You are on page 1of 21

URTeC: 2432330

Breakthrough in Hydraulic Fracture & Proppant Mapping: Achieving


Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Increased Precision with Lower Cost


Matthew Dawson* 1, 2, Günther Kampfer 2
Reveal Energy Services 1, Statoil 2

Copyright 2016, Unconventional Resources Technology Conference (URTeC) DOI 10.15530/urtec-2016-2432330

This paper was prepared for presentation at the Unconventional Resources Technology Conference held in San Antonio, Texas, USA, 1-3 August 2016.

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

Hydraulic fracturing has been instrumental in commercializing ultra-tight unconventional resources. Although
hydraulic fracturing has been used for nearly half a century in more than a million wells, understanding and mapping
hydraulic fracture growth remains a challenge for shales and ultra-tight reservoirs. A number of approaches have
been taken to better understand and characterize hydraulic fractures in the subsurface, but a technology which can
accurately map hydraulic fractures with minimal operational interference and negligible cost remains elusive.
Microseismic based mapping is arguably the most ubiquitously deployed method, but this approach is costly and
provides limited insight into characterizing hydraulic fractures. Alternative technologies for mapping hydraulic
fractures are currently being explored, but many of these technologies provide only qualitative information or
require expensive data acquisition tools.

This paper discusses a novel hydraulic fracture and proppant mapping technology (IMAGE Frac), which is
technically robust, easy to use, and low cost. The technology is founded upon basic poromechanic theory, utilizing
measurements from surface pressure gauges during the stimulation process to determine the geometry, orientation,
and spatial location of hydraulic fractures with higher precision than other traditional techniques, while providing
insight into the proppant distribution. The data acquisition approach requires only minor deviations from traditional
practices and can be implemented without impacting completions efficiency. This technology has been utilized
successfully in over 30 wells in multiple plays including the Bakken and Eagle Ford and is targeted for deployment
in several other plays in mid-2016. An overview of the technology is provided along with an in-depth discussion of
the validation studies from both the field and simulations. Two case studies are provided, which show the potential
of the technology to provide new insight and improve drilling and completions operations. The case studies illustrate
specific examples of how this technology enables better selection of landing zones, proppant size, pumped volumes,
well spacing, and overall completions strategy. A companion paper (Kampfer and Dawson, 2016) discusses the
fundamentals of the technology and provides in-depth, simulation-based studies, which demonstrate the robustness
of the technique.

Introduction

Recent advances in hydraulic fracturing, including multi-stage fracturing, novel completion tools, and use of
slickwater & hybrid fluid systems (King 2010), have enabled economic production from North American shale
reservoirs. Well productivity in these ultra-low-permeability reservoirs is largely controlled by the stimulation
effectiveness (i.e., the ability to create a large matrix-fracture contact area with optimal conductivity distribution). In
developing a field, multiple horizontal wells are drilled on a pad, so identifying the optimum well spacing to
maximize the value of a pad is critical. Currently, well spacing is often optimized based on numerous well spacing
field trials, which is a costly and inefficient approach. Developing an improved understanding of hydraulic fracture

URTeC 2016
438
2432330

geometry using systematic methods and workflows, therefore, has potential to significantly improve the value
extracted from a development.

Hydraulic fracturing has been used for decades to enhance the producibility of tight-gas reservoirs. The
fundamentals of fluid transport in fractures, matrix leakoff, and fracture mechanics during fracture propagation have
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

been well-studied, leading to the development of pseudo-3D (Settari & Cleary 1986) and planar 3D fracture models
(Barree 1983), as well as bottomhole treatment pressure analysis tools (Nolte & Smith 1981). These tools have been
widely used for estimating fracture lengths and drainage boundaries in hydraulically fractured tight-gas reservoirs.
However, despite the wealth of knowledge in tight-gas reservoirs and studies on hydraulic fracture propagation
dating back to when Sneddon (1946) developed one of the first fracture propagation models, understanding the
fracturing process in unconventional reservoirs is still in its infancy. Shale reservoirs are complex and
heterogeneous. Moreover, they often contain natural fractures, faults, and other planes of weakness, which can
complicate fracture propagation. The interaction between hydraulic and natural fractures can lead to reactivation of
natural fractures and complex fracture growth (Renshaw & Pollard 1995; Wu et. al. 2012). Although there have
been recent attempts to model complex fracture propagation (Koshelev et al., 2003; Zhang et al., 2006; Weng et. al.,
2011; McClure 2014), the mechanics of network growth is not fully understood, and reservoir characterization and
simulation in three dimensions remains challenging. This has limited the applicability of fracture models in ultra-
tight, complex plays. Many diagnostic technologies such as microseismic methods, microgravity methods,
electromagnetic imaging, fluid tracers, and proppant tracers have been employed to gain insight into the geometry of
hydraulic fractures in these complex systems. However, these methods are costly and have large interpretation
uncertainties (King 2010). For example, microseismic measurements only capture a small portion of the total rock
deformation and can easily be biased by the placement of the monitor well, inaccuracies in the velocity models, or
methods of data interpretation and are challenged with signal attenuation, noise, and distortion of the wavefield. All
of these challenges make it difficult to rely upon the resulting fluid distributions produced from microseismic
analyses (Cipolla and Wallace, 2014). Likewise electromagnetic based mapping techniques, such as resistivity
mapping tools face similar challenges and can be more cumbersome in the field to implement with particular
requirements for the fracturing fluid chemistry and downhole instrumentation. Fluid and proppant tracers are other
commonly used diagnostic tools that have proven useful in identifying communication between wells, but
unfortunately provide limited insight into fracture and proppant geometries.

This paper presents a cost-efficient, high-precision method for analyzing hydraulic fracture geometries and proppant
distribution in multi-well unconventional pads. The technology leverages a particular strategy for stimulation
designs, which enables surface pressure responses to be measured in isolated stages of wells during the stimulation
of adjacent wells. Poroelastic signals in the pressure data are then differentiated from direct fluid connectivity
signals based on the arrival time, magnitude, derivatives, and duration. These poroelastic signals can then be used in
conjunction with fully-coupled (solid mechanics & pressure diffusion) numerical models to determine fracture
geometry & complexity with sufficiently high precision while providing insight into propped geometry.

One of the key elements of this integrated approach is the use of poroelastic theory. Poroelastic theory has been
effectively studied and applied in the industry for decades, but it has yet to be fully leveraged in unconventionals.
This theory describes the mechanical behavior of a saturated porous media by combining the theory of linear
elasticity with Darcy’s law. Terzaghi (1923) established the foundation of poromechanics in his 1D consolidation
theory. Biot (1935 and 1941) later extended this framework, developing a full 3D model. Over the past century,
many authors have expounded upon this field, most recently including Detournay & Cheng (1993), Coussy (1991),
Gueguen (2009). Some of the most classic works in this field include Haimson & Fairhurst 1969, Detournay et al.
1989 and Boone & Ingraffa 1989. The analysis provided herein extends these concepts to ultra-tight, unconventional
systems, providing fresh insights into critical parameters such as fracture geometry. Daneshy et al. (2012) carried
out field measurements, detecting stress shadow effects and tried to relate it to fracture dimensions.

IMAGE Frac Background

The focus of this paper is on a low-cost approach to mapping hydraulic fractures and proppant distribution in a
horizontal well. The technique uses an Integrated Modeling Approach for Geometric Evaluation of Fractures and is
thus, referred to as IMAGE Frac. The technique relates poromechanically induced signals measured in observation
wells during the hydraulic fracturing process to hydraulic fracture and proppant distribution geometry. A detailed
review of the technology is provided in a companion paper (Kampfer & Dawson, 2016). This paper, herein, focuses

URTeC 2016
439
2432330

on qualification studies, validating the technology, and provides two selected case studies showing the potential to
use this technology to improve development of unconventional resources.

