SPE-174839-MS

Engineered Shale Completions Based On Common Drilling Data

W. D. Logan C&J Energy Services

Copyright 2015, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Annual Technical Conference and Exhibition held in Houston, Texas, USA, 28 –30 September 2015.

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
Current best practices in North America’s shale basins deliver inconsistent production results because of
a lack of reservoir understanding along the lateral. Reservoir data is difficult and expensive to obtain in
horizontal sections; therefore, most completions are designed geometrically, with little or no concern for
reservoir heterogeneity. This leads to dramatic variability in productivity between neighboring wells,
often described as the statistical nature of the play.
In this study, we demonstrate a reliable, cost effective methodology that empowers shale operators with
reservoir data on every well. This technique will enable operators to engineer their designs by using
reservoir data on every completion, resulting in significant productivity improvement without any
significant increased cost or inconvenience.
This methodology leverages commonly available drilling measurements (rate of penetration, weight on
bit, rotational speed, hole diameter, flow rate, differential pressure and standpipe pressure) and mud motor
parameters (Kn, Tmax, ⌬Pmax) to derive Mechanical Specific Energy (MSE), using well established
algorithms. The MSE parameter is then shown to be a good proxy for Unconfined Compressive Strength,
a valuable reservoir parameter commonly used in frac designs. The MSE drives a facies-based answer
product that enables the operator to position perforation clusters so that they breakdown at a common
treating pressure, resulting in uniform fracture treatment within each frac stage.
To date the technique has been applied successfully on over 60 wells. It is validated through
comparison to openhole wireline logs and by comparing productivity to offsetting wells. The technique
is shown to be superior to similar attempts done by using openhole wireline logs and production data
confirms its accuracy and reliability. This study demonstrates that drilling data can be effectively used to
derive reservoir parameters that lead to completion designs that produce superior production results.
This methodology is the first application of drilling data to derive reservoir parameters that are then
used to engineer a completion design. The significance of this development cannot be overstated, since
it provides shale operators with the guidance they need to intelligently engineer their completions. The end
result is significant productivity increases without any associated cost or inconvenience.

Introduction
Oil and gas operators in North America have “productized” the process of drilling and completing their
wells to drive down cost. This effort has been a huge success and, within a decade, this has drastically
2 SPE-174839-MS

changed the landscape of the global petroleum industry. However, many operators have observed that
identical wells drilled on the same pad often can have productivity that varies by 2- to 3- fold. This is
referred to by many in the business as the “statistical nature of the play”. Many operators are now working
towards understanding this phenomenon, not satisfied with blaming inconsistent results on statistics.
This variability has been traced down to the stage level by using tracers and microseismic monitoring
(refs 5, 10) and down to the perf cluster level by using production logs (ref 7). It is now well understood
that one of the largest contributors to the variability in production is a variation in stress between
perforation clusters within each stage (ref 6). Using full-waveform sonic data (ref 9) and imaging logs (ref
6) some operators are gaining insight into the stress profile along the lateral and then engineering their
completion designs to eliminate this source of variability.
These efforts have resulted in case studies being published that demonstrate that understanding the
heterogeneity within these laterals can be used successfully to engineer completions that lead to superior
productivity. Unfortunately, despite all of this impressive technical work, the large majority of the wells
continue to be completed geometrically because reservoir evaluation tools are generally limited to
measurement while drilling gamma ray logs and mudlogs.
The factory approach thrives on techniques that are cost effective, repeatable, and relatively easy to
deploy. Generally speaking, most of the techniques that are available for evaluating the heterogeneity
along the lateral do not meet these criteria. They tend to be difficult to deploy and therefore expensive.
Being able to prove the incremental economic value they deliver is not trivial, so operators struggle to
deploy these techniques universally. Even the most fervent technology adopters still only evaluate a small
percentage of their wells, hoping they can leverage learning on one well to improve many wells.
Unfortunately, the nature of the problem does not allow this approach to succeed.
Scope of the Study
The industry needs a reliable, low cost reservoir evaluation methodology that can be leveraged to deliver
an engineered completion on every lateral. Ideally, the measurements used would be easily deployed and
cause little or no disruption to the current drilling and completions workflows. Such a technique would
make engineered completions part of the factory approach on every lateral completed.
Our solution starts with data that already exists on every lateral. Current drilling practices include using
an Electronic Drilling Recorder that collects data such as weight on bit (WOB), torque (TOR), rotational
speed (RPM), rate of penetration (ROP), mud flow rate (Q), standpipe pressure (SPP) and differential
pressure (DIFP) on every well. These parameters are used by the drilling department to help ensure an
efficient drilling operation and the completions department can leverage this data to drive superior well
productivity through a drilling parameter known as Mechanical Specific Energy. The study will demon-
strate how, using our patent-pending technique (ref 12), Mechanical Specific Energy has been used to
improve the completion design on over 60 wells to date, in a reliable and cost-effective manner.
Adapting Mechanical Specific Energy For Reservoir Evaluation
The use of Mechanical Specific Energy (MSE) in the oil industry has been around for over 50 years, first
published by Teale in 1964 (ref 1). Drilling experts use MSE to optimize drilling operations because an
understanding of the geomechanical properties of the rock will enable them to make the right decisions
on how to drill. The completions department can take advantage of MSE by recognizing the close
relationship between it and unconfined compressive strength (UCS).
(1)

