You are on page 1of 11

t

Special section: Interpretation for unconventional resources

An integrated seismic reservoir characterization workflow to predict


hydrocarbon production capacity in unconventional plays
Gorka Garcia Leiceaga1, Mark Norton2, and Jol Le Calvez1
Abstract
Seismic-derived elastic properties may be used to help evaluate hydrocarbon production capacity in unconventional plays such as tight or shale formations. By combining prestack seismic and well log data, inversionbased volumes of elastic properties may be produced. Moreover, a petrophysical evaluation and rock physics
analysis may be carried out, thus leading to a spatial distribution of hydrocarbon production capacity. The result
obtained is corroborated with the available well information, confirming our ability to accurately predict hydrocarbon production capacity in unconventional plays.

Introduction
Within the last decade, new technologies such as horizontal drilling and hydraulic fracturing have transformed
the potential of unconventional plays to a multibillion dollar industry. In addition, extracting hydrocarbons from
tight oil/gas formations is accompanied by geoscientific
and engineering challenges which are significantly different from the ones associated with conventional resource
exploration and production. The use of seismic-derived
rock properties to evaluate reservoir quality in unconventional plays has recently become a topic of interest,
which stems from the need to use more sophisticated
methods to identify high-producing resource plays. In
this paper, a seismic reservoir characterization (SRC)
workflow is presented where inversion-derived elastic
properties have been used to determine hydrocarbon
(tight-gas) production capacity in the Montney formation, northeast British Columbia, Canada. The Montney
formation is a stratigraphic unit of Lower Triassic age
and is mostly composed of low-permeability, highly
laminated organic clay, silt, and fine sand.
The workflow begins with seismic data conditioning, then prestack, simultaneous inversion for elastic
properties, followed by a petrophysical evaluation and
rock physics analysis to determine a spatial distribution of hydrocarbon production capacity. The result
obtained is corroborated with the available wells
drilled within the survey area. In addition, a volume of
seismic discontinuities representing natural fractures
and faults is used along with microseismic data to help
improve our understanding of the relationship between
reservoir production capacity and seismic-derived elastic properties.

Methology
The available data set (Figure 1a) used in our study
includes four conditioned seismic partial stacks (8
20, 2028, 2835, and 3545), horizontal and vertical wells (some with microseismic), and a volume of
natural fractures and faults (Pedersen et al., 2002) derived from poststack seismic data. A gas production
evaluation for the horizontal wells discussed in this
study is split into three zones of interest as shown in
Figure 1b. Zone 1 contains two medium-producing horizontal wells, Zone 2 includes two low-producing wells,
and Zone 3 encompasses three horizontal wells in close
proximity; one low, one medium, and one high-producer.
In addition, Zone 3 contains a vertical well with two perforated intervals with the shallow interval providing high
production, while the bottom interval produces much
less.
The work carried out in this SRC study has been
divided into four main components (Figure 2).

The seismic data set is conditioned to preserve


the amplitude versus offset (AVO) response while
flattening the events to adequately attenuate noise
without generating artifacts.
Prestack inversion for rock properties is carried
out using simulated annealing and composed of
a globally optimized algorithm which simultaneously inverts the angle stacks and their corresponding wavelets. The model parameterization
is based on the Aki and Richards (1980) approximation of the Zoeppritz equations.
A petrophysical evaluation on three available vertical wells is initially carried out to predict mineral

Schlumberger, PTS, Houston, Texas, USA. E-mail: ggarcia23@slb.com; jcalvez2@slb.com.


Progress Energy Corporation, Calgary, Canada. E-mail: mnorton@progressenergy.com.
Manuscript received by the Editor 13 February 2013; revised manuscript received 22 April 2013; published online 24 October 2013. This paper
appears in Interpretation, Vol. 1, No. 2 (November 2013); p. SB15SB25, 10 FIGS., 1 TABLE.
2

http://dx.doi.org/10.1190/INT-2013-0007.1. 2013 Society of Exploration Geophysicists and American Association of Petroleum Geologists. All rights reserved.