Data Acquisition
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

The first step in applying this mapping technology is ensuring proper data acquisition. The IMAGE Frac approach
primarily leverages surface pressure measurements from a ‘monitor well’ adjacent to a well undergoing stimulation,
where the monitor well can be any of a number of other wells on the same pad undergoing completion around the
same time. While measuring pressure in adjacent wells during stimulation is not uncommon, extracting useful
information, beyond simply identifying direct fracture hits, can be difficult since very little is known about the
location of fluid communication along the wellbore. Subsurface gauges can be placed along the lateral to better
pinpoint communication, but without proper isolation both inside and outside the wellbore, this too can provide
limited benefit. Moreover, both permanent and retrievable subsurface pressure gauges can be very costly to deploy
in ultra-deep, ultra-long laterals. In a slickwater stimulation process, the wells are filled with water, which is highly
incompressible, allowing for a surface gauge to effectively act as a surrogate to downhole pressure.

The proposed method, therefore, uses low cost surface gauges to minimize capital expenditure. It also overcomes the
challenge of locating the origin of the pressure signals in the monitor well by isolating a single stage along the
lateral from prior stages. For instance, isolating a single stage in the monitor well can be achieved by isolating the
annulus with a packer and isolating the interior of the well with a bridge plug or by using controllable frac sleeves. It
is recognized that these approaches may not achieve complete zonal isolation, but it is believed under most
circumstances these provide sufficient isolation for this approach. After isolation, the stage in the monitor well can
be completed and monitoring of the surface pressure may begin, measuring the pressure response in a single,
‘isolated’ stage in the monitor well. Thus, the spatial location can be largely known for both the isolated stage in the
monitor well as well as any stages undergoing completions in adjacent wells. The pressure data collected in the
monitor well can then be used to analyze fracture geometry and proppant distribution.

An example of data acquisition for the technology is shown below in a study where five horizontal wells were
completed simultaneously. As the adjacent wells are hydraulically fractured, surface pressure data is recorded in
‘isolated’ stages in the monitor well (6H). An illustration of data acquisition in one monitoring stage (stage 10) is
provided in Figure 1. In this case, stage 10 is isolated from the prior stages with a bridge plug while stages 11-13 in
wells 7TFH and 5TFH as well as stages 7-13 in Wells 4H & 3TFH are stimulated.

The surface pressure response in Stage 10 of the monitor well is shown in Figure 2. Pressure changes in excess of
800 psi in less than five minutes are identified, indicating the potential for surface gauges to accurately register
downhole pressure changes. These signals show that surface gauges can act as direct analogues for subsurface
gauges in a case where the wellbore is largely filled with water (a highly incompressible fluid) and pressure is
measured in a stage that has already been hydraulically fractured (providing a sufficient mobile fluid volume to
account for the compressibility and convey the pressure signal).

Figure 1. Well layout illustrating pressure data acquisition process in stage 10 of the monitor well (6H) as stages 11-13 in wells 5TFH and 7TFH
as well as stages 7-13 in wells 3TFH, 4H are stimulated.

URTeC 2016
440
2432330
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 2. Surface pressure measurements from stage 10 of the monitor well (6H) during hydraulic stimulation of the adjacent wells.

Although using a recently completed well as the monitor well is preferred to ensure accurate conveyance of the
downhole pressure signal, the ability to use this technology in producing wells has also been examined and shown to
be effective; however, in these scenarios where the monitor wellbore is not largely filled with water, only qualitative
information can be derived.

In order to keep this method low cost, one of the key elements of the process is to ensure that there is no impact on
completions efficiency. In some cases, this is handled by completing a pad with several wells simultaneously. In the
previous example, five wells are used and grouped into two groups. The first group, comprising wells 5TFH, 6H,
and 7TFH, are completed together, allowing for optimal completions efficiency. The measurements begin after the
desired stage is completed in the monitor well (e.g. Stage 10). The monitor well is allowed to sit idle while
additional stages are completed in both the first group and the second group, comprising Wells 3TFH and 4H.
Allowing the monitor well to sit idle provides a continuous data stream which minimizes influences from operations
in the monitor well and enables measurements from many stages in several wells to be taken from a single stage in
the monitor well. Completing wells in groups also enables the monitoring stage in the monitor well to run well
ahead of the stages being completed in the second group. This provides a much more meaningful set of data.
However, in some completions approaches, completing a single well at a time is preferred. The IMAGE Frac
technology can also be applied under these scenarios with the proper design.

Identifying Poromechanically Induced Signals


After acquiring the pressure data in the monitor well(s), poromechanically induced pressure signals must be
differentiated from pressure signals caused by direct fluid communication (either through a direct fracture hit or
through an underlying high-permeabilty network, such as a pre-existing natural fracture network). A number of
analytical, numerical, and field-based studies were conducted to develop processes for identifying poromechanically
induced pressure signals. Both the analytical and numerical studies demonstrated poromechanic signatures have
different arrival times, magnitudes, derivatives, and durations than direct fluid communication signatures, and direct
fluid communication signatures cannot be designed to emulate a pure poromechanics signature. These theoretical
studies were further supported by field-based characterization studies, which included horizontal core, advanced
MicroScope Ultra-High Resolution Image logs, a natural fracture characterization study, and tracer studies.

Applicability
Once identified, poromechanic signals can be used to determine hydraulic fracture geometry and provide insight into
proppant distribution. The companion paper (Kampfer & Dawson, 2016) describes how poromechanic signals can
be related to fracture geometries and spatial orientations. A detailed review of the fundamental theory underpinning
this technique is provided in that paper. Several sensitivity studies are also provided to show the robustness of the
approach in practical scenarios where it is expected that significant geologic heterogeneity is possible, well
trajectories are complex, multiple & complex fractures may exist, and asymmetry in fracture dimensions is
expected. The paper discusses in depth where this technology can be applied and where its limitations are. A
summary of the target deployment scenario for mapping the largest fracture in each stage examined is outlined
below:

URTeC 2016
441
2432330

 ≥ 2 horizontal wells with a well spacing less than two times the average fracture half-length
 ≥ 5 stages per horizontal well (can be relaxed to fewer with lower accuracy)
 Limited variation in vertical offset of wells < 80% of frac height
 Minimal well trajectory deviation < 1 m / 10 m of lateral (may be extended with lower accuracy)
 Limited faults & natural fractures < 20% of stages connected (may be extended to higher percentage but risk
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

of inability to image is increased) – highly connected stages cannot be imaged


 Average formation permeability < 100 D (Potential in future to extend to <10 mD+)
 Fracture geometry is largely convex

IMAGE Frac Validation Studies

To validate the ability of the IMAGE Frac technology to map hydraulic fractures and proppant coverage in the field
with sufficient accuracy, twelve qualification studies were completed, as outlined below along with two additional
studies in progress. These studies listed below encompass both numerical simulations and field studies, leveraging
deployment of the technology in over 30 wells in three plays, which include the Bakken and Eagle Ford:

1) Consistency With Variable Field Inputs


2) Fracture Propagation Direction Mapping
3) Fracture Hit Analyses
4) Fluid & Proppant Tracer
5) Shut-In Pressure Interference Tests
6) Field & Simulator Validation of Signal Classification
7) Fundamental Simulation Sensitivity Studies
8) Proprietary Frac Length Validation Signals
9) Fracture Simulator vs. IMAGE Frac Geometry
10) Diverter Tests
11) Production Rate Transient Analysis
12) Uniqueness Validation Study
13) Electromagnetic Imaging (In Progress)
14) Microseismic (In Progress)

1. Consistency With Variable Field Inputs


The first qualification study to evaluate the potential of IMAGE Frac is a series of consistency tests in the field. In
these studies, the results produced by the automated IMAGE Frac analysis were compared with varying input
scenarios in the field. For instance, in one scenario, fracture geometries of three wells (Well C, Well D, and Well E)
were mapped from two different wells on a pad independently (Well A and Well B). Table 1 presents six normalized
fracture half-length parameters for the average of the largest fractures over the five stages (FHLT) observed in Wells
C, D, and E when observing from Well A and B. It is apparent that the IMAGE Frac technology yields consistent
results when mapping fractures from different wells with the standard deviation between different observation wells
being approximately 4%.

