Académique Documents
Professionnel Documents
Culture Documents
net/publication/254552182
CITATIONS READS
6 256
3 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Ajay Suri on 05 November 2015.
Summary the fracture. Some models assume that the fluid enters the well-
A new model is developed that calculates the productivity of a bore radially (Binder and Raymond 1967), while others assume an
hydraulically fractured well, including the effect of fracture-face elliptical-flow field (Prats 1961). An accurate productivity model
damage caused by fluid leakoff. Results of the new model are com- for fractured wells is needed in conjunction with current fracture-
pared with three previous models (McGuire and Sikora 1960; Prats propagation models to make better engineering decisions about
1961; Binder and Raymond 1967). The existing models assume the fracturing process.
either elliptical or radial flow around the well, with permeability
varying azimuthally. Significant differences in the calculated well Background
productivity indicate that earlier assumptions made regarding Several authors have studied the effect of vertical fractures on
the flow geometry can lead to significant overestimates of well hydrocarbon production (McGuire and Sikora 1960; Prats 1961;
productivity index (PI). Agreement with the analytical solution of Binder and Raymond 1967). Most present their results by plotting
Prats (1961) is achieved for finite-conductivity fractures and no the productivity index J (q/P), normalized by a base productivity
fracture damage. It is shown that the use of either McGuire’s model index Jo. Jo is defined as the productivity index of an unfractured,
(McGuire and Sikora 1960) or Raymond’s model (Binder and Ray- undamaged well in a circular drainage area. With some assump-
mond 1967) to estimate improvement in well PI in fractured wells tions (see Model Formulation and Definitions section), Jo for
can lead to a significant overestimation of the well PI. The new steady state can be written as
model provides a useful tool to quickly calculate the productivity
of wells that have both a finite-conductivity fracture and damage 2 kh
Jo = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (1)
in the invaded zone. The simple and discrete nature of the model ln(re / rw )
makes it ideal for implementation in spreadsheets and to connect
to fracture-performance models. J/Jo is a dimensionless measure of the expected increase in produc-
Cleanup of the damage in the invaded zone depends on the tivity compared with an unfractured, undamaged well.
capillary properties of the formation and the drawdown pressure McGuire and Sikora (1960) were the first authors to present
applied across the damaged zone during production. If capillary systematic results for fractured-well productivity. They developed
forces are small and drawdown pressure is high, the water will a graphic method to estimate the PI of a well with a constant-width
be recovered, resulting in negligible damage. It is found that the fracture in a square drainage area. Their results are displayed as
invaded zone will cause significant damage when the permeability graphs of PI vs. relative fracture conductivity.
of the damaged zone is reduced by more than 90%. For low-perme- Binder and Raymond (1967) included the effect of fracture
ability, depleted formations where water recovery is poor, the frac- conductivity in their model. They also included the effect of a
turing fluid should be energized with a gas component so that the damaged zone around the wellbore. Their model assumes radial
relative permeability damage to gas inflow can be minimized. flow with an average radial permeability (averaged azimuthally).
The radial permeability is a function of the fracture permeability,
Introduction fracture width, and the reservoir permeability. This assumption is
Hydraulic fractures increase production of oil and gas wells by valid for short fractures where the flow is near radial around the
creating a highly conductive connection between the wellbore wellbore. It is shown in this paper that their model provides an
and the reservoir. Productivity of a fractured well is a function overestimation of productivity for a fractured well.
of the reservoir drainage area, fracture dimensions, fracture con- The Prats model (Prats 1961) assumes elliptical flow around
ductivity, formation conductivity, and the characteristics of any the fractured well in an elliptical drainage area. The dimensions of
damage created during the process. Almost always some polymer the equipressure ellipses around the fractured well depend mainly
gel and water leak off into the formation during well fracturing, on a function of a dimensionless parameter aprats:
which results in a damaged or invaded zone around the fracture.
