Mineralogy Based Brittleness Prediction From Surface Seismic Data - Application On The Barnett Shale

t Technical papers
Mineralogy-based brittleness prediction from surface seismic data:

Application to the Barnett Shale
Roderick Perez Altamar1 and Kurt Marfurt1
Downloaded 02/12/15 to 129.15.127.242. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
Abstract
Differentiating brittle and ductile rocks from surface seismic data is the key to efficient well location and
completion. Brittleness average estimates based only on elastic parameters are easy to use but require empirical
calibration. In contrast, brittleness index (BI) estimates are based on mineralogy laboratory measurements and,
indeed, cannot be directly measured from surface seismic data. These two measures correlate reasonably well
in the quartz-rich Barnett Shale, but they provide conflicting estimates of brittleness in the calcite-rich Viola,
Forestburg, Upper Barnett, and Marble Falls limestone formations. Specifically, the BI accurately predicts lime-
stone formations that form fracture barriers to be ductile, whereas the brittleness average does not. We used
elemental capture spectroscopy and elastic logs measured in the same cored well to design a 2D λρ–μρ to brit-
tleness template. We computed λρ and μρ volumes through prestack seismic inversion and calibrate the results
with the λρ–μρ template from well logs. We then used microseismic event locations from six wells to calibrate
our prediction, showing that most of the microseismic events occur in the brittle regions of the shale, avoiding
more ductile shale layers and the ductile limestone fracture barriers. Our λρ–μρ to brittleness template is em-
pirical and incorporates basin- and perhaps even survey-specific correlations of mineralogy and elastic param-
eters through sedimentation, oxygenation, and diagenesis. We do not expect this specific template to be
universally applicable in other mudstone rock basins; rather, we recommend interpreters generate similar
site-specific templates from logs representative of their area, following the proposed workflow.
Introduction Although the Barnett Shale formation is capable of

Shales are described as organic-rich, fine-grained generating, retaining, and storing huge amounts of hy-
reservoirs (Bustin, 2006) and are typically dominated drocarbon, gas flow is limited if the individual microre-
by clays. The mineral composition and the presence servoir compartments cannot be connected via well
of organic matter can influence not only the distribution stimulation (Jarvie et al., 2007). Hydraulic stimulation
of pores and fluid saturation (Sondergeld et al., 2010a), connects otherwise unconnected pore spaces.
but also the effectiveness of stimulation. Analysis to differentiate brittle from ductile rocks has
Bowker (2003) reports that most of the production in been key to stimulation success in shale gas reservoirs,
the Barnett Shale comes from zones with 45% quartz especially in the Barnett Shale where brittleness is
content and only 27% clay. In general, the average mainly controlled by quartz content. Based on the min-
porosity is 6% with pore throats typically less than eralogy content, ductility (the opposite of brittleness) is
100 nm (Bowker, 2003). Field experiments show that controlled by clay, calcite, and total organic content
more effective hydraulic fracturing results in improved (TOC). Jarvie et al.’s (2007) and Wang and Gale’s
completion and higher production rates (Saldungaray (2009) brittleness index (BI) defines ductile and brittle
and Palish, 2012). In the Forth Worth Basin, Gale et al. regions in terms of its mineralogical content, generating
(2007) describe at least two sets of natural fractures a smooth transition between both regions. In contrast,
that might be cemented. However, they can be reacti- Grieser and Bray (2007) define an empirical brittleness
vated during hydraulic fracturing, providing a larger cut-off based on Poisson’s ratio and Young’s modulus.
rock volume and optimizing production. The cut-off depends on the shale play in which it is ap-
Gas is derived from thermogenic cracking of kerogen plied and on the expertise of the interpreter.
and from the cracking of any retained oil in the shale Perez (2011) combines λρ–μρ seismic logs with pro-
(Jarvie, 2003; Montgomery, 2005). Free gas is stored duction data to create templates that can be applied to
within the rock pores and natural fractures, whereas better understand the parameters that control the esti-
the adsorbed gas is stored on the organic material. mated ultimate recovery (EUR) and stimulated rock
1
ConocoPhillips School of Geology and Geophysics, Norman, Oklahoma, USA. E-mail: [email protected]; [email protected].
Manuscript received by the Editor 10 October 2013; revised manuscript received 8 March 2014; published online 18 August 2014; corrected
version published online 20 November 2014.This paper appears in Interpretation, Vol. 2, No. 4 (November 2014); p. T255–T271, 13 FIGS., 2 TABLES.
http://dx.doi.org/10.1190/INT-2013-0161.1. © 2014 Society of Exploration Geophysicists and American Association of Petroleum Geologists. All rights reserved.
Interpretation / November 2014 T255

volume in unconventional reservoirs. In the case of calculated Young’s modulus, Poisson’s ratio, λρ, and μρ.
stimulated volumes, the templates show the variation Then, we proceed to map these variables against the BI,
in λρ–μρ for a combination of mineralogical mixtures and we use this template to predict brittleness from sur-
(quartz and clay) versus total and effective porosity. face seismic inversion, which were calibrated and ana-
However, these results do not show a direct link with lyzed using microseismic event locations.
brittleness.
Brittle rocks are unable to withstand large amounts The Barnett Shale
of strain creating a plane of weakness in the rock, giving “The Fort Worth Basin is a shallow north–south elon-
rise to microseismic events when they fail. If the rocks gated foreland basin, encompassing roughly 15;000 mi2
are brittle, the injected proppant will keep these frac- in North Texas, formed during the late Paleozoic
tures open. In contrast, ductile rocks deform plastically Ouachita Orogeny” (Walper, 1982). Paleotectonic colli-
and can undergo significant strain prior to fracture. sion events in the Fort Worth Basin resulted in a north-
Fractures in more plastic ductile rocks are thought to west–southeast main stress field orientation at the time
close about the proppant, thus sealing pathways to fluid of the Barnett Shale deposition. However, the present-
flow. Rocks with high brittleness exhibit naturally oc- day regional maximum stress direction in the basin is
curring and hydraulically induced fractures, although northeast–southwest, with local deviations in intensity
in the Barnett Shale, most natural fractures are ce- and direction about the mineral wells and other minor
mented (Gale et al., 2007). faults (e.g., Simon, 2005).
Brittle rocks may lack natural fractures entirely The Barnett Shale is an organically rich and ther-
(Gale et al., 2007) if the loading history is insufficient mally mature rock deposited during Mississippian time
to drive fracture growth. This is the case most of the (≈340 Ma) in the Fort Worth Basin, characterized by
time, as documented by lack of fractures or very simple low average permeability (70 nD) and porosity (6%) dis-
fracture patterns in even highly brittle rocks (Ellis et al., tributed in a variety of depositional facies (Deacon,
2012). Subsequently, brittle rocks will only “exhibit” hy- 2011). The Viola, Forestburg, and Marble Falls Lime-
draulically induced fractures if a frac job is performed stones are hydraulic fracture barriers and are not con-
nearby, and in that case, even nonbrittle rocks may frac- sidered production targets because they are water-
ture. This situation entirely depends on the engineering bearing. The Viola Formation deposited on top of the
variables applied, such as completion and treatment karsted Ellenburger Formation (Loucks, 2008) presents
techniques, used to hydraulically stimulate the well. a potential risk of water production. In the area of
Other authors describe the complexity of the frac- study, the Forestburg Limestone divides the Upper Bar-
ture pattern in the Barnett Shale using microseismic nett and the Lower Barnett shales into two members,
event locations and correlate them with seismic attrib- which must be treated and fractured separately.
utes to highlight more brittle areas in the formation (Re- Singh (2008) and Perez (2009) describe three distinc-
funjol, 2010; Thompson, 2010, Refunjol et al., 2012). tive gamma ray (GR) log patterns: upward-increasing,
Using the same data set examined in this paper, Thomp- upward-decreasing, and constant. These GR log pat-
son (2010) finds that structural ridges are highly ce- terns are correlated to lithofacies using cored wells
mented, such that most microseismic events occur in in the Barnett Shale representing unique deposition
bowl-shaped reservoir zones. Simon (2005) finds that environments. Singh (2008) defines the upward-increas-
microseismic events form a better fracture network ing GR parasequence (GRP) to be composed of upward-
in areas that exhibit low-azimuthal velocity anisotropy, increasing amounts of clay accompanied by a decrease
resulting in high EUR. In contrast, microseismic events in calcite content.
in zones that exhibit higher azimuthal anisotropy pro- Kale (2009) and Gao (2011) combine Singh’s (2008)
duce narrow, elongated fracture patterns, resulting in 10 petrofacies into three petrotypes exhibiting similar
lower EUR. petrophysical properties. Kale (2009) bases his classifi-
The objective in this paper is to predict brittleness cation on porosity, TOC, and total carbonate measure-
from surface seismic data. To do so, we need to recon- ments. These petrotypes are then ranked in terms of
cile and link conflicting brittleness parameters defined cumulative production data from three vertically cored
by Grieser and Bray (2007), Jarvie et al. (2007), Rick- wells.
man et al. (2008), and Wang and Gale (2009).
We begin with a review of the geology and the min- Mineralogy
eralogy composition of the Marble Falls Limestone, Conventional logs such as GR, neutron porosity, and
the Upper Barnett Shale, the Forestburg Limestone, resistivity are useful to stratigraphically characterize a
the Lower Barnett Shale, and the upper section of the reservoir. However, these logs do not fully provide the
Viola Limestone formations deposited in the Fort Worth information needed to characterize organic shales in
Basin, paying special attention to the Barnett Shale for- terms of their geomechanical behavior. This addi-
mation. Next, we estimate the BI based on the rock min- tional information can come from the integration of
eralogy composition from log measurements in a fully sequence stratigraphy, special analysis techniques, spe-
cored well, located just outside the seismic survey. Using cialized logging tools, and core lab measurements. Re-
density logs and P- and S-wave sonic logs in the well, we cent availability of mineralogy logs such as elemental
T256 Interpretation / November 2014