where Deff is the efficiency of transmitting the penetration power of the rig to the rock
Both Teale (ref 1) and Kelessidis (ref 2) point to the relationship between UCS, MSE and drilling
efficiency. The efficiency of current shale drilling operations is largely dependent on factors such as the
SPE-174839-MS 3

type of bit being used, how worn that bit is and the tortuosity of the wellpath. This drilling efficiency is
not a parameter that is easy to derive; therefore, one must take this into account when attempting to
compare one interval to the next. However, over relatively short intervals (i.e., the length of a frac stage),
it is a fair assumption that the Deff factor for the rig will remain reasonably constant. Therefore, MSE can
be used as a qualitative predictor of unconfined compressive strength within each frac stage. By
calculating MSE from our drilling data we can then predict zones of comparable UCS within each stage.
This will enable us to place our perforation clusters in zones of similar UCS within each stage, which is
exactly what many completions departments are attempting to do using wireline measurements.

Deriving MSE from Common Drilling Data

Teales’ original equation includes two components, a thrust component and a rotary component. The
thrust component et is stated as Force/Area (lb/in3),
(2)

The rotary component er is stated by Teale as:

(3)

By adding the thrust component and the rotary component together we get:
(4)

where
WOB ⫽ Weight on Bit (k.lb)
N⫽ Rotational Speed (rev/min)
T⫽ Torque (k.ft-lb)
D⫽ hole diameter (in.)
ROP ⫽ rate of penetration (ft/hr)
This equation is well suited to conventional drilling in vertical wells and has been used successfully for
drilling optimization (ref 4). However, horizontal wells involve the use of a mud motor, which changes
the rotary component of the equation. The rotation seen at the bit is now the sum of the rotation of the
pipe (N) and the rotation of the mud motor.
(5)

Kn ⫽ Mud motor speed to flow ratio (rev/gal)

Q ⫽ Total Mud flow rate (gal/min)
N ⫽ Rotational Speed of drill pipe (rev/min)
In Teales’ equation the torque value comes from a surface measurement that is completely adequate for
estimating torque at the bit-face in vertical wells. In horizontal wells, this is not the case, because the
surface measurement is impacted by the many “touch-points” along the lateral where there is friction
between the pipe and the well trajectory. The result is that the surface measurement of torque will read
significantly lower than torque at the bit-face. In sliding sections it is not unusual for the torque at surface
to read zero, because the entire drill-string is touching the wellbore and the torque is completely absorbed
by the frictional losses along the drill-string/wellbore interface.
To solve this problem, one alternative is to add in a downhole torque measurement into the bottomhole
assembly (BHA), which will provide a reliable estimate of torque at the bit. Another alternative is to take
advantage of the mud motor and its’ proximity to the bit. As the drill bit engages the rock, the torque at
4 SPE-174839-MS

the bit-face manifests itself as a differential pressure across the mud motor, which can be used to derive
a good estimate of torque (ref 11) as shown in equation (6):
(6)

Tmax ⫽ Mud Motor max-rated torque (ft-lb)

Pmax ⫽ Mud Motor max-rated ⌬P (psi)
⌬P ⫽ Differential Pressure (psi)
With a mud motor in the place, Teales’ equation can now be re-stated as:
(7)

Investigating this equation further (using realistic values for each parameter), the rotary component
calculates to approximately 99% of the total value of MSE. This means that variability in WOB is not
significant to the end result. Also, in most cases the values of N and Q tend to stay reasonably constant
throughout the entire well. If you are looking for relative changes in MSE across short intervals, these two
parameters are often insignificant. The values of Kn, Tmax, and ⌬Pmax are parameters determined by the
mud motor type in use, and D is hole size. This leaves ROP and ⌬P as the two parameters in this equation
that drive the value of MSE. When considering the quality of the input drilling data, it is important to
consider the impact each parameter will have on the end result.