kL
In the invaded zone, the high water saturation reduces the relative aprats = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2)
permeability of gas/oil. In addition, the invasion of polymer can 2 k f wave
result in the formation of a low-permeability internal filter cake The model presented in this paper also assumes a similar con-
that can have a large impact on the well productivity (even if the stant-pressure ellipse around the fractured well. Prats introduced
invasion depth is small). the effect of damage in the invaded zone for infinitely conductive
Currently, several models exist that calculate the productivity fractures (but not for finitely conductive fractures), as shown in
of fractured wells producing under steady-state or pseudosteady- Eq. 3:
state conditions. However, none of these models can compute the
productivity of a fractured well with both a finite-conductivity ⎛r ⎞
fracture and damage in the invaded zone around the wellbore and ln ⎜ e ⎟
J qf q ⎝ rw ⎠
= = . . . . . . . . . . . . . . . . (3)
J o qo q f ⎛ 2r ⎞ 4 h ⎛ k ⎞
ln ⎜ e ⎟ + leak ⎜ − 1⎟
⎝ L ⎠ L ⎝ kd ⎠
Copyright © 2010 Society of Petroleum Engineers
This paper (SPE 119264) was accepted for presentation at the SPE Hydraulic Fracturing Cinco-Ley and Samaniego (1977) included damage in the
Technology Conference, The Woodlands, Texas, USA, 19–21 January 2009, and revised
for publication. Original manuscript received 10 November 2008. Revised manuscript invaded zone in addition to finite fracture conductivity. They
received 21 May 2010. Paper peer approved 24 June 10. did not assume steady-state but instead solved for the transient
b2
b1,i
h leak,i
wi /2
rw
L
a1
a2
2( 2 L4 + 4 A2 − L2 ) a prats=10
b2 = . . . . . . . . . . . . . . . . . . . . . . . . (4)
2
A
a2 = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (5) b 1,0
(b2 )
q 1
J= = , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (6)
P Rtot
where Rtot is calculated by the combination of the circuit given in 0.135 a prats=0.1
Fig. 3. The definition of each resistance is shown in Appendix A.
a prats=0
J/Jo is calculated by dividing Eq. 6 by Eq. 1 and correcting for
only one-quarter of the area: Well
0 Fracture 1
J 2 ln(re / rw )
= . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (7)
Jo khRtot Fig. 2—Inner ellipse minor axis vs. aprats conductivity.
Damage in the Invaded Zone During Production. A damaged For illustration purposes the Brooks and Corey (1966) and
zone is formed when leakoff of water occurs during creation of the Corey (1977) models were used for calculating the capillary pres-
fracture. The zone may or may not cause significant damage to the sure and relative permeability of gas/oil as a function of phase
fracture productivity on the basis of the extent of the leakoff. Fig. 4 saturations.
shows a schematic of damage in the invaded zone after hydraulic
fracturing and during production. For cases where the permeability Pc = Pc o (Sw* )−1/ b , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (8)
and drawdown pressures are high, the water is removed efficiently.
In cases with low permeability and low drawdown pressure, the Sw − Swr
where Sw* = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (9)
damage remains significant because the viscous flowing forces are 1 − Swr
not sufficient to overcome the capillary forces. We assume that
the capillary force is the main force trapping the water and that kd
= krnw = krnwo (1 − S )n, . . . . . . . . . . . . . . . . . . . . . . . . . . . (10)
the relative permeability to gas/oil is a function of the capillary k
properties of the rock and the drawdown pressure only. The fol-
lowing simple model shows how relative permeability to gas/oil Sw − Swr
where S = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (11)
is estimated in the invaded zone. 1 − Swr − Snwr
R out
R el 1
R el 1
R el 2
Rd 1
Rd 2 R el 3
Rd 3
Rf 1 Rf 2 Rf 3 R elt
(a) (b)
Fig. 4 (a) Saturation of water in and around the fracture after hydraulic fracturing; (b): steady saturation of water in and around
the fracture during production.
Note that b is the Brooks-Corey exponent that depends on the the absolute permeability, porosity, and surface forces. The results
pore-throat distribution of the rock. Swr and Snwr refer to the residual collapse onto one dimensionless plot of capillary pressure vs. the
saturation of the wetting and nonwetting phase, respectively. Water saturation of the wetting phase. At high wetting-phase saturation,
is assumed to be the wetting phase, making the recovery of water the capillary pressure approaches the value Pco. Eq. 14 shows how
by gas/oil a drainage process. the calculation could be accomplished. The 0.42 in Eq. 14 is the
If the capillary pressure is much higher in the damaged zone dimensionless capillary pressure at high wetting-phase saturation.
than in the fracture (likely to be the case), the capillary pressure The other terms redimensionalize 0.42 into units of pressure. As
in the invaded zone is equal to the drawdown pressure across the an example, if the permeability is 0.1 md, porosity is 20%, and
invaded zone. Eqs. 8 and 10 are solved for an explicit expression cos = 50 dynes/cm, the value of Pco is 133.8 psi (beware of the
that relates the damaged-zone effective permeability kd to the units, 1 dyne/cm/md1/2 = 4.617 psi).