capture spectroscopy (ECS) and dipole sonic logs en- Conventional log sets such as GR (Figure 2, track 2),
able characterization of a reservoir in terms of its min- photoelectric factor, and resistivity were acquired in the
eral content and elastic properties, providing a means area, and they differentiate shale from limestone forma-
to differentiate lithology types by their completion tions. Applying the same concept used by Singh (2008)
response. and Perez (2009), we interpret the GR log pattern (in-
Core and outcrop studies show the Barnett Shale to dicated by arrows in Figure 2, track 2) and their corre-
be dominated by clay- and silt-size sediment with occa- sponding GRP (Figure 2, track 3).
sional beds of skeletal debris. Organic and biogenic Organic-rich sediments (source rock) that contain
constituents were deposited at the time of the sedimen- significant amounts of organic matter have a higher
tation of the Barnett Shale and include algal remains, resistivity and sonic transit time than organic-lean sedi-
spores, plant remains, sponges, and radiolarians, ments. Also, organic-rich rocks are often highly radio-
among others (Slatt, 2011). After sedimentation, chemi- active, which is easily seen on GR log measurements
cal reactions lead to the generation of secondary (Schmoker, 1981a, 1981b). The TOC in this paper was
minerals including authigenic clays, calcite, dolomite, calculated using a Schlumberger proprietary methodol-
quartz, pyrite, and hydrothermal minerals (Slatt and ogy, which has been calibrated by hundreds of TOC
O’Brien, 2011). core data from other wells in the basin. TOC varies be-
Based on several wells, Jarvie (2003) reports that the tween 1% and 11% as shown on track 8 in Figure 2, and it
mean composition of the Barnett Shale in the region agrees with TOC core measurements from Singh (2008).
consists of 40% quartz, 29% illite (with minor smectite),
13% calcite, 3% organic matter, and 2% pyrite. Karasta- Brittleness
this (2007) reports that the Upper Barnett Shale shows When a rock is subjected to increasing stress, it
a higher carbonate content than the Lower Barnett passes through three successive stages of deformation:
Shale, with the main carbonate minerals present in elastic, ductile, and fracture. Based on these behaviors,
the formation being calcite, dolomite, siderite, and it is possible to classify the rocks into two classes: duc-
aragonite with calcite being dominant. Illite is the dom- tile and brittle. If the rock has a smaller region of elastic
inant clay mineral, contributing close to 70% by weight behavior and a larger region of ductile behavior, ab-
to the overall clay content (Kale, 2009). sorbing much energy before failure, it is considered
In addition to conventional logs, ECS logs were also ductile. In contrast, if the material under stress has a
acquired in the area of the study, revealing important larger region of elastic behavior but only a smaller re-
vertical and lateral mineralogy variations. The ECS gion of ductile behavior, the rock is considered brittle.
technique measures relative elemental yield based on The measurement of stored energy before failure is
neutron-induced capture GR spectroscopy, detecting known as brittleness, and it is a complex function of
silicon (Si), iron (Fe), calcium (Ca), sulfur (S), titanium rock strength, lithology, texture, effective stress, tem-
(Ti), gadolinium (Gd), chlorine (Cl), barium (Ba), and perature, fluid type (Handin and Hager, 1957, 1958;
hydrogen (H), but not magnesium (Mg). The ECS tool Handin et al., 1963; Davis and Reynolds, 1996), diagen-
sends neutrons into the wellbore wall, while a detector esis, and TOC (Wells, 2004). Brittleness is defined by
measures the counts and energy spectrum from the re-
leased GRs. The algorithm combines the resulting spec-
trum with other logs such as the bulk density and
photoelectric factor, among others, to interpret the
most likely mineralogy composition of the rock.
In our study, the ECS results have been calibrated
with several cored wells in the area. Figure 1 shows
the mineralogy ternary plot corresponding to a repre-
sentative well in the area of study where the sum of
all the clay, quartz, and calcite minerals is displayed
on each individual vertex of the ternary plot, indicating
that the mineral distribution along the wellbore agrees
with Karastathis’ (2007) and Kale’s (2009) previous find-
ings. Other authors (Sondergeld et al., 2010b; Slatt,
2011) have used similar mineralogy log tools to charac-
terize shale reservoirs. Table 1 summarizes the mean
weight mineralogy composition of the main formations
computed from ECS data corresponding to seven wells
in the area of study. The ECS data indicate that the Figure 1. A representative ternary mineralogy distribution
plot of quartz, carbonate, and clay for the Marble Falls Lime-
Lower Barnett Shale has higher quartz content than stone, Upper Barnett Shale, Forestburg Limestone, and the
the Upper Barnett Shale, agreeing with core measure- Lower Barnett Shale from ECS log measurements correspond-
ments made by Kale (2009) and Karastathis (2007). ing to well A (the well location is shown in Figure 10c).

researchers for different purposes. Hetenyi (1966) deposition and the origin and habit of mineral rock
fines brittleness as the lack of ductility. Ramsey components (such as quartz and calcite, and/or the type
(1967) states that when the internal cohesion of rocks of cement in the rock). To estimate a reliable and robust
is broken, the rocks are said to be brittle. Obert and Du- reservoir brittleness measurement, it is necessary
vall (1967) define brittleness as a property of materials therefore to combine conventional well logs with direct
that rupture or fracture with little or no plastic flow. measurements of geomechanical properties, such as
Several of the more commonly used brittleness def- Young’s modulus E and Poisson’s ratio ν (Grieser
initions neglect geologic factors such as the rock com- and Bray, 2007; Rickman et al., 2008).
Table 1. Fractional mean weight mineralogy composition of the formations from seven elemental capture
spectroscopy well logs in the area of study, where MF Lm corresponds to the Marble Falls Limestone, UB Sh to
the Upper Barnett Shale, F Lm to the Forestburg Limestone, LB Sh to the Lower Barnett Shale, and V Lm to the
Viola Limestone Formations.
Formation Quartz wgt. (%) Pyrite wgt. (%) Illite wgt. (%) Calcite wgt. (%) Dolomite wgt. (%) Clay wgt. (%) TOC wgt. (%)
MF Lm 0.000 0.000 0.067 0.925 0.000 0.064 0.000

UB Sh 0.142 0.007 0.148 0.334 0.000 0.219 0.150
F Lm 0.001 0.000 0.065 0.844 0.000 0.090 0.000
LB Sh 0.339 0.016 0.189 0.116 0.000 0.229 0.111
V Lm 0.016 0.000 0.044 0.886 0.000 0.054 0.000
Figure 2. GR, GRPs, and mineralogy logs corresponding to well A (tracks 1, 2, and 3, respectively). Track 4 shows the ECS log
corresponding to well A indicating that the mineral distribution along the wellbore agrees with Karastathis’ (2007) and Kale’s
(2009) findings, whereas tracks 5, 6, and 7 show the individual mineralogy log results decomposed into clay, quartz, and calcite
mineral content, respectively. The BI logs are calculated using Jarvie et al. (2007) (track 9) and Wang and Gale (2009) (track 10).
Track 11 shows the classification results from Figure 3d, where brittle (red) and more brittle zones (orange) are associated with
high quartz and TOC content zones.

The brittleness index tal organic carbon content, and Cly is the clay content
The BI is a relative measurement that depends on the by weight in the rock.
field of study and the purpose of the investigation (Al- Jarvie et al.’s (2007) equation 2 estimates BI using
tindag and Guney, 2010). One common BI measure is quartz in the numerator, and the sum of quartz, clay,
the ratio of compressive strength σ c to tensile strength and calcite in the denominator. In contrast, Wang and
σ t (Coates and Parsons, 1966; Aubertin and Gill, 1988; Gale’s (2009) equation 3 extends this formulation by in-
Baron, 1992; Aubertin et al., 1994; Ribacchi, 2000; Ha- cluding dolomite as a contributor to brittleness in the
jiabdolmajid and Kaiser, 2003): numerator, and TOC and dolomite in the sum of con-
stituent minerals of the rock in the denominator.
σc We calculate the BI using Jarvie et al.’s (2007) and
BI ¼ : (1)
σt Wang and Gale’s (2009) equations using the ECS log
data points and show the results in Figure 2, tracks 9
Because tensile strength and compressive strength and 10, respectively. Comparing both BI indices with
are measured only in the laboratory, it is difficult to ex- the mineralogy logs (Figure 2 track 4), we observe that
tend this definition to the reservoir scale. The higher the the zones with high quartz and calcite content are more
magnitude of the BI, the more brittle the rock. brittle than the regions with high clay content, which
Jarvie et al. (2007) and Wang and Gale (2009) pro- are less brittle (ductile).
pose BI definitions based on the mineral composition In the absence of dolomite (Table 1), Jarvie et al.’s
of the rock, dividing the most brittle minerals by the (2007) and Wang and Gale’s (2009) BI results differ
sum of the constituent minerals in the rock sample, con- subtly because Wang and Gale (2009) includes TOC
sidering quartz (and dolomite, in the case of Wang and in the equation, increasing the ductility of the rock. Be-
Gale, 2009) as the more brittle minerals: cause TOC in some zones is close to 10% (wgt.), we will
use Wang and Gale’s (2009) equation in the remaining
Qz analyses.
BIJarvieð2007Þ ¼ ; (2)
Qz þ Ca þ Cly We crossplot GR versus BI for all the formations in
Figure 3a, showing that the shale formations (Upper
and and Lower Barnett Shale) exhibit moderate to high-
GR and BI values. Exceptionally high-GR values
Qz þ Dol (>200 API) correspond to thin layers containing highly
BIWangð2009Þ ¼ ; (3)
Qz þ Dol þ Ca þ Cly þ TOC radioactive phosphatic nodules and concretions (Singh,
2008). Counterintuitively, this crossplot shows a posi-
where Qz is the fractional quartz content, Dol is the do- tive correlation between GR and BI. As expected, lime-
lomite content, Ca is the calcite content, TOC is the to- stone formations (Marble Falls Limestone, Forestburg
Figure 3. (a) GR versus BI corresponding to

well A (using Wang and Gale’s [2009] equa-
tion) differentiating the Marble Falls Lime-
stone (cyan), Upper Barnett Shale (gray),
Forestburg Limestone (navy), Lower Barnett
Shale (green), and Viola (blue). (b) GR versus
BI corresponding to the Upper and Lower Bar-
nett Shale color coded by TOC content, and
(c) Singh (2008) Barnett Shale lithofacies
definition ranked in relation to interpreted rel-
ative bottom oxygenation and organic rich-
ness. (d) Brittle (red), less brittle (orange),
less ductile (yellow), and ductile (green) clas-
sification proposed (classification results are
shown in track 12 in Figure 2).

Limestone, and Viola Limestone) show low-GR and low- tive bottom oxygenation and its organic richness. She
BI values. Notice that the Viola Limestone formation concludes that the phosphatic, siliceous noncalcareous
(light blue) exhibits a subtle positive variation in BI re- mudstone, and siliceous calcareous mudstone lithofa-
lated to clay content in the limestone formation. cies exhibit higher TOC, in contrast to the silty shale
Color-coding the previous crossplot with TOC con- and reworked shelly deposits, which exhibit lower TOC
tent reveals that high TOC values correspond to high (Figure 3c). Based on these depositional constraints
brittleness and high-GR values (Figure 3b), conflicting TOC and quartz are correlated such that high-TOC
with equation 3. It is important to highlight that TOC is a zones are more brittle in our study area.
direct measurement of organic richness, and its preser- In Figure 3d, we break the data population into four
vation depends on the bottom oxygen levels at the time equal petrotypes, setting the BI between 0 and 0.16 as
of deposition (Singh, 2008). Oxygenated strata are char- ductile (green), between 0.16 and 0.32 as less ductile
acterized by bioturbation and benthic activity, which (yellow), between 0.32 and 0.48 as less brittle (orange),
can be responsible for lowering the TOC. Based on this, and greater than 0.48 as brittle (red). We plot the clas-
Singh (2008) defines a relationship between the posi- sification results on track 11 on Figure 2. Examining
tion of the described lithofacies in relation to the rela- the GR log along with the ductile/brittle classification
Figure 4. (a) Set of elastic logs corresponding to well A, (b) Poisson’s ratio versus Young’s modulus crossplot indicating em-
pirically defined ductile-brittle regions, and the expected fracture pathway geometry (modified from Grieser and Bray, 2007),
(c) the Poisson’s ratio versus Young’s modulus values corresponding to formations in well A overplotted by Grieser and Bray’s
(2007) ductile (green)-brittle (red) regions color coded with BI from ECS mineralogy analysis. Classification results are shown in
track 13 in panel (a).
Figure 5. (a) Poisson’s ratio versus Young’s modulus crossplot corresponding to each formation in the study area. (b) Poisson’s
ratio versus Young’s modulus crossplot corresponding to Upper and Lower Barnett Shale color coded by BI, overlapped by a
proposed brittle/ductile classification and (c) the proposed classification.

results based on BI, we observe that the mid-lower part where E is Young’s modulus, and E min and E max are the
of the Lower Barnett Shale and the middle section of minimum and maximum Young’s modulus measured in
the Upper Barnett are the more brittle zones because the logged formation, and
of their higher quartz content in the vertical section.
ν − νmax
Also, we identify from the log several thin ductile layers νbrittleness ¼ ; (5)
(green) within the shale formation related to high clay νmin − νmax
mineral content.
where ν is Poisson’s ratio, and νmax and νmin are the
maximum and minimum values of Poisson’s ratio

The brittleness average
logged in the formation.
The term brittleness average is proposed by Grieser
Finally, they define a brittleness average, BA, as
and Bray (2007) as an empirical relationship between
Poisson’s ratio and Young’s modulus to differentiate E brittleness þ νbrittleness
ductile from brittle regions. They hypothesize that duc- BA ¼ : (6)
2
tile rocks exhibit a low Young’s modulus and high Pois-
son’s ratio, whereas brittle rocks exhibit a moderate to We use the elastic logs corresponding to well A (Fig-
high Young’s modulus and low Poisson’s ratio. Grieser ure 4a) to compute Poisson’s ratio and Young’s modulus
and Bray (2007) normalize Young’s modulus and Pois- logs and display the results on tracks 8 and 9 in Fig-
son’s ratio by their ranges, resulting in scaled elastic ure 4a. We calculate Young’s modulus brittleness and
parameters: Poisson’s ratio brittleness using equations 4 and 5
and display the results on tracks 10 and 11 in Figure 4a.
E − Emin Finally, we calculate the brittleness average (using
E brittleness ¼ ; (4)
Emax − Emin equation 6) and display the results on track 12 in
Figure 6. GR, V P , V S , velocity ratio, and density logs corresponding to the same representative well, and its acoustic and elastic
impedance λρ and μρ logs.