Interpretation / November 2013 SB15

fractions, porosity, total organic carbon (TOC),


and hydrocarbon saturation. Next, all wells are
split into three categories (high, medium, and
low) with respect to gas production. The obtained
petrophysical logs are used to build a relationship
between the elastic response of the subsurface
and reservoir production. The relationship is then
used to predict a spatial distribution of hydrocarbon production capacity.
The interpretation of results consists of comparing the obtained volume of hydrocarbon production capacity with the production of the existing
wells. The placement of these wells to date has
been primarily driven by targeting zones with relatively low Poissons ratio (PR), so this strategy is
examined in our analysis. Moreover, a volume of
seismic discontinuities representing natural fractures and faults is used along with microseismic
data to help improve our understanding of the relationship between reservoir production capacity
and seismic-derived elastic properties.

Seismic data conditioning


Seismic data conditioning is vital to properly invert
for rock properties, although it does not replace highquality seismic data acquisition and processing. The objective for conditioning seismic before inversion is to
ensure amplitude preservation while eliminating the
postprocessing residual effects such as anisotropy
and noise, which should make the seismic data more
AVO- and inversion-friendly.
The procedure must begin with an analysis of the
common midpoint (CMP) gathers for determining
how the data can be improved. The potential issues
may be residual multiples, linear noise, random noise,
gather flatness, and wavelet stretch. Only after determining which issues are present in the data can an adequate workflow be designed.
Figure 3 shows various types of conditioning enhancements which can be applied to improve the quality of seismic data. These conditioning processes are
here subdivided into two segments: noise attenuation
and gather flattening. A preliminary step may be the
analysis of S/N to identify whether the data have noise

Figure 2. Main components of the SRC workflow.

Figure 1. (a) Three-dimensional view of the project area displaying the drilled vertical and horizontal wells with associated mapped microseismic events, a time slice highlighting
the fracture and fault networks in the lower Montney formation as well as seismic and Poissons ratio cross sections.
(b) Seismic depth slice with fractures and faults showing
the horizontal wells discussed in this study along with mapped
microseismic hypocenters.
SB16 Interpretation / November 2013

Figure 3. Diagram showing the main types of improvements


which may be applied to CMP gathers for optimizing data quality in AVO inversion for rock properties.

problems which must be addressed. If noise is in fact an


issue, determining whether the noise is linear, random,
or comes from multiples may be achieved by analyzing
the gathers. For example, if the data exhibit residual
multiples, a weighted least squares radon transform
is effective in modeling the coherent residual multiples.
The application of such processes to model seismic
events by moveout discrimination will aid in reducing
interference from multiples with primary energy. Caution is required in this process given the risk of removing usable signal, especially at near offsets where the
moveout differentiation between primaries and multiples is difficult to achieve.
For interbed multiples, regular gap deconvolution
with zero phase correction is an option which can be
applied with a frequency bandwidth constraint. For
random noise issues, it is assumed that the stacking
process will reduce or suppress noncoherent noise,
although in specific environments such as carbonates,
there is low fold and low transmission which has the
effect of significantly lowering the S/N ratio. In this
situation, a random noise attenuation technique may
prove worthy for increasing the continuity of events.
Accurately predicting reservoir properties using simultaneous prestack inversion can be highly dependent
on the degree of flatness in seismic gather events. The
primary cause for the need to apply gather flattening
techniques to seismic data is residual moveout due to
anisotropy, where its effects typically appear as hockey
sticks in the farther offsets. If the data are stacked without correcting for residual moveout, output will be
smeared and lower in frequency content.
Another issue which may also pose a threat to attaining success in AVO inversion studies consists of
residual time shifts caused by unaccounted time shifts
among the traces. Trim statics was used in our study,
which is a commonly used technique that uses maximum crosscorrelation values with a model trace from
different computation windows. This process can dynamically shift data samples in time and preserve wavelet character across the entire offset range.
The removal of residual anisotropy and time shifts is
vital, especially in PP, PS, and time-lapse inversion,
where multiple independently acquired data sets (i.e.,
baseline and monitor surveys) will need to be aligned.
Seismic-driven characterization studies will usually
benefit from seismic data conditioning work. There
are instances where the job performed during the data
processing sequence is sufficient and no postprocessing
conditioning is required. This depends highly on the
quality of the processing algorithms available and the
experience level of the seismic processor.
Prestack seismic inversion
The use of seismic inversion is central in obtaining a
3D distribution of rock properties. By correlating the
seismic and well domains, sparse 1D borehole measurements are extrapolated into 3D space, thereby providing
a powerful tool to characterize reservoirs. Various inver-