drawdown pressure P:
⎡ −b n
Pc o = 0.42 cos . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (14)
⎛ P ⎞ ⎤ k
⎢ (1 − Swr ) ⎜ o ⎟ ⎥
kd ⎢ ⎝ Pc ⎠ ⎥
= krnwo ⎢1 − . . . . . . . . . . . . . . . . . . . . (12)
k 1 − Snwr − Swr ⎥ Results
⎢ ⎥
⎣⎢ ⎥⎦ The results in this section were obtained by using the model
presented (UT-PI) with fracture and reservoir properties shown
The dependence on Pco can be removed by defining a dimension- in Table 1. The fracture dimensions are set by the ratio L/re and
less drawdown pressure: wmax. The fracture width starts at the maximum wmax at the wellbore
P and reduces elliptically to zero at L. The ratio hleak/w, a constant,
P * = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (13) simulates the invaded-zone distances being proportional to the
Pc o fracture width at any location. This simulates a hypothetical frac-
The parameter Pco defined in Eq. 8 is a measure of the capillary tured case that is meant to represent a typical system. Situations
pressure at high water saturation. If the region around the fracture that differ from the base scenario are evaluated by changing one
is fully saturated with the wetting phase, Pco represents the mini- variable at a time and discussed in subsequent figures and are in
mum drawdown necessary for any wetting phase to be recovered the discussion.
from that region. If the drawdown is less than Pco, the water satura-
tion will remain high, causing relative permeability damage that Nondamaged-Fractured-Well Productivity. Fig. 5 shows a com-
may permanently inhibit gas from flowing into the fracture. parison of the PIs obtained from the different models (UT-PI, Prats,
Leverett (1941) discusses capillary pressure in porous solids. McGuire, and Raymond) as a function of dimensionless fracture
The study concludes that the capillary pressure is a function of conductivity, FCD.
k f wave
TABLE 1—BASE-CASE SPECIFICATIONS FCD = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (15)
kL
Constant Values
k 0.01 md
12
wmax 0.10 in. UT-PI
A 40 acre 10 Prats
rw 3 in. McGuire
8 Raymond
Ratios
J/Jo
L/re 0.5 6
kd /k 0.001
4
hleak/w 100
kf /k 1000000 2
Necessary for Drawdown Pressure Calculation
0
Swr 0.2 0.001 0.01 0.1 1 10 100
Snwr 0.2 FCD
krnwo 0.9
n 3 Fig. 5—Comparison of production models for fractured wells
with no damage.
J /J o
J/Jo
6 6
L /re=0.99, Prats
4 4
2
2
0
0
0.001 0.01 0.1 1 10 100 0.001 0.01 0.1 1 10 100
FCD FCD
Fig. 6—Comparison of UT-PI to the Prats model (finite-conduc- Fig. 7—UT-PI master plot for undamaged fractures with finite
tivity fractures with no damage). conductivity.
The value of FCD was varied by changing kf /k. It is clearly seen Fig. 7 shows a complete set of results of the UT-PI model
that the productivity increase predicted by our model is consistent without damage around the fracture. This figure can be used as a
with the Prats model. Both of these models compute a productiv- master plot to calculate J/Jo for finite-conductivity fractures with
ity enhancement that is significantly lower than that predicted by no matrix damage around the fracture face. Any gel-induced dam-
the Raymond and McGuire models. Results from these models age in the fracture can be accounted for by varying the fracture
should, therefore, be considered overly optimistic (except for FCD conductivity.
values less than unity) in terms of the expected performance of
hydraulically fractured wells. Damaged-Fractured-Well Productivity. This section discusses
The differences between the models arise because of unrea- what happens when a damaged or invaded zone around the frac-
sonable assumptions in the earlier models. The Raymond model ture is added. Fig. 8 shows results for two cases of damage, kd/k
assumes radial flow in the reservoir. The permeability is calculated of 0.01 and kd /k of 0.001. The results can only be compared to
by averaging the reservoir and fracture permeability at each radial the Prats model for infinite-conductivity fractures. The value of
location around the wellbore. Assuming radial isopotential lines J/Jo is close to the Prats solution at high conductivity for kd/k of
results in an unrealistic increase in calculated productivity. This 0.01, but not for 0.001. In the case of large damage, the UT-PI
is not the case in bilinear-flow models where the fracture helps model redirects the flow from the damaged fracture into the radial
stimulate only the part of the region that is fractured, rather than the wellbore flow resistor, giving the fluid a low-resistance pathway.
