Declining Frac Effectiveness - Evidence That Propped Fractures Lose Conductivity, Surface Area, and Hydraulic Continuity

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

SPE 168744 / URTeC 1579008

Declining Frac Effectiveness - Evidence that Propped Fractures


Lose Conductivity, Surface Area, and Hydraulic Continuity
M.C. Vincent*, SPE, Consultant; M.R. Besler*, SPE, Consultant
Copyright 2013, Unconventional Resources Technology Conference (URTeC)

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

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

Introduction

Hydraulic fractures are often believed to provide durable, highly conductive channels that remain effective
during decades of production. Traditional well test analyses often yield non-unique solutions, with poor ability
to distinguish whether steep decline curves should be attributed to insufficient reservoir quality or declining
fracture effectiveness. However, there is a growing body of evidence demonstrating that fractures lose
effectiveness over time, apparently due to a number of mechanisms.

This paper will:


• Examine evidence that fractures fail to provide durable, effective conductivity throughout the created
extent
• Tabulate suspected mechanisms causing loss of fracture effectiveness
• Summarize best practices currently used to address these concerns
• Comment on other needed data collection, research and development and efforts to introduce best
practices

This study has identified a number of field examples which convincingly indicate that fractures have failed to
sustain a conductive channel over time, including:
• Wells interconnected during a fracturing treatment, which fail to sustain interference during production
• Production results highly sensitive to lateral depth, indicating fractures fail to drain reserves from thick
pay sections
• Infill drilling through a microseismic swarm without evidence of pressure depletion
• Adjacent wells improved or unaffected by offset well “bashing”
• Field studies examining the performance of alternate frac designs, convincingly demonstrating the
initial treatment was not optimized
• Lack of agreement between “perm while pumping” and “perm after closure”
• Restimulation results which compare the durability of initial completion strategies

This paper explores a previously underappreciated concept. The results indicate that greater focus should be
placed on fracture effectiveness, instead of initial fracture length or extent. The cited field results are
encouraging, and suggest that best practices are emerging which can improve fracture effectiveness and
durability.
URTeC 1579008 2

Indications that fractures are losing conductivity and/or surface area

While many production trends fail to yield a unique interpretation as to the damage mechanism, there are a number
of observations that, when considered in conjunction, compellingly demonstrate that fractures are not recovering the
hydrocarbon resource as effectively as commonly anticipated [Vincent, 2012a]:

• Infill drilling – in many resource plays, new wells drilled through a microseismic swarm are encountering
near-virgin reservoir pressure and appear to recover >80% of the EUR of parent wells drilled in virgin rock.
This indicates original “parent” wells have not depleted the rock volume apparently covered with
microseismic events.
• Restimulation – horizontal wells in most resource plays have been successfully restimulated, indicating the
initial fracture treatments were insufficient to drain the recoverable reserves accessible from the wellbore.
• In many resource plays, the well productivity is highly sensitive to the depth at which the lateral is landed.
If fractures were highly conductive, durable, vertical fractures that penetrated the entire pay interval, well
productivity would not be significantly affected by the specific lateral depth. Evidence in the Bakken,
Three Forks, Barnett, Niobrara, Woodford, Montney and Eagle Ford convincingly demonstrate that lateral
depth significantly influences well productivity.
• Evidence that in some reservoirs, managing the drawdown seems to extend EUR, suggesting that typical
operational practices may be accelerating fracture degradation. While this degradation could possibly be
attributed to stress-dependent permeability of the reservoir, this observation likely suggests that traditional
frac treatments are insufficiently durable to withstand traditional drawdown or standard operating practices
(shut-ins, cyclic stress loading, etc).
• Laboratory testing. Even in “ideal” conditions in which proppant is arranged in wide packs with optimal
packing arrangements, tremendous degradation of conductivity over time has been noted in every published
extended duration test. While some proppants are more durable than others, all lab tests indicate that
proppants continue to crush and compact over time. Unless something magical happens when we pump
proppant down a mile of pipe, it should be logically anticipated that actual fractures will similarly lose
conductivity over time.
• Recognition of fracture complexity. In numerous mine-back and core-through studies it has been
documented that fractures are not simple, wide, planar features. The industry should recognize that it is
challenging to place a uniformly wide fracture that will retain hydraulic continuity through highly
heterogeneous or laminated intervals. Stress on proppant is greatly magnified in irregularly propped
fractures, or in horizontal fracture “stairsteps” in which the proppant must support the entire overburden.
• Cases in which hydraulic fractures connect two wellbores. In a number of reservoirs, proppant has been
pumped into adjacent wellbores completed at exactly the same depth. Interference while pumping is often
documented with pressure data, production of frac fluid into adjacent wells, liquid chemical tracers and
even solid radioactive tracer particles observed in offset wells. However, irrefutable proof is occasionally
documented when crosslinked, sand-laden slurry is pumped from one wellbore into an adjacent well
completed in exactly the same horizon/depth, filling the entire wellbore and surface equipment with sand.
In this case, it is irrefutable that a fracture was created between the two wells and proppant was transported
through the entire frac length. However, subsequent diagnostic tests (pulse tests, interference tests, PBUs,
calculation of EUR) frequently indicate ineffective retained hydraulic continuity, and often zero indication
of a sustained hydraulic connection. The fractures are not sustaining sufficient conductivity or are
collapsing at some point between the wellbores. Two of the more encouraging examples that show some
level of sustained hydraulic continuity will be reviewed in this paper. Even in these encouraging examples,
large pressure losses are apparent in the fracture network, suggesting opportunities to improve the fracture
design or implementation.
• In addition to fractures connecting horizontal wells completed at the same depth, there are some occasions
in which stacked laterals demonstrate the ability to initially create a fracture connecting wellbores
completed within 100 ft depth, such as Bakken to Three Forks. However, after being brought onto
production, these stacked laterals often show little to no sustained continuity, and commonly produce
distinctly differing watercuts, indicating that the wells are draining unique reserves and the hydraulic
continuity and conductivity within the fractures is compromised during drawdown and production. Clearly
there is a large pressure loss within whatever fracture network connects the two laterals.
• In some situations, adjacent wells can be permanently stimulated when an offset well is fracture stimulated.
URTeC 1579008 3