sion methodologies exist to predict subsurface rock


properties, which can be applied to various types of seismic data including conventional, multicomponent, timelapse, crosswell, and azimuthal. The available outputs
from the inversion process depend on the log data type
available as well as whether there is prestack or poststack seismic to be used. The workflow discussed in this
paper focuses on deterministic, simultaneous AVO inversion with key inputs of prestack seismic data, extracted
wavelets, and three a priori models.
The primary inputs into the inversion kernel include
the seismic partial stacks, extracted wavelets and a priori models. In the model-based inversion approach, independent wavelets are extracted for each partial angle
stack and estimated from the correlation between the
well reflectivity and the seismic trace at the well location. Variations in amplitude, frequency, and phase between the different seismic input stacks are captured
by the wavelets so there is no need for scaling, phase
rotation, or frequency balancing of the seismic data.
The parameters which have the most impact on wavelet
quality include the log calibration into the time domain,
the method of extraction (e.g., amplitude and phase assumptions), the analysis window, and to some degree the
length of the wavelets. If multiple wells exist within the
seismic coverage area, a multiwell estimation approach
should be used to better capture the lateral variations in
acoustic and elastic properties of the subsurface. Moreover, time-varying changes in wavelet frequency due to
attenuation must also be taken into account. This attenuation factor may be determined by building an attenuation model derived from the seismic data.
The a priori or low-frequency model (LFM) input
data include the calibrated wells, seismic horizons,
and dip information obtained from the seismic data
(Rasmussen, 1999). The LFM is an integral part of
the inversion iteration process and although its contribution is a small part of a full-bandwidth inversion, its
role is crucial. Without a correct background trend, accurate rock property values cannot be obtained in the
inversion no matter how precisely the mid-to-high frequency information is predicted. The structural complexity of the study area and the number of available
wells dictate how the LFM is generated, because various methodologies exist for generating such models. If
enough spatially properly distributed well data are
available, geostatistical methods such as kriging, cokriging, sequential Gaussian simulation, and moving
average may be appropriate to extrapolate the low
frequencies from the wells using the stacking velocities
as a guide. In many reservoir characterization studies
with limited well information, an extrapolation algorithm where the weight of the extrapolated well log data
is inversely proportional to the square of the offset distance from the well is a common practice.
The main inversion outputs in the deterministic approach include volumes of acoustic impedance, VP/VS,
and density. Through algebraic manipulation, other
elastic properties (Table 1) may be generated from
Interpretation / November 2013 SB17

Table 1. Various types of inversion processes along with their respective methodologies, well and seismic data
requirements, and possible volume outputs. For time lapse inversion, n-vintages refer to the unlimited number of
monitor surveys which may be used to determine the ratio change with respect to the baseline survey.
Seismic inversion processes and applications
Inversion processes