Observation Well Stimulated Well FHLT

Well C 1.00 X

Well A Well D 1.05 X

Well E 0.90 X

Well C 0.98 X

Well B Well D 0.90 X

Well E 0.90 X
Table 1: Comparison of six normalized fracture half-length parameters for the average of the largest fracture in each stage over the five stages
(FHLT) observed for Wells C, D, and E when observing from Well A and Well B.

URTeC 2016
442
2432330

Another set of consistency tests used variable pumped volumes into wells on the same pad and across the same field
to evaluate the robustness of the IMAGE Frac technology. IMAGE Frac consistently mapped fractures which scaled
with the volume pumped per stage, despite having no knowledge of the volumes pumped into the well. In this 18
well study, the eight wells with the smallest volume per stage were identified by IMAGE Frac as the eight wells
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

with the smallest average max fracture lengths. Likewise, IMAGE Frac identified the two wells with the largest
pumped volumes per stage to be the two wells with the largest overall average max fracture lengths. A single pad
study is shown in Table 2 where the volume pumped per stage was varied across four wells. The normalized total
pumped volume per stage is shown for each well along with the resulting average max normalized fracture length
(FL M_A) generated by IMAGE Frac. Table 2 shows a consistent trend of IMAGE Frac fracture length scaling with
the pumped volume per stage without any knowledge of the volume of fluid pumped into the well.

Well Name Total Pumped Volume/Stage (Normalized bbls) FL M_A

Well B6 1.0 X 1.0 X

Well B8 1.2 X 1.3 X

Well B7 1.3 X 1.5 X

Well B4 1.7 X 1.7 X


Table 2: Comparison of the normalized average total fracture length of the largest fracture in each stage over the observed stages generated by
IMAGE Frac (FL M_A) versus the normalized total pumped volume per stage for four wells on a pad.

In a third set of consistency tests, fracture lengths in wells adjacent to three partially depleted wells were examined.
In all cases fractures emanating from wells in proximity to partially depleted wells grew substantially more on the
side of the depleted well than the undepleted side. This is consistent with what would be expected as producing
wells reduce the local minimum horizontal stress as volumes are removed from the reservoir. Hence, as fractures
from adjacent wells grow into these zones the fractures are prone to grow more rapidly towards the depleted zones
than the undepleted zones. Again, in these studies, the IMAGE Frac mapping methodology has no information on
the presence or absence nearby producing wells and yet automatically it identified fractures growing more toward
the depleted wells than the more native reservoir. Not only did the IMAGE Frac technology identify fracture growth
toward depleted wells, but the fracture lengths observed were proportional to the amount of depletion that has
occurred and the proximity to the depleted well. In one example, fracture lengths from two wells in proximity to a
depleted well are examined, where one well is in the same formation as the depleted well and another is closer from
a well spacing perspective but in a deeper formation. In this example, the fractures from the well in the same
formation were found to grow more rapidly toward the depleted well than those in the deeper formation. Figure 3
shows the average normalized length of the largest fracture in a stage over the observed stages for Well TH4 and
BH7, both in proximity to Well BH1, which had been producing for approximately five years. Well BH7 and BH1
are in the same formation while Well TH4 is in the deeper formation. IMAGE Frac identified the propensity of the
fractures in Well BH7 to grow more toward BH1 than those in Well TH4.

Figure 3: Plan view (x/y plane) of average max hydraulic fracture lengths generated by IMAGE Frac in stimulated wells BH7 and TH4 in
proximity to an approximately five-year old producing well BH1. BH7 and BH1 are in the same formation while TH4 is in a different formation
at a different depth. Fractures in both wells are observed to grow preferentially toward the side of the depleted well.

URTeC 2016
443
2432330

In another example, a well in the same formation at the same spacing as the previous example was placed next to a
well that had been producing for approximately two years. The ratio of the fracture length on the depleted side of the
well to that of the undepleted side was similar in the same formation as that in the five-year case, but substantially
less than that in an adjacent formation. In the depleted well consistency studies, IMAGE Frac produced results
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

consistent with expectations despite having no knowledge of the presence or absence of a depleted well, the
proximity to such a well, or the age of the well.

Despite, IMAGE Frac mapping having no knowledge of observation location, pumped volumes, and development
history of the pad, it repeatedly produced fracture geometries that were consistent with expectations for each
scenario. The reproducibility of the geometries observed in the first study when observing from different wells, the
scaling of fracture length proportional to the volume of fluid pumped, and the repeated identification of fractures
growing toward depleted wells in accordance with depletion history and proximity all help to lend credence toward
the validity of the IMAGE Frac mapping approach.

2. Fracture Propagation Direction Mapping


A second study area used to validate IMAGE Frac is the direction of fracture propagation (i.e. identification of
minimum horizontal stress using IMAGE Frac). The largest poroelastic responses occur with the largest fracture
overlap and closest proximity of the observation fracture to the stimulated fracture (Kampfer & Dawson, 2016).
Therefore, by observing the stage location of the peak poroelastic pressure signal in a given well, one can estimate
the fracture propagation direction. Figure 4 shows the poroelastic pressure responses observed in Stage 16 of Well 2
while stimulating Well 1 and Well 3. The peak signal from Well 1 is observed in Stage 14 while the peak signal
from Well 3 is observed in Stage 20. In both cases the estimated fracture propagation direction is determined to be
48o E of N, which is consistent with log measurements in the area (45o – 60o E of N) and microseismic
measurements in the area (45o – 55o E of N). This consistency observed between known fracture propagation
directions and IMAGE Fracs mapped fracture propagation direction further assists in validating the IMAGE Frac
technology.

A) B)
Figure 4: A) Peak poroelastic pressure response as a function of the stage number stimulated for Well 1 and Well 3. B) Plan view (x/y plane) of
well layout and interpreted fracture orientation of 48 oE of N (Note: Fractures not to scale).

3. Fracture Hit Analyses


Two fracture hit studies were carried out in conjunction with IMAGE Frac tests. The first was a two well test with
both wells in the same formation at similar depth, separated by approximately 1,200 feet. Both wells were not
cemented, allowing for identification of any fractures intersecting a well from an adjacent well. Among all the stages
examined, IMAGE Frac showed substantially shorter fractures than would be required to intersect the adjacent well.
In this study zero fracture hits were identified, consistent with IMAGE Frac geometries.

A second five well study was carried out with wells spaced approximately 600 feet apart landing in a deeper (‘T’)
and shallower formation (‘B’), respectively. The vertical separation between wells was proximal enough and the
stress contrasts limited enough that fractures are expected to grow substantially in both formations (Plan view of
well layout show in Figure 5). Pressure was monitored in three stages of Well 6B (Stages 5, 10, and 20) while
several stages in the adjacent wells were stimulated (for illustration, Figure 5 shows stages 7 through 13 stimulated
in the adjacent wells while monitoring Stage 10 in Well 6B).

URTeC 2016
444
2432330
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 5: Plan view (x/y plane) of well layout for five well, uncemented liner, fracture hit test, illustrating one monitor stage (Stage 10 in Well
6B) while several adjacent wells are stimulated (Stages 7 through 13 in Wells 3T, 4B, 5T, and 7T). Note: Fracture lengths are not to scale but for
illustrative purposes only of the number of stages examined from Stage 10 Well 6B.

