expected to lead to confined fracture- that, after observing a unit-slope line on the Nolte-Smith chart, a
height growth. As seen in Fig. 6, the fracturing fluid has a flow- fracturing engineer needs to wait at least one-quarter of a log
behavior index (n) of 0.5, for which the slope of the fracture-prop- cycle of time to confirm fracturing Mode III. In case of Example
agation mode (i.e., Nolte-Smith Mode I) ranges from 1=5 to 1=6. 1, this necessary precaution will delay recognition of fracture
Analysis With the Nolte-Smith Technique. The log-log plot packing until t ¼ 35.6 minutes. Table 1, which compares the time
of Pnet vs. time is shown in Fig. 10 compared with Case 2 of taken by the Nolte-Smith and the moving-reference-point techni-
Fig. 5 of Nolte and Smith (1981) (see Fig. 11) and is composed of ques to detect fracture packing, shows that the latter technique
two distinct periods: takes approximately one-fourth of the time taken by the former
technique to detect the onset of fluid storage.

Lock up →
Case 1
Pressure above closure (psi)

4000 II
I I
2000
* Case 2
II
1000
I Case 3

*
*
I II I

1 1
Slope: to 0 1.0 Negative
8 4
Mode: I II III IV
* - Start Proppant

20 40 60 100 200 400


Time (min)

Fig. 11—Examples with different characteristic slopes in Nolte and Smith (1981).

June 2015 SPE Journal 513

ID: jaganm Time: 15:51 I Path: S:/3B2/J###/Vol00000/140069/APPFile/SA-J###140069


J166132 DOI: 10.2118/166132-PA Date: 4-June-15 Stage: Page: 514 Total Pages: 10

1.5

1.25

e 0.75

0.5

0.25

0
0 5 10 15 20 25 30 35 40
Time (min)

Fig. 12—Example 1: Analysis with the moving-reference-point technique.

Since the Beginning of Injection Since the Start of Proppant


The Nolte-Smith The Nolte-Smith Moving-Reference-Point
Moving-Reference-Point Technique
Technique Technique Technique
35.6 minutes 23 minutes 172 minutes 4.6 minutes

Table 1—Fracture-packing identification times.

Example 2: High-Permeability Gas-Well Frac Pack. A frac- test, the closure stress of the pay zone is approximately 6,500 psi,
pack treatment performed in a high-permeability sandstone for- which is substantially lower than the closure stresses of the
mation (see Fig. 13) consumed 21,000 gal of a 25 lbm/1,000 gal bounding shale.
seawater-based fracturing fluid (same as Example 1 fracturing Analysis With the Nolte-Smith Technique. On the log-log
fluid; see Figs. 7 and 8) and 90,600 lbm of a 12/18 lightweight plot of Pnet vs. time of this example (see Fig. 15), one can identify
synthetic proppant. A constant injection rate of 18 bbl/min was three distinct periods:
maintained throughout the treatment, and the proppant was Period I: ( t < 1 minute) Pressure grows with a small slope of
injected in a ramped manner, as described in Fig. 14. Bound by 0.1, so the major fracturing event of this short period is Perkins-
two thick shale layers, the pay zone consists of two high-leakoff Kern-Nordgren-type fracture propagation.
sandstone layers separated from each other by several layers of Period II: [(1 minute < t < 3 minutes) and (3 minutes < t < 9
shale and silty sand. From Fig. 13 and as confirmed by a minifrac minutes)] With a slope of approximately –0.35 and –0.1,
Shale
sandstone
Shale

6400 7800 0.0 1.0 0.000 0.020


ft
Closure pressure (psi) Young’s Modulus (106 psi) Fluid-loss coefficient ( )
min

Fig. 13—Example 2: Geology.