Method

Required
seismic data

Required well data

Acoustic inversion
AVO inversion

Deterministic
Deterministic

Fullstack
Prestack

VP, RHOB
VP,VS, RHOB

Acoustic inversion
AVO inversion

Stochastic
Stochastic

Fullstack
Prestack

VP, RHOB
VP,VS, RHOB

Acoustic inversion
AVO inversion

Discrete spike
Discrete spike

Poststack
Prestack

VP, RHOB
VP.VS, RHOB

Poststack multicomponent inversion


Multicomponent
AVO inversion
Poststack time-lapse
inversion

Deterministic

VP, RHOB

Time-lapse AVO
inversion

Deterministic

AVO azimuthal
inversion

Deterministic

PP Fullstack, PS
Fullstack
PP Prestack, PS
Prestack
Baseline
Fullstack,
n-vintages
monitor
fullstack
Baseline
prestack,
n-vintages
monitor
prestack
Prestack
azimuthal stacks

Crosswell inversion
AVO azimuthal
inversion
Inversion
applications
Lithology prediction

Deterministic
Deterministic

Fullstack
Prestack

VP, RHOB
VP.VS, RHOB

Method

Seismic data

Wall data

Rock physics

Prestack

Crossplot
analysis
Rock physics

Poststack,
prestack
Prestack

VP, VS, RHOB, Vd,Sw,


PHIT
VP, RHOB, PHIT (VS
for Prestack)
VP, VS, RHOB, PHIT.
Sw

Neuial
networks
Rock physics,
neural
Geomechanics

Prestack

VP,VS,RHOB,Sw

Prestack

VP, VS.RHOB.Vsh, Vd

Prestack

VP, VS.RHOB.Vsh, Vd

Geomechanics

Prestack

Geomechanics

Prestack

VP, VS,RHOB, PHIT,


PHIE, GR, Vd
VP, VS,RHOB, PHIT,
PHIE, GR, Vsh, Vd

Porosity
Joint porosity and
saturation
Water saturation,
resistivity
Volume of shale
Volume of clay
Pore pressure (high
resolution vels)
Mechanical
earth modeling
Wellbore
stability

Deterministic
Deterministic

VP.VS, RHOB
VP, RHOB

Output volumes
Al
Al, SI, VP/VS, PR, RHOB, LR,
MR, LM, FF, K, E, M, G, lambda
Al
Al, SI, VP/VS, PR, RHOB, LR,
MR, LM, FF, K, E, M, G, lambda
Al
Al, SI, VP/VS, PR, RHOB, LR,
MR, LM, FF, K, E, M, G, lambda
Al, SI
Al, SI, VP/VS, PR, RHOB, FF, K,
E, M, G, lambda
Baseline Al, n-vintages ratio
change Al

VP.VS, RHOB

Baseline At, n-vlntages ratio


change Al, SI, VP/VS, PR,
RHOB, LR, MR, LM, FF, K, E,
M,G, lambda

VP, VS Isotropic, VS
Fast VS Slow, VS Fast
Azimuth, RHOB

Al, RHOB [vertical isotropic],


SI fast,
SI slow, SI fast azimuth, SI
slow/SI fast ratio
Al, PHIT, VP
Al, SI, VP/VS, PR, RHOB, LR,
MR, LM, FF, K, E, M, G, lambda
Output volumes
User defined lithology and
associated probabilities
Total porosity
Total porosity, water
saturation and associated
probabilities
Water saturation, resistivity
Volume of shale, volume of
clay
Pore presssure
Stress
Shear failure gradient, fracture
gradient, breakout and
breakdown failures

Note: AI = acoustic impedance, SI = shear impedance, PR = Poissons ratio, LR = lambda rho, MR = mu rho, LM = lambda mu, FF = fluid factor, K = bulk modulus, E =
Youngs modulus, M = P-wave modulus, G = shear modulus, lambda = Lams first parameter.

