You are on page 1of 11

SPE-169590-MS

Impact of Fracture Closure on Productivity Decline of


Unconventional Wells
A. Sarna, Q. Xing, J. Mork and I. Ershaghi, University of Southern California

Copyright 2014, Society of Petroleum Engineers


This paper was prepared for presentation at the SPE Western North American and Rocky Mountain Joint Regional Meeting held in Denver, Colorado, USA, 1618 April 2014.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
With the rapid pace of drilling wells to tap unconventional oil and gas resources in the U. S.,
there is a need for further understanding of the fundamentals of reservoir mechanics that
control well productivities. Wells with more than 5 years of data provide the opportunity to
understand flow mechanisms affecting well productivity. What is particularly important is to
test the validity and appropriateness of decline curve forecasting methods advocated in the
literature over the last several years. To explain the unique and similar performance trends
from unconventional wells, there are a number of issues that are gradually getting attention.
Among these are the causes of rapid production decline seen after the initial peak. Certainly
the physics of flow conditions causing the decline is not captured when regression analysis of
production data is used with various decline models. Based on earlier studies, there is selfsimilarity among fractures formed during hydraulic fracturing. As discussed in the literature;
the transient pressure behavior of complex self-similar fractures is equivalent to that for a
single fracture. As such it is the flow capacity of the equivalent fracture that controls the
behavior seen in actual performance of hydraulically fractured wells.
To test the applicability and weaknesses of commonly practiced regression models, we
examined the performance of number of wells with production histories exceeding five years
We used more than 40 examples from various shale plays all showing the loss of fracture
quality resulting in lower flow capacity within the fracture network. The dynamic change in flow
capacity is certainly not captured when decline curves are used that do not honor the flow
regimes changes. As such, there can be major uncertainties associated with estimated EURs
from decline analysis techniques
Introduction
Hydraulic fracturing is one of the most effective methods for enhancing productivity from tight
shale reservoirs. Combined with high surface contact horizontal wells, with laterals often
exceeding several thousand feet, U. S. producers have achieved phenomenal success in
producing from tight shale gas and shale oil resources. A review of published or available online well histories from different plays often indicate substantial performance variations among

SPE-169590-MS

wells in a given play area. Variations are seen in initial rate, duration of stabilized rate and
follow up decline trends. In an effort to attach an estimation of EUR (Estimated Ultimate
Recovery), practitioners have borrowed several concepts from conventional reservoir
engineering. One of which is the use of decline curves for well productivity forecasting. There
are several publications that include discussions about the application of these decline curve
methodologies. Nobakht et al (2012) review a number of proposed methods for projecting
production from wells producing from unconventional oil and gas reservoirs.
These include
various type curve approaches, empirical methods and hybrid methods.
In this study we examine typical histories for a large number of wells with 5 years or more life
histories. Physics of flow as affected by rock and fluid quality, fracture flow capacity,
interporosity flow and completion practices is primarily the cause affecting behavior of well
performance data. Our study objectives were to examine the merits of the decline curve
methodologies suggested in the literature for their ability to capture the physics of flow in early
time data and to predict what has actually been observed on wells with a long history.
Statement of the Problem
As shown in Fig. 1, typical performance plots for hydraulically fractured horizontal wells
producing from shale oil and gas resources include 3 distinct intervals. Production data on
such wells start with an interval I representing the highest initial rate. Based on information
available on various plays, this interval usually can have a short duration often less than 9
months. Interval II representing some rapid decline can last one or more years and it may
exhibit
several decline trends. Finally interval III, called here a stabilization period, may
continue for some years while rates are still economical.