While “bashing” has also been documented to result in damage in some adjacent wells, the observations of
positive influence of specific well pairs in the Bakken, Eagle Ford, Niobrara, and Barnett demonstrate that
the created fractures are NOT behaving as infinitely conductive conduits between the wells.
• Production decline curves are often steeper in many resource plays than anticipated with best estimates of
reservoir and fracture parameters. The steeper than anticipated declines should be at least partially
attributed to insufficient conductivity, collapse of fracture conductivity over time, or loss of surface area
effectively contacted by the treatment.

Possible Mechanisms Responsible for Steep Production Decline

Traditional techniques to analyze production decline from individual wells are rather ambiguous, with poor ability to
uniquely determine whether steep decline should be attributed to insufficient reservoir quality, or deficiencies in the
stimulation (Vincent 2012b). In many cases, a good reservoir with a poor completion can be misinterpreted to be a
poor reservoir with a presumed excellent and durable fracture treatment. In highly permeable formations completed
with vertical wells, often pressure transient testing or other diagnostics can be used to narrow the range of plausible
matches. However, in multi-fractured horizontal wells in ultra tight reservoirs, it is impossible to claim any unique
solution via standard decline curve analysis of isolated wells. In that situation steep decline curves have been
blamed on a number of mechanisms, including:

Table 1: Partial list of mechanisms that are believed to account for steep well declines

Reservoir Low permeability


Poor lateral extent or continuity of reservoir (lenticular)
Perm anisotropy yielding high aspect ratio drainage areas
Highly laminated with poor vertical perm/recovery
Pressure dependent permeability or collapse of natural fractures with drawdown
Other mechanisms
Frac Design – Fracture network too sparse (insufficient frac density or complexity); or conversely,
Stage Spacing Transverse fractures too close together (competition between adjacent fracs)
Frac Design – Insufficient initial conductivity
Proppant selection Degradation of proppant over time
Thermal degradation of sand-based proppants
Uninformed use of extremely low quality sand and low quality ceramics during past decade.
[Enormous variation in proppant quality – poor awareness by operators]
Complex frac geometry requiring stronger or more conductive proppant in the turns and “pinch points”
Insufficient proppant concentrations, resulting in discontinuous proppant packs after frac closure. This
problem is compounded when operators specify intermediate or high density ceramics but pump the
same mass concentration, resulting in reduced fracture width and 20% to 30% smaller frac
geometry.
Frac Design – Overflushing of proppant from the near-wellbore area in transverse fracs
Implementation or operation Flowback of proppant from near-wellbore area in transverse fracs
Frac Design – Failure to place sufficient proppant concentrations throughout the created network (both lateral and
Conductivity throughout pack vertical placement)
Insufficient conductivity to accommodate high velocity hydrocarbon flow due to convergence near-
wellbore, especially in liquid-rich formations
Embedment of proppant, especially in softer layers
Frac Design – Fluid sensitivity – evidence that some frac fluids “soften” the formation allowing more significant
Fluid Systems embedment and/or spalling
Durable gel filtercakes that may completely occlude narrow propped fractures
Other Failure to recover water from liquid-submerged portions of the fracture below the wellbore elevation
Industry rush to secure acreage as “held by production” without adequate attention to completion
effectiveness or optimization. Frenetic development pace has reduced many completion engineers’
primary focus to be on scheduling and assuring materials are available, with less time devoted to
optimization of well productivity
Potential for chemical diagenesis of proppant (controversial and conflicting laboratory studies). To
date, proppant samples recovered from wells do not appear to indicate formation of zeolites;
however, it is clear that zeolite precipitation can be encouraged in laboratory tests under specific
conditions
Precipitation of salt, asphaltenes, barium sulfate and calcium carbonate scales
Other unrecognized mechanisms
URTeC 1579008 4