Drilling Data Quality

Using drilling data to perform reservoir evaluation is a novel concept but does not come without its share
of issues to consider. As with all measurements, including even the most modern wireline logs, there are
environmental effects that are not related to the properties of the reservoir. These effects need to be
corrected before doing further interpretation steps and this is particularly important when working with
drilling data.
Averaging: Wireline logs are acquired at a rate of one sample every 6 in. and then averaged over three
samples. Drilling data is normally sampled at a rate of one sample per foot, and we have found that
averaging it over three samples tends to give the best result. Since this application is predominately
focused on horizontal laterals, where vertical resolution is not an issue, we chose to average across 3-ft
intervals. Application of MSE in a vertical well might warrant less averaging to achieve a better vertical
resolution.
De-Spiking: There are some wireline/LWD measurements (i.e., sonic logs) that are prone to spiking,
particularly in gas zones. There are software algorithms available to help remove these spikes. Drilling
data is much less prone to spiking, but when spurious data occurs they can be manually removed by using
the same software tools.
Missing data “variable” data channels: Occasionally there will be missing/null values over a range
of depths. These can be manually repaired by using patching algorithms that have been developed for
wireline applications. Drilling data tends to be more prone to these effects. If it is over relatively short
intervals (⬍ 5 ft) it might suffice to patch the curve manually. More care must be taken when the issue
persists over longer sections. In this case, we look at other data (i.e., LWD GR and mudlogs) to help guide
decisions on how to repair missing data. If there is a high degree of uncertainty, then the best practice is
to mark the interval accordingly.
Missing data “constant” data channels: Some of the data channels tend to be very stable, rarely
changing across the lateral. Often the RPM and Q channels rarely change while the driller is rotating. If
data is missing, even over longer rotating sections, it is not a bad assumption to patch them with the same
constant value. Sliding intervals are different and are discussed in more detail below.
SPE-174839-MS 5

ROP approaching zero: In the rotational component of the equation ROP is in the denominator. As
ROP approaches zero the equation will be unstable and deliver extremely high estimates for MSE. Sliding
intervals are particularly prone to this happening because progress is slow when only drilling with the mud
motor. If this effect is causing an issue, then we set a limit on ROP (i.e., if ROP ⬍ 3 ft/hr, then set
ROP⫽3).
RPM in sliding intervals: When sliding, the drill pipe does not rotate and therefore RPM⫽0.
However, some drillers like to “rock” the pipe when sliding to reduce the static friction. This rocking
effect shows up as RPM values that are non-zero. Since this movement of the pipe does not translate to
additional rotational force at the bit, for the purposes of calculating MSE we set RPM⫽0 when sliding is
taking place.
Differential Pressure (DIFP): The calculation of MSE involves six drilling measurements, with two
that are the primary drivers: ROP and DIFP. The ROP channel tends to be a high-quality measurement
that requires very little editing. Unfortunately, the same cannot be said for DIFP. Standard practice in the
oilfield is that the directional driller re-zeroes the differential pressure as he sees fit, but that varies from
driller to driller. In many cases, the DIFP is not zeroed properly, which results in very low or even
negative values of DIFP in sliding intervals. The issue also occurs in rotating intervals, but it is more
difficult to detect, because DIFP values are higher and the errors can easily be misinterpreted as changes
in rock properties. If not handled properly, this can be problematic and lead to significant errors in
reservoir evaluation.
During this study the relationship between Standpipe Pressure (SPP) and DIFP was investigated. Both
of these measurements contain a reservoir-related component and a non-reservoir-related component. The
non-reservoir component is impacted primarily by three effects: the frictional forces caused by the
interaction of the drilling fluids and the drillpipe, which increases with depth; changes in Q from the mud
pumps; and changes in density of the drilling fluid inside the pipe. The impact of these effects makes it
necessary for the driller to re-zero the DIFP measurement repeatedly while drilling. For the purposes of
reservoir evaluation, it would be preferable to avoid a measurement that contains these unpredictable
events.
The best practice developed during this study was to avoid using the field measurement of DIFP, and
instead correct the SPP for the effects previously listed. All three effects are predictable and easy to
observe. After the SPP channel is corrected for these effects, it is then calibrated to DIFP in a rotating
interval directly off bottom. This channel is now equivalent to the DIFP channel we need to calculate MSE
without the side effects caused by re-zeroing in the field. The result is a reliable DIFP channel that delivers
a superior MSE calculation.
Interpretation Workflow
After the data has been properly corrected and prepared for interpretation, we can apply the appropriate
Mechanical Specific Energy (MSE) equation and compute an MSE channel across the entire lateral
section. The next step is to categorize the lateral into a set of geomechanical-based facies according to
their MSE, which will facilitate the process of placing perforations optimally. The dynamic range of the
MSE channel generally tends to vary according to lithology. In general, shales are easier to drill (lower
MSE) while sands tend to be tougher to drill (higher MSE). With that being said, this is not universally
true and we have seen cases in which adjacent limestones and shales drill in a similar fashion with
comparable MSE values.
For ease of use and consistency, a standard facies classification has been deployed, as shown in Fig.
1. This standard classification could be fine-tuned to a facies classification for each basin. The basis for
a customized facies classification would be corroborative information (i.e., core data, log data) that guides
the adjustment of the cutoffs for the color index by using this ground truth information. This is a standard
practice in the industry and could be extended to our drilling-based facies classification scheme.
6 SPE-174839-MS

