You are on page 1of 9

SPE 84491

Closing the Gap: Fracture Half Length from Design, Buildup, and Production Analysis
R. D. Barree, Barree & Associates; S. A. Cox, J. V. Gilbert, M. Dobson, Marathon Oil Company
Copyright 2003, Society of Petroleum Engineers Inc.
This paper was prepared for presentation at the SPE Annual Technical Conference and
Exhibition held in Denver, Colorado, U.S.A., 5 8 October 2003.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300
words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract
It is commonly observed that hydraulically fractured wells
perform as though the effective fracture half-length were
much less than the designed half-length. This observation has
been explained by various models including poor fracture
height containment, poor proppant transport, proppant falling
out of zone (convection), ineffective proppant pack cleanup,
capillary phase trapping, multi-phase flow, gravitational phase
segregation, and non-Darcy flow, with combinations of any of
these mechanisms. With recent improvements in diagnostic
measurements of fracture geometry, some of these
explanations have lost credibility, but the problem of low
effective fracture length persists.
This paper presents detailed evaluation of hydraulically
fractured well behavior using continuous production analysis,
pressure transient (buildup) analysis and fracture treatment
evaluation using actual field data from a tight-gas reservoir in
the Rocky Mountain Region. The various analyses explain the
observed producing behavior of the well and lead to a
consistent determination of the actual effective fracture halflength compared to the physically created or propped length.
Problems relating to semantics and inconsistent fracture and
reservoir description, especially the physical processes
encompassed by various analytical techniques, will
be addressed.
Methods will be outlined for predicting the useful effective
length from available proppant conductivity data. The process
outlined helps to close the gap between designed frac length
and producing length and points out the causes for remaining
system bottlenecks that limit post-frac well productivity.
Finally, the understanding of these mechanisms provides a
means to arrive at an economic optimum fracture treatment
design for a reservoir, once key parameters are known.

Introduction
The integration of fracture design lengths with actual well
performance can provide valuable insight into the
effectiveness of the fracture stimulation. This process requires
the effective integration of several analytical tools. The
evaluation process starts with the pre-stimulation design and
ends with an evaluation of the wells production performance.
The actual rate and pressure response from the stimulation
should be history matched to determine the placed fracture
half-length. The resulting length should then be compared to
both the pre-job estimates and to actual well production
performance. Several analytical techniques are currently
available to perform post production analysis including
pressure transient testing and production analysis. Production
analysis is a technique which incorporates the wells rate and
flowing pressure into a type curve matching process to provide
a consistent post stimulation analysis. The resulting well
performance can then be evaluated by comparing the wells
producing capacity with its actual performance.
The
evaluation of past performance has proven to be the best
method of improving future performance.
In many cases the resulting fracture half-length calculated
from post production analysis is much shorter than planned.
This discrepancy can be a source of contention between the
team responsible for completing the well and the team
responsible for the optimization of field performance.
Frequently, these discrepancies can be quantified through a
comprehensive, consistent analysis of the available
information. Identifying the problems that result in short
effective fracture lengths allows appropriate design changes to
be made to improve future well performance.
The example well case-histories described represent a
subset of a more extensive field study. In general, the study
identified the need for a change in the stimulation design in
the field which has been successfully implemented and has
resulted in significant production performance improvements.
Well selection criteria for case study
The two wells selected for this case study each have a
comprehensive data set including detailed fracture stimulation
records, long term production histories, including flowing
pressures and rates, and pressure transient tests. Both wells
produce from sandstone formations, but not the same
reservoir. Reservoir and completion properties for each well
are summarized in Table 1. Both wells produce from a single
zone during the evaluation period. Long-term production
shows the influence of boundary dominated flow behavior.
Fracture stimulation data includes the treating-pressure

SPE 84491

history, pumping rate, and proppant schedule. These data


allow for a detailed history-match of the stimulation treatment,
and development of inferred fracture geometry from the
pressure match. The pressure transient tests are of sufficient
quality to estimate fracture half-length.
Table 1 - Reservoir and Fluid Properties

Rock Type
Initial Reservoir Pressure, psia
o
Reservoir Temperature, F
Thickness, ft
Porosity
Initial Water Saturation
Wellbore Radius, ft
Gas Specific Gravity
o
Condensate API
Formation Compressibility, 1/psi
Water Compressibility, 1/psi
Formation Depth, ft

Well 1
Well 2
Sandstone Sandstone
3,500
2,500
190
190
24
37
0.11
0.11
0.35
0.35
0.354
0.354
0.62
0.62
55
55
3E-06
3E-06
3E-06
3E-06
9,570
8,450

Fracture treatment evaluation


The fracture treating-pressure history for both wells was
modeled using a fully 3-D planar fracture geometry model
(GOHFER).1-4 Digital treatment data was used to generate the
input pumping schedule for the simulation studies, along with
rock mechanical properties and stress profiles derived from
digital well log data files. Pore pressure and reservoir
permeability estimates were derived from transient pressure
build-up and production analysis results.
Well Case 1
Well 1 was treated down 4.5, 11.6 #/ft casing using a 20Q
N2 assisted low-polymer borate-crosslinked fracturing fluid.
The job was pumped at an average rate of 25 bpm and placed
193,000 lbs of 20/40 white sand. The treating pressures, rates,
and proppant concentrations throughout the job are shown in
Figure 1, along with the model history match.