Evolving Best Practices – Resource Plays

The goal of this paper is to investigate and raise awareness of a number of the controllable variables in fracture
design that can reduce the decline in fracture effectiveness, resulting in superior sustained production. It is evident
that there is no single recipe or ideal frac design that can be optimal for all of the resource plays. “Changes in
kerogen content, total organic content, rock brittleness, thermal maturity, in situ pressure, formation thickness,
permeability, porosity, degree and variability of natural fracturing all influenced well spacing, lateral lengths and
frac design.” – Fisher, 2012. However, there are a number of common observations that appear to apply broadly to
resource plays developed with transversely fractured horizontal wells:
• Increased number of stages
o Dividing a treatment into more stages is clearly increasing production in most plays developed with
multiple transverse fracs. This is in contrast to what is predicted with production models, which
incorrectly forecast that placing more, but smaller stages (reduced frac length) would hurt long-term
productivity. This evidence strongly suggests that at least one of the following factors is true:
 Pressure losses are high within transverse fracs, and dividing the hydrocarbons between multiple
entry points is advantageous
 Fractures are losing hydraulic continuity at some location within the fracture, so putting the
investment nearer the wellbore increases the proportion that remains productive over time
 The connection between the wellbore and fracture is tenuous and increasing the number of
connections (fractures) increases the durability
 The overflushing of proppant, or flowback of proppant is jeopardizing well productivity, and
therefore increasing the number of stages somewhat compensates for our completion practices
• Some reservoirs appear to be tolerant of overflushing and can be economically developed with industry-
standard practices that require flushing the wellbore clean between stages, typically resulting in
overflushing proppant some distance away from the wellbore in some of the created fractures. Other
formation types appear to be damaged with overflushing. Several operators have disclosed their efforts to
reduce and/or eliminate overflushing, and some operators are choosing to screenout specific stages to
evaluate the production benefit achieved with an improved connection between the wellbore and fractures.
• In wells susceptible to proppant flowback, the use of curable resins near the wellbore have been beneficial
in some plays, while the use of uncoated ceramic proppants has eliminated proppant flowback in others
prone to flowback of less durable frac sand. In general, field results demonstrate that sufficiently strong,
larger diameter proppants are more resistant to proppant flowback than smaller diameter and weaker sand.
• There is growing recognition that placing proppants in diffuse concentrations in crosslinked fluids is
yielding discontinuous proppant packs after closure. There are two potential solutions that are being
pursued. Some operators are switching to low viscosity banking fluids (slickwater) that build a continuous
proppant bank via the mechanism of saltation. The retained fracture width and conductivity is insensitive
to proppant concentration in a slickwater, duning/banking placement strategy. Stages placed in slickwater
also appear to be less damaged by overflushing. Other operators are trying to improve continuity by
specifying high proppant concentrations (often exceeding 6 to 9 ppa) in crosslinked fluids in efforts to
place a sufficient number of proppant grains such that hydraulic continuity can be improved. Additional
efforts to reduce or eliminate overflush in these gelled treatments have proven effective in some plays.
Given an “industry standard” of ~ 4 ppa and overflushing even when using crosslinked treatments in many
resource plays, it appears that either approach (slickwater at any concentration, or increasing to ~8 ppa in
crosslinked fluids) can improve continuity of the created fracture.
• Use of diagnostics. A growing number of operators are reporting the use of detectable proppants
(radioactive or non-radioactive particles which can be identified with logging tools) that demonstrate
location of proppant near the wellbore and also indicate which fracturing stages contact offset wells.
Additionally, the use of oil-soluble and water-soluble chemical tracers can indicate which injected fluids
reach adjacent wells, and document the relative productivity of stages using tracer recovery as proxies for
load recovery and hydrocarbon productivity. Efforts to integrate the interpretation of these diagnostics with
other data such as seismic attributes, microseismic frac mapping, mud logging, wellbore depth, production
logs, DTS/DAS data from fiber optic monitoring and other data are improving the understanding of which
parameters most directly correlate with sustained fracture productivity.
• Several industry-leading operators are demonstrating recognition that proppants are not equivalent. There
are >60 mines supplying frac sand to the North American frac industry, >90 plants supplying ceramic and
>18 plants supplying resin coated proppant. Vast variation in quality and durability of these proppants has
URTeC 1579008 5

been documented, with variation of 500% between various sands all marketed as “top quality white sand”,
and variation of more than 400% between “premium ceramics” misleadingly described as interchangeable
commodities. Given the enormous variation in quality provided by different sources, it is foolhardy for
operators to accept generic proppant identifiers without careful documentation of performance. Several
industry-leading operating companies are beginning to specify exactly which material and which quality
they will accept, and keeping careful records so that wells stimulated with lower quality proppants can be
identified for restimulation.
• Review of isolation procedures. The collapse of natural gas prices in the USA [attributed to the
outstanding success of developing resource plays with hydraulic fracturing of horizontal wells] has brought
some welcome relief in the number of stimulation treatments performed, as the industry has refocused
investments into liquid-rich resource plays. This reduced activity level has given some operators their first
opportunity to consider the use of more time-consuming isolation procedures that were previously
unavailable or not feasible given the intense development pace.
• Managed drawdown. In some reservoirs, compelling data are available suggesting that aggressive
drawdown results in premature damage to the completion and/or reservoir resulting in loss of reserves. In
other reservoirs, it appears that aggressive drawdown can only be pursued with specific completion and/or
proppant types. In other reservoirs, drawdown procedures have not yet been proven to affect EUR.
However, industry efforts to determine the best operational practices are ongoing in all resource plays.
• Restimulation. Refracs appear to be required and profitable in most resource plays, again indicating the
opportunity to improve the quality and durability of initial stimulation treatments. It would be more cost-
effective and profitable to improve the initial treatment, if possible, to avoid the need for subsequent
restimulation efforts.

Brief Summary of Evidence that Fracture Conductivity and Durability Significantly Impact Productivity

In vertical wells, there is ample evidence that well productivity can be improved by increasing fracture width,
proppant quality and conductivity. Traditional fracture designs frequently constrain well potential, as evidenced by
history matching results from more than 200 field trials (Vincent 2009). However, with ultra-tight resource plays,
many researchers mistakenly claim that fracture conductivity is no longer of concern. In reality, transverse fractures
initiated from horizontal wells typically result in extremely high hydrocarbon velocity due to converging flow, and
the concern for fracture conductivity near wellbore can be of extreme importance in resource plays (Shah, 2010;
Besler, 2007). An increasing number of papers have been published demonstrating the production impact of altering
the fracture conductivity in transverse fracs, a portion of which are summarized in Table 2.