Figure 1—Standard hardness index used for facies classification

We have also found it useful to deliver what we refer to as a “macro-facies” and a “micro-facies,”
shown in Fig. 2. The macro-facies uses the standard classification just described. The micro-facies does
not use hard cut-offs but is a continuous spectrum of color that helps the user understand the details as
the section transitions from one facies to the next. This can be particularly useful when faced with having
to perforate different macro-facies within a given stage.

Figure 2—Facies logs based on drilling data

After the facies log is complete, the next step is the engineered completion. Typically, we start with
the standard geometrical design that the operator is most comfortable with. This usually involves a set
SPE-174839-MS 7

stage length (i.e., 225 ft), number of perfs per stage (i.e., four clusters/stage), and fixed distance between
clusters (i.e., 45 ft). Next, the frac plugs depths are positioned according to the start/stop depths for the
overall completion. After the geometric perf clusters are determined, we color code each cluster as per the
macro-facies classification at that depth, as shown in Fig. 3. Note that clusters that are left blank are those
that land in sliding intervals in the geometric design. This happens less than 10% of the time (5 out of 120
in Fig. 3) and when possible, we attempt to move the perf clusters to avoid sliding intervals, since the MSE
calculation has a much higher uncertainty in these intervals.

Figure 3—Example of a geometric perf design color coded as per the macro-facies plot

The geometric design is then investigated stage by stage. At each stage, the heterogeneity of the MSE
is evaluated and options are investigated for re-positioning perf clusters to minimize the impact of this
heterogeneity. Fig. 4 shows a typical geometric design distribution with its’ associated MSE classification.
In the example, stages 2, 7, and 9 are of particular concern, since there is one perf cluster in each with
a lower MSE value (HD1 yellow) than the other nine. This design is sub-optimal, because there is a high
likelihood that the HD1 perf cluster will be over-treated while the others will be under-treated by the frac.
To avoid this situation, the perf clusters are relocated so that they are situated in facies with similar MSE
values. For example, in stage 7 of this example Perf Cluster #2 has a high likelihood of being over-treated.
The macro-facies plot for this stage is shown in Fig. 4 with the geometric perf design and Fig. 5 shows
an alternative to the geometric design, where the clusters are moved around so that eight of the ten clusters
are now orange (HD2), while the other two are red (HD3). This optimized perf design will increase the
odds of getting an evenly spaced treatment across at least 8 of the 10 perf clusters, which will improve
the productivity of this particular stage. While a number of perfs were recommended for re-positioning,
in this example the most important change was moving Cluster #2 from HD1 (yellow) to HD2 (orange).
8 SPE-174839-MS