entire circular zone. The differences between the McGuire model The Prats model assumes that flow must go though the fracture
and our model are more difficult to quantify because insufficient in order to get to the wellbore; therefore, the higher the damage,
details of the model are provided in the original paper (McGuire the more the Prats model and UT-PI will disagree. UT-PI yields a
and Sikora 1960). Differences could be a result of the assumptions more accurate prediction of J/Jo because it will redirect the fluid
of a square drainage area and constant-width fracture, which are into the nondamaged wellbore instead of forcing it through a highly
different from the assumptions in the Prats and the UT-PI model. damaged fracture.
The primary reason for the discrepancy might be that the McGuire Fig. 9 shows both the size and permeability of the invaded zone
model is based on physical experiments that do not represent the contributing to the reduction of productivity in fractured systems.
fractured-well system accurately. Both the McGuire and Raymond The reduction can be of paramount importance if either property
models have been quoted extensively in the literature. Field reaches an extreme value, shown by the PI ratio dropping to near
observations for the productivity of fractured wells almost always 1 in some cases. For most cases, the reduction is negligible if the
underperform predictions made by these models. Our results here ratio kd /k > 0.1. In other words, the invaded zone has to reduce in
show that at least a part of the explanation lies in the fact that the permeability by more than 90% in order to cause any significant
models overpredict the productivity enhancement because of the impact on well PI. This means that some water leakoff is accept-
creation of the hydraulic fracture. able as long as it does not reduce the hydrocarbon relative perme-
The agreement between the Prats model and the model pre- ability by more than an order of magnitude. In low-permeability
sented here is to be expected because the isopotential ellipses
are calculated in the same way. The model presented here, how-
ever, allows us to specify any distribution of damage around the 7
fracture (this is not possible in the Prats model). For L/re = 0.5, UT-PI, No damage
the results from the two models are shown to match. Results for 6
extreme values of L/re (0.1 and 0.99) are also shown in Fig. 6. UT-PI, kd /k=0.01
5
Both models give similar results for all values of L/re, even though UT-PI, kd /k=0.001
b1,0 is estimated for an L/re value of 0.4166 in the UT-PI model. 4 Prats, kd /k=0.01
J/J o
5 5
4 4
J /Jo
J /Jo
3 3 b=0.25
h leak /w =1. b=0.5
2 h leak /w =10. 2 b=1.0
h leak /w =100.
h leak /w =1000. 1
1
0 0
0.0001 0.001 0.01 0.1 1 0.01 0.1 1 10 100
kd /k ΔP*
sands and sandstones (tight gas sands and shales), the gas relative subsection discusses the effect of damage around the wellbore that
permeability has been shown to decrease by up to three or four may exist before a fracture is created: the type of damage that is
orders of magnitude as the water saturation is increased. In addi- present from drilling and well-completion fluids around the well.
tion, the trapped water is difficult to remove because the capillary Fig. 11 shows the PI ratio vs. fracture conductivity using the
pressure is high. parameters in Table 1.