Table 2: Selected references suggesting more attention should be focused on fracture conductivity and durability in ultra-
tight resource plays

Lab Testing showing proppants degrade over time, but some proppants are much more durable SPE 136757, 12616, 14133, 15067,
than others. Sand and resin coated sand suffer from thermal degradation. 110451, 128612, 134330, 139875
Refrac summaries indicating the initial frac treatment failed to capture available reserves SPE 134330
Wells stimulated with durable ceramic proppants may delay or avoid the need for restimulation SPE 136757
Evidence that improved frac conductivity increases well productivity and profitability SPE 134595, 146376, 160206, 110679,
Bakken 136757 + many unpublished cases
Eagle Ford 138425, 152533, 152839, 155779
Haynesville 125079, 164043, 134165, 147436
Granite Wash 164043
Evidence that more durable proppants affect EUR, not just acceleration of reserves SPE 147436
Wells connected by fractures that show insufficient conductivity SPE 144702, 145463, 140463, Murray
(2013 ATW), 151980, 160480,
139774, Taylor (Mathistad testing)
Evidence that stacked Bakken and Three Forks laterals produce at distinctly different watercuts unpublished

Field studies documenting incremental productivity achieved with improved frac designs are convincing to some
operators. However, other evidence may be more compelling to specific audiences. There are numerous wells in
the Barnett, Bakken, Three Forks, Marcellus and Eagle Ford in which diagnostics proved conclusively that there was
full communication of fracturing pressure between two adjacent horizontal wellbores during the fracturing
treatment. In many well pairs, it is evident that liquid chemical tracers, solid tracers, and even full wellbore volumes
URTeC 1579008 6

of frac sand have been conveyed during the fracturing treatment into the adjacent wellbore. However, a great
number of these wells appear to have essentially zero hydraulic continuity retained six months after the frac
treatment. It appears that as the wells are produced, drawdown causes some portions of the fracture to collapse.

Two examples will be reviewed which show more encouraging, but incomplete continuity sustained over time. In
the Marcellus, Mayerhofer (2011) provided a thorough dataset demonstrating excellent communication between
adjacent wellbores on 950 ft spacing, as documented with downhole pressure monitoring, microseismic mapping,
and chemical tracer recovery in adjacent wells (Figure 1). Each of the wells produces at similar rates. However,
when one well is shut in, the other will increase in productivity by approximately 20%. If the fracture network were
infinitely conductive with zero pressure losses, the adjacent well would double in production. This is evidence of
significant pressure losses, (thousands of psi) within the connected fracture networks.

Marcellus Fractures Intersecting Offset Laterals


Marcellus - Slickwater
Microseismic, DFITS, downhole pressure
gauges, PTA, chemical tracers, production
interference

950 ft spacing. 1H treated 5 weeks after 2H

Cemented, 7 stage PnP


Slickwater 100 mesh, 40/70 and 30/50 sand
~6000 ft TVD

Pressure communication in 6 of 7 stages


Chem tracers from 2,3,5,6,7 recovered in 2H

When one well is shut in, the other well


increases in rate by ~20% demonstrating
some degree of connection, but
clearly imperfect after 6 months.
Large pressure losses inside the
fractures.

Mayerhofer SPE 145463 – Nov 2011

Figure 1 – Adjacent wellbores in the Marcellus showed communication during the fracturing injection treatment, as
demonstrated with pressure monitoring, chemical tracers, and microseismic mapping However, after six months of
production, interference testing demonstrates that pressure losses within the fracture network must be thousands of psi.
Figure adapted from Mayerhofer et al 2011

Elsewhere in the Marcellus, Edwards et al (2011) documented the results of drilling wells on 500 ft spacing.
Production data to date indicate that the wells do not appear to share reserves, and production analyses suggest the
effective fracture lengths may be merely 150 feet after six months, despite mapped lengths exceeding 1000 ft. Wells
drilled on 500 ft spacing appear to be similar in productivity to those drilled on 1000 ft spacing, suggesting that they
are not competing for reserves. While there are a variety of possible interpretations, these results strongly suggest
that the entire created fracture geometry does not remain effective.

Similar data have been shown by Murray et al (2013) in the Eagle Ford. Excellent data gathering demonstrated
connection during the stimulation between a horizontal well and parallel laterals at 500-600 ft spacing. Chemical
tracers and solid radioactive proppant were observed in most stages in both immediate offsets. Soon after cleanup
and initial flowback, pressure communication testing was performed by bringing the center well onto production
while parallel laterals remained shut-in with pressure gauges recording the pressure influence. Lag times of 43 and
55 minutes elapsed before any pressure influences were documented in the offset wells. However, when a similar
interference test was conducted three months later, the lag times increased to 76 and 81 minutes. While some of this
may be due to compressibility changes in the formation, it may also suggest that the “flowpath length” between the
wells has increased over time, which would indicate further deterioration of the fracture network.
URTeC 1579008 7

It is interesting to note that if the connecting fractures were envisioned to behave as an infinitely conductive pipe, a
pressure pulse in one well could be observed at a 500 ft distance with a lag time of less than one second. Further,
the pressures would equilibrate through this infinitely conductive pipe, and there could be no difference in
bottomhole pressures between the wells. Clearly, it is incorrect to envision these fractures as infinitely conductive,
open channels. It is also compelling that when testing was repeated three months after initial stimulation, despite
production of the central well, the bottomhole pressure in offset wells continued to increase, convincingly
demonstrating the reservoir was capable of supplying hydrocarbons to those wellbores much more rapidly than the
hydraulic fractures connecting the wells could deplete that pressure.