SB18 Interpretation / November 2013

the main inversion outputs for consideration in the seismic petrophysics part of this SRC workflow. As the inversion estimates the physical properties, the reflection
coefficients for each partial stack are calculated and
convolved with the appropriate wavelet to compare
the modeled seismic with the measured seismic data.
The mid-to-far stack strongly contributes to the estimation of VP/VS ratio and density. The high-resolution
contribution from the near- and midstacks is used without contradicting the low-frequency information in the
far-stack. For all input stacks, the application of a separate wavelet ensures that the synthetic seismic for
each partial stack has frequency content and phase
comparable to that of the measured seismic data.
The objective function Z of the inversion algorithm
for each layer property p may be written as
Z p arg min Eseismic E prior E horizontal E vertical ;
where:

Eseismic Penalty for the differences between the


seismic data and the synthetic seismic is accumulated over all partial stacks; related to the S/N of
each PP seismic angle stack.
Eprior Penalty for deviation of the estimated layer
properties from the LFMs are accumulated over
all layer properties; guides the solution toward/
against the a priori model.
Ehorizontal Penalty for horizontal changes are accumulated over all layer properties; controls the
horizontal continuity of the results.
Evertical Penalty for vertical changes are accumulated over all layer properties; controls the threshold for reflection coefficients.

The results obtained from the inversion carried out


in our study match nicely with the measured logs
(Figure 4). The same goes for all wells used in the inversion process. Such achievement gives a high level of
confidence in the postinversion work subsequently
carried out.

Seismic petrophysics
Tight gas reservoirs are associated with a high degree of lateral and vertical heterogeneity, low matrix
porosity, and low permeability (Boyer et al., 2006).
To determine a proper understanding of the reservoir,
a simultaneous global error minimization solver is used
to perform the petrophysical analysis. Using standard
triple combo logs (gamma ray, resistivity, and neutron-density), core data, and available cuttings descriptions, we determine mineralogy, TOC, porosity, and
hydrocarbon saturation over the Montney formation.
The mineralogy (Figure 5) consists of a mix of clay, silt,
quartz, carbonates, coal, anhydrite, kerogen, and heavy
minerals. Quartz volume shows a high correlation with
porosity and is likely to be a key driver in determining
reservoir quality in the Montney formation.
Subsequently, a production classification is carried
out where all wells are split into three categories (high,
medium, and low) with respect to gas production. The
classification is based on a 30-day average, initial production (IP) rate, where a medium producer falls within
35 MMCFday. A good correlation is observed when
comparing 30-day IP averages versus the long term production of the well. Moreover, production rates are normalized to the lateral length of the horizontal wells.
A comparison between petrophysical properties and
production reveals that effective porosity, effective
water saturation, and clay volume best correlate with
production. In an attempt to represent the three production classifications in terms of petrophysical properties,
cutoffs are applied to these petrophysical logs. Although
all three affect reservoir quality, clay volume also influences the ability to effectively induce fractures in
the formation.
Next, an n-dimensional probability density function
(PDF) is established from a cluster analysis on the
log data as a representation of the variability in the formation properties. Bayesian decision theory is then
used to establish probability density functions for
solving rock physics classification problems (Duda
et al., 2000). The theorem allows for expressing the

Figure 4. Comparison of well logs (red curves), low-frequency model (green curve), and prestack inversion traces (blue curves)
at a vertical well for acoustic impedance (a), lambda/mu (b), and density (c). The Montney formation is highlighted in yellow. The
correlation between inversion results and well logs is high, giving confidence in the postinversion work carried out.
Interpretation / November 2013 SB19

probability of a particular class given an observed x using Bayes theorem as


PC j jx

Px; C j PxjC j PC j

;
Px
Px

where Px; C j is the joint probability of x and


C j ; PxjC j denotes the conditional probability of x
given C j (Avseth et al., 2005). To establish a class x
(i.e., high producer, medium producer, low producer),
a rock diagnostic is performed where, using the
available well log data, a set of crossplots is generated
using the inversion output attributes (i.e., acoustic
impedance, PR, density, etc.) and petrophysical properties. The objective is to determine which seismic-derived elastic properties best separate out the desired
classes in the crossplot analysis.
In this analysis, a 2D PDF using Lambda/Mu and
density is used to predict the reservoirs hydrocarbon
production capacity. Lambda/Mu () may be derived
from PR () using the equation
2