Table 2. The λρ and μρ values of the three most Figure 4a. Notice the differences in brittleness average
common minerals in the Barnett Shale (calcite, clay, in the formations, showing that the lower section is
and quartz) using the moduli, densities, and velocities
published by Mavko et al. (2009). more brittle than the upper section of the Lower Barnett
Shale, and the upper section of the Upper Barnett Shale
is more brittle than the lower section. The brittleness
Mineral λρ (GPa) (g/cm3) μρ (GPa) (g/cm3) average (track 12 in Figure 4a) is somewhat smoother
than the BI (track 2 in Figure 4a) because the ECS log
Quartz 20.32 116.60
tool has a finer resolution (≈1.5 ft) than the sonic and
Clay 48.45 22.95
dipole sonic log (≈3 ft).
Calcite 149.41 88.08
Poisson’s ratio versus Young’s
modulus brittleness crossplots
Crossplotting Poisson’s ratio versus Young’s modu-
lus, the shales with low average BI values fall into
the ductile (green) region proposed by Grieser and Bray
(2007) (Figure 4b). Using this template, we plot Pois-
son’s ratio and Young’s modulus corresponding to all
the formations corresponding to well A, color coded us-
ing the BI calculated previously. Shales with high BI fall
into the brittle region (red) agreeing with Grieser and
Bray’s (2007) definition. We plot the classification re-
sults in track 13 in Figure 4a. Notice that a cloud of
low-BI (ductile) points (color coded in purple) falls into
the brittle region, generating a conflict. To clarify these
conflicting results, we classify the shale and limestone
formations in terms of its Poisson’s ratio and Young’s
modulus in Figure 5a. The Upper and Lower Barnett
Figure 7. The λρ–μρ values corresponding to solid pure min-
erals of quartz, calcite, and clay are calculated using rock lab- Shale exhibit Poisson’s ratio ranging from 0.14 to
oratory measurements published by Mavko et al. (2009). 0.36 and Young’s modulus from 20 to 50 GPa.
Arrows represent the expected change of λρ and μρ with in- Wang and Gale’s (2009) BI definition assumes that
creasing porosity (ϕ), fluid content, and crack density (t). quartz and dolomite are the most brittle minerals in
Figure 8. The λρ–μρ crossplot (a) color

coded by GR, (b) color coded by TOC,
(c) color coded by BI (extracted from miner-
alogy) overplotted by a proposed brittle/duc-
tile classification, and (d) the proposed BI
classification.

the rock. Those rocks with low negligible amounts of present λρ–μρ lithology templates, which have been
these two minerals, such as limestone or shaly lime- very useful for reservoir characterization and correlat-
stones, will be considered to be ductile. Limestone for- ing brittleness to rock properties. Goodway et al. (2010)
mations are not a completion target in this study. They use these templates to estimate the minimum closure
do not have hydrocarbon potential but rather serve as a stress from seismic estimates of λρ and μρ.
frac barrier. For these reasons, they are displayed in P-wave sonic, dipole sonic, and density logs were
white in Figure 4a, track 13. Although Grieser and used to calculate acoustic impedance, elastic imped-
Bray’s (2007) crossplot is useful in the absence of min- ance, λρ, and μρ logs corresponding to well A (Figure 6).
eralogy logs, it fails to differentiate between brittle We also compute and plot the λρ and μρ of the three
quartz-rich shales and ductile limestones. Given the most common minerals in the Barnett Shale: calcite,
transition between shale and limestone through the clay, and quartz (Table 2, see Figure 7) using the
Upper Barnett Limestone, the direct use of the template moduli, densities, and velocities published by Mavko
shown in Figure 4b is problematic. et al. (2009). Connecting the three vertices of each min-
eral generates a mineralogy ternary plot in the λρ–μρ
The λρ versus μρ brittleness crossplots space (Figure 7). These calculations assume zero poros-
Goodway et al. (1997) demonstrate that it is possible ity (ϕ ¼ 0); introducing cracks and pores is nontrivial
to extract lithology and pore fluid information from and is left for future investigation. As porosity (ϕ), fluid
λρ to μρ crossplots derived from surface seismic and content, and crack density (t) increase, the results will
well logs. Goodway et al. (1997) and Perez (2011) mimic a triangular form but the values will be shifted
Figure 9. (a) GR, density, V P , and V S logs

corresponding to well B (the well location
is shown in Figure 10c), its corresponding
(b) Poisson’s ratio versus Young’s modulus
crossplot, and (c) λρ–μρ crossplot overlapped
by the proposed brittle/ductile template.

horizontally along the x-axes as lines of constant μρ. Brittleness prediction and calibration
This shift is caused by the gas saturation effects that High-quality long-offset surface seismic data were
decrease the values of λρ toward the origin (Goodway acquired after more than 400 wells were drilled and
et al., 2006). Berryman et al. (1999, 2002) report similar completed in the area of study. Using commercial soft-
results and created several λρ–μρ crossplots discrimi- ware, we simultaneously inverted the data to estimate
nating porosity and saturation effects. We expect that λρ and μρ following the workflow presented by Good-
the lateral displacement rate is not the same between way et al. (1997).
clastic and carbonate rocks. Usually, clastic rocks have Microseismic mapping is a passive seismic technique
a higher crack density than carbonates; in contrast, that records events generated as a consequence of proc-
cracks in carbonates tend to be wider and larger. esses within the reservoir such as fluid flow and hy-
Using the λρ–μρ template described previously for draulic fracturing (McGillivray, 2005; Daugherty and
the three most common minerals in the area of study Urbancic, 2009; Noe, 2011). Microseismic data are used
as a reference, the λρ–μρ well log results were cross- to map fracture growth (Waters et al., 2009) and have
plotted and color coded by the GR values, resulting been useful in mapping the effectiveness of the
in high-GR regions (shale) and low-GR values (lime- hydraulic fracturing job (e.g., Simon, 2005). Typical
stone) (Figure 8a). As discussed previously, high-TOC microseismic data products include the approximate
zones (Figure 8b) are depositionally related to high coordinates in space, local time of occurrence, and
quartz content and to high BI. We will use the template magnitude of each event. Using microseismic events
in Figure 8c to predict BI from λρ–μρ estimates made Refunjol et al. (2012) find that fracture-prone rocks are
from surface seismic data. consistently characterized by a specific range of P- and
Because well A is located out of the seismic survey S-impedance, suggesting that properties extracted from
area, the same workflow discussed previously is repeated surface seismic properties can be useful in highlighting
for well B. The corresponding well log results are shown zones where the rock is prone to failure during hy-
in Figure 9a. An ECS well log is not available in well B. draulic stimulation.
We therefore use the same brittle-ductile regions in the Event locations corresponding to wells C and D
λρ–μρ defined from well A in Figure 9b and 9c. are shown in Figure 10a and 10b, respectively. The
Figure 10. Map view of microseismic event locations corresponding to (a) well C and (b) well D; the orientation of the fracture
lineaments formed by the microseismic events align with the current maximum horizontal stress direction in the Fort Worth Basin
(northeast–southwest). (c) Horizon slice along the top Viola Limestone through the most positive curvature seismic attribute
volume. Most of the microseismic event locations fall into the areas with negative curvature values (bowl shapes). Red vectors
indicate velocity anisotropy in which the length of the vector is proportional of the degree of anisotropy, and the direction indicates
the azimuth of maximum anisotropy (modified from Thompson, 2010). The seismic data were acquired after 400 wells were stimu-
lated, such that the velocity anisotropy represents the postfrac stress regime. The location the cored and ECS shown in Figure 2 is
indicated as well A, northeast of this seismic survey.

hydraulic fracture job started at the toe and ended at positive or into a negative curvature zone. Because the
the heel of each well. In both cases, the map view shows seismic survey was acquired after the reservoir was in-
that some of the events preferentially aligned (dashed tensively fractured, we interpret the velocity anisotropy
lines) with the current northeast–southwest maximum seen in Figure 10c to be correlated to the open induced
horizontal stress in the Fort Worth Basin (Heidbach fracture network (H. Lynn, personal communication,
et al., 2009). 2010). Zhang (2010) finds the azimuthal anisotropy to
In the case of well C, most of the microseismic events be highly variable, but locally organized.
extend to the southwest of the well at stage 1, in contrast Figure 11a shows the lateral view of microseismic
to stage 2, in which the events are more centered about events about well C, while Figure 11b shows the depth
the wellbore. For stage 1 in well D, most of the events are histograms at each stage revealing that most of the
northeast of the wellbore. Notice from Figure 10c that events are concentrated in the Lower Barnett Shale,
microseismic events corresponding to well C and well close to the wellbore. More microseismic events are re-
D trend toward negative values of the most positive cur- corded in stage 2 than in stage 1. Stage 2 may have been
vature (green) indicating bowls, avoiding the positive influenced by the fractures generated by stage 1. The
values (orange) indicating ridges and domes, exhibiting previous map view (Figure 10a) reveals that the event
the same behavior described by Browning (2006) and locations corresponding to stage 1 are confined into a
Thompson (2010). Browning (2006) observes that micro- smaller region than those from stage 2.
seismic event locations occur more often in negative In addition to mineralogy and elastic parameters,
curvature zones, whether the well was drilled onto a there are other parameters that can affect brittleness,
Figure 11. (a) Lateral view and (b) microseismic vertical (depth) histograms, GR, BI log, and BI classification (brittle [red], less
brittle [orange], less ductile [yellow], and ductile [green]) corresponding to individual stages of microseismic event locations cor-
responding to well C. Vertical histogram shows a decrease in events recorded in the upper section of the Lower Barnett Shale
toward the Forestburg Limestone, possibly due to the increment in clay minerals and therefore ductility, creating a ductile zone.

such as the degree of interbedding (Sierra et al., 2010) that the rock preferentially fractures in the more brittle
and the scale of stratigraphy (Slatt and Abousleiman, regions.
2011). Sierra et al. (2010) and Slatt and Abousleiman
(2011) report that the degree of lamination plays a Limitations
significant role in rock strength. Laboratory measure- Shales are anisotropic. The BI to elastic parameters
ments concluded that core-plug-sized samples of lami- template is based on P- and S-wave sonic measurements
nated shale break more easily when stress is applied carried out in vertical wells, which measures the
parallel and less easily when the stress is applied per- vertical component of elastic impedance. The horizontal
pendicular to the lamination orientation (Slatt, 2011). component of velocity (and hence impedance) is not
The vertical histogram and BI calculation using elastic measured at all. Surface seismic inversion is sensitive
logs and the template in Figure 8c indicate a decrease in to vertical components of impedance at near angle
events toward the Forestburg Limestone, correspond- incidence. Anisotropic effects influence reflections at
ing to an increase in clay minerals and ductility (Fig- larger angles of incidence. Although a more accurate
ure 11b). The decrease in microseismic events might BI to elastic parameters template should account
be associated to the increase in interbedding in the for anisotropy, simple modeling experiments discussed
top section of the Lower Barnett Shale, where we ob- in Perez (2011) show that ignoring anisotropy for a lim-
serve an increment in the clay content in these thin ited angle (θ < 30°) prestack impedance inversion is
layers. As the clay content increases, it begins to sup- acceptable.
port neighboring grains and impacts the velocities, cre- Furthermore, in this research, the ECS logs do not
ating a ductile zone in the reservoir. indicate any dolomite content in the seven wells, which
To validate the brittle-ductile regions results, we ex- disagrees with observations by Kale (2009), Karastathis
tract the λρ and μρ values at each microseismic event (2007), and Singh (2008), with Singh (2008) reporting
and plot them against our brittleness template. Note dolomitic mudstone lithofacies with 30%–40% of dolo-
that most of the points fall into the area that we defined mite embedded in a clay matrix. Dennie (2010) and
to be more brittle (Figure 12), validating our hypothesis Dennie et al. (2012) find evidence of postdepositional
Figure 12. Extracted λρ–μρ from each micro-

seismic event location corresponding to
(a) well C and (b) well D (well locations
are shown in Figure 10c). (c) The λρ–μρ cross-
plot from the area around the well. Comparing
the microseismic events distribution to those
in panels (a and b) shows that most of micro-
seismic events occurs in the area that we de-
fine as brittle (red) and less brittle (orange).

