SPE 168749 / URTeC 1579514

A Systematic Approach to Understanding Well Performance in

Unconventional Reservoirs: A Wolfcamp Case Study
Kshitij Mohan*, SPE, Kyle D. Scott*, SPE, Gene D. Monson*, SEG, and Paul A.
Leonard*, SPE, Pioneer Natural Resources USA Inc.
Copyright 2013, Unconventional Resources Technology Conference (URTeC)

This paper was prepared for presentation at the Unconventional Resources Technology Conference held in Denver, Colorado, USA, 12-14 August 2013.

The URTeC Technical Program Committee accepted this presentation on the basis of information contained in an abstract submitted by the author(s). The contents of this paper
have not been reviewed by URTeC and URTeC does not warrant the accuracy, reliability, or timeliness of any information herein. All information is the responsibility of, and , is
subject to corrections by the author(s). Any person or entity that relies on any information obtained from this paper does so at their own risk. The information herein does not
necessarily reflect any position of URTeC. Any reproduction, distribution, or storage of any part of this paper without the written consent of URTeC is prohibited.

Summary

This paper describes a systematic approach to analyzing hydraulically fractured horizontal wells (HFHW) in the
Permian Wolfcamp shale. With increased shale development via multiple transverse fractures in horizontal wells, it
becomes imperative to develop techniques to better characterize well performance which ultimately improve our
understanding for well spacing and completions optimization. The workflow presented here integrates fracture
treatment analysis (FTA), rate transient analysis (RTA) and numerical simulation to analyze well performance data.
Neither FTA nor RTA can be used alone to analyze well performance due to complex flow regimes, multiphase flow
effects, and reservoir compaction. However, they are important building blocks in developing numerical simulation
models for understanding the quality of stimulation and the effectiveness of reservoir drainage. By studying flow
characteristics using RTA, we can obtain useful estimates of the effective fracture surface area and effective
permeability of the stimulated volume that effectively contributes to production. These fracture properties and
dimensions are used to constrain inputs for numerical simulation, which additionally accounts for complex phase-
behavior, multi-phase flow, and rock compaction effects. Production history-matching is used to additionally
estimate the effective properties of the complex fracture network, namely the width of the enhanced zone (Wenh) and
the associated permeability within it (Kenh) that was created by stimulation processes. In this paper, we present a
field case study in which we have successfully used this approach to characterize well performance. The method
demonstrates how analytical techniques can be used before undertaking detailed numerical simulation in order to
constrain the ranges of fracture dimensions that explain observed performance.

Introduction

With technological advances in well design and stimulation techniques, numerous shale plays have become
profitable to develop in recent years. This has been demonstrated in the Bakken, Eagleford and Woodford shales.
The latest addition to these plays are the Wolfcamp shales located in the Permian Basin of West Texas. The
Wolfcamp formation lies beneath the Spraberry and Dean formations which form the Spraberry Trend field, and
which until recently had been the principal target for reservoir development in the Permian Basin. Within the last 2
decades development had been extended to the Wolfcamp (and deeper) formations via vertical wells completed in
up to 10 different zones. Pioneer Natural Resources is the largest acreage holder in the Spraberry field, and operates
approximately 900,000 gross acres with a current producing well count over 7000 vertical wells. The majority of
the acreage appears prospective for the Wolfcamp Shale. To accelerate production and maximize economic
recovery, Pioneer recently drilled a number of horizontal wells targeting the Wolfcamp with tremendous success.