514 June 2015 SPE Journal

ID: jaganm Time: 15:51 I Path: S:/3B2/J###/Vol00000/140069/APPFile/SA-J###140069


J166132 DOI: 10.2118/166132-PA Date: 4-June-15 Stage: Page: 515 Total Pages: 10

Bottomhole Pressure Slurry Rate Proppant Concentration


9000 40

8000 30
Slurry Rate
(bbl/min)
BHP 7000 20
(psi) Proppant
Concentration
6000 10 (PPG)

5000 0
0 10 20 30
Time (min)

Fig. 14—Example 2: Treatment schedule.

Net Pressure
(psi) 1000

100
0.1 1 10 100
Time (min)

Fig. 15—Example 2: Analysis with the Nolte-Smith technique.

1.5

0.5

e 0
0 5 10 15 20 25 30
–0.5

–1

–1.5
Time (min)

Fig. 16—Example 2: Analysis with the moving-reference-point technique.

respectively, these periods meet the conditions of Mode IV (i.e., These interpretations are in accord with the results of our frac-
fracture-height growth), as described in Fig. 11. turing-simulation study, which was able to replicate the observed
Period III: (t > 10 minutes) During this period, pressure net pressure acceptably. As shown in Fig. 17, the created fracture
increases with a slope larger than unity, indicating blockage of has a length/height ratio ratio of approximately 0.75, proving the
fluid-flow paths by the injected proppant. predominance of height growth during the first 9 minutes. Fig. 17
Analysis With the Moving-Reference-Point Technique. The also shows that the fracture profile remains almost constant from
moving-reference-point technique (see Fig. 16) confirms the find- t  9 minutes until the end of the frac pack. It also shows that the
ings of the Nolte-Smith technique. This implies that an elongated major event during this period is fluid storage marked by rapid
period with an average e(t) of approximately –0.70 prevails increase in pressure and fracture width.
through the first 10 minutes of the treatment. This period is fol- To obtain a more-detailed understanding of the behavior of
lowed by a prolonged period of fluid storage with an e(t) of the induced fracture, we adjusted the parameters of the numerical
approximately 1.0 that lasts until the end of injection. procedure outlined in the data-analysis flow chart. The results

June 2015 SPE Journal 515

ID: jaganm Time: 15:51 I Path: S:/3B2/J###/Vol00000/140069/APPFile/SA-J###140069


J166132 DOI: 10.2118/166132-PA Date: 4-June-15 Stage: Page: 516 Total Pages: 10

Shale 0.000 0.000


0.050 0.301
0.101 0.603

Width - Total in

Width - Total in
0.151 0.904
0.202 1.206
0.252 1.507
0.302 1.809
0.353 2.110
0.403 2.412
0.454 2.713
0.504 3.015
Shale

1.513 m/sec 1.513 m/sec

6475 7025 19 38 56 19 38 56
Stress (psi) Fracture Penetration (ft) Fracture Penetration (ft)

t = 9 minutes t = 27.89 minutes

Fig. 17—Example 2: The results of our simulation study confirm that fracture propagation is negligible after t 5 9 minutes.

1.5

0.5

e 0
0 5 10 15 20 25 30
–0.5

–1

–1.5
Time (min)

Fig. 18—Example 2: A higher resolution with the moving-reference-point technique.

Since the Beginning of Injection Since the Start of Proppant


The Nolte-Smith Moving-Reference-Point The Nolte-Smith Moving-Reference-Point
Technique Technique Technique Technique
17.8 minutes 8 minutes 11. 11.55 minutes 1.75 minutes
Table 2—Fracture-packing-identification times.