Figure 4 —Stage 7 macro-facies with geometric perf design (10 perf clusters)

Figure 5—Stage 7 showing optimized perf placement

Calculating MSE in Sliding Intervals

Fig. 6 is an example of a typical sliding interval, as marked by the red bar in the depth track. When sliding,
it is normal for RPM and TOR to go to zero and WOB to be more erratic in nature. You will also typically
see ROP and DIFP both drop dramatically. There are huge changes in the mechanics of sliding vs rotating
and this has a significant effect on the calculation of MSE.
SPE-174839-MS 9

Figure 6 —Example of drilling data in a sliding interval

When calculating MSE in a sliding interval, you must perform tasks to prepare the data:
1. Check to ensure that the RPM has gone to zero. If the RPM is not reading zero, then it must be
zeroed out. When sliding, none of the RPM that is measured at surface is being transmitted to the
bit. The only component of rotation that the bit will see comes from the mud motor, (Kn*Q).
2. Perform any necessary corrections to DIFP. Depending on how the driller calibrated the differ-
ential pressure, DIFP might drop erroneously low. Sometimes, you can even see it go negative,
which is a physical impossibility. To account for this we add an offset to the DIFP channel across
each sliding interval. This offset will vary across each sliding interval. We set this offset to ensure
that the jump in MSE (as we transition from rotating to sliding) is minimized.
For this example 50 psi was the offset applied to the DIFP channel in the sliding interval. This value
ensured that the MSE calculated in the upper half of the sliding interval (10,115 to 10,140 ft) was in
agreement with the MSE in the rotating section directly above this section. This decision is backed up by
the GR (in the depth track), which suggests that there is a facies change at 10,140 ft. In the lower half of
the sliding interval (10,170 to 10,140 ft) the MSE values are consistent with the section below the sliding
interval. Again, this is in good agreement with the GR log.
Perforating Depth Control
Over the last 10 years the shale boom has helped the oil industry revolutionize operations in North
America. One unfortunate by-product is that now we have an entire generation that has never had to be
concerned with perforating depth control. The assumption has been that the reservoir was homogeneous
and isotropic; therefore, being off-depth with perforating guns was of little or no consequence. Just get it
10 SPE-174839-MS

close and try not to shoot any collars. In most current completions wireline CCL depth is tied directly into
the pipe tally. A CBL/GR/CCL might be available but this does not always cover the lateral. The CBL
is referenced to pipe tally through short joint(s). In cases in which there is sufficient dymanic range across
the lateral section, using a gamma ray helps. Very often, there is little change in GR across the lateral,
making the tie-in difficult.
Engineered completions will demand more from our wireline crews, because now we need to be on
depth within a few feet. This is the case whether we are using drilling data or wireline data to identify the
“like rock”. It also means we will need to add in more short joints, preferably one every 1000 ft, so that
the perforators can tie in close to where they are shooting. Fig. 7 demonstrates just how significant this
can be. For this particular example, the perforating company tied into a crossover located at 7,250 ft. On
each of the 28 stages the perforators used the crossover to correlate to and then proceeded to perforate
assuming they were on depth. The crossplot shows the resulting difference between the casing tally and
wireline depth at each collar. The difference at 15,000 ft was 40 ft, and it gradually improved at each collar
all the back to the cross-over where the two match. This difference is essentially the difference between
stretch in a wireline cable and stretch in a casing joint. This discrepancy is well understood and has always
been accounted for when perforating vertical wells. To reap the rewards of an engineered completion
design, we will need to treat perforating depth control in horizontal wells with the same care and diligence.

Figure 7—Depth discrepancy between perf CCL log and casing tally

One possible alternative would involve running GR/CCL in memory mode while doing TCP as a cost
effective way to log the GR and collars, and then compare them to pipe tally and LWD results. This will
tie in all three measurements and identify depth correction issues without making extra runs in the well
or increasing time spent at the well site.
Results, Observations and Conclusions
During the early phases of this study, the focus was on being able to reliably deliver a consistent product
on every well. After the technical testing was completed, the next step involved validating the approach.
To evaluate the effectiveness of this technique, comparisons were done with a variety of reservoir
evaluation tools that are well understood in the industry. The following case studies are intended to aid
the reader in understanding how reservoir evaluation based on drilling data compares to a variety of
standard reservoir evaluation tools currently accepted in the industry.
SPE-174839-MS 11