The Wolfcamp is about 1750 ft thick in the central part of the Midland Basin and can be sub-divided into a least 4
major intervals, designated the Wolfcamp A, B, C, and D. All intervals are principally “shale” across the vast
URTeC 1579514 2

majority of the basin, and all have organic content to one degree or another. However, within that framework there
is significant variability. The A member is more carbonate rich and overlies the B, which contains less carbonate
and more quartz, and tends to be organic-rich. The C tends to be more clay rich and quite thick. The D is cyclical
containing multiple para-sequence boundaries that are correlative across large parts of the basin. To date the A, B,
and D have been demonstrated to have economic rates and estimated ultimate recoveries (EURs), with the B being
the most prolific. In addition to stratigraphic variability there is significant variation spatially especially near the
basin margins. Correlative carbonate rich intervals exist within the Wolfcamp that are thought to be derived from the
Central Basin Platform and deposited as debris flows. There are thick carbonate beds near the Horseshoe Atoll. Also
at some points around the basin significant quartz rich clastic input is found. For instance, along the far eastern shelf
the Wolfcamp D changes facies into more sand rich intervals. These variations notwithstanding, the central part of
the basin contains significant potential in all four intervals. The bottom line is the Wolfcamp is a remarkably thick
and correlative stacked sequence of unconventional reservoirs with potential for thousands of drilling locations.
With so much potential and capital requirement, it becomes imperative to better characterize and understand well
performance.

Problem Statement

As the Wolfcamp horizontal well program moves into a development phase, many critical questions need to be
answered involving selections of development well spacing, completion designs, and other optimizations that will
affect long-term production and EURs. In recent years, many authors have presented different techniques to
understand these issues. Mayerhofer, Warpinski and Cipolla et al. (2008) introduced the concept of stimulated
reservoir volume (SRV), where they used microseismic data to generate a fracture network model for the Barnett.
They demonstrated that the drainage area is constrained to the created SRV in ultra-tight shale reservoirs and that 3
month cumulative gas production is proportional to the SRV. Simultaneously several authors (Wattenbarger et al.
(1998), Nobakht et al. (2010), Anderson et al. (2010), Economides et al. (2011)) have applied RTA to approach
these problems. While some of the methods outlined above are time and data intensive, others seem to be too
simplistic. Nevertheless, they all provide key insights into completions and reservoir performance. This paper
outlines a hybrid approach which integrates FTA, RTA, and numerical simulation to improve understanding of
contributing fracture dimensions, drainage volume, and EUR, and to optimize field development through well
spacing and completion design. Models built and calibrated using this methodology are then used to forecast the
relative benefits of various possible development alternatives.

Methodology

The proposed methodology consists of three main steps.

1. The first step is to estimate fracture dimensions with FTA, using a 3-D fracture model simulator. Rock
properties estimated from density and sonic logs, fracture-treatment pump schedules and injection volumes and
rates, and treatment pressures are used as inputs in the 3-D fracture simulator to predict geometries created by
hydraulic fractures, and the distributions and conductivities of the proppant placed within those fractures.
Typically fracture model simulators predict large fracture dimensions that are in agreement with proppant and
water-phase tracer data, but which would be over-estimated when compared to observed production. The key
question that needs resolving is how much of the created fracture surface area actually contributes to
production. We assumed that the contributing fracture height (HCF) to be equal to the adequately propped height
from the fracture simulator, and we use this value in combination with results from RTA to estimate the
corresponding contributing fracture half-length (XCF). Also, a goal of fracture modeling is to estimate fracture
conductivity and porosity at the initial net confining stress.

2. The second step is to utilize RTA to determine contributing fracture half-length and drainage volume. Several
authors (Meideros et al. 2007; Economides et al. 2011) have extensively studied flow regimes experienced by
URTeC 1579514 3

HFHW using analytical methods and numerical simulation. Depending on the flow regimes exhibited, they
devised techniques to estimate key parameters, such as fracture half-length (XF) and EUR. Samandarli et al.
(2011a) have shown the main flow regimes in HFHW are pseudo linear flow (PLF) and pseudo steady state
flow (PSSF), with the latter being more dominant and prolonged. Large stimulation treatments and tight cluster
spacing result in short-lived PLF as pressure transients from adjacent fractures quickly interfere at the midpoint
between perforation clusters, marking the onset of PSSF. Phenomena such as well cleanup, supercharging due
to stimulation, and stress dependent permeability can easily mask early PLF, making it harder to identify and
analyze with any confidence. Consequently, in many instances we are left with only the PSSF regime for
analysis.