alteration in authigenic minerals, the carbonate miner- The influence of kerogen in brittleness
als (e.g., calcite, dolomite, siderite, and ankerite) being The scanning electron microscopy (SEM) image
the most common authigenic phases in the Barnett from a Barnett Shale core (Figure 13) shows that the
Shale. ECS and other standard mineralogy log tools kerogen is located inside the grains in the sample,
are calibrated to detect the presence of silica (SiO2 ), and it exhibits quasicircular pores (Sondergeld et al.,
but they cannot deduce its origin. Using petrographic 2010a). At a high pressure and temperature, we do
methods including secondary electron, backscattered not expect to find quasicircular pores. This circular
electron, and cathodoluminescence imaging, Papazis structure suggests that the grains around the intragra-
and Milliken (2005) identify detrital and authigenic nular kerogen support the stress. In this example, a high
quartz in the Barnett Shale. Detrital quartz makes TOC does not significantly imply high ductility and the
up to 40% of the quartz in the formation, and it kerogen does not affect the elastic properties of the
occurs as silt-size grains and particles aggregated rock. Slatt (2011) describes and classifies a variety of
within the tests of agglutinated foraminifera. Also, au- pore types that exist in the Barnett and the Woodford
thigenic quartz is found in veins, biologically derived Shale, and Slatt and Abousleiman (2011) conclude that
replacements of skeletal debris (radiolaria), sealed frac- different pore types significantly influence the geome-
tures within individual detrital grains, and as cement chanical properties.
within agglutinated foram tests (Papazis and Milli- Lucier et al. (2011) conclude that the gas saturation
ken, 2005). decreases the V P ∕V S ratio in unconventional shales,
Other studies have analyzed the relationship be- which leads to a lower Poisson’s ratio calculation.
tween brittleness and the mineralogical composition. The particularity about the Barnett Shale is that the ker-
In the case of the Austin chalk, the grain rimming ogen is encapsulated in the pores of the rock, which
versus pore filling correlates with different mechanical does affect the mechanical properties of the rock. Given
properties and natural fracture attributes, even though the granular support and the correlation of TOC with
the modal composition is the same (Laubach et al., quartz in this study area, the effect of TOC is minimized
2009). such that high TOC rocks are “brittle” stratigraphically.
Figure 13. Backscattered SEM image of an ion-milled Barnett Shale sample imaged and ion milled in a dual-beam SEM (modified
from Sondergeld et al. [2010a] and courtesy of OU MPGE Integrated Core Characterization Center). Larger silt (orange) and calcite
(yellow) grains are mixed with clay (magenta) particles, and intragranular dark objects are interpreted as kerogen (white). The
smaller circular darker objects within the kerogen are pores (cyan).

In this special case scenarios we found that those rocks Bowker, K. A., 2003, Recent developments of the Barnett
that are more “frackable” are the ones that contain a Shale play, Fort Worth Basin: West Texas Geological
higher amount of TOC. Society Bulletin, 42, 4–11.
Browning, D. B., 2006, Investigating correlations between
Conclusions microseismic event data, seismic curvature, velocity
The BI and brittleness average are commonly used to anisotropy, and well production in the Barnett Shale,
define brittleness in quartz-rich shales. In the case of the Fort Worth Basin, Texas: Master’s thesis, The University
Barnett Shale, brittleness is dominated by quartz, of Oklahoma.

whereas ductility is dominated by clay content and cal- Bustin, M. R., 2006, Geology report: Where are the high-po-
cite. Although we anticipated that log and core descrip- tential regions expected to be in Canada and the US?
tion with the two more popular brittleness estimation Capturing opportunities in Canadian shale gas: Pre-
techniques would show a consistent brittleness estimate, sented at The Canadian Institute’s 2nd Annual Shale-
this was not the case. In particular, the brittleness aver- Gas Conference.
age does not accurately predict the ductile behavior of Coates, D. F., and R. C. Parsons, 1966, Experimental crite-
limestones. We therefore modified a previous brittleness ria for classification of rock substances: International
concept to construct a template based on mineralogy Journal of Rock Mechanics and Mining Sciences and
and elastic parameters measured in the study area. This Geomechanics Abstracts, 3, 181–189, doi: 10.1016/
template allows us to use a single brittleness for all lith- 0148-9062(66)90022-2.
ologies in the target area: brittle shales, ductile shales, Daugherty, J., and T. Urbancic, 2009, Microseismic moni-
and ductile limestone fracture barriers. toring of a carbon sequestration field test: Frontiers and
Our template is validated through a suite of micro- innovation: Presented at Frontiers + Innovation —
seismic experiments, which shows most events occur- CSPG CSEG CWLS Convention.
ring in brittle zones, fewer events in the ductile shale, Davis, D., and S. J. Reynolds, 1996, Structural geology of
and fewer events still in the limestone fracture barriers. rocks and regions, 2nd ed.: Wiley.
Deacon, R. J., 2011, Barnett provides future returns: E&P
Acknowledgments Magazine, 2011, 71–75.
We would like to thank Devon Energy for providing Dennie, D., R. D. Elmore, J. Deng, E. Manning, and J. Pan-
the seismic and well data and for the financial support nalal, 2012, Paleomagnetism of the Mississippian Bar-
to complete this project, along with Schlumberger (Pet- nett Shale, Fort Worth Basin, Texas: Journal of the
rel) and CGG Veritas (Hampson-Russell) for providing Geological Society, London, 371, 89–106, doi: 10
the software licenses Petrel and Strata to the University .1144/SP371.10.
of Oklahoma. Dennie, D. P., 2010, An integrated paleomagnetic and dia-
genetic investigation of the Barnett Shale and underly-
References ing Ellenburger Group carbonates, Fort Worth Basin,
Altindag, R., and A. Guney, 2010, Predicting the relation- Texas: Ph.D. dissertation, ConocoPhillips School of
ships between brittleness and mechanical properties Geology and Geophysics.
(UCS, TS and SH) of rocks: Scientific Research and Es- Ellis, M. A., S. E. Laubach, P. Eichhubl, J. E. Olson, and P.
says, 5, 2107–2118. Hargrove, 2012, Fracture development and diagenesis
Aubertin, M., and D. E. Gill, 1988, Une methodologie of Torridon Group Applecross Formation, near
d’evaluation du potentiel de coups de terrain dans les An Teallach, NW Scotland: Millennia of brittle deforma-
mines d’Abitibi: Colleque sur le Controle de Terrain tion resilience?: Journal of the Geological Society,
(AMMQ), 47–77. London, 169, 297–310, doi: 10.1144/0016-76492011-086.
Aubertin, M., D. E. Gill, and R. Simon, 1994, On the use of Gale, J. F., R. M. Reed, and J. Holder, 2007, Natural frac-
the brittleness index modified (BIM) to estimate the post- tures in the Barnett Shale and their importance for hy-
peak behaviour of rocks: in P. P. Nelson, and S. E. Lau- draulic fracture treatments: AAPG Bulletin, 91, 603–
bach, eds., Proceedings of the 1st North American Rock 622, doi: 10.1306/11010606061.
Mechanics Symposium, AA Balkema, 945–952. Gao, C., 2011, Multiscale ensemble Kalman filter for
Baron, L. I., 1992, Determination of properties of rocks: history matching and characterization of Barnett
Gozgotekhizdat (in Russian). Shale: M.S. thesis, Mewbourne School of Petroleum
Berryman, J. G., P. A. Berge, and B. P. Bonner, 1999, Role and Geological Engineering: The University of
of λ-diagrams in estimating porosity and saturation from Oklahoma.
seismic velocities: 69th Annual International Meeting, Goodway, B., T. Chen, and J. Downton, 1997, Improved
SEG, Expanded Abstracts, 176–179. AVO fluid detection and lithology discrimination using
Berryman, J. G., P. A. Berge, and B. P. Bonner, 2002, Es- Lamé parameters λρ, μρ and μλ fluid stack from P- and S-
timating rock porosity and fluid saturation using only inversion: 67th Annual International Meeting, SEG, Ex-
seismic velocities: Geophysics, 67, 391–404, doi: 10 panded Abstracts, 183–186.
.1190/1.1468599.

Goodway, B., M. Perez, J. Varsek, and C. Abaco, 2010, Seis- Karst Waters Institute, Special Publication, vol. 14,
mic petrophysics and isotropic-anisotropic AVO meth- 130–135.
ods for unconventional gas exploration: The Leading Lucier, A., R. Hofmann, and L. Bryndzia, 2011, Evaluation
Edge, 29, 936–944. of variable gas saturation on acoustic log data from the
Goodway, B., J. Varsek, and C. Abaco, 2006, Practical ap- Haynesville Shale gas play, NW Louisiana, USA: The
plications of P-wave AVO for unconventional gas re- Leading Edge, 30, 300–311, doi: 10.1190/1.3567261.
source plays — I: Seismic petrophysics and isotropic Mavko, G., T. Mukerji, and J. Dvorkin, 2009, The rock phys-
AVO: CSEG Recorder, 31, 90–95. ics handbook: Cambridge University Press.
Grieser, B., and J. Bray, 2007, Identification of production McGillivray, P., 2005, Microseismic and time-lapse seismic
potential in unconventional reservoirs: Presented at monitoring of a heavy oil extraction process at Peace
SPE Production and Operations Symposium. River, Canada: CSEG Recorder, 5, 5–9.
Hajiabdolmajid, V., and P. Kaiser, 2003, Brittleness of rock Montgomery, S. L., 2005, Mississippian Barnett Shale, Fort
and stability assessment in hard rock tunneling: Tunnel- Worth Basin, North-Central Texas: Gas-shale play with
ing and Underground Space Technology, 18, 35–48, doi: multi-trillion cubic foot potential: AAPG Bulletin, 89,
10.1016/S0886-7798(02)00100-1. 155–175, doi: 10.1306/09170404042.
Handin, J., and R. V. Hager, 1957, Experimental deforma- Noe, J., 2011, Microseismic frac mapping: Moving beyond
tion of sedimentary rocks under confining pressure: the dots from an engineering perspective: Presented at
Tests at room temperature on dry samples: AAPG Bul- OGS Shales Moving Forward Workshop, Oklahoma
letin, 41, 1–50. Geological Society.
Handin, J., and R. V. Hager, 1958, Experimental deforma- Obert, L., and W. I. Duvall, 1967, Rock mechanics and the
tion of sedimentary rocks under confining pressure: design of structures in rock: Wiley.
Tests at high temperature on dry samples: AAPG Bulle- Papazis, P. K., and K. Milliken, 2005, Cathodoliminiscent
tin, 42, 2892–2934. textures and the origin of quartz in the Mississippian
Handin, J., R. V. Hager, M. Friedman, and J. N. Feather, Barnett Shale, Fort Worth Basin, Texas: Presented at
1963, Experimental deformation of sedimentary rocks the AAPG Annual Convention and Exhibition.
under confining pressure: Pore pressure tests: AAPG Perez, M., 2011, Developing templates for integrating quan-
Bulletin, 47, 717–755. titative geophysics and hydraulic fracture completions
Heidbach, O., M. Tingay, A. Barth, J. Reinecker, D. Kurfe, data: Part I — Principles and theory: 81st Annual
and B. Muller, 2009, The world stress map based on the International Meeting, SEG, Expanded Abstracts,
database release 2008, equatorial scale 1:46,000,000: 1794–1798.
Commission for the Geological Map of World. Perez, R., 2009, Quantitative analysis of gamma ray para-
Hetenyi, M., 1966, Handbook of experimental stress analy- sequences in the Barnett Shale: M.S. thesis, ConocoPhil-
sis: Wiley. lips School of Geology and Geophysics: The University
Jarvie, D. M., 2003, Evaluation of unconventional natural of Oklahoma.
gas prospects: The Barnett Shale fractured shale gas Ramsey, J. G., 1967, Folding and fracturing of rocks:
model: Presented at 21st International Meeting on Or- McGraw-Hill.
ganic Geochemistry. Refunjol, X. E., 2010, Hydraulically-induced microseismic
Jarvie, D. M., R. J. Hill, T. E. Ruble, and R. M. Pollastro, fracture characterization from surface seismic attrib-
2007, Unconventional shale-gas systems: The Mississip- utes and seismic inversion: A North Texas Barnett Shale
pian Barnett Shale of North-Central Texas as one model case of study: M.S. thesis, ConocoPhillips School of
for thermogenic shale-gas assessment: AAPG Bulletin, Geology and Geophysics.
91, 475–499, doi: 10.1306/12190606068. Refunjol, X. E., K. M. Keranen, J. H. LeCalvez, and K. J.
Kale, S., 2009, Petrophysical characterization of Barnett Marfurt, 2012, Integration of hydraulically induced
Shale play: M.S. thesis, Mewbourne School of Petroleum microseismic events location with active seismic
and Geological Engineering: The University of Oklahoma. attributes: A North Texas Barnett Shale case of study:
Karastathis, A., 2007, Petrophysical measurements on tight Geophysics, 77, no. 3, KS1–KS12, doi: 10.1190/geo2011-
gas shale: M.S. thesis, Mewbourne School of Petroleum 0032.1.
and Geological Engineering: The University of Oklahoma. Ribacchi, R., 2000, Mechanical tests on pervasively jointed
Laubach, S. E., J. E. Olson, and M. R. Gross, 2009, Mechani- rock material: Insight into rock mass behaviour: Rock
cal and fracture stratigraphy: AAPG Bulletin, 93, 1413– Mechanics and Rock Engineering, 33, 243–266, doi:
1426, doi: 10.1306/07270909094. 10.1007/s006030070002.
Loucks, R. G., 2008, Origin and modification of Lower Rickman, R., M. Mullen, E. Petre, B. Grieser, and D. Kun-
Ordovician Ellenburger Group paleokarst breccias dert, 2008, A practical use of shale petrophysics for
and fractures in Central and West Texas, in I. D. stimulation design optimization: All shale plays are
Sasowsky, C. T. Feazel, J. E. Mylroie, A. N. Palmer, not clones of the Barnett Shale: Presented at SPE An-
and M. V. Palmer, eds., Karst from recent to reservoirs: nual Technical Conference and Exhibition.