IMAGE Frac showed the average maximum fracture half-length of Well 4B over the stages examined was
approximately 1,000 feet, propagating at approximately ~48o E of N. The distance between Well 4B and Well 6B at
that orientation was approximately 1,620 feet. Zero fracture hits from Well 4B were identified in all stages observed
in Well 6B, which is consistent with the predicted fracture lengths from IMAGE Frac. The maximum fracture half-
lengths observed for Wells 3T, 5T, and 7T in the ‘B’ formation were between 700 feet and 800 feet based on
IMAGE Frac. The distance from Well 6B to Well 5T and 7T to Well 6B is approximately 810 feet along the fracture
orientation. In the three stages observed in Well 6B, there were zero fracture hits identified. The absence of fracture
hits in Well 6B from Well 3T, 4B, 5T, and 7T is consistent with IMAGE Frac’s evaluated fracture lengths in each
well.

4. Fluid & Proppant Tracer


A five-well fluid tracer study with uncemented liners was used to further evaluate the ability of IMAGE Frac to map
hydraulic fracture geometry. The study was in the same pad as shown in Figure 5 where 22 different chemical
tracers were injected into over 30 stages in Well 4B. Tracer was measured in Well 4B, 5T, and 6B. All chemical
tracers were recovered in Well 4B, as would be expected, demonstrating each chemical tracer could be effectively
recovered from the formation. Tracer from several stages in Well 4B were found in Well 5T. This could be due to
several faults that were identified locally in Well 6B with a MicroScope Ultra-High Resolution Image log and with
dipole sonic logs. Alternatively it could be due to fractures emanating from either Well 4B or Well 5T reaching the
other well (noting that the well separation along the fracture orientation was approximately 810 feet and IMAGE
Frac showed fracture half lengths ranging from 700 to 800 feet. No tracer from Well 4B was observed in Well 6B.

Although the ability to map proppant coverage is still under exploration and uncertainty remains, several
qualification tests have indicated the potential for IMAGE Frac to also map proppant distribution. The IMAGE Frac
proppant mapping tool was carried out on a five well pad. Figure 6 presents a cross-sectional view of an IMAGE
Frac fracture for Well C3H (red outline) overlaid on a modeled fracture geometry using a standard a finite-element
based, 3D planar fracture modeling software (with the number of fractures per stage selected to be two, which most
closely matched the IMAGE Frac results). The corresponding proppant distribution generated by the fracture
modeling software after the proppant has settled is also shown along with the IMAGE Frac proppant distribution
after stimulation (red shaded region). Within the first 500 feet of the well, the proppant distributions for IMAGE
Frac and the fracture design software are reasonably consistent. It is well recognized that fracture design software is
challenged at simulating proppant transport and often over estimates the depth of proppant penetration due to the
inability to capture the true tortuosity in an actual hydraulic fracture. However, the ability for IMAGE Frac to
estimate proppant effectively is still uncertain and remains under development. There is still a chance that IMAGE
Frac is unable to pick up the tail end of the proppant signature, if it is truly there, due to the small dimensions and
resulting shallow depth of influence. A proppant tracer log was run in Well C4H to identify any tracer in proximity
to C4H emanating from fractures in C3H. The proppant log is shown in an inset in Figure 6. Along the well length
there is a variation in the vertical separation between Well C3H and C4H, as noted in the inset. A significant amount
of proppant was identified in the 55 to 60 ft vertical separation range while virtually no proppant was identified at
vertical separations greater than 60 feet. This may indicate at this distance from the well, this is the approximate
transition zone of the proppant. This location is marked with a yellow star in the cross-sectional image of the C3H
fracture and proppant distribution. IMAGE Frac showed proppant in this area and also indicated the potential of the
top of the proppant zone to be near this marker. However, it is not clear if the proppant log readings are showing the
top of the proppant distribution at the C4H location or if another phenomenon is at play in the heel half of the well.

URTeC 2016
445
2432330

But, the registration of proppant in that region does correspond with the IMAGE Frac proppant distribution at that
location.
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 6: Cross-sectional view of IMAGE Frac geometry and proppant geometry (red outline and red shaded region, respectively) for Well C3H
vs. the fracture and proppant distribution generated by a frac modeling software (with number of fractures selected to most closely match IMAGE
Frac geometry). Proppant tracer log results in C4H shown as inset. Yellow star indicates location of proppant identified in C4H log.

5. Shut-In Pressure Interference Tests


A three-well shut-in interference test was conducted on a pad to compare apparent fracture and proppant lengths
with those observed from IMAGE Frac. IMAGE Frac was not run on this pad but was run on a proximal pad with
very similar pumping schedule, so it is assumed the study is applicable. In the first shut-in test, all three wells were
shut-in shortly after the onset of production (early-life) and allowed to build-up. The pressure in each well was
monitored as Well W1, W3, and then W2 were brought back on production, respectively (Figure 7A). It is evident
from Figure 7A that there is strong, direct fluid communication between W2 and W1 as well as between W2 and
W3. However, there is no clear indication of direct fluid communication between Well W1 and W3. The derivative
curves indicate potential for very minor interference between Well W1 & W3, which may be attributable to
poroelastic effects rather than direct fluid communication. Therefore, this early-time interference test indicates the
hydraulic fracture half-lengths at the corresponding well depths are likely 400 > FHL B > 900 ft.

A second interference test was carried out after substantial reservoir depletion, where stress on the unpropped zones
likely increased to substantially reduce their productivity. Thus, this late-life interference test is believed to give an
indication of the effective propped zone around a well. The results from this test are shown in Figure 7B. No
indications of interference were seen among any of the wells in the late-life test, indicating the entire fracture length
between wells was likely not effectively propped (i.e., the propped half-length was likely less than 440 feet, FHLBP
< 440 feet). IMAGE Frac indicated a propped half-length in an adjacent pad in the range of 350 feet < FHLBP < 500
feet, which is consistent with the information from the interference test (noting, however, that the interference test
does not provide a lower bound to compare the IMAGE Frac proppant half length). Therefore, the IMAGE Frac
fracture and proppant distributions are both consistent and in-line with the range observed from the early-time and
late-life interference tests.

A) B)
Figure 7: A) Early-Life shut-in interference test providing insight into fracture geometry for Wells W1, W2, and W3. B) Late-life shut-in
interference test providing insight into proppant distribution in Well W1, W2, and W3.

URTeC 2016
446
2432330

6. Field & Simulator Validation of Signal Classifications


Analytical, numerical, and field validation studies were used to evaluate whether poroelastically induced pressure
signals could be consistently differentiated from direct fluid communication signals. It was determined that
poroelastically induced pressure signals can be identified based on the arrival time, magnitude, derivatives, and
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

duration of the signal. Both analytical studies and numerical studies were conducted to determine if a direct fluid
communication signal (a direct fracture hit, a high-permeability pathway, or a low-permeability natural fracture
network) could emulate a poroelastic signal under conditions which mimicked fracturing scenarios. In both studies
poroelastic signals could not be emulated by direct fluid connectivity signals. An example comparing a
poromechanics signal to that of three numerical models with low-permeability natural fracture networks of varying
density and permeability is shown in Figure 8. All three numerical signals as well as the poroelastic signal yield
approximately the same magnitude of pressure response but the signal arrival time, signal derivatives, and duration
of the signal could not be matched.

Figure 8: Comparison of poroelastic signal to three simulated direct fluid communication signals. The arrival time, magnitude, derivatives, and
duration of poroelastic signals under fracturing scenarios could not be emulated by direct fluid communication signals in numerical models under
similar boundary conditions.