Fig.1 Three stages commonly seen on production plots of wells producing from shale
reservoirs
In computing EUR, information primarily from intervals I and II and sometimes III is used to
make performance forecasting to some economic limit. To examine the merits of decline curve
projection techniques, we examined more than 40 wells with production histories of 4 years
and longer. These wells were from shale plays including the California Monterey, North
Dakota Bakken, West Virginia/Pennsylvania Marcellus and Texas Barnett plays. Before a
projection methodology is used, it is important to understand the causes of rapid decline seen
in Interval II. In conventional oil reservoirs, production decline in wells, after initial high rates, is
attributed to the drop in reservoir pressure, evolution of gas and a drop in oil relative

SPE-169590-MS

permeability. For conventional gas reservoirs, decline of gas rate for a given choke setting
corresponds to depletion of reservoir energy. But in unconventional oil and gas reservoirs, the
rapid start of declines in Interval II, after only a few months of production, needs other causes
beside reservoir pressure drop. Measurements reported on wells producing from shale
reservoirs indicate that, because of low depletion rates, compared to the body of rock
constituting the drainage volume, static pressures measured on wells shut down for a short
period of time can approach initial static reservoir. Flow rates, however, are not restored
indicating potential losses in fracture flow capacity. To explain the cause of decline, we should
examine the nature of fracture clusters.
Nature of Fractures in Shale Rocks
Fundamentally, success in enhancing productivity from shale gas and oil resources depends
on the ability of induced fractures connecting pre-existing natural fractures. In general, flow
capacity of fracture network depends on fracture aperture and tortuosity of the flow path. For
an established network caused by hydraulic fracturing, the focus needs to be on dynamic
changes of fracture aperture or gradual control of flow by lower flow capacity members of the
fracture network. What can also cause the change includes local stress effects during a
reduction of pore pressure caused by production and gradual changes related to clay swelling
in certain rock compositions because of exposure to frac fluid, Hyattdavoudi et al (2013). Loss
of fracture flow capacity will accordingly result in a decline of production rates.
Various studies conducted at different scales ranging from examining cores, outcrops and
seismic monitoring have indicated patterns of self similarities in fractures formed in natural
rocks. Tchalenko (1970) indicated that fracture pattern in crustal shear zones is developed
over a wide range of scales. Experimental data, on how the rocks fracture, shows a pattern of
self similarity covering a wide range of scales. Barenblatt and Botvina (1986) discussed the
concept of self-similarity in physics of fracture formation. Barenblatt (1993) discussed selfsimilarity in fracture mechanics. Studies by Barton and Hsieh (1989), and Sammis et al.
(1991), have made similar observations for natural fractures respectively in the Yucca
Mountain project and those contributing to steam flow in the Geysers geothermal system.
When fracturing in rocks occurs over a wide range of scales, it can best be expressed by
fractal geometry where there is a power law increase in the number of smaller size fractures.
Zhao et al (2010) also discussed the fractal nature of fractures formed during hydraulic
fracturing. The subject of percolation threshold in fractures has also been discussed by several
authors such as Nolte et al (1989).
To formulate a prediction mode for fluid flow, we need to examine the consequences of
formation breakdown under hydraulic fracturing. Barton and Scholz (1995) indicated that the
finite nature of fracture volume for any given volume of the rock makes the contribution of
smaller fractures less important to flow. Study of pressure transient response in fractured
system representing self similarity was published by Chang and Yortsos (1990), Acuna and
Yortsos (1991), and Acuna et al (1995). In the latter study, data on the naturally fractured
greywacke rock of the Geyser geothermal reservoir indicated that in spite of the presence of
hundreds of fractures represented by successive steam entries, the entire well pressure
transient responses behave like having been cut by a single fracture. Single fractures of high
conductivity exhibit a linear flow represented by a slope of on a log-log plot of pressure (for
constant rate or (1/q) for constant pressure cases. It is also known that fractures choked or
those with low permeabilities can exhibit slopes less than 0.5 and often times down to 0.2 or
even less, Agarwal (1979). A study recently published by Cossio et al (2013), shows that for a
fractal system the slope change can take values of 0.5 or less depending on the degree of