RTA studies conducted previously (Wattenbarger et al. 1998; Bello et al. 2010) have focused mostly on the
constant rate or constant pressure solution to identify and analyze flow regimes. While these idealizations may
be realized in the later part of a well’s life, production data from the early part of a well’s life suggest that the
wells typically experience variable rates and pressures. To accommodate changing rates and pressures, Song et
al. (2011) and Liang et al. (2011) proposed plotting Rate Normalized Pressure (RNP) versus Material Balance
Time (MBT) (Eqs. 1 and 2). By using MBT instead of a superposition time, they converted the variable
pressure/rate scenario to an equivalent constant rate solution. As a result, the unique characteristics of both
flow regimes are preserved. PSL is indicated by a half-slope on a log-log plot of RNP versus MBT, and PSSF
is reflected by a unit slope on a log-log plot of RNP versus MBT. If a PSSF flow regime is identified in this
way, then a Cartesian plot of RNP versus MBT can be used to calculate a slope and an intercept which
determine the parameters indicated in Eq.3. Combining the slope and intercept with HCF estimated from
fracture modeling and reservoir and fluid properties from petrophysical measurements and fluid sampling, we
can calculate XCF and effective system permeability (Ksys).

Song et al. (2011) and Siddiqui et al. (2012) have successfully developed the RTA theory and applied it to
analyze reservoirs where flow is predominantly single phase. With significant frac load recoveries and
increasing GORs, the above RTA approach needs to be modified to account for multi-phase flow in the
Wolfcamp. Our approach is to combine the production of 3 phases into an equivalent total fluid volume (Eq.
4), that can be analyzed using the RNP vs. MBT plot. For this approach, we sum the reservoir voidage of each
fluid phase, and convert this to a surface volume of the principal phase using the initial formation volume factor
for that phase.

{( ) }

Physically, PSSF indicates the pressure transient has reached a virtual no-flow boundary. This boundary can
either be the mid-point between adjacent fractures or the edge of the zone of enhanced permeability surrounding
the primary hydraulic fracture. Similar to Stalgorova et al. (2012), a diagram of the enhanced permeability zone
and its representation of broken rock are shown in Fig.1. Associated with this zone of enhanced permeability is
a width, Wf, representing the dendritic growth of micro-fractures away from the primary propped fracture and a
permeability magnitude representing complexity of the fracture network.
URTeC 1579514 4

When PSSF is interpreted from production data by RTA, there is no way to distinguish where or what is causing
the virtual no-flow boundary, only that the drainage volume within a boundary of unknown shape can be
estimated. RTA also interprets a single value for system permeability without separating the region of
enhanced permeability from the background unstimulated matrix permeability. These as well as frac fluid
injection, multiphase flow effects, and stress-dependent permeability are important characteristics in the
Wolfcamp that are not adequately addressed in RTA. Furthermore, RTA overlooks dependence of fluid
properties on pressure and petrophysical heterogeneity by assuming constant, averaged properties. Thus to
better account for these complexities, to predict the width and permeability of the enhanced zone, and to
improve long-term forecasting, we then move to numerical simulation for more robust analysis. Obtaining
fracture dimensions and effective permeability from RTA has proven to be a fast and efficient means to
determine the starting point for reservoir simulation history-matching.

Horizontal Well Top View

Broken Rock
kenh
Perf Cluster
Propped Fracture

kmatrix

wenh

xf

Fig. 1—Zone of enhanced permeability representing fracture network complexity

3. The third step is to utilize the fracture dimensions and system permeability determined from RTA to condition a
geocellular model of the rock matrix for the conditions caused by the hydraulic fracturing. Numerical
simulation is used to estimate the width of the enhanced zone and distinguish difference in permeability
between it and the surrounding unstimulated matrix. By running multiple sensitivities, we can conclude whether
the zone of enhanced permeability is fully connected between adjacent fractures or if the zones are isolated.
Once the above dimensions are determined through a successful history-match, the reservoir simulation model
can be used to forecast long-term production decline and predict EURs. Completion optimization may also be
performed by running sensitivities on treatment design and cluster spacing. Additional wells may also be added
to the model to predict optimal well spacing for field development. A diagram of this proposed workflow is
shown in Fig. 2 below.
URTeC 1579514 5

Fig. 2—Workflow of the methodology described above

Example Application

To illustrate application of the method described above, an example case study is presented here. Well A was drilled
and completed in the Wolfcamp formation of the Permian Basin. Details about Well A and other key reservoir
properties can be found in Table 1.