Case Study 1 Naturally Fractured Reservoir

This first case study comes from a well that was drilled in West Texas in May 2014. This well was a 4,700
ft lateral drilled through the Wolfcamp shale to a total depth of 14,600 ft. The use of reservoir evaluation
to engineer completions is particularly popular in this play because this reservoir tends to be highly
heterogeneous in nature (ref 9). Unlike most horizontal shale wells, this well was drilled with water-based
mud to be able to run an image log in the lateral. The evaluation included a comprehensive suite of
openhole wireline logs including nuclear porosity logs, a spectral GR log and a formation micro-imager
(FMI) log. For many years, the FMI log has proven to be an excellent reservoir evaluation tool in
horizontal shale plays (ref 6). The goal of this reservoir evaluation was to quantify the heterogeneity along
the lateral and identify intervals with open, natural fractures. These naturally fractured intervals would be
the target of the completion design for this well.
The open-hole log data is represented alongside the drilling-based facies logs in Fig. 8. The interval
shown is 50 ft in length. The spectral GR is shown in the depth track at the bottom of the figure, and it
suggests a rather homogeneous section. In the track just above the depth track, are curves that indicate
results from image interpretation that quantify the presence of drilling-induced fractures (shaded green),
healed natural fractures (light blue), and open natural fractures (royal blue). Just above that, in the middle
of the figure, are the static and dynamic images from the FMI log. The interpretation based on the image
log shows that the majority of the open natural fractures (royal blue) reside only in the middle of this
particular interval. If placing a perforation cluster in this 50-ft section, the recommendation based on
image interpretation would be to place it somewhere in the middle.

Figure 8 —Comparison of drilling-based facies logs to image logs

The facies logs prepared from drilling data are shown directly above the image logs. The macro-facies
and micro-facies logs suggest that the middle of this section (red) is different from the facies at either end
(yellow and orange). The middle section, where the FMI interpretation identifies open natural fractures,
was the most difficult to drill and exhibited the highest MSE values. This demonstrates that by focusing
on the red facies in this lateral, the operator could have successfully identified the best intervals to place
their perforation clusters. This section is typical of the rest of the well and demonstrates that MSE is a
reliable indicator of “like rock” in the Wolfcamp.
12 SPE-174839-MS

Case Study 2 Production Results in Offset Wells

This case study compares two wells drilled by Peregrine Petroleum in Ellis County, OK. The formation
is the Lower Cleveland sand, a silty sandstone. These two wells were selected by the operator for
evaluation because the two wells were in close proximity (Fig. 9), were completed in a similar fashion and
yet had significantly different production results (Fig. 10).

Figure 9 —Map showing location of the two Lower Cleveland wells (0.7 miles apart)

Figure 10 —MSE based macro-facies logs on the Lower Cleveland 1H and 4H wells
LOWER CLEVELAND 1H LOWER CLEVELAND 4H

Both wells were completed geometrically with identical schemes (20 stages, 4 clusters per stage, stage
length of 235 ft) and were treated with similar frac programs.

Well Name Proppant Fluids

Lower Cleveland 1H 3.9 MM lb 40/70 73,000 bbl slickwater
Lower Cleveland 4H 4.0 MM lb 40/70 77,000 bbl slickwater

A 3,300-ft section of the macro-facies log from each of these two wells are shown in Fig. 10, and it
shows that the 1H well has significantly less lateral heterogeneity than the offsetting 4H well. The
variability in rock strength at each perf cluster is shown in Fig. 11, and this gives us valuable insight into
variability in production between the two wells.
SPE-174839-MS 13

Figure 11—The perf clusters in the 1H well show far less variability within each stage

Assuming the weakest facies in each stage will be effectively treated, the table on the left in Fig. 11
predicts that 63 of the 80 perf clusters in the 1H will be effectively fractured. The table on the right, for
the 4H well, would suggest that 42 of the 80 perf clusters would be effectively fractured. This means that
the 1H well is interpreted to have 50% more fractures contributing, which is in excellent agreement with
the actual production data (Fig. 12).