In addition to evidence that fractures collapse at a lateral distance from the wellbore, it appears that wellbores
stacked vertically show similar loss of hydraulic connection over time. It has been repeatedly documented that it is
possible to inject fracturing fluids through the lower Bakken Shale, apparently creating fractures connecting stacked
laterals landed in the overlying Middle Bakken and underlying Three Forks. The vertical separation of these
wellbores is often less than 100 feet. Frac fluid, chemical tracers, solid radioactive tracers and full wellbores of sand
have been recovered from offsets, corroborating pressure monitoring and microseismic data, conclusively
demonstrating that at least in some portions of the Bakken development, it is possible to initially create fractures that
connect Middle Bakken and Three Forks laterals. However, wells appear to operate independently, with little
sustained production interference. It is common for Three Forks wells to produce at a distinctly different water cut
and GOR from overlying Middle Bakken completions, easily demonstrating that the wells are not redundant, and are
draining unique reserves. The connecting fractures are NOT infinitely conductive as envisioned by some. At least
five Bakken operators have announced plans to evaluate “high density”, “vertical down-spacing”, or “array
fracturing” in which laterals will be landed into multiple targets in recognition that fractures are insufficiently
capturing resources throughout the section. A conceptual figure is shown in Figure 2.

Figure 2 – A number of operators are investigating “vertical downspacing” in the Bakken petroleum system. [Slide from
CLR Investor Presentation, Continental, 2012]. Similar efforts are underway in the Niobrara, Woodford, Montney, and
several Permian Basin plays, in recognition that conventional hydraulic fractures sustain insufficient continuity and
conductivity to drain the thick pay sections.

That operators have elected to spend hundreds of millions of dollars to vertically down-space wells in these
formations is sufficient evidence that conventionally-executed fractures fail to adequately drain the vertical section
in many laminated plays.
URTeC 1579008 8

New Evidence that Supports the Above Interpretations


An extensive review spanning more than 100 pages was completed [Vincent 2012a] to investigate the damage
mechanisms and identify specific examples of frac design changes that appear to improve the sustained performance
of fracture treatments in resource plays. This paper will not reproduce that large volume of evidence, but will
instead highlight a few specific examples of the need to focus on the connection between the wellbore and the
fracture.

Is Overflushing Acceptable in All Plays?


Following the success of completing Barnett Shale wells with large slickwater treatments that were overflushed, it
has become “conventional wisdom” that all horizontal resource plays can be completed with similar strategies. It is
estimated that more than 90% of all transverse fractures pumped by our industry have been overflushed, because it is
operationally necessary to sweep the wellbore clean of proppant before dropping balls or pumping plugs to isolate
stages. It is acknowledged that in some plays, there is no evidence that overflushing causes any problems with well
productivity. In some formations, overflushing may actually result in superior production.

However, the authors question the wisdom of blindly following that strategy without investigating whether it is
appropriate in each specific reservoir. Consider some of the evidence:
• In many plays, stages initiated in more “brittle” rock (or highly siliceous) are more productive than those in
more ductile rock. It has been argued the mechanism is that more complex fracturing develops in “brittle”
rock. But could an alternative explanation be that this is an artifact of the way wells are overflushed? Is it
possible that more rigid, competent rock better accommodates the unpropped or poorly propped void caused
by overflushing? If our treatments were designed differently, might it be possible to achieve high
productivity even in stages initiated in more ductile rock?
• Numerous operators and service companies have reported that large percentages of perforation clusters fail to
produce measurable hydrocarbons, and in many cases tracer logs indicate there is minimal or no proppant in
some perforations. There are a large number of explanations that have been proposed by the industry.
However, it is worth questioning whether a portion of this observation can be attributed to overflushing
between stages?
• In some plays, operators must manage the drawdown (choke management) because aggressive production
appears to result in a loss of reserves. The highest drawdown is near the wellbore, likely concentrated on the
tenuous connection between the wellbore and the fracture. This critical connection, when overflushed, may
be inadequately propped and may be vulnerable to collapse.
• Restimulation attempts have been successful in nearly all resource plays, demonstrating that initial frac
treatments failed to capture the reserves accessible to the well.
Is it possible that overflushing proppant might possibly be responsible for some of these observations in certain
reservoirs?

Proposed Experiment
Instead of blindly overflushing all stages as the customary approach, these authors often recommend that their
clients experiment with completion strategies. In many wells, it is common for the operator to drill out the frac
plugs or ball seats after the fracturing treatments are completed. In those wells, the incremental time and expense to
intentionally screenout the heel stage is relatively negligible (merely wasting a casing volume of proppant). The
heel-most stage can be designated as a trial stage, comparing the productivity of this stage to the immediately
adjacent stage which was overflushed per common techniques. Often, operators will choose to try several things on
this heel-most stage, substituting a stronger, durable proppant, pumped at higher concentrations, with intentional
screenout to dilate the near wellbore fractures to the maximum extent, pack the area without overflushing, and
perhaps divert fluid or initiate additional entry points as a result of the increased treating pressure. This intentional
screenout allows the operator to calibrate expectations and better investigate how aggressive a proppant size or
concentration the formation will accept. When the individual stage productivity can be determined via production
logs or chemical tracer recovery, in a number of reservoirs, this heel-most stage which focused on frac conductivity
and improved connection between the wellbore and fracture network is frequently evaluated to be twice the
productivity of adjacent stages with less attention to the near-wellbore connection and conductivity.
While this experiment has been compelling to some individuals, without exception there have been others in the
respective organizations who discount the validity of the test. Surely, it is a single stage, it is closer to the wellhead,
the formation geology is variable—the results admittedly may just be dumb luck. Plus, the operator is unwilling to
URTeC 1579008 9

intentionally screenout multiple stages anyway. One technique to resolve this disagreement has been to identify
specific stages in a future well to be treated with differing displacement methods for investigation.

