American Oil and Gas Reporter - July 2015 - 66

SpecialReport: Horizontal Well Operations
FIGURE 2
Brittle Lithofacies Based on FMI Data (Left) versus Dipole Sonic (Right)
Brittle lithology based
on dipole sonic-derived
brittleness

Poisson's Ratio

Poisson's Ratio

Possible fracture
area based on
brittle lithology

Other
Brittle
Dolomite
Porous
Young's Modulus

Young's Modulus

FIGURE 3
Geostatistical Facies Modeling using Seismic-Derived Information
Seismic-derived facies
(only high reliability-high probablity)

Most likely
facies
SIS

Modeled facies
Other
Brittle
Facies
reliability

Low

High

Dolomite

Facies logs,
facies proportions curves

Limestone

FIGURE 4
Local Orientations and Intensities of Fracture Families (Left)
And Comparison of Orientations from FMI and Seismic Data (Right)

0.14
0.12

FMI data
Seismic data

Proportion

0.10
0.08
0.06
0.04
0.02
0.00
0

66 THE AMERICAN OIL & GAS REPORTER

20

40

60

80 100 120 140 160
Azimuth

180

cross-plots of elastic properties, which
can hinder using seismic data to guide
facies mapping.
If facies are defined without "human
intervention," using only statistical methods such as multivariate cluster analysis
from log or seismic data, their consistency
should be checked from log- to seismicscale as well as their relationships to the
actual production drivers that control
flow behavior.
If natural fracture information is available, it should be used when defining facies. Figure 1 shows an example of facies
definition using a neutron-density crossplot. Only points that correspond to conductive or partially conductive fractures
from formation microimaging (FMI) data
are plotted. Notice that most observed
natural fractures in the image occur in
lower-porosity intervals and these rocks
are called "brittle." Other rocks with considerably fewer or no fractures are called
"porous" in this example.
However, the brittle facies defined by
the presence of fractures do not necessarily
coincide with rocks with the highest brittleness coefficient estimated from dipole
sonic data. For this reason, a careful calibration with log data is important to understand the actual relations between defined facies and production drivers in
unconventional reservoirs. As an example,
Figure 2 shows a comparison of brittle
lithofacies based on FMI data (left) versus
brittle rock from brittleness derived from
dipole sonic (right).
Once facies have been defined and
rock physics analysis shows that they
can be mapped spatially using elastic
properties derived from seismic prestack
inversion, different methods can be used
to estimate facies probabilities across the
reservoir. Most probabilistic approaches
to map facies using seismic data are
based on an approach to statistical rock
physics for reservoir characterization proposed by Stanford University professor
Tapan Mukerji, et. al. (2001). This approach requires probability density function modeling on cross-plots of elastic
properties at log scale colored by facies.
Unlike the common implementation
of the Mukerji approach, the new workflow works on cross-plots of inverted attributes at seismic scale, and yields not
only estimates of the spatial variability
of facies probabilities, but also estimates
of reliability of such probabilities (even
in cases when there is total overlap between the target facie and the background
American Oil and Gas Reporter - July 2015

Table of Contents for the Digital Edition of American Oil and Gas Reporter - July 2015