Table 1—Key well and reservoir properties

Reservoir & Fluid Properties Well Properties
Thickness, ft 470 Depth, ft 9650
Total Porosity, fraction 0.096 Lateral Length, ft 5600
Matrix Permeability, nD 46 # of Stages 30
Pressure, psia 5760 Clusters/Stage 3
Temperature, ˚F 183
Oil gravity, ° API 43.6
Bo, res bbl/STB 1.56
Oil Viscosity (μo), cp 0.3
Initial GOR, scf/STB 1060
Oil Compressibility (co), psi-1 13E-5

Fig. 3 shows results of modeling one of the many hydraulic fracture stages using a 3-D fracture simulator. The
model predicts HCF of 160 ft with fractures extending up to 2000 ft laterally from the wellbore. These results are in
agreement with interpretations of the microseismic data that was acquired during stimulation of Well A. As
URTeC 1579514 6

mentioned above, getting these high fracture half-lengths to honor production data is very difficult. To resolve this,
we used RTA to estimate the contributing fracture half-lengths.

Proppant Concentration, lb/ft2

0.0 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.0

+100

+50
Height, ft

-50

-100

200 400 600 800 1000 1200 1400 1600 1800 2000
Length, ft

Fig. 3—Modeled fracture geometry

The production data for Well A are shown in Fig. 4. The well was flowed naturally for the first 270 days and
then placed on gas lift. The high water production rates are primarily due to flowback of stimulation fluids; the
formation has little to no mobile water.

Well A
2,400 6,000
Oil, STB/D
2,200 Water, STB/D 5,500

FBHP, psia
2,000 5,000
GOR, SCF/STB
Production Rates, STB/D GOR, SCF/STB

1,800 4,500

1,600 4,000

1,400 3,500

FBHP, psia
FBHP
1,200 3,000

1,000 2,500

800 2,000
Water
Oil
600 1,500

400 1,000
GOR
200 500

0 0
0 50 100 150 200 250 300 350 400 450 500 550
Time, days

Fig. 4—Production profile of Well A

The production data were used to compute equivalent total fluid production volumes, material balance time, and
rate-normalized pressure variables, as described above. Upon plotting RNP versus MBT on a log-log plot, a
unit slope indicating PSS flow was identified (Fig. 5). Note that no half-slope which would indicate PLF can be
identified in the early-time data on this plot. The unit slope was verified by plotting the derivative of RNP with
MBT, which is shown as the open symbols on the log-log plot in Fig 5. Once we identified PSS on the log-log
plot, we obtained the slope and intercept from the Cartesian plot of RNP versus MBT (Fig. 5). The slope and
URTeC 1579514 7

intercept were used along with Eq. 3 to calculate contributing fracture half-lengths and an estimate of effective
system permeability (Ksys). For Well A, XCF was calculated to be 170 ft, and the system permeability was
estimated to be 350 nD. We note that although there is an excellent correlation coefficient (r 2 = 0.978) in this
case, there typically remains significant uncertainty in the intercept of this plot, which reflects uncertainty in the
number of fractures (stimulation clusters) actually contributing to production as well as uncertainty in the
effective value of the system permeability. Since RTA makes no distinction between the enhanced region and
unstimulated matrix, it is necessary to utilize reservoir simulation to quantify specific geometries and
permeabilities of each region.