Saldungaray, P., and T. Palish, 2012, Hydraulic fracture Wells, F., 2004, A new method to help identify unconven-
optimization in unconventional reservoirs: World Oil, 233. tional targets for exploration and development through
Schmoker, J. W., 1981a, Organic content of Devonian shale integrative analysis of clastic rock properties: Houston
in western Appalachian Basin: AAPG Bulletin, 64, 2156– Geological Society Bulletin, 47, 34–49.
2165. Zhang, K., 2010, Seismic attribute analysis of unconven-
Schmoker, J. W., 1981b, Determination of organic matter tional reservoirs, and stratigraphic patterns: Ph.D. dis-
content of Appalachian Devonian shales from gamma- sertation, ConocoPhillips School of Geology and
ray logs: AAPG Bulletin, 65, 1285–1298. Geophysics: The University of Oklahoma.
Sierra, R., M. H. Tran, Y. N. Abousleiman, and R. M. Slatt,
2010, Woodford Shale mechanical properties and im-
pacts of lithofacies: Presented at 44th U.S. Rock Me- Roderick Perez is currently working
chanics Symposium and 5th U.S.–Canada Rock as a technical service consultant for
Mechanics Symposium, ARMA10-461. DrillingInfo — Transform Software
Simon, Y., 2005, Stress and fracture characterization in a and Services, Inc., in Houston, Texas.
shale reservoir, North Texas, using correlation between Born in Caracas, Venezuela, he com-
pleted his undergraduate degree in
new seismic attributes and well data: M.S. thesis, De-
geophysics engineering at the Univer-
partment of Geosciences, University of Houston. sidad Simon Bolivar (Caracas, Ven-
Singh, P., 2008, Lithofacies and sequence stratigraphic ezuela) in 2007, after spending a
framework of the Barnett Shale: Ph.D. dissertation, year as an exchange student at the University of Oklahoma
ConocoPhillips School of Geology and Geophysics: (OU) during the fall of 2005 to the spring of 2006. He com-
The University of Oklahoma. pleted his M.S. in geology at OU in 2008 under the direction
Slatt, R. M., 2011, Important geological properties of of Dr. Slatt, and he completed his Ph.D. in geophysics
unconventional resource shales: Central European Jour- under the guidance of Dr. Marfurt in 2013. His dissertation
nal of Geoscience, 3, 435–448, doi: 10.2478/s13533-011- research focuses on seismic attributes applied to reservoir
0042-2. characterization of unconventional reservoirs, with results
Slatt, R. M., and Y. N. Abousleiman, 2011, Merging se- additionally calibrated with available microseismic data.
He has authored several articles with his two advisors,
quence stratigraphy and geomechanics for unconven-
and his dissertation is under review for publication as a
tional gas shales: The Leading Edge, 30, 274–282, doi: book. He has also reviewed several papers for Interpreta-
10.1190/1.3567258. tion, among other geoscience magazines. As part of a
Slatt, R. M., and N. R. O’Brien, 2011, Pore types in the Bar- group effort, He won first place in the 2008 AAPG Imperial
nett and Woodford gas shales: Contribution to under- Barrel Award (IBA) competition along with his four team-
standing gas storage and migration pathways in fine- mates representing the OU. He also was the recipient of the
grained rocks: AAPG Bulletin, 95, 2017–2030, doi: 10 SEG Scholar–Charles C. McBurney Memorial Award 2010–
.1306/03301110145. 2011, among other scholarships. In addition to his aca-
Sondergeld, C. H., R. J. Ambrose, C. S. Rai, and J. Mon- demic duties as a full-time research assistant at OU, he par-
crieff, 2010a, Micro-structural studies of gas shales: Pre- ticipated and accepted leadership roles in several
sented at SPE Unconventional Gas Conference. extracurricular activities on campus, such as president
of the OU–SEG student chapter (2009–2010), vice-
Sondergeld, C. H., K. E. Newsham, J. T. Cominsky, M. C.
president (2007–2008), and secretary (2006–2007) of the
Rice, and C. S. Rai, 2010b, Petrophysical considerations Association Friends of Venezuela. He is the IBA–Latin
in evaluating and producing shale gas resources: Pre- America Coordinator for Venezuela, Mexico, and Trinidad
sented at SPE Unconventional Gas Conference. and Tobago, and he participates in several committees in
Thompson, A. M., 2010, Induced fracture detection in the AAPG and SEG. He continues to stay informed on ad-
Barnett Shale, Fort Worth Basin, Texas: M.S. thesis, vancements in the oil and gas community and always looks
ConocoPhillips School of Geology and Geophysics. forward to learning about new data sets and finding new
Thomsen, L., 1986, Weak elastic anisotropy: Geophysics, ways to provide value back value back to companies with
51, 1954–1966, doi: 10.1190/1.1442051. which he collaborates.
Walper, J. L., 1982, Plate tectonic evolution of the Forth Kurt Marfurt began his geophysical
Worth Basin: Dallas Geological Society. career as an assistant professor teach-
Wang, F. P., and J. F. W. Gale, 2009, Screening criteria for ing mining geophysics at Columbia
shale-gas systems: Gulf Coast Association of Geological University’s Henry Krumb School of
Societies Transactions, 59, 779–793. Mines in New York. After five years,
Waters, G., H. Ramakrishnan, J. Daniels, D. Bently, J. Bel- he joined Amoco at its Tulsa Research
hadi, and D. Sparkman, 2009, Utilization of real time Center. Through successive reorgan-
microseismic monitoring and hydraulic fracture diver- izations at Amoco, Marfurt obtained
diverse experience in seismic model-
sion technology in the completion of Barnett Shale
ing, migration, signal analysis, basin analysis, seismic-
horizontal wells: Presented at OTC. attribute analysis, and multicomponent analysis. Through

Amoco, he won five patents, two in seismic coherence Marfurt joined the University of Oklahoma as the Frank
technology. Marfurt joined the University of Houston in and Henrietta Schultz Chair and Professor of Geophysics.
1999 as a professor in the Department of Geosciences His recent work has focused on applying coherence, spec-
and as director of the Center for Applied Geosciences tral decomposition, structure-oriented filtering, and volu-
and Energy, where his primary emphasis is on develop- metric curvature to mapping fractures and karst as well
ment and calibration of new seismic-attribute technology. as attribute-assisted processing.

t Technical papers
Identification of brittle/ductile areas in unconventional reservoirs using

seismic and microseismic data: Application to the Barnett Shale
Roderick Perez Altamar1 and Kurt J. Marfurt1
Abstract
Brittleness in unconventional reservoirs is mainly controlled by mineralogy, and it increases with quartz and
dolomite content, whereas an increase in the clay content represents an increase in ductility. To generate
regional brittleness maps, we have correlated the mineralogy-based brittleness index to elastic parameters mea-
sured from well logs. This correlation can then be used to predict the brittleness from surface seismic elastic
parameter estimates of λρ and μρ. We applied the workflow to a 3D seismic survey acquired in an area where
more than 400 wells were drilled and hydraulically fractured prior to seismic acquisition. Combining λρ and μρ
into a single 3D volume allowed the combination of both attributes into a single 3D volume, which can be con-
verted to brittleness using a template based on the well log and core data. Neither of these seismic estimates
were direct measures of reservoir completion quality. We, therefore, used production logs and extracted surface
seismic estimates at microseismic event locations to analyze the completion effectiveness along several hori-
zontal wellbores in the reservoir. We defined four petrotypes in λρ and μρ space depending on their brittleness
and gas saturation, and we found that most of the microseismic events fell into the zone described as brittle in
the λρ-μρ crossplots. These observations supported the well-known idea that regardless of where the well was
perforated, microseismic events appeared to preferentially grow toward the more brittle areas, suggesting the
growth of hydraulic fractures into the brittle petrotype.
Introduction et al. (2012) observe that microseismic event locations

The economic success of shale plays depends not in the Barnett Shale occurred in areas of low seismic
only on the formation thickness and the total organic impedance. Rutledge and Phillips (2003) observe a cor-
carbon (TOC) content of the reservoir but also on the relation between shear activation of fractures and low
completion quality. Because of their negligible perme- seismic impedance.
ability, almost all shale reservoirs need to be hydrauli- Perez (2010a, 2010b, 2011) demonstrates that using
cally fractured. Successful hydraulic fracturing requires elastic rock physics parameters, such as Young’s
targeting the most brittle rocks in which the fractures modulus E and Poisson’s ratio ν or alternatively Lamé
can be induced, thereby draining the highest amount parameters λ and μ can characterize a reservoir geome-
of gas from the rock. chanically. Most of the engineering literature (e.g.,
The study area includes a high-quality 3D seismic sur- Grieser and Bray, 2007) estimates geomechanical
vey that was acquired after more than 400 wells had been behavior using Young’s modulus and Poisson’s ratio.
drilled over a 10-year period using different stimulation Because density is difficult to measure from surface
fluids (gel, water, and surfactants), and varying numbers seismic data, Sharma and Chopra (2012) use ρE and
of stages, in vertical and horizontal wells. ν, resulting in 3D volumetric estimates of geomechani-
Poststack seismic attributes, such as coherence and cal behavior.
curvature, are routinely used to avoid drilling geoha- Although hydraulically induced fractures can propa-
zards, such as faults and karst features, which can gate through ductile and brittle rocks, the goal of the
connect to nearby water-bearing formations. Although proppant is to better preserves induced fractures in
curvature attributes may not illuminate individual frac- more brittle rocks. Alzate (2012) combines production
tures, they do quantify the amount of strain (Nelson, logs and microseismic data with simultaneous seismic
2001). Thompson (2010) and Zhang (2010) show that inversion attributes to analyze completion effectiveness
in the Barnett Shale seismic impedance, curvature, along four horizontal wells in the Barnett Shale. He con-
and other attributes can be visually correlated with mi- cludes that the hydraulically generated fractures prefer-
croseismic events and reservoir performance. Refunjol entially grow toward the more brittle rock, generating a
1
ConocoPhillips School of Geology and Geophysics, Norman, Oklahoma, USA. E-mail: [email protected]; [email protected].
Manuscript received by the Editor 1 March 2013; published online 19 August 2015. This paper appears in Interpretation, Vol. 3, No. 4 (November
2015); p. T233–T243, 13 FIGS.
http://dx.doi.org/10.1190/INT-2013-0021.1. © 2015 Society of Exploration Geophysicists and American Association of Petroleum Geologists. All rights reserved.