obtained are shown in Fig. 18. As seen in this figure, fracturing sixth of the time taken by the Nolte-Smith technique to detect the
behavior is more complex than previously thought. For example, inception of the new fracturing event (see Table 2).
there is a short period of constant pressure that took place in the
middle of the initial pressure-decline period; this point is marked
by a sign on Fig. 18. The physical meaning of this constant Conclusions
pressure is not exactly known to us yet, but it could mark propa- The Nolte-Smith technique relies on assumptions, such as continu-
gation through the two close shale layers that separate the top and ous fracture propagation and fixed injection rate, which do not nec-
bottom sandstone formations. In addition, according to the Nolte- essarily hold true in all fracturing operations. In addition, the Nolte-
Smith technique, the major event during t > 8 minutes is pure Smith technique requires prior accurate knowledge of formation-
fluid storage, whereas this period actually includes some short closure stress. Compression of data, imposed by the log-log plot of
periods of possible fracture propagation. These short periods are this technique, makes interpretation of fracture behavior rather slow
marked by signs on Fig. 18. and quite insensitive to minor fracturing events.
More importantly, Fig. 18 permits accurate determination of By taking the derivative of the power-law fracture-propagation
the onset of fracture packing/screenout. With the moving-refer- equation, we were able to develop the moving-reference-point
ence-point technique, inception of fracture packing is identified at technique in this study. This new technique eliminates the major
t ¼ 8 minutes in which the plot falls completely in the red zone limitations of the Nolte-Smith technique, as noted previously.
(on Fig. 18, this point is marked with a sign). With the Nolte- Application of the new technique in the analysis of two frac-pack
Smith technique, the fracturing engineer would have had to wait cases proved successful, which implies that, in addition to provid-
at least until t ¼ 17.8 minutes to confirm fracture packing. In other ing a fairly detailed understanding of fracture behavior, this tech-
words, the moving-reference-point technique took less than on- nique helped to accelerate identification of fracturing problems.