In addition to the analytical and numerical studies a field study was conducted to further verify that the identified
poroelastic signals were not caused by direct fluid connectivity. The field study focused on characterizing the natural
fracture network on a pad using an integrated investigation program, consisting of logging, coring, pressure testing,
and fluid tracing. A more comprehensive review is given by Dawson, et. al, 2015. The logging program used triple
combo and caliper logs in concert with more advanced logs, such as a high-resolution image log (Microscope
UHRI) and a dipole sonic log, to evaluate the prevalence of faults in the Middle Bakken. These tools are useful in
identifying small faults or significant fracturing along the lateral, but in this area did not pick up signals indicating
the presence of a natural fracture network; however, it is important to note that despite the high image quality, the
logs were not suitable for capturing very small scale fractures. Therefore, to better understand the natural fracture
distribution in the reservoir, a 180-ft inclined horizontal core was carefully taken. Two groups of fracture
characterization experts examined the core. The results of this study indicates this region had some minor natural
fracturing present, but the network was sparse and most of the natural fractures were small in scale with ~40%
terminating within the 2 9/16-in diameter core itself (Dawson, et. al, 2015). A vast majority of the natural fractures
were also found to have the same orientation, indicating even further the low probability of a connected natural
fracture network even after stimulation. This was further confirmed by pressure testing and fluid tracer studies.

The analytical, numerical, and field studies all demonstrate that the identified poroelastically induced signals are
distinct from direct fluid communication signals and readily differentiated. This result is very important as the
IMAGE Frac technology depends on the ability to identify poroelastically induced pressure signals.

7. Fundamental Simulation Sensitivity Studies


In depth numerical studies were carried out to evaluate the robustness of the IMAGE Frac technology. A more
detailed discussion is provided in a companion paper by Kampfer and Dawson (2016). However, several critical
studies are given below to demonstrate the robustness of the mapping technology. In the first study the influence of
geologic uncertainty on the ability to map fractures was evaluated. The impact on hydraulic fracture geometry of
variations to the base case matrix permeability, matrix porosity, fracture porosity, Young’s modulus, Poisson’s ratio,
Biot’s coefficient, and minimium horizontal stress were evaluated. Figure 9 shows the impact of these variations on
a plot of the expected normalized pressure signal in the observation well versus the two normalized geometric

URTeC 2016
447
2432330

fracture parameters. It is evident from Figure 9 that significant uncertainty in the petrophysical properties of the play
issue in acceptable levels of error, well within the target ranges of uncertainty for this mapping technology.
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

A) B)

Figure 9: A) The impact of variations in geologic properties on a normalized pressure in the observation well versus normalized fracture offset
plot. B) The impact of variations in geologic properties on a normalized pressure signal in the observation fracture versus normalized fracture
overlap plot.

Since the IMAGE Frac technology is based upon numerical simulation, it is important to investigate the sensitivity
of the solutions to mesh refinement. Figure 10 shows the negligible sensitivity of the IMAGE Frac mapping
technology to mesh refinement.

Figure 10: A) The impact of mesh refinement on a normalized pressure in the observation well versus normalized fracture overlap plot.

It was previously noted that the IMAGE Frac technology targets mapping of the geometry of the largest fracture in a
stage with high precision. In part, it is able to achieve this since the solution space is largely independent of
secondary fractures. Figure 11 shows two simulation studies that demonstrate the impact of secondary fractures on
the ability to map the largest fracture in a stage. It is clear from Figure 11 that placement of secondary fractures does
not have an appreciable impact on the ability to map the largest fracture in a stage with IMAGE Frac.

The simulations shown here and in the companion paper (Kampfer & Dawson, 2015) demonstrate the robustness of
the IMAGE Frac technology and provide insight into how the technology can achieve relatively high precision in
mapping the geometry of the largest fracture within a stage.

8. Proprietary Frac Length Validation Signals


A method was developed to determine when hydraulic fractures have grown passed adjacent wells. This method is
applicable regardless of whether the adjacent wells are cemented or not and independent of whether or not the fluid
in the fracture interacts with the fluid in the wellbore of the adjacent wells. This approach was used to further qualify
the IMAGE Frac technology in two separate studies.

In one study five wells were used with approximately 250’ well spacing. The local orientation of the maximum
horizontal stress relative to the well was found to be at approximately 65 o to the well, giving an effective well

URTeC 2016
448
2432330
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

A) B)

C)
Figure 11: A) The impact of relative position and presence of secondary fractures on the normalized pressure signal in the observation fracture
versus the fracture offset plot for 1.0 overlap of large fracture with observation fracture. B) The impact of relative position and presence of
secondary fractures on the normalized pressure signal in the observation fracture versus the fracture overlap plot for 0.15 offset of large fracture
to the observation fracture. C) Illustrations of the position and presence of induced fractures (black) and the observation fracture (red).

spacing along the frac direction of ~ 275’. The first stage of Well C1H was stimulated and examined from stage 1 of
C2H along with stages 1 through 3 of C3H, C4H, and C5H. C1H, C3H, and C5H were at approximately the same
depth while C2H and C4H were 50’ to 100’ shallower. Signals were received in C2H Stage 1, C3H Stages 1, 2, and
3 indicating the C1H fracture had passed those wells at those depths. No signals were received in stages 1 through 3
of C4H and C5H, indicating the C1H fracture had not passed those wells. Therefore, these signals indicate that the
fracture half-length of C1H stage 1 at the depths associated with C3H and C4H, respectively, was at least 550’ but
less than 830’. These results were compared with the IMAGE Frac geometry, which measured fracture half lengths
at the respective depths of ~600’ (for C3H depth) and ~750’(for C4H depth), both of which were consistent with the
proprietary signals suggested > 550’ (for C3H depth) and < 830’ (for C4H depth). This test was repeated with C2H
stage 1, observing from stages 1 through 3 of C3H, C4H, and C5H. Signals were observed for C2H stage 1 fracture
passing C3H and C4H (i.e., > 550’ @ C4H depth) but not reaching C5H (< 830’ @ C5H depth), again consistent
with the IMAGE Frac geometry at those depths of (~600’ and ~500’). A similar study was carried out on a pad with
larger well spacing (e.g., 600 ft) and the results are consistent with IMAGE Frac geometries; however, the close
proximity of the wells in the study described herein provides a narrower band of uncertainty for comparing the
IMAGE Frac results.

9. Fracture Simulator vs. IMAGE Frac Geometry


A finite-element based, 3D planar fracture design tool was used to simulate planar fractures in two different plays. A
detailed reservoir characterization allows for increased confidence in fracture modeling tools, but it is recognized
that there is significant uncertainty with leak-off rates, fracture widths and number of clusters effectively stimulated
per stage. Therefore, this comparison was primarily to examine the geometric shape predicted by the fracture
simulator, which has knowledge of the stress distribution in the reservoir, to that identified by the IMAGE Frac
mapping and to ensure the IMAGE Frac geometry was within the uncertainty band of the fracture simulator
predictions. A comparison of the geometries of the fracture simulator and IMAGE Frac are shown in Figure 12 for
two different formations and completions approaches. It is noteworthy that the sizes of the fractures in the first
comparison are quite similar, but attention should be given to the large uncertainty range on the planar fracture
model predictions, and it should be recognized that slightly different assumptions on the number of effective clusters
per stage or the leak-off rate would result in a significant change, as the uncertainty band shows.

Two key observations can be taken from this study. The first is that the general fracture geometry mapped by
IMAGE Frac (without knowledge of the stress state or formation) is similar to that predicted by the simulator (with

URTeC 2016
449
2432330
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

A) B)
Figure 12: A) A comparison of fracture geometries predicted by the fracture simulator to that of IMAGE Frac in formation 1. B) A comparison of
fracture geometries predicted by the fracture simulator to that of IMAGE Frac in formation 2.

knowledge of the stress state). The second is that the IMAGE Frac fracture size is within the range of uncertainty in
the fracture design software. Both observations support the potential viability of the IMAGE Frac approach.