;
1 2
where

:
2 2

The incompressibility term Lambda () represents


Lams coefficient and Mu () is the ratio of shear stress
to shear strain, also termed rigidity or shear modulus.
Lambda/Mu in rocks behaves similarly to PR, although
for a change in PR, the numerator increases as twice the
change while the denominator is reduced by 12,
thereby enhancing the change to a maximal asymptotic approach for 0.5, where is infinite (Goodway, 2001).
Last, the generated probability density function is applied to the selected seismic-derived elastic property
output cubes to produce a spatial distribution of the reservoirs capacity to produce hydrocarbons. At each
sample, there is an associated probability which represents the likelihood of each defined class based on the
crossplot analysis explained above. Several examples,
as well as a more comprehensive explanation of Bayesian statistics, are given in Bachrach et al. (2004), Avseth et al. (2005), and Sengupta and Bachrach (2007).
Interpretation
Although every unconventional play is unique, the
key properties governing production potential generally
include porosity, TOC, stiffness, natural fractures, and
closure stress (Ouenes, 2012). Moreover, current drilling decisions in this field have been highly influenced by
targeting zones of relatively low PR with the belief that
these zones are easier to fracture, and therefore produce more gas. This strategy has proved useful in many

Figure 5. Montney formation depth plot of mineralogy, compressional (DTP) and shear (DTS) slowness, water saturation (SW),
effective porosity (PHIE), and TOC for one of the vertical wells used in the seismic petrophysics analysis. The corresponding
mineralogy is labeled on the left-most track, where VCL is volume of clay, VSIL is volume of silt, VSAN is volume of sand, VCLC
is volume of calcite, VDOL is volume of dolomite, volume of coal is solid black, VANH is volume of anhydrite, VSM1 is kerogen,
VSM2 is heavy minerals, and VXBW is bound water. The perforated interval depths are indicated in the depth reference track
(upper perf at 1915 m, lower perf at 2050 m) as solid horizontal lines.
SB20 Interpretation / November 2013

cases, but not always. We hypothesize that seismicderived elastic properties, together with microseismic
data and a volume of seismic discontinuities representing natural faults and fractures can provide an understanding to hydrocarbon production capacity, leading
to more successful drilling decisions. Microseismic
monitoring of hydraulic fracturing treatments in tight
oil and gas reservoirs can provide useful information
about the results and the level of success of the rock
volume stimulated. For example, the length of the generated fracture systems and the geometry of the induced fracture system may be determined using
mapped hypocentral locations. In a naturally fractured
reservoir, hydraulic treatments may reactivate natural
fracture systems and therefore locally enhance permeability.
The color scheme used in the resulting hydrocarbon
production capacity (HPC) volume follows a traffic-light
approach where green corresponds to expected high
production, yellow to medium, and red to low; nonclassified samples are indicated in black. In Figure 6, a depth
slice comparison between the HPC volume and PR is
shown where two medium-producing horizontal wells
are located (cf. Figure 1b, Zone 1). The zone is dominated by relatively low PR. The HPC volume is primarily
yellow, indicating the expected production from a well
drilled in this zone is medium. Figure 7 documents the
mapped microseismic events as well as an estimated
stimulation volume (ESV). Most of the microseisms
are clustered within a zone where 35 MMCFday production is expected. The ESV uses microseismic event

Figure 7. HPC volume depth slice (a) and vertical intersection (b) through a medium-producing well showing microseismic events and an ESV where most of microseismic events are
clustered within a zone where medium production is expected.