A
10000

Casing Pressure (psi)


Foam Proppant Concentration (lb/gal)
GOHFER Surface Pressure (psi)
GOHFER Surface Prop Conc (lb/gal)

A BH Foam Rate (bpm)


C GOHFER Bottom Hole Pressure (psi)
A GOHFER Slurry Rate (bpm)
C

B
A
B

B
40

9000

C
10
9

8000

8
30

7000

6000

5000

20

4000

5
4

3000

3
10

2000

1000

00:50

5/21/2003

01:00

01:10

01:20

01:30

Time

01:40

01:50

5/21/2003

Figure 1: Treating pressure match for Well 1

The job was pumped to completion using a very small pad


volume and the final proppant concentration reached 6 ppa.
The early-time pressure rise is consistent with contained
height growth. During the 5 ppa stage the treating pressure
trend flattens out as some growth out-of-zone begins. The

relatively low matrix permeability and lack of accelerated


pressure-dependent leakoff allowed the job to be placed
without screenout. The total net pressure build was limited by
the partial height containment.
Figure 2 shows the final proppant distribution at the end of
the job, as modeled by the simulator. The proppant
concentration is represented by color-fill over the surface of
the created fracture. The color scale at the bottom of the figure
gives proppant concentration in lb/ft2. The fracture half-length
is indicated by the distance markers superimposed on the
color-map. The log-track at the left edge of the plot shows the
gamma-ray (GR) trace over the zone of interest. The
perforation interval is shown by the green bar in the GR track.
The model accurately matched the onset of downward height
growth corresponding to the change in the treating
pressure trend.

Figure 2: Final proppant distribution for Well 1

The fracture treatment of Well 1 resulted in a propped


fracture half-length of 1460 feet with an average proppant
concentration of 1.3 lb/ft2. Proppant was relatively uniformly
distributed throughout the fracture length and height.
Volumetric fluid efficiency at shut-down was calculated to
be 55.4%.
Well Case 2
Well 2 was also treated down 4.5, 11.6 #/ft casing. The
fluid was also a nitrogen assisted low-polymer boratecrosslinked guar. The job was pumped at a total rate of about
24 bpm and placed approximately 210,000 lbs of 20/40 mesh
white sand. The target pay interval was partially pressure
depleted by offset production. The pore pressure depletion,
along with other rock properties and stress contrasts, resulted
in effectively perfect fracture height containment. Figure 3
shows the simulator pressure history match of the job.
Confined height growth is indicated by the constantly
rising treating.5 The model accurately matches the observed
surface pressure through the post-job flush, including the rate
variations at the end of the job. The field-calculated
bottomhole pressure, shown as the green curve in Figure 3, is
subject to field interpretations of pipe friction, slurry density,
and N2 quality and may not be accurate. The job was flushed
to completion with no indication of impending screenout.

SPE 84491

Casing Pressure (psi)


BH Foam Rate (bpm)
GOHFER Bottom Hole Pressure (psi)

A
B
A
B

A
GOHFER Slurry Rate (bpm)
12000

Calc'd BH Pressure (psi)


Foam Proppant Concentration (lb/gal)
GOHFER Surface Pressure (psi)
GOHFER Surface Prop Conc (lb/gal)

A
C
A
C

B
30

C
12

10000

25

10

8000

20

6000

15

4000

10

2000

10:40

10:50

5/21/2003

11:00

11:10

11:20

11:30

Time

Job Date: for Well 2


Figure 3:Customer:
Treating pressure match
UWI:
Well Description:

Ticket #:

11:40

5/21/2003

StimWin v4.3.0
21-May-03 10:40

Figure 4 shows the final proppant distribution and created


fracture geometry corresponding to the pressure history match
shown in Figure 3. In this case the fracture is contained to the
pay zone and has a propped half-length of 1700 feet with an
average proppant concentration of 1.0 lb/ft2 over the
entire length.

iterative in nature, coupling material balance, decline curves


and traditional pressure transient techniques into a
comprehensive tool.
GPA uses a plotting technique presented by Agarwal and
Gardner9,10 to estimate effective drainage area. This technique
uses a plot of reciprocal dimensionless wellbore pressure,
1/PwD, from equation (3) verses dimensionless cumulative
production, (QDA) from equation (2), to generate the ratecumulative-decline curve. The plot forms a straight line
tending towards 1/2 during boundary dominated flow. The
benefit of this plot is that the resulting estimate of drainage
area is independent of the reservoir permeability. Therefore
the formation flow capacity, or reservoir transmissibility, need
not be known to estimate effective drainage area.
Q DA =