10. Diverter Tests


Intra-stage diverter is designed to impede fracture growth during a stimulation job. A single-well, five-stage test was
conducted with four of five stages using diverter. The stages were mapped with IMAGE Frac. In one of the four
stages where an intra-stage diverter drop was used, IMAGE Frac showed the primary fracture in the stage
completely ceased growing after the diverter was used. In two of the four stages where diverter was used, the rate of
fracture growth was observed to significantly slow while in the last stage where diverter was used, no appreciable
impact was observed. As expected and observed already in hundreds of other stages where no diverter was used,
there was no evidence of fracture growth rate impediment during the fracturing process in the stage where no
diverter was used. Although the sample data set is too small to make any statistically significant conclusions, it is
recognized that IMAGE Frac identified fracture growth impediment in 75% of stages where diverter was used and
0% of stages where diverter was not applied. This study is another indication of the potential of IMAGE Frac to map
hydraulic fracture geometry.

11. Production Rate Transient Analysis


A comprehensive rate transient analysis on production from nearly 50 wells was used to evaluate the approximate
average fracture half-length for both slickwater and hybrid completions in a particular formation. The analysis was
carried out towards the beginning of production including flow black data, where uncertainty in bottom hole
pressures is reduced. Also, stress on the unpropped fractures is low enough (in this relatively stiff formation), that it
is assumed the entire fracture (propped and unpropped) is contributing effectively to production. For hybrid jobs the
estimated fracture half-length based on RTA was 510 feet. The RTA based fracture half-length for slickwater jobs
was 870 feet. The average fracture half-lengths from IMAGE Frac mapping for hybrid and slickwater jobs in the
area was 600 feet and 880 feet, respectively. It’s important to note that the average pumped volume for the IMAGE
Frac hybrid completions was approximately 1.14 times larger than that for the wells examined by RTA. Therefore,
scaling the frac length linearly with volume, the normalized RTA estimated average frac length for hybrid
completions is 580 feet. Thus, the resulting differences between IMAGE Frac and RTA base average fracture half-
lengths for slickwater and hybrid completions were 1% and 3%, respectively. It is recognized that there is
uncertainty in the IMAGE Frac geometry mapping and significant uncertainty in RTA based methods for
determining fracture half-lengths; therefore, the small errors between these approaches are considered somewhat of
a coincidence. Nevertheless, the RTA-based fracture length evaluation provides additional evidence that the IMAGE
Frac technique has potential.

12. Uniqueness Validation Study


The analysis of poroelastic pressure signals in the IMAGE Frac workflow to identify fracture geometries is
cumbersome without an efficient software solution. As part of the software development a system self-test is
implemented to evaluate if the IMAGE Frac technology yields a consistent, reproducible and unique solution when
analyzed in an automated fashion. The software enables identification of the poroelastic signal, runs a normalization

URTeC 2016
450
2432330

process to extract critical information, and automatically identifies the fracture half-length, and height using a pre-
developed solution space (additional details of the methodology can be found in Kampfer & Dawson, 2016).
Simulation data was developed to test the IMAGE Frac software analysis for determining possible errors in the
analysis framework. By nature, field data has a degree of uncertainty without any way to verify the exact
geometries, so simulation data was selected to be the most accurate way to validate the ability of the technology to
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

identify a unique solution. Using simulation data enables the conclusion that in a “perfect” case, the resulting
geometry values are unique once 2 or more pressure signals are available.

Figure 13: Simulated set-up for uniqueness simulation test. Plan view of six stages of a fractured well (FW) and a monitor stage emanating from
an observation well (Well).

Consider the setup in Figure 13, a fractured well in orange (FW) and an observation well (Well). The setup has one
observation stage (red), and six fractured stages (green). Each of the stages includes one single fracture, all fractures
have exactly the same shape and their position is exactly known, since this data set was obtained numerically with a
FEM simulator. The first step in the analysis is to use the software to select the poroelastic pressure changes from
the measurements in the observation well. Figure 14, presents the artificial pressure signal observed in the
observation well generated by a numerical simulator. The sequence of fractured stages is from left to right in Figure
13.The pressure signals are small at larger distances and increase as they approach the observation fracture before
decreasing again as they move farther away from the observation fracture. The oblique nature of the fractures with
respect to the well trajectories results in an asymmetric distribution in the pressure profile as one would expect.

Figure 14: Artificial poroelastic pressure signal monitored from the observation stage during the fracturing of six stages in an adjacent well. The
colored regions represent the pump time

The second step is converting the stage coordinates from a global coordinate system into a local coordinate system
with its origin at the intersection of the observation stage and the observation well and the y-axis oriented in the
direction of the fracture length. Having accomplished that, the transferred coordinates can be plotted during the
matching process of the fracture shape, half-length, and height with the solution space. The IMAGE Frac software
identifies the optimal fracture shape, half-length (FHL), half-length to height ratio (FFR), and orientation angle that

URTeC 2016
451
2432330

minimizes the overall error from an expected solution. This solution are pre-calculated poroelastic pressure signals,
which need to match the space of measured pressure changes. For the solution provided by the IMAGE Frac
software, a cross plot showing the measured pressure (generated by a numerical simulator), the expected pressure
response (generated by the IMAGE Frac mapping software), and the error between the measured and expected
pressures for all stages are presented in Figure 15. This plot shows the robustness of the IMAGE Frac mapping
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

approach and the ability to identify a ‘unique’ solution (Note: the error between the expected and measured
pressures in this example is zero for all stages).

Figure 15: Cross plot showing the measured pressure change on the vertical and the stage names on the horizontal axis. The green bullets are the
measured pressures and the thin red bars are the expected pressure, once the match is “perfect”. The error is shown as number next to each of the
green circles, which is 0 for all the stages.

A sensitivity study about the identified solution is presented in Figure 16. The free parameters are adjusted and the
error is assessed. Note, that every adjustment to a geometric parameter changes the relative spatial position between
the observation and stimulated fractures. Hence, the pattern of the expected pressure signals (in the cross plot) in
relation to one another changes as well. It was determined that it is not possible to get a different combination of
parameters which provides a high confidence match as shown in Figure 15, let alone one with zero error in all
stages. It can be shown, that if everything is known with enough accuracy, an automated pressure matching solution
using IMAGE Frac provides an exact geometry inversion with two or more pressure measurements. This by itself
may not be a sufficient proof of uniqueness of the methodology but it is a necessary condition which is fulfilled.

Figure 16: Sensitivity of the cross plot towards changes in orientation angle of the fractures, the FHL and FFR

URTeC 2016
452
2432330

The results of this study only hold for a “perfect” data set though. In reality, there is an uncertainty connected to all
parameters needed for the IMAGE Frac analysis, and it is likely impossible to achieve a zero-error match in a field
study even with more than two stages. Therefore, it is recommended that more than two measurements are used.
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Even with multiple measurements, a series of solutions with similar small error magnitudes may be possible.
However, based on practical field studies, it seems the solutions are relatively well constrained with a sufficiently
high degree of accuracy. A visualization tool is under development, which will quantify the error of all potential
solutions to the free parameters (e.g., fracture angle, FHL, FHT, and shape). This will enable even clearer
identification of the uncertainty in the IMAGE Frac geometry solution in applied field cases.

Field Deployment Case Studies & Value Proposition

The IMAGE Frac technology has been applied in over 30 wells multiple plays including the Bakken and Eagle Ford.
The following two examples are brief case studies showing the potential value proposition this technology can
provide.

Bakken Deployment
In one deployment in the Bakken, 4 wells were completed on a pad with one legacy well, which had been producing
for approximately five years. IMAGE Frac was used to map the fractures of the new wells. Figure 17 shows the
average of the maximum hydraulic fracture length of the stages examined for each well. It is evident from Figure 17
that the fracture growth in well BH7, which was in the same formation as the legacy well BH1, was strongly
influenced by depletion.