516 June 2015 SPE Journal

ID: jaganm Time: 15:51 I Path: S:/3B2/J###/Vol00000/140069/APPFile/SA-J###140069


J166132 DOI: 10.2118/166132-PA Date: 4-June-15 Stage: Page: 517 Total Pages: 10

The new technique uses a numerical procedure to analyze fractur- Nolte, K.G. 1979. Determination of Fracture Parameters From Fracturing
ing pressure that needs to be implemented with computer pro- Pressure Decline. Paper presented at the SPE Annual Technical Con-
gramming. The final outcome of this procedure is a Cartesian plot ference and Exhibition, Las Vegas, Nevada, 23–26 September. SPE-
of e(t) vs. time by studying the changes in fracture behavior that 8341-MS. http://dx.doi.org/10.2118/8341-MS.
one can identify quite easily and quickly. Nolte, K.G. and Smith, M.B. 1981. Interpretation of Fracturing Pressures.
It is strongly recommended that a downhole gauge with sur- J Pet Technol 33 (9): 1767–1775. SPE-8297-PA. http://dx.doi.org/
face readout be used to monitor pressure changes. This type of 10.2118/8297-PA.
pressure monitoring gives accurate representations of the periods Nordgren, R.P. 1972. Propagation of a Vertical Hydraulic Fracture. SPE J.
of growth and ballooning. Surface-pressure gauges may be less 12 (4): 306–314. SPE-3009-PA. http://dx.doi.org/10.2118/3009-PA.
expensive to implement; however, they do not present an accurate Perkins, T.K. and Kern, L.R. 1961. Widths of Hydraulic Fractures. J
measure of the fine changes in pressures. Pet Technol 13 (9): 937–949. SPE-89-PA. http://dx.doi.org/10.2118/89-
PA.
Nomenclature Soliman, M.Y., East, L., and Adams, D. 2008. Geomechanics Aspects of
Multiple Fracturing of Horizontal and Vertical Wells. SPE Drill &
BHPest: ¼ estimated bottomhole pressure (psi)
Compl 23 (3): 217–228. SPE-86992-PA. http://dx.doi.org/10.2118/
C ¼ power-law fracture-propagation-equation constant, di-
86992-PA.
mensionless
1
Cff ¼ fracturing-fluid compressibility, psi
e ¼ the exponent of the power-law fracture-propagation Elias Pirayesh is a doctoral-degree student at Texas Tech
University. His research focuses on the development of consti-
equation, dimensionless tutive material models for hydrocarbon-bearing weak forma-
E ¼ Young’s modulus tions. Pirayesh has authored or coauthored five technical
E0 ¼ plane-strain Young’s modulus
pffiffiffiffiffiffi articles. He also holds one US patent. Pirayesh earned an MS
KIC ¼ fracture toughness, psi in: degree from Texas Tech University and a BS degree from Sharif
Lf ¼ fracture length (tip to tip), ft University of Technology, Iran, both in petroleum engineering.
n ¼ flow-behavior index of a power-law fluid M.Y. Soliman is the George P. Livermore Professor of Petro-
Pnet ¼ net pressure, psi leum Engineering at Texas Tech University. Before joining
Pclosure ¼ closure stress, psi Texas Tech University, he worked for Halliburton Energy for
qi ¼ injection rate into one wing of the fracture, ft3/min more than 30 years in several technical and managerial posi-
q1 ¼ leakoff rate of one wing of the fracture, ft3/min tions. Soliman is an SPE Distinguished Member and a licensed
t ¼ time, minutes professional engineer in the State of Texas. He has authored
ti ¼ time of start of a new period, minutes or coauthored more than 200 technical papers. Soliman also
Vf ¼ fracture volume, ft3 holds 26 US patents. He earned a BS degree in petroleum en-
 ¼ Poisson’s ratio gineering with top honors from Cairo University in 1971. Soli-
man earned MS and PhD degrees from Stanford University in
1975 and 1978, respectively. His areas of interest include frac-
References turing, well-test analysis, diagnostic testing, and numerical
simulation.
Ayoub, J.A., Bourdet, D.P., and Chauvel, Y.L. 1988. Impulse Testing.
SPE Form Eval 3 (3): 534–546. SPE-15911-PA. http://dx.doi.org/ Mehdi Rafiee currently works as a Senior Researcher—Com-
10.2118/15911-PA. pletions in the Research, Development, and Innovation
Ayoub, J.A., Brown, J.E., Barree, R.D. et al. 1992. Diagnosis and Evalua- department at Statoil. His research focuses on developing
tion of Fracturing Treatments. SPE Res Eng 7 (1): 39–46. SPE-20581- new technologies to acquire more data and knowledge from
fracturing operations. Rafiee is also involved in fracture model-
PA. http://dx.doi.org/10.2118/20581-PA.
ing and data interpretation of stimulated unconventional
Jeffrey, R.G., Zhang, X., and Thiercelin, M.J. 2009. Hydraulic Fracture wells in Statoil US onshore assets. He holds a PhD degree and
Offsetting in Naturally Fractured Reservoirs: Quantifying a Long-Rec- an MS degree, both in petroleum engineering, from Texas
ognized Process. Paper presented at the SPE Hydraulic Fracturing Tech University.
Technology Conference, The Woodlands, Texas, 19–21 January. SPE-
Ali Jamali is a doctoral-degree student at Texas Tech Univer-
119351-MS. http://dx.doi.org/10.2118/119351-MS.
sity, where he works on development of capacitance/resistive
Mayerhofer, M.J., Stegent, N.A., Barth, J.O. et al. 2011. Integrating Frac- models and their application to enhanced-oil-recovery proc-
ture Diagnostics and Engineering Data in the Marcellus Shale. Paper esses. He has coauthored four technical papers. Jamali holds
presented at the SPE Annual Technical Conference and Exhibition, an MS degree from Texas Tech University and a BS degree
Denver, Colorado, 30 October–2 November. SPE-145463-MS. http:// from Sharif University of Technology, Iran, both in petroleum
dx.doi.org/10.2118/145463-MS. engineering.

June 2015 SPE Journal 517

ID: jaganm Time: 15:51 I Path: S:/3B2/J###/Vol00000/140069/APPFile/SA-J###140069

You might also like