Fig. 10 shows the same data as Fig. 9 for hleak/w =100 with the There are six different scenarios shown in Fig. 11. Three levels
x-axis converted to the dimensionless drawdown, P* (Eqs. 12 of damage around the wellbore (skin = 0, 10, 100) are shown, and
and 13). Fig. 10 shows the results for three different values of b, each is shown with a damaged and undamaged fracture. The skin,
the Brooks-Corey (Brooks and Corey 1966) exponent for capillary which is a dimensionless measure of pressure drop around the
pressure. The effect of Pco is removed by reporting the drawdown wellbore, is commonly calculated by combining the permeability
pressure in dimensionless terms. The results confirm that the and depth of the damage:
higher drawdown pressures will effectively recover all the water
and cause no damage. The opposite will be true for low drawdown ⎛k ⎞
pressures. If the drawdown pressure is less than Pco, then the water s = ⎜ − 1⎟ ln(rs / rw ), . . . . . . . . . . . . . . . . . . . . . . . . . . . . (16)
will remain trapped and it may be necessary to energize the frac- ⎝ ks ⎠
turing fluid by adding a gas component. The gas component will
keep the saturation of water low in the invaded zone and, therefore, where rs and ks are the radius and permeability of the damaged zone
keep the gas relative permeability high. around the wellbore, respectively. At low fracture conductivity, the
PI ratio is below unity for wells with skin. Remember that Jo is
Fractured-Well Productivity With an Initially Damaged Well. defined as a PI for an unfractured, undamaged well in a circular
The previous subsection discussed damage around the fracture that drainage area. Wells with damage around the wellbore and low
occurred because of leakoff during the creation of the fracture. This fracture conductivity will produce less than that amount, resulting
6
Skin=0, damage in
invaded zone
Skin=0, no damage
5 in invaded zone
Skin=10, damage
in invaded zone
4 Skin=10, no damage
in invaded zone
Skin=100, damage
J /Jo
3 in invaded zone
Skin=100, no damage
in invaded zone
0
0.001 0.01 0.1 1 10 100
FCD
2 ⎛ a2 + b2 ⎞ b1,t
Rout = ln Rel ,t =
hk ⎜⎝ a1 + b1,0 ⎟⎠
. . . . . . . . . . . . . . . . . . . . . . . . . . (A-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-9)
kh(a1 − L )
The factor in Eq. A-3 is 2/ and not 1/2 because the model
is only for one-quarter of the drainage area. A radial-flow resistor Kyle Friehauf is a completions engineer for ConocoPhillips in
connects the wellbore to the inner ellipse. This resistor is the only the Completions and Production Technology group. Friehauf’s
path for the fluid to get to the wellbore without going through responsibilities include development of hydraulic fracturing
the fracture: and other stimulation technologies for unconventional gas
resources. He began work with ConocoPhillips in the Fall of
2009. Friehauf holds a BS degree from the Colorado School of
2
Rel ,0 = ln(b1,0 / rw ). . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-4) Mines and a PhD degree from the University of Texas in Austin,
hk both in chemical engineering. Mukul M. Sharma is a profes-
sor and holds the “Tex” Moncrief Chair in the Department of
If damage around the wellbore is present, the wellbore resistor has Petroleum and Geosystems Engineering at the University of
two parts, (1) the resistance from the wellbore through the damage Texas at Austin where he has been for the past 25 years. He
and (2) the resistance from the damage to the inner ellipse, served as chairman of the department from 2001–05. His cur-
rent research interests include improved oil recovery, injection-
2 2 water management, hydraulic fracturing, formation damage,
Rel ,0 = ln(rs / rw ) + ln(b1,0 / rs ). . . . . . . . . . . . . . . (A-5) and petrophysics. He has published more than 200 journal
hks hk articles and conference proceedings and has 12 patents.
Sharma has a bachelor’s of technology in chemical engi-
In Eq. A-5, rs and ks are the radius and permeability of the damaged neering from the Indian Institute of Technology and MS and
zone around the wellbore, respectively. PhD degrees in chemical and petroleum engineering from
For the flow to go through the fracture, it must go through the University of Southern California. Among his many awards,
three sections of linear-flow resistance. The first connects the inner Sharma is the recipient of the 2009 Lucas Gold Medal, the
ellipse to the damaged zone of the fracture face with the reservoir 2004 SPE Faculty Distinguished Achievement Award, the 2002
permeability: Lester C. Uren Award, and the 1998 SPE Formation Evaluation
Award. He served as an SPE Distinguished Lecturer in 2002, has
(b1,i − hleak ,i ) served on the editorial boards of many journals, and taught
Rel ,i = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-6) and consulted for industry worldwide. Ajay Suri is a consultant
khX for water- and polymer-injection wells for reservoir-pressure
maintenance and disposal. He is currently enhancing and
The second is the damaged zone perpendicular to the fracture: applying the University of Texas Well Injectivity Decline simula-
tor to model deposition of particles and growth of fractures
hleak ,i around injection wells. He is involved in several water injection
Rd ,i = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-7) projects operating in the Gulf of Mexico and North Sea. Before
kd hX
this, he was a post-doctoral fellow in the department of petro-
The third is flow inside the fracture: leum and geosystems engineering at the University of Texas at
Austin. He holds a bachelor’s in technology in petroleum engi-
neering from Indian School of Mines and MS and PhD degrees
2X
R f ,i = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-8) in petroleum engineering from University of Texas at Austin. He
k f hwi has published his findings in the various SPE Journals.