Figure 17: Average of the maximum hydraulic fracture length of the stages examined for four newly stimulated wells in proximity to a legacy
well BH1.

This deployment along with several others in the Bakken have yielded a number of important observations. Several
of these observations and potential methods of mitigation are listed below (Note, these observations are only
applicable for the wells used in these studies and should not be taken as general recommendations):

Key Findings from IMAGE Frac Deployments in the Bakken


 Proppant is not effectively filling the entirety of the fractures (~30-60% near depleted wells and 50-80% in
native zones in the wells examined)
o Recommendations were made to change the proppant size distribution to get more effective
coverage
 Low pumped volume can result in fractures not effectively covering the pad (i.e., not overlapping)
o Minimum pumped volumes per stage have been identified to ensure fracture overlap
 A vast majority of fractures in certain areas can be designed to significantly overlap w/o intersecting
o Well to well flooding with produced water in wells with significant fracture overlap but minimal
fluid communication may prove economically attractive
 Fracture orientation clearly trends in the ~50o E of N direction
o Orientation of wells NW-SE would provide for favorable development (barring lease issues)

URTeC 2016
453
2432330

 Fractures are strongly influenced by depletion in adjacent wells


o For older legacy wells, re-fracturing them prior to stimulating new wells on a pad may be
favorable while for younger legacy wells, use of diverter or adjustments in pumped volumes in
adjacent wells is advised
 Fracture growth rates are rapid, particularly near depleted wells and diverter is often applied too late
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

o Timing of intra-stage diverter drops should be accelerated, particularly near depleted wells

A new completion approach was tested for the study in Figure 18 where a different concentration of smaller
proppant was used in accordance with the IMAGE Frac recommendation, along with a larger total pumped volume.
Figure 18 shows the inverse productivity plot of the newly completed well (New Design) versus that of two
comparative wells with more historical designs. The inverse productivity plot is a measure of how much pressure
draw down is required to produce a given volume of fluid. Figure 18 clearly shows the New Design significantly
outperforms the others in during the early-life of the well, requiring a much smaller drop in pressure to produce an
equivalent volume of fluid. It is important to note there is always a degree of uncertainty on performance from well
to well and not all of this uplift may be attributed to the completions approach alone. It is also important to note that
the New Design utilized more fluid and costs more today than the historical designs cost. Moreover, only a portion
of the production uplift is likely due to the changes in proppant design; however, it is believed that this improved
proppant design based on the IMAGE Frac recommendation has yielded significant benefit.

Figure 18: Inverse productivity index plotted versus linear superposition time comparing two Historical Designs to a New Design. The New
Design substantially outperforms Historical Designs, showing more oil production for a given pressure drawdown in the New Design, relative to
the Historical Designs.

Eagle Ford Deployment


IMAGE Frac was used in a five well pad in the Eagle Ford. Figure 19A shows a cross-sectional view of the average
geometry of the largest fracture in a stage among the stages examined in the five well deployment. The fracture
geometry is the outlined area while the shaded area represents the proppant distribution. It is evident from Figure
19A that there is significant fracture overlap and some amount of proppant overlap, particularly in the center of the 5
well pad.

This amount of overlap points to possible reduction of wells while maintaining the total recovery factor. A
hypothetical redesign of the pad is shown in Figure 19B where the center well is removed and the well landing
zones along with the proppant schedules are slightly adjusted. In this theoretical example, it is hypothesized that the
improved proppant design and well landing zones would increase proppant coverage and overall EUR, while the
removal of the center well would not adversely impact long-term EUR. The reduction of a well results in significant
cost savings during the pad development (i.e., 20% savings) while having only a minor negative impact on early
production rates from the pad (it is expected that rates in the remaining four wells would all increased due to more
effective fracture growth into the zone where the center well would have induced stress shadowing, more effective
landing zones, and more effective proppant coverage. Therefore, the total rate reduction for the pad is expected to be
much less than the ~20% reduction one might expect from removing 1/5 th of the wells).

URTeC 2016
454
2432330
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 19A: A cross-sectional view looking down a lateral of the average geometry of the largest fractures in a stage among the stages examined
in a five-well pad in the Eagle Ford. Fracture geometry is outlined and propped geometry is shaded.

Figure 19B: A hypothetical redesign of the five-well pad, removing the center well, adjusting the proppant schedule, and improving the landing
zones and spacing. It is theorized that this pad would only yield a slight reduction in early stage rates while yielding a significant cost reduction
(~20%) and a significant increase in cumulative ultimate recovery from the pad.

Conclusions
A new approach (IMAGE Frac) for evaluating hydraulic fracture geometry and proppant distribution has been
discussed. This method utilizes poromechanically induced pressure signals acquired from surface pressure gauges
during the stimulation of a multi-well pad. The IMAGE Frac mapping technology can be readily executed in the
field, as it requires only minimal deviation from standard completions practices. With this new approach,
determining the optimal spacing of wells, target zones, and completions designs for a development can be done at a
fraction of the cost of traditional methods while yielding more reliable results. IMAGE Frac can be used in low
permeability formations with two or more multi-stage horizontal wells where the wells are in reasonable proximity
both in the vertical and lateral directions. Limited variations in the horizontal well trajectory and limited natural
fracturing and faulting are also desired. It is believed that IMAGE Frac is likely applicable to a majority of the
acreage in the major ‘shale’ plays in North America.

A detailed review of twelve qualification studies evaluating the IMAGE Frac technology is provided. IMAGE Frac
results are consistent with all twelve validation studies shown below, which indicates the strong potential of this
technology.

1) Consistency With Variable Field Inputs


2) Fracture Propagation Direction Mapping
3) Fracture Hit Analyses
4) Fluid & Proppant Tracer
5) Shut-In Pressure Interference Tests
6) Field & Simulator Validation of Signal Classification
7) Fundamental Simulation Sensitivity Studies
8) Proprietary Frac Length Validation Signals
9) Fracture Simulator vs. IMAGE Frac Geometry
10) Diverter Tests
11) Production Rate Transient Analysis
12) Uniqueness Validation Study
13) Electromagnetic Imaging (In Progress)
14) Microseismic (In Progress)

This technology has been successfully deployed in over 30 wells in multiple formations. Two case studies are
provided for the Bakken and Eagle Ford formations. The IMAGE Frac technology revealed a number of
opportunities for improvement, including opportunities around proppant design, pumped fluid volume design,
fracture design for enhanced oil recovery, intra-stage diverter use & timing, well orientation, and treatment design
near legacy wells. An example of adjusting the completions design based on IMAGE Frac results shows significant
potential for value creation with an improved understanding of proppant & fluid design. In this particular example,
the early-time productivity of the redesigned wells had a dramatic improvement over proximal wells with more
traditional designs. Likewise, the second case study also showed potential for significant value creation with
redesign of landing zones, well spacing, and proppant design based on IMAGE Frac results.

URTeC 2016
455
2432330

Overall, the novel approach provided in this paper has shown potential to provide new insight into hydraulic fracture
geometry and proppant distribution. This technique provides a low-cost, reliable method of determining the optimal
development strategy in tight plays and may position industry to extract more value from unconventional resources.
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Acknowledgments

The authors would like to thank Reveal Energy Services and Statoil for permission to publish this paper. The authors
would also like to thank Gene Beck, Darren Schmidt, and Joar Moberg who pioneered the first field deployments of
this technology and Lee Nguyen and Pandurang Kulkarni for their contributions to this work. Special thanks to
software development team in ITV DIS RD in Statoil, Trondheim, Norway for developing the software, which
enables an effective processing and matching process.