4.5Tz i Gi m( p ) ....... (2)


t DA
=
p wD
hAp i m( p )

p wD =

khm( p )
.... (3)
1422Tq (t )

Where

Production history analysis


The post-frac production history of the two wells was
analyzed to determine the behavior of the coupled reservoir
and fracture using Gas Production Analysis (GPA). GPA is a
type curve matching technique which provides a consistent
estimate of reservoir and completion parameters.6 It is based
on earlier work by Fraim and Wattenberger7, Palacio and
Blasingame8 and Agarwal and Gardner9,10 Flowing bottomhole pressures and production rates are treated as a long term
production draw-down test providing an estimate of effective
drainage area, flow capacity and effective fracture half length
through type curve analysis.
A pseudo equivalent time function9, ta, given by equation
(1) is used to allow gas well production with both varying rate
and pressure to be analyzed using constant-rate type
curve solutions.
ta =

(c )

(c )
q(t )dt
=

q(t ) ( p)c ( p) q(t )


g i

g i

zi Gi
[m( p)] ..(1)
2 pi

0.006328 kt ..(4)
(ct )i A

Effective drainage area will continue to increase through


the transient flow period until all flow boundaries have been
reached. However, it is important to recognize the effects of
the pressure depletion in the near well area when performing
GPA. Once the effective drainage area has been determined,
an estimate of total flow capacity and completion parameters
can then be obtained.
Early time production trends provide qualitative insight
into completion efficiency and can also provide an indication
of near-well flow geometry. Figure 5 shows the characteristic
shapes of the linear, uniform, and infinite acting flow
geometries. The linear flow response can be a result of
channel flow or an indication of directional reservoir
permeability heterogeneities.
Infinite Conductivity Fracture

100
Kh = 3.27 md-ft
Xf = 200 ft

10

Boundary Dominated
Uniform Flow

Infinite Acting Flow

Boundary Dominated
Linear Flow

PwD

Figure 4: Final proppant distribution for Well 2

t DA =

0.1
0.0001

0.001

0.01

0.1

10

100

tDA

The average reservoir pressure must be determined to


calculate the correct pseudo equivalent time. From material
balance, the average reservoir pressure can be calculated if the
effective drainage area is known. Therefore GPA is inherently

Figure 5: Representation of various flow geometries on the


diagnostic type-curve

SPE 84491

Effective Drainage Area 182 acres - Well 1

0.9

Actual
Analytical

0.8
0.7
0.6
1/PwD

Data in Figure 5 were calculated for a reservoir flow


capacity of 3.27 md-ft and an effective fracture half length of
200 ft. The type-curves show that for dimensionless times less
than 0.01 all three geometries perform similarly.
Figure 6 shows the characteristic type-curve shapes for a
well producing in pseudo-radial flow and boundary dominated
fractured-well flow. The plot also includes actual well
performance data from an East Texas producer. The well was
produced naturally for 90 days at an average rate of 420
Mscf/D. The well was then fracture stimulated and placed on
production at rates in excess of 2,000 Mscf/D. The early-time
radial flow behavior and the later-time fractured-well behavior
are both shown on the type-curves in Figure 6. The data points
are the actual well data and the lines represent the GPA model
results. All reservoir parameters are the same for both the prefracture and post-fracture models. The change in type-curve
shape is caused only by the near-well and fracture properties.

0.5
0.4
0.3
0.2
0.1
0.0
0

0.02

0.06

0.08

0.1

0.12

0.14

Figure 7: GPA rate cumulative decline curve, Well 1

Effective Drainage Area 130 acres - Well 2

0.1
Kh = 3.3 md-ft
Boundary Dominated
Flow

Inifite Acting
Pseudo Radial Flow

0.1

1/PwD

PwD or PwD'

Actual
Analytical

0.1

10

0.16

QDA

East Texas Example 2 to 1 Rectangular Boundary at 1,017 Days

100

0.04

Uniform Flux Fracture


Xf = 380'

0.1
0.1
0.0

Actual PwD
Actual PwD'
Analytical PwD
Analytical PwD'
0.1
0.0001

0.0
0.0

0.001

0.01

0.1

10

100

0.02

0.04

0.06

tDA

GPA Analysis of example wells


The initial step in the GPA analysis is to estimate the
effective drainage area affected by the cumulative production
from the well. This is determined from the rate-cumulativedecline curve. Figures 7 and 8 show the plots for the two
examples wells. The field production data, reduced to 1/PwD
and QDA, are shown as the data points. The red line represents
the rate-cumulative-decline curve for the reservoir effective
drainage area and estimated transmissibility used in the GPA
model. The data points plot along a line that extrapolates to
the y-intercept at QDA=1/2 as required for consistency. The
slope of the line is proportional to the assumed
reservoir transmissibility.
Once the effective drainage area has been determined the
reservoir flow capacity, or transmissibility, and effective
fracture half length can be derived from a semi-log plot of
normalized pressure versus adjusted time11 and from type
curve matching. The use of the log-log diagnostic type-curve
and other plots (not shown) assist in arriving at a more unique
model of the reservoir. The difference in behavior between
effectively fracture-stimulated wells and radial-flow wells is
clearly shown on the type curve in Figure 6.