SPE-169590-MS

connectivity. Given the negligible drop in reservoir pressures when the well productivity
exhibits a decline, one may consider the possibility of fractures approaching limited flow
capacity. This can be caused by lowering of pore pressure and closing of fracture aperture
under geomechanical stress field, losses in fracture conductivity caused by clay-water
interaction, or depletion in larger fractures and contribution to flow by lower flow capacity
branches of fracture network. As such, expectations are that for shale gas or shale oil
horizontal wells with multiple frac stages, loss of productivity can be caused by losses in
fracture network flow capacity. High flow capacity fracture exhibit linear flow graphically
represented by a 0.5 slope indicating fracture linear flow on a log-log plot of q vs. time. Then
there seems to be a change resulting in slopes to more than 0.5. (Slopes more than 0.5 are
the same as slopes of less than 05 when 1/q is plotted on the log-log scale). The increase in
the slope is related to decreasing fracture network conductivity. A flat period developed in
interval III of Figure 1, could be long lasting or a short duration indicating the strength of
interporosity flow over a large surface area generated during hydraulic fracturing.
Discussion
To examine this hypothesis, we reviewed production performance data on a number of wells
with histories exceeding 4 years. Log-log diagnostics plots of q vs. time, show slope changes
from 0.5 (linear fracture flow) to numbers less than 0.5 when there is a stabilization. But in
some cases we also observe an early slope of 1 prior to the onset of the 0.5 slope. This
according to the concept of wellbore unloading relates to unification of a portion fluid in the
major fractures and immediate communication with the wellbore fluid.
Taking this expected response model, we prepared a comparison of the merits of this
sequential slope changes vs. trends on the Fetkovich (1980), type curve representing decline
curves. Figures 2-11 show 5 pairs of well production histories from among the 40 wells
analyzed in this study. In Figure 2-3, we see some influence of unloading large fractures as
evidenced from a unit slope during early period. Forcing a decline curve would have totally
misrepresented the actual pattern experienced with updated production data. Late data
indicate a larger network of lower flow capacity fractures contributing to sustained flow. In
Figures 4-5, production plots for a well 6 in Bakken are compared and errors associated with
early rush to use decline data are obvious when examining real production trends developed
after 4 years of production. Similarly Figures 6-7 show data for well 3 from the Barnett.
Projection of data for the first 42 months would have totally misrepresented and overestimated
the performance trend. Now we see in Figures 8-9 performance data for well 14 producing
from North Shafter. A significant drop in the production after 8 months could not have been
projected from the early time data. Another example is the first 20 month plot for a Marcellus
Well E as shown in Figure, 10. Again extrapolation of that data into the future would have
significantly deviated from later actually observed performance data.
Summary and Conclusions
Cluster of fractures generated during hydraulic fracturing is a means to connect available
natural fractures in shale reservoirs. The concept of self similarity suggested by a number of
studies for the fractures generated in rocks means (as proposed by Acuna et al) that the
entire cluster can be equivalent to a single fracture. Loss of fracture flow capacity as
discussed by Agarwal and Cinco is characterized by slopes of less than 0.5 on a log-log plot of
1/q or more than 0.5 on log-log plots q vs. time. Examination of performance data from wells
producing from various shale plays show that the behavior of production data under the
condition of constant reservoir pressure indeed follows the behavior of a single representative
fracture. Early time high fracture flow capacity as evident from the slope of the