Suppose the engineering and geological team has sufficient information to estimate the formation potential of
proposed stages. For sake of argument, suppose that stages 4 and 5 seem to be in identical rock. Stages 8 and 9
seem to have equivalent potential. Stages 12 and 13 appear equivalent according to all available information.
Stages 4, 9, and 12 are designated to receive “Frac Design A” and Stages 5, 8, and 13 will receive Frac Design B.
Subsequent production logging compares the productivity of the stages. If the productivity of strategy A surpasses
that of strategy B in 3 of 3 tests, then there is a decent statistical likelihood that something important has been
learned, and that strategy A is truly superior. Clearly there is at least a 15% chance that unseen variation or “dumb
luck” influenced the outcome, even when one strategy is superior 3 of 3 times. [Note that if the results are not
unanimous, there is almost no statistical merit to the interpretation with this small sample size.] While this approach
is not foolproof, it is at least an attempt to optimize based on actual production data instead of by intuition.

Table 3: Proposed experiment to compare productivity of specific treatment designs on selected stages

If electric logs, drilling trajectory, mud logs allow estimation of If no information is available to designate stages of comparable
stage potential potential

A minimum of three stage pairs are selected that appear to have Every odd-numbered stage receives Frac Design A
equivalent production potential. e.g., stages 4&5, 8&9, and Every even-numbered stage receives Frac Design B
12&13 are designated as evaluation pairs. Relative productivity evaluated with chemical tracers or
Stages 4, 9, and 12 receive Frac Design A production logging
Stages 5, 8, and 13 receive Frac Design B
Relative productivity evaluated with chemical tracers or
production logging

Figure 3 is provided with permission of an operator under the conditions of anonymity and specific prohibition to
release details of the reservoir and treatment schedule. Although this well was not treated in the careful systematic
manner proposed, it is presented to show conceptually what may be observed from a production log.

Figure 3 – If production logs or chemical tracers can be used to evaluate individual stage contribution, it is possible to
gather interesting evidence that may affect future design strategies. This log was gathered in a horizontal well in a liquid-
rich formation, not the Bakken. Other details are withheld at operator request.

The evidence shown in Figure 3 was far from an optimal experiment. A casing obstruction prevented access to
planned Stage 1, and Stage 2 therefore does not have the adjacent stage for comparison. However, from the data
presented, the three stages that did achieve screenouts (green bars) appear to be approximately twice the productivity
of standard overflushed stages at the time of this production log. Unpropped treatments in stages 3 and 4 do not
look encouraging in this formation. Curiously, the operator did not screenout the “free” heel stage on this well. In
some other examples, the decision to NOT screenout the heel stage can counter skeptical claims that “the heel is
always better” and distinguish between damage introduced by intentional overflush compared to unintentional
URTeC 1579008 10

damage incurred with isolation of subsequent stages. There is no statistical merit to the interpretation of stage 10, in
which the same mass of proppant was pumped, but the frac fluid volume was reduced 25% [ie, presumably a
shorter, fatter frac]. If this interpretation were not supported by the observations of stages 2, 7, and 13 it would be
discounted entirely. However, coupled with previous wells in this formation in which intentionally screened out
heel stages were of higher productivity, these results might be compelling that an operator should focus on the
conductivity of the fracture in the near-wellbore area in this particular resource play.

The authors are aware of the tremendous industry reluctance to even consider this evidence. Many operators are
adamant that they will not intentionally screenout regardless of this finding, since it is operationally impractical to
perform multiple flowbacks or cleanouts on wells. Even when confronted with the possibility of doubling well
productivity, it is common for most operators to ignore these results and instead retreat to the conventional wisdom
that “overflushing works just fine, is fast, and inexpensive”.

It is recognized that there are both advantages and disadvantages to screenouts.


• Probable advantages to screenouts
• Wider frac (more net pressure)
• Better connection to wellbore
• Treatment diversion into other perforation clusters
• Reduced proppant flowback
• A screened-out fracture may be “immune” to subsequent overdisplacement when pumping
plug/dropping ball
• May be “immune” to subsequent refrac injection?
• Disadvantages to screenouts
• Standby time and cost to cleanout/flowback
• Higher pressures may induce more gel damage
• Stress on equipment and tubulars during treatment
• Higher stress must be borne by proppant
• Never screenout wells with ULWP or deformable proppant
• May crush cleats in CBM, delicate formations
• High net pressure may induce unwanted height growth, sacrificing propped length
In one of the most statistically significant field trials published in our industry [Vincent 2007], it was evident that
screenouts in the Pinedale Anticline in stages using ceramic proppant consistently resulted in superior productivity
and superior apparent fracture length, while stages that screened out using lower strength sand or RCS were
consistently of poor productivity. This demonstrates that careful evaluation may be necessary to determine the best
strategy in a specific field. However, if the relatively inexpensive heel-most screenout strategy with strong proppant
is tested and production logs indicate improved productivity, there are steps that can be taken to capture part of the
benefit, while perhaps avoiding the cost and delay of multiple flowbacks or cleanouts.