Oil + Water + Gas, STB' Oil + Water + Gas, STB'

1,000.0 35

RNP' RNP
30
RNP' (dRNP/dlnMBT)

RNP
Linear (RNP)
Unit Slope
100.0
25

RNP, psi/bbl/day
20
y = 0.016x + 1.141
10.0
RNP (ΔP/q), psi/bbl/day;

R² = 0.978
15

10
1.0

0.1 0
10 100 1,000 10,000 0 200 400 600 800 1000 1200 1400 1600 1800
Material Balance Time, days Material Balance Time, days

Fig. 5—RNP vs MBT plots identifing PSSF

The fracture dimensions and properties so determined were used to modify a geocellular model of the matrix
properties of the reservoir, using local grid refinements to describe the extent of regions representing the
enhanced volumes around planar fractures. Numerical simulations were conducted to aid in determining the
width and permeability of enhanced zone surrounding each fracture.

The geocellular model in this study was created using well log formation tops and 3D seismic interpreted
horizons for structural control. Reservoir properties of porosity, net-to-gross, and water saturation were derived
from upscaled well log data that were calibrated to core plug analysis and distributed throughout the geologic
model. Matrix permeability was estimated by a porosity-to-permeability relationship derived from core plug
analyses, and then adjusted to effective permeability to oil and distributed via a cloud transform in the geologic
model. A flow-based method was then used to upscale matrix permeability to the simulation scale. Porosity,
net-to-gross, and water saturation were upscaled to the simulation model using volume-weighted arithmetic
averaging. The average matrix properties were 9.6% total porosity, 4.8% hydrocarbon-filled effective porosity,
and an average matrix permeability of 46 nD.

The history-match of the first 570 days of production is shown in Fig. 6. Oil, watercut, and GOR are each
matched reasonably well, and our experience has shown that it is essential to match flowback in order to obtain
a good representation of the bottomhole pressure data. We obtained this history-match chiefly by adjusting the
enhanced zone to 7 ft width and by adjusting the enhanced zone effective permeability to a value of 750 nd. The
effective fracture height of 160 ft and half-length of 170 ft were retained from our fracture treatment and rate
transient analysis processes.
URTeC 1579514 8

Fig. 6—History match results for Well A

Fig. 7 shows a schematic of how the isolated enhanced region surrounds each fracture in the numerical
simulator.

Matrix
• Kmatrix = 46 nd
• ΦT = 9.6% Enhanced Zone
• Wenh = 7 ft Propped Fracture
• Kenh = 750 nd • XCF = 170 ft
• HCF = 160 ft
• Wfracture = 0.78 ft
• FC = 2.5 md-ft (FCD = 105)

Time: 50 years
Pressure

Fig. 7—Pressure depletion around the contributing fractures and the enhanced zones associated with them
URTeC 1579514 9

Conclusions

Based on the work shown, we conclude:

1) A systematic approach has been developed to understand horizontal well performance in the Wolfcamp
Shale.
2) Using total equivalent fluid volumes improves interpretation of contributing fracture and drainage volume
dimensions using the RNP and MBT methodology.
3) Horizontal Wolfcamp wells exhibit an extended PSSF that is useful for estimating effective fracture
properties.
4) Horizontal Wolfcamp wells do not exhibit a well-developed PLF period that is useful for analysis; if a PLF
exists, it is either short-lived or masked by other effects.
5) Based on the integrated approach combining FTA with RTA and reservoir simulation, we can characterize
important fracture parameters (HCF, XCF, Wenh, and Kenh).
6) A history-matched reservoir simulation model which adequately describes the average contributing
fracture, drainage volume, and enhanced permeability zone can then be used to answer key field
development questions involving long term production decline, EUR, completion optimization, and well
spacing.