Nomenclature
σ : stress tensor
C : stiffness tensor δ : Kronecker delta
ε : strain tensor 𝜙 : porosity
α : Biot coefficient K : drained bulk modulus
pf : pore pressure Ku : undrained bulk modulus
M : Biot modulus Ks : bulk modulus of the solid constituent
k : matrix permeability fhl : fracture half length
μ : viscosity of the fluid fht : total fracture height
γ : fluid source density P : hydrostatic part of the stress
f : gravity term in the pressure equation Qvm : von Mises stress (deviatoric part of stress)
F : gravity term and external forces in the t : time
solid mechanics equation

References

1. Barree, R.D. 1983. A Practical numerical simulator for three-Dimensional Fracture Propagation in
Heterogeneous Media. Paper SPE 12273presented at the SPE Reservoir Simulation Symposium, San
Francisco, CA, USA, 15-18 November.

2. Biot, M.A. 1935. “Le problème de la consolidation des matières argileuses sous une charge,” Ann. Soc.
Sci. Bruxelles, B55, 110-113.

3. Biot, M.A.. 1941. “General theory of three-dimensional consolidation”, J. Appl. Phys., 12, 155-164.

4. Boone, T.J., Ingraffea, A.R. 1989. An investigation of poroelastic effects related to hydraulic fracture
propagation in rock and stress measurement techniques. In: Rock mechanics as a Guid for Efficient
Utilization of Natural Resources, Khair (ed.) Rotterdam (ISBN 9061918715

5. Boussinesq, J., 1885. Application des potentiels a l'etude de l'equilibre et due mouvement des solid
elastiques. Gauthiers-Villars, Paris, (as referenced by Holtz and Kovacs, 1981).

6. Cipolla, C.L. and Wallace, J. 2014. Stimulated reservoir volume: A misapplied concept? Paper SPE
168596, presented at the Hydraulic Fracturing Technology Conference, Woodlands, Texas, USA, 4-6 Feb.

7. Coussy, O. 1991. Mécanique des Milieux Poreux, p.437. Editions Technip, Paris.

8. Daneshy, A., Au-Yeung, J., Thompson, T. And Tymko, D., 2012. Fracture Shadowing: A Direct Method
for Determining of the Reach and Propagation Pattern of Hydraulic Fractures in Horizontal Wells. SPE
151980, presented at the SPE Hydraulic Fracturing Technology Conference held in The Woodlands, Texas,
USA, 6-8 February 2012.

URTeC 2016
456
2432330

9. Dawson, M., Nguyen, D., Li, H. 2015. Designnig an Optimized Surfactant Flood in the Bakken. SPE
175937, presented at SPE CSUR Unconventional Resources Conference, Calgary, Alberta, 20-22 October.

10. Detournay, E., Cheng, A.H-D., Roegiers, J-C. and McLennan, J.D.. 1989. “Poroelasticity considerations in
in situ stress determination by hydraulic fracturing,” Int. J. Rock Mech. Mining Sci. Geomech, 26, 507-513.
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

11. Detournay, E. and Cheng, A.H.-D. 1993. Fundamentals of poroelasticity. Chapter 5 in Comprehensive
Rock Engineering: Principles, Practice and Projects, Vol. II, Analysis and Design Method, ed. C. Fairhurst,
Pergamon Press, pp. 113-171.

12. Detournay, E. and Cheng, A.H-D. 1991. “Plane strain analysis of a stationary hydraulic fracture in a
poroelastic medium”, Int. J. Solids Struct., 37, 1645-1662.

13. Guéguen, Y., Boutéca M.2009. Mechanics of fluid saturated rocks. International Geophysics series,
Volume 89., Elsevier Academic Press.

14. Haimson and Fairhurst. 1969. “Hydraulic fracturing in porous-permeable materials”, JPT, July, 811-817.

15. Kampfer, G. and Dawson, M. 2016. “A Novel Approach to Mapping Hydraulic Fractures Using
Poromechanic Principles”, Proceedings American Rock Mechanics Association. 50th US Rock
Mechanics/Geomechanics Symposium, Houston, Texas, 26-29 June 2016. In Press.

16. King, G.E. 2010. Thirty years of shale gas fracturing: What have we learned? Paper SPE 133456 presented
at the SPE Annual Technical Conference and Exhibition, Florence, Italy, 19-22 September.

17. Koshelev, V., and A. Ghassemi (2003), Hydraulic Fracture Propagation near a Natural Discontinuity,
Proceedings, Twenty-Eight Workshop on Geothermal Reservoir Engineering.

18. Makhnenko, Labuz 2013. Saturation of Porous Rock & Measurement of the B coefficient, ARMA 13-468.

19. McClure, M. 2014. The Potential Effect of Network Complexity on Recovery of Injected Fluid Flowing
Hydraulic Fracturing. Paper SPE 168991 presented at the SPE Unconventional Resources Conference -
USA held in The Woodlands, Texas, USA, 1-3 April.

20. Newmark, N.M. 1935. Simplified computation of vertical pressures in elastic foundations. Circular No 24,
Engineering Experiment Station, Published by the University of Illinois.

21. Nolte, K.G. and Smith, M.B. 1981. “Interpretation of Fracturing Pressures.” JPT (September 1981) 1767.

22. Perkins, T.K. and Kern, L.R. 1961. Widths of Hydraulic Fractures. J Pet Tech. 13 (9): 937–949. SPE-89PA.

23. Renshaw, C.E. and Pollard, D.D. 1995. An experimentally Verified Criterion for Propagation across
Unbounded Frictional Interfaces in Brittle Linear Elastic Materials. Int. J. Rock Mech. Min. Sci. &
Geomech. Abstra., 32: 237-249.

24. Sneddon, I. N. (1946), The Distribution of Stress in the Neighborhood of a crack in an Elastic Solid,
Proceedings of the Roy. Soc. London. Series A, Mathematical and Physical Sciences, 187 (1009), 229-260.

25. Settari, A.S. and Cleary, M.P. 1986. “Development and Testing of a Pseudo-Three-Dimensional Model of
Hydraulic Fracture Geometry.” SPE Production Engineering, 1 (06): 449-466.

26. Skempton, A.W. 1954. “The pore pressure coefficients A and B”, Geotechnique, 4, 143-147.

27. Terzaghi, K.. 1923. “Die Berechnung der Durchlassigkeitsziffer des Tones aus dem Verlauf der
Hydrodynamischen Spannungserscheinungen”, Sitz. Akad. Wis., Wien Math. Nat..Kl., IIa, 132, 105-124.

URTeC 2016
457
2432330

28. Weng, X., Kresse, O., Cohen, C., Wu, R. and Gu, H. 2011. Modeling of Hydraulic Fracture Network
Propagation in a Naturally Fractured Formation. Paper SPE 140253 presented at the SPE Hydraulic
Fracturing Technology Conference and Exhibition held in The Woodlands, Texas, USA, 24-26 January.

29. Westergaard, H.M., 1938. A problem of elasticity suggested by a problem in soil mechanics: A soft
Downloaded 12/26/16 to 132.239.1.231. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

material reinforced by numerous strong horizontal sheets. In Contributions to the Mechanics of Solids,
Stephen Timoshenko 60th Anniversary Volume, MacMillan, New York, pp. 260 – 277.

30. Wu, R., Kresse, O., Weng, C., and Gu, H. 2012. Modeling of Interaction of Hydraulic Fractures in
Complex Fracture Networks. Paper SPE 152052 presented at the SPE Hydraulic Fracturing Technology
Conference and Exhibition held in The Woodlands, Texas, USA, 6-8 February.

31. Zhang, X., and R. G. Jeffrey (2006), The role of friction and secondary flaws on deflection and re-initiation
of hydraulic fractures at orthogonal pre-existing fractures, Geophysical J. International, 166, 1454-1465.

32. Zeng, Z. 2008. Geomechnaical Study of Bakken Formation in eastern Williston Basin, North Dakota.
Progress Report NDIC Contract No. G015-031

URTeC 2016
458

You might also like