difference in gas production. Because the 3D seismic P-wave sonic and density logs provide a direct measure
survey was acquired after more than 400 wells were of Lamé parameters λ and μ at the well (Goodway,
drilled and completed in the area, we expect that the 2007).
findings show evidence of the intensive hydraulic frac- Compressional and shear velocities V P and V S can be
turing. written in terms of Lamé’s parameters λ and μ and bulk
Jarvie et al. (2007) and Wang and Gale (2009) pro- density ρ:
pose brittleness index (BI) definitions based on the min- pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
eral composition of the rock, dividing the most brittle V P ¼ ðλ þ 2μÞ∕ρ (1)
minerals by the sum of the constituent minerals in

the rock sample, considering quartz (and dolomite, in and
the case of Wang and Gale, 2009) as the more brittle pffiffiffiffiffiffiffiffi
mineral. However, it is very important to highlight that VS ¼ μ∕ρ: (2)
the term BI is a relative measurement that depends on
the field of study and the purpose of the investigation
(Altindag and Guney, 2010). Seismic reflections are sensitive to changes in P- and
The proposed methodology integrates different S-impedances, Z P and Z S . We can therefore estimate
tools, such as petrophysics, well log analysis, and seis- Lamé impedances (moduli-density relationships) from
mic simultaneous inversion, and it is a continuation of surface seismic data as follows:
the research done by Perez (2013) and Perez and Mar-
furt (2014) who described brittle/ductile template based μρ ¼ ðρV S Þ2 ¼ Z 2S ; (3)
on the mineralogy log measurements and elastic param-
eters, using the same area of study and data set in the and
Barnett Shale. The Barnett Shale is well known to be
an organically rich and thermally mature formation de- λρ ¼ ðρV P Þ2 − 2ðρV S Þ2 ¼ Z 2P − 2Z 2S : (4)
posited during Mississippian time in the Fort Worth Ba-
sin. It is described to be an unconventional reservoir
because of its low average permeability (70 nD) and Prestack simultaneous inversion estimates P-wave
porosity (6%) distributed in a variety of depositional fa- impedance, S-wave impedance, and density, which in
cies (Deacon, 2011). turn can be used to predict lithology and geomechanical
Prestack simultaneous inversion was performed to behavior. Following Goodway et al. (1997) and Good-
estimate λρ and μρ seismic volumes in an effort to way (2007), we estimate the P- and S-reflectivities from
detect and highlight brittle and ductile regions in an un- prestack seismic angle gathers using Fatti et al.’s (1994)
conventional reservoir. The 2D color bars and interac- method. Then, we estimate P- and S-impedances, Z P
tive 2D crossplotting technology are used to estimate and Z S , and finally calculate the Lamé impedances λρ
geomechanical behavior from λρ and μρ estimated from and μρ using equations 3 and 4.
surface seismic inversion. Wells are used with micro- A representative vertical slice (north–south) through
seismic experiments and surface seismic estimates of λρ and μρ seismic volumes (Figure 1a and 1b) shows the
λρ-μρ to quantify damaged rock. At each microseismic difference between shale and limestone formations.
location, we extract the corresponding and μ values and Limestone formations exhibit a higher λρ (related to
crossplot the results using a 2D color bar, providing a incompressibility) and μρ (related to rigidity) than the
link between discrete interactive crossplotting and the shale formations, which show low λρ and low μρ. The
continuous variability of the data. Finally, we visually Viola Limestone exhibits a higher μρ than the Marble
correlate anisotropy intensity volumes with microseis- Falls and Forestburg Limestones in the section indicat-
mic event locations and production wellbore mea- ing that it is more rigid than the surrounding shale for-
surements. mations, confirmed by microseismic measurements
that show the Viola to be an excellent fracture barrier.
Seismic inversion The gamma ray log (Figure 1c) reveals low gamma
In an isotropic, linear elastic medium, only two elas- ray values in the limestone formations, and high gamma
tic constants (such as Lamé parameters λ and μ) are ray values in the shale formations. Notice the increase
necessary to completely specify the stress-strain rela- in gamma ray values at the middle of the Lower Barnett
tion. Calibrated by mineralogy logs, these elastic param- Shale formation and its correlation with the decrease in
eters can be used empirically to predict whether a λρ and μρ (white arrow). We interpret the high gamma
reservoir will deform plastically (for a “ductile” rock) ray values to be from an increase in the amount of ra-
or cataclastically (for a “brittle” rock). dioactive minerals (clay). Singh (2008) finds the TOC to
Lamé’s incompressibility parameter λ relates uni- be a function of oxygen levels, exhibiting an increase
axial and lateral strain to uniaxial stress. The λ is pri- toward the bottom of each parasequence.
marily a longitudinal measure and hence “orthogonal” To set the template theoretical limits, Perez and
to Lamé’s rigidity parameter μ a quantity that relates Marfurt (2014) compute λρ and μρ in the three most
shearing stress to strain. Dipole sonic coupled with common minerals in the Barnett Shale: calcite, clay,

and quartz. Connecting the three vertices of each min- on its limestone (magenta, blue, and purple), quartz
eral generates a mineralogy ternary plot in λρ-μρ space. (yellow and red), and clay (green) content, and there-
We crossplot λρ on the x-axis and μρ on the y-axis, fore its geomechanical behavior.
using a 2D color bar with 4096 colors (64 × 64) to color Figure 4 shows six stratal slices (stratal slice loca-
code the output volume. To be consistent with the pre- tions are shown in white dashed lines in Figure 3c)
vious scale ranges, we set the initial λρ and μρ axes to through crossplotted λρ versus μρ seismic volumes cor-
range between 0 and 160 GPa g∕cm3 (Figure 2a). responding to each formation in this study. Limestone
However, the 2D histogram shows that the data do formations, such as Marble Falls Limestone, Forestburg
not extend through this entire range (Figure 2b); to bet- Limestone, and Viola Limestone exhibit magenta, blue,
ter use the whole spectrum of colors, we therefore clip and purple colors, respectively. In contrast, shale for-
the λρ values from 20 to 100 GPa g∕cm3 , and μρ from mations, such as the Upper Barnett Shale and the Lower
10 to 90 GPa g∕cm3 as shown in Figure 3d. Barnett Shale (upper and lower sections) are shown in
Figure 3a and 3b shows representative slices (east– green, yellow, and red, respectively. This plot enhances
west) through λρ and μρ seismic volume, respectively. the mineralogical composition between the upper and
Figure 3c shows the same vertical slice through cross- lower section of the Lower Barnett Shale, indicating
plotted λρ versus μρ seismic volumes using the 2D color that the upper section of the Lower Barnett Shale has
bar as shown in Figure 3d. This type of attribute allows a higher quartz content than the lower section of the
the mineralogical discrimination of the reservoir based same formation, which exhibits a higher clay content.
Figure 1. Vertical slices A-A′ through (a) λρ and (b) μρ seismic volumes and their corresponding histograms. Notice that the shale
formations exhibit lower values of λρ and μρ (red and yellow) than the limestone formations (cyan and blue). Location of the line is
shown in Figure 4a, and (c) λρ-μρ crossplot color coded by gamma ray from logs indicating that shale formations exhibit low λρ and
low μρ (Perez and Marfurt, 2014). (d) Gamma ray versus BI indicating that in the Barnett Shale high gamma ray values represent
high brittleness and TOC, confirming the core analysis by Singh (2008).

This 2D color bar is a useful tool to interpret geome- stratal slices corresponding to different layers in the
chanical properties in the reservoir. Lower Barnett (Figure 6). Clipping the color bar enhan-
We then isolate the λρ and μρ values corresponding ces the horizontal and vertical mineralogy variation of
to the Lower Barnett Shale section and generate a spe- the formation. In the upper section of slice B exhibits a
cific 2D color bar and histogram displayed in Figure 5a higher quartz content than in the south, and at the same
and 5b. This clipped color bar is used to display four time, slice D exhibits a higher limestone content be-
cause it is closer to the Viola Limestone
formation. From previous analysis, low
λρ and low μρ indicate less brittle zones,

and low λρ and medium to high μρ indi-
cate more brittle zones. Using the tem-
plate described in Perez (2013), we
use commercial software to generate
stratal slices similar to those in Figure 7.
Analysis and discussion

Microseismic event analysis
Microseismic monitoring techniques
are based on the same principles as earth-
quake seismology. A complete analysis
includes the detection, location, and esti-
mation of magnitude and moments of
the microearthquakes induced by hy-
draulic fracturing and reservoir depletion
Figure 2. (a) The 2D color bar and its equivalent in 1D, using 4096 (64 × 64) processes (Warpinski et al., 2005). Micro-
colors, and (b) 2D histogram corresponding to the λρ and μρ values from the seismic data can be used to evaluate ef-
entire seismic section. The data range is selected to be consistent with figures fectiveness of completion designs and
presented previously. The white rectangle represents the clipped color bar used map the development of fracture patterns
in Figure 3 to fit all the data. in the reservoir.
Figure 3. Vertical slices B-B′ through (a) λρ and (b) μρ seismic volumes and (c) through the crossplotted λρ versus μρ volumes
using a (d) 2D color bar. The location of line B-B′ is shown in Figure 4a. The range of the 2D color bar enhances the differences
between quartz- (yellow and red), clay- (green), and limestone-rich (magenta, blue, and purple) formations, providing an estimate
of lithology and geomechanical behavior.