0.1

0.12

0.14

0.16

QDA

Figure 8: GPA rate cumulative decline curve, Well 2

Figures 9 and 10 present the final type curve matches for


the two example wells. Note that both match curves have a
much flatter slope, and the curves plot higher on the type
curve than the fractured-well type-curves shown in Figure 6.
The general shape and position of the type-curves in Figures 9
and 10 indicates marginally stimulated or radial flow behavior
for both example wells.
Infinite Conductivity Fracture in 1 to 1 Rectangular Boundary at 384 Days

100

Kh = 3.27 md-ft
Xf = 22 ft

10
PwD or PwD'

Figure 6: Field example of type-curve signatures

0.08

1
Actual PwD
Actual PwD'
Analytical PwD
Analytical PwD'
0.1
0.0001

0.001

0.01

0.1
tDA

Figure 9: GPA type-curve match, Well 1

10

100

SPE 84491

Radial Flow and 1 to 1 Rectangular Boundary

2500

Radial Flow in 1 to 1 Rectangular Boundary at 650 Days

Forecast Rate

1000
kh = 31.45 md-ft
skin = 0.5

2500

Actual Rate
2000

Bottomhole Pressure 2000

1500

1500

1000

1000

500

500

10

Actual PwD
Actual PwD'
Analytical PwD
Analytical PwD'

0.1

0.01
0.001

0
0

0.01

0.1

10

100

200

100

400

500

600

0
700

Time (Days)

tDA

Figure 12: GPA history match, Well 2

Figure 10: GPA type-curve match, Well 2

The combination of independent and redundant curve


matches used in the GPA analysis gives a relatively unique
model of the coupled reservoir and well geometry. The GPA
analysis results for the two example wells are summarized in
Table 2, along with the results of the subsequent pressurebuild-up (PBU) analysis. The model indicates very short
effective fracture half-lengths for both wells. Well 1 has an
apparent effective length of 22 feet. The production analysis
for Well 2 shows pseudo-radial flow behavior, with no
apparent fracture stimulation. Both wells were fracture treated
with more than 200,000 lbs of proppant.
The final step in the GPA process is to compare the
predicted production performance from the type curve match
to the actual well performance. Figures 11 and 12 show the
actual rate and pressure histories for each well, compared to
that predicted from the GPA model.

Infinite Conductivity Fracture and 1 to 1 Rectangular Boundary


1400

Pressure transient results


Pressure build-up (PBU) tests were run on both wells after
significant post fracture-stimulation production. Neither test
was run long enough to reach pseudo radial flow, therefore an
independent estimate of both reservoir flow capacity and
fracture half-length could not be obtained. This shows one
drawback to the use of post-fracture build-up analysis in low
permeability formations.
The shut-in time required to reach pseudo radial flow can
be estimated using equation (5) for a well producing from an
infinitely conductive fracture.12 For the example wells a buildup time of more than 20 weeks would be required to reach
pseudo-radial flow. The extended time required to reach
pseudo radial flow following a fracture stimulation treatment
highlights the need for a pre-frac pressure transient test or the
incorporation of the GPA technique into the post fracture
evaluation process. For consistency the fracture half-length
used in equation (5) should be based on the static PBU
analysis or the designed half-length.

2500
Forecast Rate

t Dxf =

Actual Rate

1200

Bottomhole Pressure

2000

1500

800
600

1000

400
500
200
0
0

50

100

150

200
Time (Days)

Figure 11: GPA history match, Well 1

250

300

350

0
400

Pressure (psia)

1000
Rate (Mscf/D)

300

Pressure (psia)

Rate (Mscf/D)

PwD or PwD'

100

0 . 006329 kt

c t x

3 ....(5)

2
f

Figures 13 and 14 show the log-log type curve match for


each well. The type-curve matches were achieved using the
GPA predicted reservoir transmissibility in lieu of the pseudoradial flow value from the build-up test. Use of the GPA
reservoir flow capacity ensures that the reservoir description
used in all models is consistent. The fracture half-lengths
predicted by the PBU analyses are much longer than those
given by GPA. As shown in Table 2, the PBU fracture length
for Well 1 is 396 feet, and for Well 2 is 588 feet. Both
fractures
are
characterized
by
finite-conductivity
flow behavior.