Application of these Concepts in the Bakken


One of the authors has advocated and implemented the following strategy in the Bakken and Three Forks formations
in an effort to improve fracture conductivity:
• High conductivity ceramic proppants are used to maximize flow capacity throughout the fracture,
commonly reaching 6 ppa in the majority of the fracture treatments. No use of sand or resin coated sand in
these completions.
• At the end of each stage, a high concentration proppant slug, regularly exceeding 12 ppa and often reaching
15 to 20 ppa is introduced.
• Stages are flushed only to the top perforation, with efforts to “call flush” aggressively to underdisplace each
stage with the goal of ensuring high concentration slurry remains near the perforations and is not
overflushed into the fracture.
• Aggressive quality control is performed to ensure water quality and cleanliness, fluid rheology, breakers,
pH, scale inhibition, and proppant quality, with the goal of maximizing fracture conductivity and
connection between the wellbore and fracture.

The authors acknowledge that this strategy does not guarantee that a highly concentrated slurry remains near the
wellbore in all perforation clusters, and that some damage likely occurs during subsequent wellwork to pump the
URTeC 1579008 11

required plug and perforation gun assembly before treating the next stage. It is not argued this strategy achieves an
identical proppant arrangement to stages that are screened out. Instead, this is an operational compromise that
increases the likelihood of retaining a durable, conductive connection in the near-wellbore area while minimizing
expensive and time consuming flowback/cleanout following screenouts.

Production Comparison of Various Strategies


One Bakken operator completed 34 horizontal wells in close proximity with two primary strategies, shown as blue
and red dots on Figure 4. Both blue and red categories of wells utilized ceramic throughout the entire pump
schedule, with small amounts of Ottawa frac sand used in portions of only five red wells. One completion strategy
(blue) was designed by another consultant focused on proppant transport to ensure superior fracture lengths, which
was achieved with greater fluid volumes, higher gel concentrations and reduced breaker concentrations. This
category of wells also had the advantage of greater spacing between the stages to delay production competition
between adjacent stages. The red completion strategy focused on improving the near-wellbore conductivity by
placing high concentration proppant slugs and taking extra precautions to reduce or eliminate overflushing of
proppant where possible. The majority of these treatments were plug and perf completions in uncemented laterals
isolated with swellable elements in the casing/borehole annulus to compartmentalize frac stages. In the more recent
wells, between 24 and 30 stages were specified per ~9500 ft lateral draining 1280 acre spacing units. Over time, this
second category of wells had the added advantage of longer drilled lateral lengths and much higher stage count
achieved with more closely spaced swell packers.

Figure 4 – Study area bordering McKenzie and Dunn Counties of North Dakota. Blue wells emphasized proppant transport,
red wells emphasized near-wellbore conductivity, and green wells completed by offset operators used less expensive frac
sand.

Production results from the ceramic categories are shown in Figure 5. This figure shows total well rates, not
corrected for stage count.
URTeC 1579008 12

450000 Conductivity - Well A


Conductivity - Well B

Prop Transport - Well 7 Conductivity - Well C


Conductivity - Well D
400000 Conductivity - Well E
Average of Conductivity Design Conductivity - Well F
Conductivity - Well G

350000 Average of Transport Design Conductivity - Well H


Conductivity - Well I
Cumulative Production, barrels of oil

Conductivity - Well J
Conductivity - Well K
Conductivity - Well L
300000
Conductivity - Well M
Conductivity - Well N
Conductivity - Well O

250000 Conductivity - Well P


Conductivity - Well Q
Conductivity - Well R
Conductivity - Well S
200000 Conductivity - Well T
Conductivity - Well U
Conductivity - Well V
Conductivity - Well W
150000 Conductivity - Well X
Conductivity - Well Y
Conductivity - Well Z
Conductivity - Well AA
100000
Prop Transport - Well 1
Prop Transport - Well 2
Prop Transport - Well 3

50000 Prop Transport - Well 4


Prop Transport - Well 5
Prop Transport - Well 6
Prop Transport - Well 7
0 Average of Conductivity Design

0 5 10 15 20 Average of Transport Design

Months of Production

Figure 5 – Seven wells (blue) completed with techniques to enhance proppant transport underperform 27 wells (red)
completed with strategies that emphasize the near-wellbore conductivity. In this comparison, blue wells typically were
treated in 5 or 6 stages while red wells averaged 28 stages. Future figures show comparisons normalized for stage count.

Adjacent operators have completed dozens of surrounding wells, with 23 of these wells treated only with frac sand
according to public records of the North Dakota Industrial Commission. The cumulative production trends of these
23 wells known to have been treated only with frac sand are shown in Figure 6. These wells contained an average
of 20 frac stages each.
URTeC 1579008 13

450000 Known Sand - Well S1

Known Sand - Well S2

Known Sand - Well S3


400000
Average of SAND Design Known Sand - Well S4

Known Sand - Well S5

350000 Known Sand - Well S6


Cumulative Production, barrels of oil

Known Sand - Well S7

Known Sand - Well S8


300000
Known Sand - Well S9

Known Sand - Well S10

250000 Known Sand - Well S11

Known Sand - Well S12

Known Sand - Well S13


200000
Known Sand - Well S14

Known Sand - Well S15

150000 Known Sand - Well S16

Known Sand - Well S17

Known Sand - Well S18


100000
Known Sand - Well S19

Known Sand - Well S20

50000 Known Sand - Well S21

Known Sand - Well S22

Known Sand - Well S23


0
Average of SAND Design
0 5 10 15 20
Months of Production

Figure 6 – Twenty-Three wells were drilled in close proximity to the study wells in Figure 5, but were treated with frac sand
and no use of ceramic. These green wells were treated in an average of 20 stages.