Figure 6. Depth slice comparison (924 m TVD) between the HPC volume (a) and PR (b), where two medium-producing horizontal
wells are located (cf. Figure 1b Zone 1). The zone is dominated by relatively low PR. This observation is currently driving drilling
decisions in this asset play.
Interpretation / November 2013 SB21

density to compute the volume where the data are


normalized by magnitude to compensate for distance effects. This is a qualitative indicator of reservoir contact,
possibly correlating to the distribution of fracturing fluids within the fractures. Moreover, it gives an indication
of fluid-affected volumes compared to fracture extraction using planar geometry.
In Figure 8 (cf. Figure 1b, Zone 2), a comparison between HPC and PR is shown along with seismic-derived
fractures and faults where two neighboring, lowproducing horizontal wells exist. Both wells were targeted based on PR. The HPC cube accurately predicts
low production from these wells. The available microseismic data event size is proportional to S/N.
Figure 9 shows a series of depth slices for the wells
located in Zone 3 having different production ratings
a high producer, a medium producer, and a low producer which are within close proximity. In addition,
a blind vertical well is displayed. The result from the
HPC cube (Figure 9a) shows that the high (green trajec-

Figure 8. Depth slice comparison of the hydrocarbon production capacity volume (a)
and PR (b), both with seismic-derived fractures and faults, where two low-producing
horizontal wells exist (cf. Figure 1b, Zone
2). Both wells were targeted based on PR.
The HPC cube accurately predicts the observed low production from these wells.
The available microseismic data event size
is proportional to S/N. The perforation locations are indicated by black disks.

SB22 Interpretation / November 2013

tory) and medium (yellow trajectory) producing wells


(200 m apart) fall within a zone of 35 MMCFday expected production, although the high producer penetrates a small segment where high production is
expected. The corresponding PR result is also displayed
where both wells are in a zone of relatively low PR. The
high-producing well was drilled along a preexisting
natural fault or fracture, which potentially contributed
to higher production in comparison to the adjacent
medium producer. There was no significant drop in production when the medium producer was hydraulically
fractured (neighboring high producer was drilled first),
although the presence of small amounts of sand and
water were observed, which means there was some
connection between the two wells. Figure 9b shows
a depth slice corresponding to the low producer (red
trajectory) in Zone 3, where the HPC volume prediction
matches nicely with the observed well production of
less than 3.0 MMCFday. The PR result shows the well
to be in a relatively high PR zone.

Figure 9. Depth slice comparison between


the reservoirs capacity to produce versus
PR with seismic-derived fractures and faults,
for the depths of (a) 931 mcompare with
green and yellow trajectories, (b) 951 mcompare with red trajectory, (c) 998 mcompare
with gray trajectory (upper perforation), and
(d) 1,126 mcompare with gray trajectory
(lower perforation). The vertical well has
two perforation sets with different well production: upper perforated interval has high
production while the lower perforated interval produces poorly.

Figure 10. Hydrocarbon production capacity


section at a blind well where the upper perforation set gave high production (green) while
the lower perforation set is low (red). The
HPC volume shows a good correlation with
the observed production in both perforated intervals.

Interpretation / November 2013 SB23

Figure 9c and 9d shows the results at the vertical


well used as a blind test where the upper perforated interval produces more than 5 MMCFday while the
lower perforated interval produces very poorly. Hydrocarbon production capacity correlates nicely with the
observed production values at the perforated intervals.
Moreover, low PR at this blind test well is indicative of
high production. In Figure 10, the vertical well HPC result is shown in section view.
Conclusions
In this paper, we propose that seismic-derived elastic
properties be used to predict an unconventional reservoirs capacity to produce hydrocarbons. In addition,
horizontal drilling, hydraulic fracture monitoring, and
a volume of seismic discontinuities representing natural
fractures and faults can assist in the interpretation by
improving the understanding of the result obtained.
In our study, a predictability of 70% is observed when
validating the HPC volume with 27 vertical and horizontal wells. The workflow presented in this paper illustrates how many pieces of the geoscience puzzle may
fit together to predict hydrocarbon production capacity, and therefore enhance recovery rates from optimized well placement.
Acknowledgments
Our gratitude goes to Progress Energy Corporation
for permission to publish the paper. Also, we thank
Joe Leonard and Wayne Hovdebo from Progress Energy
for their contribution to the project. From Schlumberger, we thank Innocent Kalu and James Johnson
for their respective roles in the work carried out.
References
Aki, K., and P. G. Richards, 1980, Quantitative seismology
Theory and methods, Volume 1: W. H. Freeman and
Company.
Avseth, P., T. Mukerji, and G. Mavko, 2005, Quantitative
seismic interpretation: Applying rock physics tools to
reduce interpretation risk: Cambridge University Press.
Bachrach, R., M. Beller, L. Chu Ching, J. Perdomo, D. Shelander, N. Dutta, and M. Benabentos, 2004, Combining
rock physics analysis, full waveform prestack inversion
and high-resolution seismic interpretation to map lithology units in deep water: A Gulf of Mexico case study:
The Leading Edge, 23, 378383, doi: 10.1190/1.1729224.
Boyer, C., J. Kieschnick, R. Suarez-Rivera, R. E. Lewis, and
G. Waters, 2006, Producing gas from its source: Oilfield
review/Schlumberger, 3649.
Duda, R. O., P. E. Hart, and D. G. Stork, 2000, Pattern classification: John Wiley and Sons.
Goodway, B., 2001, AVO and Lam constants for rock parameterization and fluid detection: CSEG Recorder, 26,
3960.
Ouenes, A., 2012, Seismically driven characterization of
unconventional shale plays: CSEG Recorder, 37, 2228.
SB24 Interpretation / November 2013