SPE 84491

1.E+09

m(p) or m(p)', psi/cp

m(p)
1.E+08

m(p)'
regression

1.E+07

1.E+06

1.E+05
0.01

0.1

10

100

1000

t, hours

Figure 13: Type-curve match of pressure buildup, Well 1


1.E+09

m(p) or m(p)', psi/cp

m(p)
1.E+08

m(p)'
regression

k f v
(6)
Fnd = 1 +

1.E+07

1.E+06

1.E+05
0.01

0.1

10

100

1000

t, hours

Figure 14: Type-curve match of pressure buildup, Well 2

Using consistent reservoir transmissibility values the GPA


and PBU analyses give very good type-curve matches of their
respective data sets. There remains an apparent discrepancy
between the created fracture half-length of more than 1000
feet, as predicted by the fracture treatment history match, the
PBU length of 400-600 feet, and the effective length of less
than 30 feet for both wells. Roughly 200,000 lbs of proppant
were placed in both wells with good fracture height
containment. Material balance constraints alone strongly
suggest that a long fracture was created and propped. It is also
clear that the production history of the well, which is matched
by the short effective lengths, cannot be matched with the
longer fracture lengths unless fracture conductivity is
extremely small.
Table 2 - Summary of Transient Analysis Results
GPA Well 1 PBU Well 1 GPA Well 2 PBU Well 2
Infinite
Finite
Radial
Finite
Conductivity Conductivity
Conductivity
Type Curve Model:
Fracture
Fracture
Fracture
Permeability, md
0.155
0.155
0.85
0.85
Area, Acres
182
na
130
na
Half length, ft
22
369
na
558
Skin
0
0
0.5
0.414
Fracture Conductivity, md-ft
na
582
na
844
Well:

Prediction of effective length


The apparent length discrepancy can be partially explained by
understanding the various ways that effective fracture length
can be described. Effective or apparent fracture length can be
described by an infinite conductivity effective length, uniform
flux effective length, and finite-conductivity effective length.
Within each of these descriptions there are variations in the
handling of non-Darcy, or turbulent, pressure losses. The
way a fracture is described is primarily dependent on the
reservoir modeling software application used to forecast postfracture production, and there is no consistency in the
industry. This leads to considerable confusion solely based
on semantics.
Most type-curve models ignore, or are unable to rigorously
handle, non-Darcy effects and require that the fracture be
described by an apparent Darcy-conductivity that accounts for
the rate-variable increase in pressure drop along the fracture,
or a corresponding drop in conductivity. The appropriate
multiplier (Fnd) to the Darcy pressure-drop, or divisor to the
actual proppant pack conductivity can be derived from the
Forchheimer equation.13

In equation (6) the permeability is the in-place


permeability of the proppant pack under closure stress and
containing gel residue, filter-cake, and multiple flowing
phases. The fluid properties, and , refer to the average
flowing phase, the velocity (v) is the superficial velocity in the
fracture and is the Forchheimer turbulence factor for the
proppant pack. To account for the velocity profile along the
length of the fracture, a velocity of 2/3 the sandface velocity
should be used.
In the GPA and type-curve modeling presented here the
effective fracture length has been described by the effective
infinite-conductivity half-length. This results in a single
parameter to describe the behavior of the well-fracture
coupled flow system. When using a finite conductivity
fracture model, as in the PBU analysis, the conductivity and
length can be traded-off in a number of combinations to
describe the system. This leads to non-unique descriptions of
fracture effectiveness and can add to the confusion. While the
use of an effective infinite-conductivity length does not do
justice to the actual created length, it does give a valid basis
for comparison of treatment effectiveness. Still, it is necessary
to understand the causes and effects of the various damage
mechanisms that impact ultimate fracture conductivity and
effective length.
The sources of fracture conductivity damage are covered
in detail in a companion paper.13 The various damage
mechanisms will be enumerated for the two example wells,
and their individual and collective impact on fracture
performance quantified. The estimation of the impact of
damage mechanisms is determined by an iterative solution
wherein the reservoir deliverability is coupled to the fracture
conductivity and determines the degree of clean-up in the
fracture. The fracture conductivity cannot be determined from

SPE 84491

material properties (proppant and fluid) alone, but must be


coupled to the flow rate delivered from the reservoir. The
reservoir production model coupled to the clean-up simulator
uses the same formulation outlined for GPA.
Well 1 fracture clean-up estimates
The reservoir properties for Well 1, as shown in Table 1,
were input to the fracture clean-up model (Predict-K). The
model was run using the observed well flowing pressure and
approximately matching the observed well flow-rate history.
The summary of fracture conductivity damage results for Well
1 is shown in Table 3. Table 4 shows the fractional loss
(multiplier) of conductivity caused by each major damage
mechanism. The combined product of all these conductivity
losses is given as the Overall Multiplier.
Proppant:

Table 3: Fracture conductivity damage mechanisms for Well 1


20/40 White Sand
Proppant conc. (lb/sq ft):

1.30

Fluid:

Low Polymer Borate

Frac. half length (ft):

1460.0

Pack permeability (d):


Internal Pack width (in):
External Pack width (in):
Gas velocity (cm/sec):
Oil velocity (cm/sec):
Water velocity (cm/sec):
Beta (atm*sec2/g):
Non-Darcy factor:
Conductivity (md*ft):