During the hydraulic fracturing process, water and mic events are not distributed uniformly along the
any other injected materials (usually sand) are pumped wellbore.
into the wellbore and into the formation through holes To validate the previous seismic inversion results,
that have been perforated through the well-bore casing. we extract the λρ and μρ values at each microseismic
The water and other injectables squeeze
into the available spaces in the rock un-
til the rock fails. This failure state is
reached quicker when the rock is con-
sidered brittle (commonly rocks with a

high quartz content), resulting in a more
effective hydraulic fracture job.
In the Barnett Shale, it is well under-
stood that shear slippage occurs prefer-
ably on preexisting planes of weakness
as a result of the change in stress and
pore pressure that are induced by the
fracturing process (Scholz, 1968, 1990).
Paleotectonic events in the Fort Worth
Basin resulted in a northwest–southeast
main stress field orientation. However,
the current regional maximum stress di-
rection in the basin is northeast–south-
west, with local deviations in intensity
and direction due to the presence of the
main fault system in the basin, the Min-
eral Wells–Newark East fault system,
and other minor faults. Given this stress
regime in the Barnett Shale, we expect
that the fractures generated from most
of the hydraulic stimulation process are
going to have preferred orientation,
northeast–southwest.
Eleven wells (nine horizontal and two
vertical) with microseismic data are
available in the area of study. At the time
a fracture occurs S- and P-waves are Figure 4. Stratal slices through λρ versus μρ crossplot volumes corresponding
emitted, which, after processing, provide to (a) Marble Falls, (b) Upper Barnett Shale, (c) Forestburg Limestone, (d) upper
an estimated position, time, and magni- Lower Barnett Shale, (e) lower Lower Barnett Shale, and (f) Viola Limestone
tude of each event (Bennett et al., 2005). using the 2D color bar indicated in Figure 3d. Stratal slice locations are shown
in white in Figure 3c. Limestones appear as magenta, blue, and purple, whereas
Because microseismic data analysis is
quartz-rich shales appear as yellow and red, and clay-rich shales appear as green.
susceptible to problems associated with
sensor placement bias, Alzate (2012) ap-
plies several procedures to quality con-
trol and filter the microseismic data to
avoid issues related to the attenuation
of events at large distances.
Most the horizontal wells were drilled
along a northwest–southeast azimuth to
generate a northeast–southwest-trend-
ing fracture pattern, using a different
number of stages in each well. Figure 8
shows the direction of four horizontal
wells with their microseismic events
color coded by stage number. In gen-
eral, microseismic data show that most
of the area around the wellbore was Figure 5. (a) The 2D color bar using 4096 colors and (b) 2D histogram corre-
stimulated. However, low activity around sponding to λρ and μρ values from the Lower Barnett Shale section. The white
the toe of well C indicates that microseis- rectangle represents the clipped color bar used in Figures 6, 8, and 11.

event for three wells and display the results in a cross- Using the same surface seismic data examined in this
plot in Figure 9. Then, we calculate a histogram for λρ article, Zhang (2010) calculates the P-wave azimuthal
and μρ at each stage. In addition, we superimpose in the anisotropy intensity and represents microseismic event
background λρ-μρ voxels falling within boxes in Fig- clouds with polygons, exhibiting a high compartmental-
ure 8, and its corresponding histogram in dark gray. ized at the reservoir level, with the compartment edges
Even when the background cloud of voxels is scattered being defined by structural features (Figure 10). Zhang
along the plot, a greater population of the microseismic (2010) calculates the P-wave azimuthal anisotropy
events falls in the zone that we defined as a less brittle using a model based acoustic impedance inversion,
zone (orange) and brittle (red). which is tightly coupled to the geology by wells, remov-
ing most of the seismic wavelet and
thin-bed tuning effect. Notice that the
microseismic events occur in the low
anisotropy zones, which is the opposite
of what we expected. These observations
suggest that good hydraulic fracturing
zones are causing multiple induced frac-
ture orientations in the reservoir. These
observations are consistent with the
study made by Thompsen (2010) in the
same area.
These low-anisotropy zones suggest
that effective hydraulic fracturing re-
laxes the rock, thereby locally reducing
the anisotropy. Bowker (2007) assumes
the Barnett Shale as an overpressured
and fully saturated (in terms of sorption)
state. Before the reservoir is drilled and
Figure 6. Four stratal slices corresponding to the Lower Barnett Shale indicat- hydraulically stimulated, it is at a state
ing the location of the limestone- (magenta and blue), quartz- (yellow and red), of equilibrium, with a balanced pore
and clay-rich shales (green) regions using the 2D color bar as shown in Figure 5a. pressure gradient (3.58 KPa∕ft in the
The location of each slice is shown in Figure 4a. gas saturated part of the play) by the
Figure 7. The same four stratal slices presented in Figure 6, corresponding to the lower Barnett Shale indicating the location of
the more brittle and less brittle regions using the BI template in the λρ-μρ space calculated from a well with mineralogy logs 16 mi
northeast from the area of study (Figure 12b). Location of each slice is shown in Figure 4a.

capillary pressure of the rock (Bowker, 2007). When going to be different due to the fractures generated hy-
the rock is hydraulically stimulated the equilibrium is draulically.
disturbed, and gas diffuses from the matrix into the hy- Figure 11 shows that the well locations of more 400
draulic fractures, thereby flowing to the wellbore, caus- wells had already been acquired, and the microseismic
ing the relaxation of the rock. events corresponding to zones that had been stimulated
We expect that the fractures generated hydraulically (green). This indicates that there are still zones in which
create a (local) fracture set perpendicular to the regional there exists the possibility of having bypass pay areas.
reservoir stress field, causing an attenuation in the We observe that the population of microseismic events
anisotropy intensity. The change in the fracture geom- is higher in the more brittle areas because the fractures
etry caused a change in the anisotropy intensity, in some propagates to the direction of the less stresses exist and
cases from a vertical transverse isotropic (VTI) sym- in which they need to do the least amount of work.
metry to orthorhombic resulting in a stronger anisotropy
signature. However, we assume that after the rock is in- Calibration with production logs
tensively fractured, it goes from strong horizontal trans- Perez et al. (2011) present “heuristic” rock physics
verse isotropic (HTI) media to orthorhombic exhibiting templates that can be used to correlate rock composi-
more isotropy, concluding that the velocity anisotropy is tion with λρ and μρ and production, with the goal of
estimating expected ultimate recovery
in unconventional reservoirs. Alzate
(2012) and Alzate and Devegowda
(2013) define four petrofacies, each rep-
resenting 25% of the data, drawing iso-
Poisson’s ratio lines assuming that the
variation in the Young’s modulus along
these lines is a consequence of the TOC
and porosity (Figure 12a). The brittle
and rich (red) petrofacies represent
areas in the Barnett Shale with low Pois-
son’s ratio and low Young’s modulus,
the ductile and rich (yellow) petrofacies
are those regions with high Poisson’s ra-
tio and low Young’s modulus, the brittle
and poor (green) petrofacies represent
those shales with low Poisson’s ratio
Figure 8. Four microseismic wells (well locations shown in Figure 10) plotted and high Young’s modulus, and the duc-
on the top of a stratal slice color coded using the 2D color bar (low right). The red
colors represent quartz-rich areas, and therefore they are more brittle than the tile and poor (blue) petrofacies are those
green areas, which correspond to more ductile zones. Notice that most of the shales with high Poisson’s ratio and high
microseismic events are located in the more brittle region, avoiding the ductile Young’s modulus.
zones.
Figure 9. Crossplot in gray of λρ-μρ of falling voxels within boxes as shown in Figure 8 for the Lower Barnett Shale.

Continuing the work done by Perez and Marfurt (2014). At the same time, BI is the ratio between the ten-
(2014), using mineralogy well logs measurements, BI sile strength and compressive strength. Because these
was calculated. Then, we proceed to classify the data measurements can only be measured at the laboratory,
population in terms of BI into four petrotypes (brittle, Jarvie et al. (2007) and Wang and Gale (2009) propose BI
less brittle, less ductile, and ductile). definitions based on the mineral composition of the
Brittleness can be defined as the measurement of rock. Both authors divide the most brittle minerals by
the stored energy before failure, and it depends on the the sum of the constituent minerals in the rock sample,
rock strength, lithology, texture, effective stress, among considering quartz (and dolomite, in the case of Wang
others parameters. Other brittleness definitions found and Gale, 2009) as the more brittle minerals.
in the literature were reviewed by Perez and Marfurt Using elastic logs, we generate a BI λρ-μρ as shown in
Figure 12b. The rock laboratory mea-
surements are published in Table A-1
by Mavko et al. (2009) (fluid content, di-
pole logs, and mineralogy logs). Finally,
we combine both templates into one
(Figure 12c) agreeing with the Alzate
(2012) and Alzate and Devegowda
(2013) interpretation. Notice that both
independent templates agree in limits
in which the data points exist between
the iso-Poisson’s ratio lines between ν ¼
0.15 and ν ¼ 0.34. Rock with Poisson’s
ratio greater than ν ¼ 0.23 is considered
brittle, and less than ν ¼ 0.23 is classi-
fied as ductile, in which the ν ¼ 0.23
limit was selected as the midpoint be-
tween both boundaries (ν ¼ 0.15 and
ν ¼ 0.34)
The study area includes four produc-
tion horizontal wells, which were re-
corded five months after a multistage
Figure 10. Anisotropy intensity and microseismic event locations from several hydraulic fracture treatment. Produc-
microseismic events in the Lower Barnett Shale (modified from Zhang, 2010).
Notice that none of the microseismic events occur in areas of high anisotropy.
tion logs measure flow and provide a
Because 400 additional wells have fractured the rock, we hypothesize that the quantitative measure of production from
area of high anisotropy (yellow and red) corresponds to bypassed pay. each perforation. Alzate (2012) and Al-
zate and Devegowda (2013) interpret
the temperature, differential gas produc-
tion, and gas and water hold up logs for
four wells.
Gas production rates from each per-
foration are plotted in a λρ-μρ crossplot
(Figure 13a and 13b). Alzate (2012) and
Alzate and Devegowda (2013) observe
that the most prolific zones show low
Poisson’s ratio and low Young’s modu-
lus corresponding to this brittle-rich
petrofacies. Figure 13b shows the well
trajectories of the four wells with pro-
duction logs and how they penetrate
different layers, which affects their per-
formance. Based on the microseismic
events, Alzate (2012) concludes that no
matter in which the well is completed,
the fracture will preferentially grow to-
ward the area with the most brittle rock,
Figure 11. Stratal slice C shown in Figure 6, with the well locations in the area
which explains the difference in gas pro-
of study using the color bar as shown in Figure 5, in which green represents clay- duction between the wells.
rich zones, red represents quartz-rich zones, and purple represents calcite-rich Geoscientist community in the oil
zones. and gas industry geomechanically clas-

Figure 12. (a) Lower Barnett reservoir quality classification based on the seismically inverted rock properties defining four pet-
rotypes (modified from Alzate, 2012), (b) λρ versus μρ from well logs (Perez and Marfurt, 2014), and (c) λρ versus μρ crossplot
template after combining panels (a) and (b) classifications. Notice that both templates agree in the region defined as more brittle
and with higher fluid content.
Figure 13. (a) Seismic λρ-μρ crossplots extracted along the wellbores corresponding to production logs. Each point indicates gas
rate at each individual perforation. (b) East–west vertical slides through the reservoir quality volume showing the location of the
wells having production logs (modified from Alzate, 2012).
sify a rock in terms of its Poisson’s ratio, defining a rock failure behavior. Surface seismic measurements of elas-
with low and high Poisson’s ratio as brittle or ductile tic properties λ and μ are made between 5 and 150 Hz,
rocks, respectively. However, in this specific Barnett whereas those in borehole are made at 10–20 kHz.
Shales case, our results indicates that we need to For this reason, direct seismic estimates of λ and μ
redefine these concepts calling brittle those rocks with are not direct measures of nonlinear behavior. However,
high Poisson’s ratio and ductile the rocks with low Pois- through the use of mineralogy logs, Perez (2013) showed
son’s ratio, contrary to the industry wide definitions. that these elastic measurements can be used to predict
Perez and Marfurt (2014) extensively describe the brit- inelastic behavior.
tleness and BI definitions available in the literature. In
this research, we calculate the BI in terms of the min- Conclusions
eralogical content of the rock, and not in terms of its Simultaneous inversion of surface seismic data not
geomechanical properties, using Jarvie et al. (2007) only differentiates shale from limestone but also brittle
and Wang and Gale (2009) 1definitions. and ductile shale intervals. The change from brittle to
Rocks are fractured at near static conditions (0 Hz) ductile is transitional based on the mineralogy logs and
and thus depend on static nonlinear elastic, plastic, and calibrated with microseismic data.