Nomenclature

EUR = Estimated ultimate recovery

FTA = Fracture treatment analysis
Bo = Oil formation volume factor, res bbl/STB
Bw = Water formation volume factor, res bbl/STB
Bg = Gas formation volume factor, res bbl/STB
BT = Total formation volume factor, res bbl/STB
ct = Total system compressibility, psi-1
HCF = Contributing fracture height, ft
Kenh = Permeability of the enhanced zone, md
Km = Matrix permeability, md
Ksys = System permeability, md
Lf = Perf cluster spacing, ft
Nf = Number of perf clusters
NT = Cumulative fluid production, STB
Pi = Initial reservoir pressure, psi
Pwf = Flowing Bottom hole pressure, psi
qo = Daily oil production rate, STB/d
qw = Daily water production rate, STB/d
qg = Daily gas production rate, SCF/d
QT = Daily total fluid production rate, STB/d
SRV = Stimulated reservoir volume
tMB = Material Balance time, days
Wenh = Width of the enhanced zone, ft
XCF = Contributing fracture half length, ft

Greek Symbols
= porosity, fraction
μ = viscosity, cp

Subscript

CF = contributing fracture
enh = enhanced
f = fracture
g = gas
i = initial
URTeC 1579514 10

m = matrix
MB = material balance
o = oil
T = total fluid properties
w = water

References

1) Song, B and Ehlig-Economides, C.: “Rate-Normalized Pressure Analysis for Determination of Shale Gas
Well Performance,” SPE 144031, presented at the SOE North American Unconventional Gas conference
and Exhibition, The Woodlands, Texas, USA, 14-16 June, 2011.
2) Stelgorova, R. and Mattar, L.: “Analytical Model for History Matching and Forecasting Production from
Multifrac Composite Systems,” SPE 162516, presented at the SPE Canadian Unconventional Resources
Conference held in Calgary, Alberta, Canada, 30th October to1st November, 2012.
3) Mayerhofer, M.J., Lolon, E.P., Warpinski, N.R., Cipolla, C.L., Walser, D. and Rightmire, C.M.: “What is
Stimulated Reservoir Volume (SRV),” SPE 119890, presented at Shale Gas conference in Fort Worth,
Texas, 16-18 November, 2008.
4) Medeiros, F., Kurtoglu, B., Ozkan, E., Kazemi, H.: “Pressure Transient Performances of Hydraulically
Fractured Horizontal Wells in Locally and Globally Naturally Fractured Formations,” IPTC 11781,
presented at International Petroleum Technology Conference in Dubai, U.A.E., 4-6 December, 2007.
5) Samandarali, O., Valbuena, E., Ehlig-Economides, C.: “Production Data Analysis in Unconventional
Reservoirs with Rate Normalized Pressure (RNP): Theory, Methodology and Applications,” SPE 155614,
presented at the Americas Unconventional Resources Conference, Pittsburgh, Pennsylvania, USA, 5-7 June
2012.
6) Anderson, D.M., Nobakht, M., Moghadam, S., and Mattar, L.: “Analysis of Production Data from
Fractured Shale Gas Wells,” SPE 131787, presented at Unconventional Gas Conference, Pittsburgh,
Pennsylvania, USA, 23-25 February, 2010.
7) Siddiqui, S.K., Ali, A., Dehghanpour, H.: “New Advances in Production Data Analysis of Hydraulically
Fractured Tight Reservoirs,” SPE 162830, presented in Canadian Unconventional Resources Conference,
Alberta, Calgary, Canada, 30th October to 1st November 2012.
8) Clarkson, C.R., Pederson, P.K.: “Tight Oil Production Analysis: Adaptation of Existing Rate-Transient
Analysis Techniques,” SPE 137352, presented at the Canadian Unconventional Resources, Calgary,
Alberta, Canada, 19-21 October 2010.
9) Bello, R.O. and Wattenbarger, R.A.: “Modeling and Analysis of Shale Gas Production with Skin Effect,”
CIPC Conference, Calgary, Alberta, Canada, 16-19 June 2009.
10) Clarkson, C.R., Jensen, J.L., Blasingame, T.A.: “Reservoir Engineering for Unconventional Gas
Reservoirs: What Do We Have to Consider,” SPE 145080, presented at North American Unconventional
Gas Conference and Exhibition, The Woodlands, Texas, USA, 14-16 June 2011.