86.7
0.063
0.134
12.1
0.008
0.015
0.00364
5.16
34.5

Gas fractional flow:


Rel. perm. factor:
Regained perm. (%):
Xf at FCD (ft):
Gas flow rate (Mscf/d):
Flow rate ratio:
Inf. half length (ft):
FCD:
Net Closure Stress (psi):

0.998
1.19
46.8
45.0
1430.
0.17
120.0
0.158
5820.

pseudo-yield-point. The tendency for the gel residue to act as a


solid, when at rest, requires a minimum pressure gradient to
establish shear and clean-up in the fracture. Presently this is
accounted for using an empirical relation to a characteristic
critical incremental FCD for fracture clean-up. The adjusted
effective fracture length is given by Xf at FCD in Table 3. The
Xf at FCD represents the final apparent fracture length under
flowing conditions and should be equivalent to that observed
from production analysis. For Well 1 the created and propped
fracture length of 1460 feet yields an apparent fracture length,
during production, of 45 feet according to this model. The
GPA analysis (Table 2) gives a length of 22 feet. While not
exact, the model comes close to representing the actual
effective fracture length.
The amount of damage associated with each mechanism is
shown in the summary in Table 4. The table shows that the
most significant losses to conductivity are caused by nonDarcy flow, filter-cake, and gel residue (regained perm). The
loss of effective fracture length by tip plugging, caused by
insufficient pressure gradient to initiate clean-up, is also
significant. If the dynamic conductivity losses are assumed to
approach zero after shut-in, the effective fracture length and
conductivity can be estimated from equations (7) and (8)
where kfwf is adjusted for only static damage mechanisms,
such as width loss to filter-cake and spalling.

Table 4: Summary of Damage for Well 1


Damage Mechanism
Width Loss
Rel Perm Effect
Regained Perm
Non-Darcy Loss
Overall Multiplier
Xf loss to Tip Plugging

Conductivity
Loss
0.47
0.84
0.468
0.194
0.036
0.375

The various damage mechanisms accounted for include


loss of pack width and permeability due to closure stress on
the proppant pack. The effects of pack compression are shown
as Pack permeability and External Pack Width. The Internal
Pack width results from additional width losses due to
dynamically deposited filter-cake residue and formation
spalling into the proppant pack. These are internal width
losses and are not apparent from outside measurements of
pack width. The amount of filter-cake depends on polymer gel
loading, type of polymer and breaker, fluid efficiency, fracture
geometry and producing flow rate. High flow rates tend to
abrade and erode the filter-cake. The Non-Darcy flow factor
(Fnd) has already been defined.
The Rel-Perm factor accounts for loss of total mobility in
the proppant pack caused by multiphase flow conditions. In
the case of Well 1 the produced gas was accompanied by
about 5 bbl/MMSCF condensate and 9 bbl/MMSCF water. For
the observed gas rate, this results in a loss of about 20% of the
pack conductivity. The Regained Perm factor quantifies the
permeability loss caused by gel residue dispersed throughout
the pack (not in the filter-cake) and is the typical measure of
fluid clean-up a proppant pack.
The Infinite Conductivity Half-Length gives the effective
fracture length after accounting for these damage factors. One
mechanism that is left un-accounted for is the effect of gel

FCD =

k f wf

.(7)

kr X fd
X fa =

X fd
FCD

1 .0 +

1 .7

1.01

..(8)

The value of Xfd is the designed or created half-length and


kfwf represents the proppant pack conductivity for the
appropriate damage condition (static damage only). In the case
of Well 1, the apparent fracture length under non-flowing
conditions, accounting for static damage mechanisms, is 301
feet with a conductivity of 455 md-ft. This compares well with
the results of the pressure build-up analysis in Table 2.
Well 2 fracture clean-up estimates
The results for a similar analysis of the Well 2 data are
shown in Tables 5 and 6. The degree of damage for this case is
similar to the first example.
Table 5: Fracture conductivity damage mechanisms for Well 2
Proppant:

20/40 White Sand

Proppant conc. (lb/sq ft):

Fluid:
Pack permeability (d):
Internal Pack width (in):
External Pack width (in):
Gas velocity (cm/sec):
Oil velocity (cm/sec):
Water velocity (cm/sec):
Beta (atm*sec2/g):
Non-Darcy factor:
Conductivity (md*ft):

Low Polymer Borate


99.3
0.042
0.104
13.6
0.012
0.008
0.00281
6.71
22.4

Frac. half length (ft):


Gas fractional flow:
Rel. perm. factor:
Regained perm. (%):
Xf at FCD (ft):
Gas flow rate (Mscf/d):
Flow rate ratio:
Inf. half length (ft):
FCD:
Net Closure Stress (psi):