Cutoff from one zone to another is empirical, which Alzate, J. H., and D. Devegowda, 2013, Integration of sur-
is not a serious shortfall in fields with a hundreds of face seismic, microseismic, and production logs for
wells. The use of 2D color bars of λρ versus μρ helps shale gas characterization: Methodology and field appli-
to visualize and quantify the smooth transition between cation: Interpretation, 1, no. 2, SB37–SB49, doi: 10.1190/
brittle and ductile areas. INT-2013-0025.1.
Microseismic is an indirect measurement to interpret Bennett, L., J. LeCalvez, D. R. Sarver, K. Tanner, W. S. Birk,
how the fractures are distributed in the reservoir. We G. Waters, J. Drew, G. Michaud, P. Primiero, L. Eisner,
demonstrate that microseismic and production logs R. Jones, D. Leslie, M. J. Williams, J. Govenlock, R. C.
show that hydraulically induced fractures preferentially Klem, and K. Tezuka, 2005, The source for hydraulic
populate brittle regions. fracture characterization: Oilfield Review/Schlum-
Combining the results of the extracted seismic attrib-
berger, 17, 42–57.
utes and measured production logs, we can conclude
Bowker, K. A., 2007, Development of the Barnett Shale
that more brittle and more fractured the zone is the
Play, Fort Worth Basin: AAPG Search and Discovery
gas production is higher. Results showing zones in
Article 10126.
which the anisotropy intensity decreases in areas with
high concentration of wells are evidence that the seis- Deacon, R. J., 2011, Barnett provides future returns: E&P
mic response could be affected after an intense drilling Magazine, 7175, 71–75.
program. In addition, we can conclude that the simul- Fatti, J., G. Smith, P. Strauss, and P. Levitt, 1994, Detection
taneous inversion is a very effective tool to discriminate of gas in sandstone reservoirs using AVO analysis: A 3D
between brittle and ductile zones in unconventional res- seismic case history using the Geostack technique: Geo-
ervoirs. physics, 59, 1362–1376, doi: 10.1190/1.1443695.
In this specific case in the Barnett Shale, we are de- Goodway, B., 2007, A tutorial on AVO and Lamé constants
fining as brittle those rocks with high Poisson’s ratio for rock parameterization and fluid detection, http://
and ductile rocks with low Poisson’s ratio which is con- gsa.seg.org/pdf_forms/RecorderJune2001LMRAVO_
trary to the industry wide definitions for brittle and duc- new2007july.pdf, accessed 23 January 2013.
tile rocks, which assume a low Poisson’s ratio for brittle Goodway, B., T. Chen, and J. Downton, 1997, Improved
and high Poisson’s ratio for ductile rocks because the AVO fluid detection and lithology discrimination using
calculated brittleness is based on the rock mineralogi- Lamé parameters λρ, μρ and μλ fluid stack from P- and
cal content, and not on its geomechanical properties. S-inversion: 67th Annual International Meeting, SEG,
Expanded Abstracts, 183–186.
Acknowledgments Grieser, B., and J. Bray, 2007, Identification of production
We would like to thank Devon Energy for providing potential in unconventional reservoirs: Presented at
the seismic and well data and for the financial support SEP Production and Operations Symposium.
to complete this project, along with Schlumberger (Pet- Jarvie, D. M., R. J. Hill, T. E. Ruble, and R. M. Pollastro,
rel) and CGG (Hampson-Russell) for providing the soft- 2007, Unconventional shale gas systems: The Mississip-
ware licenses Petrel and Strata to the University of pian Barnett Shale of North-Central Texas as one model
Oklahoma. for thermogenic shale-gas assessment: AAPG Bulletin,
91, 475–499, doi: 10.1306/12190606068.
Appendix A Mavko, G., T. Mukerji, and J. Dvorkin, 2009, The rock phys-
ics handbook: Cambridge University Press.
Nomenclature Nelson, R. A., 2001, Geologic analysis of naturally frac-
Moduli, densities, and velocities of common tured reservoirs 2nd ed.: Gulf Professional Publishing.
minerals. Perez, M., D. Close, B. Goodway, and G. Purdue, 2011,
λ = Lamé’s incompressibility parameter Developing templates for integrating quantitative geo-
μ = Lamé’s rigidity parameter physics and hydraulic fracture completions data: Part
ρ = Lamé’s bulk density parameter I — Principles and theory: 81st Annual International
E = Young’s modulus
Meeting, SEG, Expanded Abstracts, 1794–1798.
v = Poisson’s ratio
Perez, R., 2010a, Determining brittleness in the Barnett
Shale using simultaneous impedance inversion: Pre-
References sented at Student Expo, AAPG-SEG.
Altindag, R., and A. Guney, 2010, Predicting the relation- Perez, R., 2010b, Application of LMR inversion and cluster-
ships between brittleness and mechanical properties ing analysis in the Barnett Shale: 80th Annual Inter-
(UCS, TS and SH) of rocks: Scientific Research and Es- national Meeting, SEG, Expanded Abstracts, 2236–2239.
says, 5, 2107–2118. Perez, R., 2011, Application of LMR and clustering analysis
Alzate, J. H., 2012, Integration of surface seismic, micro- in unconventional reservoirs: Presented at AAPG Geosci-
seismic, and production logs for shale gas characteriza- ence Technology Workshop, International Shale Plays.
tion: Methodology and field application: M.S. thesis, The Perez, R., 2013, Brittleness estimation from seismic mea-
University of Oklahoma. surements in unconventional reservoirs: Application

to the Barnett Shale: Ph.D. dissertation, The University under the direction of Roger Slatt, and he received a
of Oklahoma. Ph.D. (2013) in geophysics under the guidance of Kurt Mar-
Perez, R., and K. Marfurt, 2014, Mineralogy-based brittle- furt. He is currently working as a seismic interpretation
ness prediction from surface seismic data: Application specialist at Pacific Rubiales in Bogota, Colombia. His dis-
sertation research focuses on seismic attributes applied to
to the Barnett Shale: Interpretation, 2, no. 4, T255–T271,
reservoir characterization of unconventional reservoirs,
doi: 10.1190/INT-2013-0161.1.
with results additionally calibrated with available micro-
Refunjol, X. E., K. M. Keranen, J. H. LeCalvez, and K. J. Mar- seismic data. He has authored several articles with his
furt, 2012, Integration of hydraulically induced microseis-
two advisors, and his dissertation is under review for pub-

mic events location with active seismic attributes: A lication as a book. He has also reviewed several papers for
North Texas Barnett Shale case of study: Geophysics, Interpretation, among other geoscience magazines. As
77, no. 3, KS1–KS12, doi: 10.1190/geo2011-0032.1. part of a group effort, he won first place in the 2008 AAPG
Rutledge, J. T., and W. S. Phillips, 2003, Hydraulic stimu- Imperial Barrel Award (IBA) competition along with his
lation of natural fractures as revealed by induced micro- four teammates representing OU. He was also the recipient
earthquakes, Carthage Cotton Valley gas field, east of the SEG Scholar–Charles C. McBurney Memorial Award
Texas: Geophysics, 68, 441–452, doi: 10.1190/1.1567214. 2010–2011, among other scholarships. In addition to his
academic duties as a full-time research assistant at OU,
Scholz, C. H., 1968, Mechanism of creep in brittle rock:
he participated and accepted leadership roles in several
Journal of Geophysical Research, 73, 3295–3302, doi:
extracurricular activities on campus, such as president
10.1029/JB073i010p03295. of the OU SEG student chapter (2009–2010), vice president
Scholz, C. H., 1990, The mechanics of earthquakes and (2007–2008), and secretary (2006–2007) of the Association
faulting: Cambridge University. Friends of Venezuela. He is the IBA Latin America Co-
Sharma, R. K., and S. Chopra, 2012, New attribute for de- ordinator for Venezuela, Mexico, and Trinidad and Tobago,
termination of lithology and brittleness: 82nd Annual and he participates in several committees in AAPG and
International Meeting, SEG, Expanded Abstracts, doi: SEG. He continues to stay informed on advancements in
10.1190/segam2012-1389.1. the oil and gas community and always looks forward to
Singh, P., 2008, Lithofacies and sequence stratigraphic learning about new data sets and finding new ways to pro-
framework of the Barnett Shale: Ph.D. dissertation, vide value back to companies with which he collaborates.
The University of Oklahoma.
Thompson, A. M., 2010, Induced fracture detection in the
Kurt J. Marfurt began his geophysi-
Barnett Shale, Fort Worth Basin, Texas: M.S. thesis, The
cal career teaching geophysics and
University of Oklahoma.
contributing to an industry-supported
Wang, F. P., and J. F. W. Gale, 2009, Screening criteria for consortium on migration, inversion,
shale-gas systems: Gulf Coast Association of Geological and scattering (project MIDAS) at Co-
Societies Transactions, 59, 779–793. lumbia University’s Henry Krumb
Warpinski, N. R., R. C. Kramm, J. R. Heinze, and C. K. Walt- School of Mines in New York City. In
man, 2005, Comparison of single- and dual-array micro- 1981, he joined Amoco’s Tulsa Re-
seismic mapping techniques in the Barnett Shale: search Center and spent the next 18
Presented at SPE Annual Technical Conference and Ex- years doing or leading research efforts in modeling,
hibition, 95568. migration, signal analysis, basin analysis, seismic attribute
Zhang, K., 2010, Seismic attribute analysis of unconven- analysis, reflection tomography, seismic inversion, and
multicomponent data analysis. In 1999, he joined the Uni-
tional reservoirs, and stratigraphic patterns: Ph.D. dis-
versity of Houston as a professor in the Department of
sertation, The University of Oklahoma.
Geosciences and as director of Allied Geophysics Labora-
tories. He is currently a member of the Geophysical Soci-
eties of Tulsa and Houston, SEG, EAGE, AAPG, AGU, and
Roderick Perez Altamar received SIAM, and he serves as an assistant editor for GEOPHYSICS.
an undergraduate degree (2007) in His current research activity includes prestack imaging,
geophysics engineering from the Uni- velocity analysis and inversion of converted waves, com-
versidad Simon Bolivar (Caracas, puter-assisted pattern recognition of geologic features on
Venezuela), after spending a year as 3D seismic data, and interpreter-driven seismic processing.
an exchange student at the University His research interests include seismic signal analysis, 3D
of Oklahoma (OU) during the fall of seismic attributes, seismic velocity analysis, subsurface
2005 to the spring of 2006. He received imaging, and multicomponent data analysis.
an M.S. (2008) in geology from OU

Mineralogy Based Brittleness Prediction From Surface Seismic Data - Application On The Barnett Shale

Uploaded by

Copyright:

Available Formats

Mineralogy Based Brittleness Prediction From Surface Seismic Data - Application On The Barnett Shale

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mineralogy Based Brittleness Prediction From Surface Seismic Data - Application On The Barnett Shale

Uploaded by

Copyright:

Available Formats

t Technical papers

Mineralogy-based brittleness prediction from surface seismic data:

Introduction Although the Barnett Shale formation is capable of

Interpretation / November 2014 T255

T256 Interpretation / November 2014

Interpretation / November 2014 T257

MF Lm 0.000 0.000 0.067 0.925 0.000 0.064 0.000

T258 Interpretation / November 2014

Figure 3. (a) GR versus BI corresponding to

Interpretation / November 2014 T259

T260 Interpretation / November 2014

maximum and minimum values of Poisson’s ratio

Interpretation / November 2014 T261

Figure 8. The λρ–μρ crossplot (a) color

T262 Interpretation / November 2014

Figure 9. (a) GR, density, V P , and V S logs

Interpretation / November 2014 T263

T264 Interpretation / November 2014

Interpretation / November 2014 T265

Figure 12. Extracted λρ–μρ from each micro-

T266 Interpretation / November 2014

Interpretation / November 2014 T267

Barnett Shale, brittleness is dominated by quartz, of Oklahoma.

T268 Interpretation / November 2014

Interpretation / November 2014 T269

T270 Interpretation / November 2014

Interpretation / November 2014 T271

Identification of brittle/ductile areas in unconventional reservoirs using

Introduction et al. (2012) observe that microseismic event locations

Interpretation / November 2015 T233

minerals by the sum of the constituent minerals in

T234 Interpretation / November 2015

Interpretation / November 2015 T235

λρ and low μρ indicate less brittle zones,

Analysis and discussion

T236 Interpretation / November 2015

sidered brittle (commonly rocks with a

Interpretation / November 2015 T237

T238 Interpretation / November 2015

Interpretation / November 2015 T239

T240 Interpretation / November 2015

Interpretation / November 2015 T241

T242 Interpretation / November 2015

two advisors, and his dissertation is under review for pub-

Interpretation / November 2015 T243

You might also like