While it is challenging to compare wells with differing completion strategies and different number of stages,
Figure 7 shows the cumulative production trends over time for wells after normalizing for stage count, with an inset
table showing the average parameters of each category. A handful of wells were removed from the larger dataset,
including Proppant Transport Wells 5 and 7 with poor stage compartmentalization, and Wells Q and AA (wellbore
failure and re-drill). Wells with incomplete data available to determine number of stages, fluid volume, and
proppant mass were excluded from Figure 7.
URTeC 1579008 14

14000
Cumulative Production per Stage, barrels of oil per stage

Average of Ceramic Conductivity Design


12000

Average of Ceramic Transport Strategy

10000 Average of SAND completions

8000

6000

4000

2000 Parameter Ceramic - Near Wellbore Conductivity Strategy Ceramic - Proppant Transport Strategy Sand Completions
Fluid per stage (bbls) 2597 3241 2571
Proppant per stage (lbs) 113,832 90,837 126,204
Stage spacing (feet) 320 755 482

24 month Cum bo per stage 11,288 10,222 7,423


0
0 5 10 15 20
Months of Production
Figure 7 – When normalized for stage count, both ceramic strategies outperformed offset wells completed with frac sand,
despite greater proppant mass utilized in the sand stages. Ceramic stages focusing on near-wellbore conductivity appear to
have superior sustained production, despite shorter presumed fracture length and greater competition from more closely
spaced fractures. For context, 3865 bbls of incremental oil at a conservative price of $50/bbl is $193,000, easily providing a
500% return on the investment to upgrade the proppant quality in a stage.

Summary
There is a growing body of evidence indicating that fractures are not as effective and durable as commonly assumed.
It appears that a large number of damage mechanisms may influence fracture performance and different strategies
will be required to optimize treatments in each resource plays. This paper tabulated a range of suspected
mechanisms, but focused on the influence of near-wellbore connectivity in the Bakken resource play. In the Middle
Bakken and Three Forks formations, efforts to place more durable ceramic proppants and increase the concentration
in the near wellbore area appear to result in superior sustained production based on 24 months of production data
available to date.

In this study area, it is evident that more durable proppants and attention to the near-wellbore connection is justified
and generating significant increases in well production and profitability. There is not yet production evidence that
closely spaced stages are causing competition detrimental to production in the first two years. On the contrary,
increased number of initiation points improve the extent of connection between the wellbore and fractures, which
appears to improve sustained productivity.
URTeC 1579008 15

References
Besler, M.R. et al, 2007: Improving Well Productivity and Profitability in the Bakken—A Summary of Our Experiences Drilling,
Stimulating, and Operating Horizontal Wells. Paper SPE 110679 presented at the 2007 Annual Technical Conference,
Anaheim, Nov. 11-14.
Continental Resources, Inc. 2012: CLR Vision, 2012 Investor’s Day presentation, Oct 9, 2012.
Edwards, K. L., Weissert, S., Jackson, J., Marcotte, D., 2011: “Marcellus Shale Hydraulic Fracturing and Optimum Well Spacing
to Maximize Recovery and Control Costs”, SPE 140463 presented at the 2011 SPE Hydraulic Fracturing Technology
Conference: The Woodlands, 24 – 26 Jan.
Fisher, K. 2012: Trends Take Fracturing ‘Back to the Future’. AOGR Aug 2012.
Mayerhofer, M.J. et al 2011: Integrating Fracture Diagnostics and Engineering Data in the Marcellus Shale. Paper SPE 145463
presented at the Annual Technical Conference, Denver, Oct 30-Nov 2.
Murray, M. et al 2013: Understanding Shale Performance: What do the data say? Eagle Ford Shale. SPE ATW Production
Decline in Oil and Gas Shale Reservoirs: What Can We Do to Improve the Performance? Feb 21-Mar 1, Santa Fe, NM.
Shah, S.N., Vincent, M.C., Rodriquez, R., Palisch, T. 2010: Fracture Orientation and Proppant Selection for Optimizing
Production in Horizontal Wells. SPE paper 128612 presented at the SPE Oil and Gas India Conference, Mumbai, Jan 20-22.
Taylor, A: “Mathistad Communication Testing,” Report of tests conducted by Continental Resources,
https://cms.oilresearch.nd.gov/image/cache/g-018-039-f_2.pdf
Vincent, M.C., Huckabee, P., Conway, M. 2007: Field Trial Design and Analyses of Production Data from a Tight Gas
Reservoir: Detailed Production Comparisons form the Pinedale Anticline. SPE paper 106151 presented at the 2007 Hydraulic
Fracturing Technology Conference, College Station, TX Jan 29-31.
Vincent, M.C. 2009: Examining our Assumptions – Have Oversimplifications Jeopardized Our Ability to Design Optimal
Fracture Treatments? Paper SPE 119143 presented at the 2009 Hydraulic Fracturing Technology Conference, The Woodlands,
Jan 19-21.
Vincent M.C. 2012a: Review of Best Practices – Emphasis on Haynesville, Eagle Ford, Bakken, Marcellus, Barnett. Appendix
to RPSEA Sponsored Project on Sustaining Fracture Area and Conductivity of Shale Reservoirs for Enhancing Long-Term
Production and Recovery.
Vincent, M.C. 2012b: The Next Opportunity to Improve Hydraulic-Fracture Stimulation. JPT Distinguished Author Series,
March 2012 118-127, SPE 144702.

You might also like