1700
0.999
1.15
49.4
5
2570
0.26
15
0.016
5420

SPE 84491

The effective flowing fracture half-length for Well 2 was


estimated accounting for all dynamic damage mechanisms.
The result is an effective length of 5 feet, although the
propped half-length was 1700 feet. This agrees with the GPA
production analysis that indicates that the well can be
described as in radial flow with no effective stimulation.
Under static damage conditions, the data for Well 2 yields an
effective length of 207 feet and conductivity of 347 md-ft.
This is somewhat more pessimistic than the pressure build-up
analysis results shown in Table 2 but demonstrates the
difference between flowing and non-flowing conditions.
Summary of results
The results of the two well cases are summarized in Table 7.
The table shows the apparent fracture half-lengths, in feet,
arrived at from each type of analysis. The static pressure
build-up analyses consistently give the longest apparent
fracture half-lengths, but these are shorter than the inferred
propped lengths from the fracture treatment modeling. The
difference between the static PBU lengths and model propped
lengths may be related to gel plugging and pseudo yield-point
behavior which eliminates flow contributions from a portion
of the fracture tip.
The effective producing length, as shown by GPA
production analysis, is much shorter than the apparent PBU
length. The difference is attributed to dynamic flow effects
associated with multiphase and non-Darcy flow. The
magnitude of the impact of these cumulative damage effects is
approximated by the fracture clean-up model. These results
represent long-term stimulation response after unloading and
clean-up of the fracturing fluids. Different results may be
observed for short-term flow tests, when clean-up and
saturation profiles have not stabilized, or in wells subject to
extensive liquid loading.
Table 7 - Comparison of Fracture Lengths
Method
Well 1
Well2
Fracture Model Xf
1460
1700
Pressure Build-Up Xf
369
558
GPA Production Xf
22
0.35
Est. Static Xf
301
207
Est. Flowing Xf
45
5

In some cases a short infinite-conductivity fracture length


can be used to describe a longer finite-conductivity fracture. In
the case of the examples shown here, the difference in flow
behavior is much larger than can explained by the differences
in the fracture descriptions. Figures 14 and 15 show the actual
production history of the two wells. The modeled gas flow
rates using the fracture descriptions from GPA, PBU, and the

5000

1000

4500

900

GPA Rate, MSCF/day


Actual Rate, MSCF/day
PBU Rate, MSCF/day
PDK Rate, MSCF/day
WBHP psia

4000
3500

800
700

3000

600

2500

500

2000

400

1500

300

1000

200

500

100

0
0

50

100

150

200

250

300

350

BH Pressure (psi)

0.026
0.333

0
400

Time (Days)

Figure 14: Comparison of predicted post-frac rates for Well 1


5000

1000
GPA Rate, MSCF/day
Actual Rate, MSCF/day
PBU Rate, MSCF/day
PDK Rate, MSCF/day
WBHP psia

4500
4000

900
800

3500

700

3000

600

2500

500

2000

400

1500

300

1000

200

500

100

0
0

100

200

300

400

500

600

BH Pressure (psi)

Overall Multiplier
Xf loss to Tip Plugging

Conductivity
Loss
0.4
0.869
0.494
0.149

Gas Rate (MSCF/D)

Damage Mechanism
Width Loss
Rel Perm Effect
Regained Perm
Non-Darcy Loss

Predict-K analyses are shown for comparison. The same


flowing pressure history was used to drive each production
calculation. In the figures the bottomhole flowing pressure is
shown by the dashed line, the actual gas rate is shown by the
data points, and the calculated rates from each model are
shown by the lines. The GPA analysis results match the actual
production nearly exactly. The Predict-K results slightly overpredict the well rate history. The PBU result, even using a
finite-conductivity fracture model, over-predicts the observed
flow rates by more than a factor of two. The convergence of
late-time rates is caused by depletion of the drainage area of
the well, as predicted by the model.

Gas Rate (MSCF/D)

Table 6: Summary of Damage for Well 2

0
700

Time (Days)

Figure 15: Comparison of predicted post-frac rates for Well 2

Conclusions
The integration of available completion, reservoir and well
performance information using appropriate analytical tools
results in a consistent evaluation method which can be used to
evaluate post fracture-stimulation production performance.
The analytical techniques presented in this paper have been
used as a basis for recommending operational changes
concerning fracture treatments performed in this field. Figure
16 demonstrates the results of the stimulation improvements
by showing the first 21 months production for recent infill
wells, compared to the same producing period for older offset
wells. The new wells are the 3rd and 4th infill wells per section,
but their performance is still better than the offset wells. The
improvement in well performance obtained by implementing
treatment design changes suggested by this analysis process

SPE 84491

shows that understanding past performance is the key to


improving future performance.
3rd and 4th well per section down spacing
2000

Mscf/D Average Per Well

1800
8 Recent Completions

1600
1400
1200
1000
800

27 Offset completions

600
400
200
0

10 11 12 13 14 15 16 17 18 19 20 21

FIRST 21 MONTHS PRODUCTION (Time Normalized)