Pedersen, S. I., T. Randen, L. Snneland, and . Steen,


2002, Automatic fault extraction using artificial ants:
72nd Annual International Meeting, SEG, Expanded Abstracts, 512515.
Rasmussen, K. B., 1999, Use of dip in seismic inversion:
61st Annual International Conference and Exhibition,
EAGE, Expanded Abstracts, 49.
Sengupta, M., and R. Bachrach, 2007, Uncertainty in seismic-based pay volume estimation: Analysis using rock
physics and Bayesian statistics: The Leading Edge,
26, 184189, doi: 10.1190/1.2542449.

Gorka Garcia Leiceaga received a


B.S. (2001) in geophysics and a minor
in mathematics and an M.S. (2005) in
geophysics, with a focus in controlledsource electromagnetics, from the
University of Houston. In 2006, he
joined Schlumberger and specialized
in seismic inversion for rock properties and reservoir characterization
studies. Since joining Schlumberger, he has been expatriated in Mumbai, India; Villahermosa, Mexico; Rio de
Janeiro, Brazil; and Calgary, Canada. Prior to Schlumberger, he worked for Odegaard as an inversion geophysicist and as an adjunct mathematics instructor
at the University of Houston. He is currently based in
Houston, holding a senior geophysicist position in the Microseismic Services group where his focus is on the integration of surface seismic and microseismic.

Mark Norton received a B. S. (Honors) in geophysics from the University


of Manitoba, where he specialized in
electromagnetics, completing his
undergraduate thesis on magnetotellurics. He joined ExxonMobil in 2000 as
a staff geophysicist in their Calgary office. In 2003, he worked for ExxonMobil out of their Houston office in the
deepwater Gulf of Mexico. Since leaving ExxonMobil in
2005, he has worked for Husky Energy and Real Resources,
and now works at Progress Energy as a senior geophysicist.
He is a member of SEG, CSEG, CSPG, and PESGB.

Jol Le Calvez received a B.Sc. in


physics, an M.Sc. in geosciences,
and a Ph.D. in salt tectonics. He is
the North America microseismic domain expert for Schlumberger and
manages the processing groups in
U.S. Land while providing training
and support to the international
processing and interpretation groups.
He actively participates to the development of the processing, visualization, and interpretation software Schlumberger currently uses in relation to the monitoring of

induced microseismicity coupled to hydraulic fracture treatment and other applications (e.g., CO2 sequestration,
geothermal injection, etc.) using downhole, shallow wellbores, and surface arrays. Prior to joining Schlumberger,

he worked for the Bureau of Economic Geology at the


Applied Geodynamics Laboratory and Etudes et Recherches Gotechniques.

Interpretation / November 2013 SB25

You might also like