SPE-169590-MS

representative fracture on a log-Log plot is clearly indicated on such plots. Slopes exceeding
0.75 indicate losses of flow capacity resulting in decline of productivity.
Loss of flow capacity can be caused primarily by fracture closing, caused by the loss of pore
pressure and loss of integrity of proppants or gradual clay-interaction with the fluid causing
losses of well productivity. The stabilized period observed in some cases are indicative of a
larger network of fractures containing gas that contribute to sustained production but with an
overall lower fracture flow capacity.
Because of various causes and various flow regimes evolving during the early history of wells
producing from shale gas and shale oil reservoirs, use of decline curves that obviously cannot
honor such flow regime changes are not warranted for estimation of EUR.
References
Acua, J.A., Ershaghi, I., and Yortsos, Y.S. 1995. Practical Application of Fractal PressureTransient Analysis in Naturally Fractured Reservoirs. SPE Formation Evaluation 10 (3): 173179. SPE 24705-PA. http://dx.doi.org/10.2118/24705-PA
Acua, J.A. and Yortsos, Y.C. 1991. Numerical Construction and Flow Simulation in Networks
of Fractures using Fractal Geometry. Paper SPE 22703-MS presented at the SPE Annual
Technical
Conference
and
Exhibition,
Dallas,
Texas,
6-9
October
1991.
http://dx.doi.org/10.2118/22703-MS.
Agarwal, R. G., Carter, R. D., & Pollock, C. B. (1979, March 1). Evaluation and Performance
Prediction of Low-Permeability Gas Wells Stimulated by Massive Hydraulic Fracturing. Society
of Petroleum Engineers. doi:10.2118/6838-PA
Barenblatt, G.I. (1993). Some general aspects of fracture mechanics. In: Modeling of Defects
and Fracture Mechanics (Edited by G. Herrmann), Springer-Verlag, Wien, New York, 2959.
Barenblatt, G.I. and Botvina, L.R. (1986). Similarity methods in mechanics and physics of
fracture. Fiz.-Khim. Mekh. Mater. 1, 5762.
Barton, C.C., and Scholz, C.H., 1995, The fractal size and spatial distribution of hydrocarbons:
in Barton, C.C., and La Pointe, P.R., eds., Fractals in Petroleum Geology and Earth
Processes: Plenum Press, New York, p. 13-33.
Barton, CC and Hsieh, P.A.: A Field Guide for the 1989 International Geol. Congr. AGY.
Chang, J. and Yortsos, Y. 1990. Pressure Transient Analysis of Fractal Reservoirs. SPE
Formation Evaluation 5 (1). SPE 18170-PA. http://dx.doi.org/10.2118/18170-PA
Cossio, M., Moridis, G., & Blasingame, T. A. (2013, January 24). A Semianalytic Solution for
Flow in Finite-Conductivity Vertical Fractures by Use of Fractal Theory. Society of Petroleum
Engineers. doi:10.2118/153715-PA
Fetkovich, M. J. (1980, June 1). Decline Curve Analysis Using Type Curves. Society of
Petroleum Engineers. doi:10.2118/4629-PA.

SPE-169590-MS

Hayatdavoudi, A., Boukadi, F., & Hassanpoor, D. (2013, April 8). The Effect of Shale
Chemistry on Gas Production Decline in Hydraulically Fractured Shale Gas Wells. Society of
Petroleum Engineers. doi:10.2118/164093-MS
Nobakht, M., Clarkson, C. R., & Kaviani, D. (2012, June 1). New and Improved Methods for
Performing Rate-Transient Analysis of Shale Gas Reservoirs. Society of Petroleum Engineers.
doi:10.2118/147869-PA
Nolte, D.D., Pyrak-Nolte, L.J and Cook, N.G.W.: The Fractal Geometry of Flow Path in
Natural Fractures in Rock and the Approach to Percolation, PAGEOPH, Vol. 131, Nos.
(1989)
Sammis, C, An, Lin Ji and Ershaghi, I Fracture Patterns in Graywacke Outcrops at The
Geysers Geothermal Field . PROCEEDINGS SIXTEENTH WORKSHOP GEOTHERMAL
RESERVOIR ENGINEERING July 23-25, I991
Zhao Wanchun, Wang Tingting, Ju Guoshuai, and Zheng Dachun: The Study on Fractal
Damage Of Rock Under Hydraulic Fracturing Basing On Conversation Of Energy, Applied
Mechanics and Materials Vols. 29-32 (2010) pp 1363-1368

SPE-169590-MS

Fig. 2 Example for Bakken Well 8 with 48 month rate performance Data

Fig. 3- Bakken Well 8 with Real Performance after 48 Months

SPE-169590-MS

Fig. 4 Bakken Well 6 with 48 Months of Production

Fig. 5 Bakken Well 6 with Updated History

SPE-169590-MS

Fig. 6 Barnett Well 3 with 42 months Data

Fig. 7 Barnett Well 3 with Updated Production Data

10

SPE-169590-MS

Fig. 8 North Shafter Well with 48 Months of Production Data

Fig. 9 The North Shafter Well 14 with Updated Production Data

SPE-169590-MS

11

Fig. 10 Marcellus Well E with 20 Months Production Data.

Fig. 11 Marcellus Well E with Updated Information