Figure 16: Comparison of offset production to recent infill


well performance

References
1. Barree, R. D.: A Practical Numerical Simulator for ThreeDimensional Fracture Propagation in Heterogeneous Media,
paper SPE 12273 presented at the Reservoir Simulation
Symposium held in San Francisco, CA, November 15-18, 1983.
2. Barree, R. D. and Conway, M. W.: Experimental and Numerical
Modeling of Convective Proppant Transport, paper SPE 28564
presented at the SPE 69th Annual Technical Conference and
Exhibition held in New Orleans, LA, U.S.A., 25-28
September 1994.
3. Barree, R. D. and Winterfeld, P. H.: Effects of Shear Planes and
Interfacial Slippage on Fracture Growth and Treating Pressures,
paper SPE 48926 presented at the 1998 SPE Annual Technical
Conference and Exhibition held in New Orleans, Louisiana,
27-30 September.
4. McGowen, J. M., Barree, R. D. and Conway, M. W.:
Incorporating Crossflow and Spurt-Loss Effects in Filtration
Modeling Within a Fully 3D Fracture-Growth Simulator, paper
SPE 56597 presented at the 1999 SPE Annual Technical
Conference and Exhibition held in Houston, Texas, 3-6
October 1999.
5. Gidley, J. H., Holditch, S. A., Nierode, D. E., and Veatch, R. W.:
Recent Advances in Hydraulic Fracturing, SPE Monogaph
Volume 12, Chap 14, SPE, Richardson, Texas, 1989.
6. Cox, S. A., Gilbert, J. V., Sutton, R. P., and Stoltz, R. P: Reserve
Analysis for Tight Gas, paper SPE 78695 presented at the 2002
SPE Eastern Regional Meeting, Lexington Kentucky,
23-26 October.
7. Fraim, M.L. and Wattenbarger, R.A.: Gas Reservoir Decline
Curve Analysis Using Type Curves with Real Gas Pseudopressure
and Normalized Time, SPEFE (Dec 1987) 671-82.
8. Palacio, J.C. and Blasingame, T.A.: Decline-Curve Analysis
Using Type Curves Analysis of Gas Well Performance Data,
paper SPE 25909 presented at the 1993 Rocky Mountain Regional
Meeting/Low Permeability Reservoirs Symposium and Exhibition,
Denver, 26-28 April.
9. Agarwal, R.G., Gardner, D.C., Kleinsteiber, S.W. and Fussell,
D.D.: Analyzing Well Production Data Using Combined-TypeCurve and Decline-Curve Analysis Concepts, SPEREE (October
1999) 478.
10. Gardner, D.C, Hager, C.J. and Agarwal, R.G: Incorporating
Rate-Time Superposition Into Decline Type Curve Analysis,
paper SPE 62475 presented at the 2000 Rocky Mountain Regional

Meeting/Low Permeability Reservoirs Symposium, Denver,


12-15 March.
11. Earlougher, R.C. Jr.: Advances in Well Test Analysis, Monograph
Series, SPE, Richardson, TX (1977) 5.
12. Lee, W.J. and Wattenbarger, R.A.: Gas Reservoir Engineering,
SPE Textbook Series Volume 5 (1996).
13. Barree, R. D., Cox, S. A., Barree, V. L. and Conway, M. W.:
Realistic Assessment of Proppant Pack Conductivity for Material
Selection, paper SPE 84306 presented at the SPE Annual
Technical Conference and Exhibition held in Denver, Colorado,
U.S.A., 5 8 October 2003.

Acknowledgements
The authors would like to acknowledge the support of
Marathon Oil Company and its permission to publish this
paper, with the associated field data. We would also like to
acknowledge the work of Stim-Lab, Inc. for its development
of the Predict-K model and its continue efforts to understand
fracture clean-up and damage mechanisms.
Nomenclature
A = drainage area, ft
cg = gas compressibility, psi-1

cw =
ct =
FCD=
Gi =
k =
m( p) =
m ( p ) =
m ( p ) =

water compressibility, psi-1


total compressibility, psi-1
dimensionless fracture conductivity
original gas in place, Mscf
effective permeability to gas, md
real gas pseudo pressure, psi2/cp
m( pi ) m( p ) , psi2/cp
m ( pi ) m ( pwf ) , psi2/cp

pwf = bottomhole producing pressure, psia


p = average reservoir pressure, psia
pi = initial reservoir pressure, psia
p wD = dimensionless wellbore pressure
q(t ) = flow rate, Mscf/D
Q(t ) = cumulative production, Mscf
Q DA = dimensionless cumulative production based
on area (A)
t = time, days
T = reservoir temperature, R
tDxf = dimensionless time based on fracture half length
t a = pseudoequivalent time, days
t DA = dimensionless time based on area (A)
w = fracture width
xf, Xf = fracture half length, feet
z i = gas compressibility factor at pi
= porosity, fraction
= viscosity, cp
= 3.14159

a
d
i
f
r

Subscripts
= apparent
= design
= initial
= fracture
= reservoir

You might also like