Figure 12—In the first year of production the Lower Cleveland 1H well delivered 58% more oil and 74% more gas than the neighboring
Lower Cleveland 4H well.
14 SPE-174839-MS

Conclusion
This study demonstrates the successful deployment of a methodology that leverages commonly available
drilling data to produce a reliable engineered completion design. As shown in the first case study the
results provided are every bit as reliable as high end openhole wireline logs but without the associated rig
time and expense. Since the data already exists on every well drilled this technique can be utilizied for
every completion design, unlike most reservoir evaluation techniques currently available. This study also
demonstrates that there is excellent agreement between the answer products delivered and well produc-
tivity. The results confirm that this technique can be relied on to guide decisions for engineered
completions in shale plays in a way that is both timely and cost effective. By adding this technology to
the completion design process, oil and gas operators will now be able to enhance well productivity on
every well, making this a new efficiency gain in the shale factory process.

Acknowledgements
The authors would like to thank Peregrine Petroleum and Brigham Resources for their consent to publish
this work and C&J Energy Services management for their guidance and support. We also want to thank
Pat McKinley, Roger Barker and Michiel de Rijk for their mentoring and expertise in drilling data and
practices and Vorpal Energy Solutions for their data processing services.

References
1. Teale, R.: “The Concept of Specific Energy in Rock Drilling,” International Journal of Rock
Mechanics and Mining Sciences, Vol. 2, 1965 pp 57–73
2. Kelessidis, V.C.: “Rock Drillability Prediction From In Situ Determined Unconfined Compres-
sive Strength of Rock,” The Journal of the South African Institute of Mining and Metallurgy,
Volume 111, June 2011 pp 429 –436
3. Bevilacqua, M., Ciarapica, F. and Marchetti, B.: “Acquisition, Processing and Evaluation of
Down Hole Data for Monitoring Efficiency of Drilling Processes,” Journal of Petroleum Science
Research (JPSR), Volume 2, Issue 2, April 2013 pp 49 –56
4. Caicedo, H., Calhoun, W., and Ewy, R.: “Unique Bit Performance Predictor using Specific
Energy Coefficients as a Function of Confined Compressive Strength Impacts Drilling Perfor-
mance,” WPC-18-0861 presented at 18th World Petroleum Congress, Johannesburg, South
Africa, 25-29th Sept, 2005
5. Wutherich, K., Walker, K., Iroh, A., Ajayi, B. and Cannon, T.: “Evaluating an Engineered
Completion Design in the Marcellus Shale Using Microseismic Monitoring,” SPE 159681
presented at SPE ATCE in San Antonio, TX Oct. 8-10th, 2012
6. Waters, G., Heinze, J., Jackson, R., Ketter, A., Daniels, J. and Bentley, D.: “Use of Horizontal
Well Image Tools To Optimize Barnett Shale Reservoir Exploitation,” SPE 103202, presented at
the SPE ATCE in San Antonio, TX Sept 24-27th, 2006
7. Miller, C., Waters, G. and Rylander, E.: “Evaluation of Production Log Data from Horizontal
Wells Drilled in Organic Shales,” SPE 144326 presented at SPE Unconventional Gas Conference
in The Woodlands, TX, June 14-16th, 2011
8. Ejofodomi, E., Baihly, J., Malpani, R., Altman, R., Huchton, T., Welch, D. and Zieche, J.:
“Integrating All Available Data to Improve Production in the Marcellus Shale,” SPE 144321
presented at the SPE Unconventional Gas Conference in The Woodlands, TX, June 14-16th, 2011
9. Ajisafe, F., Pope, T., Azike, O., Reischman, R., Burkhardt, C., Helmreich, A. and Phelps, M.:
“Engineered Completion Workflow Increases Reservoir Contact and Production in the Wolfcamp
Shale, West Texas,” SPE 170718-MS presented at the SPE ATCE held in Amsterdam, The
Netherlands, 27-29th October, 2014
SPE-174839-MS 15

10. Crawford, E., Senters, C., Bullard, K., Leonard, R. and Woodroof, R.: “Tracers Validate
Production Techniques,” The American Oil & Gas Reporter, March 2015, pp 66 –72
11. Lyons, W.C. and Plisga, G.J.: “Standard Handbook of Petroleum and Natural Gas Engineering”
pp 4 –284 –4 –290
12. US Patent Application 14/734,290: Determining One or More Parameters of a Well Completion
Design Based on Drilling Data Corresponding to Variables of Mechanical Specific Energy