American Oil and Gas Reporter - February 2017 - 47

SpecialReport: Shale Resource Science
isotropic model makes the results difficult
to interpret, particularly at the toe of the
treatment well. The decision was made
to conduct further analysis to determine
whether the results and the model could
be improved by analyzing anisotropy.
This investigation determined that
while azimuthal anisotropy could explain
the absolute mismatch between two measurements of perforation locations for
stages 13 to 24, the error could not be
explained fully by azimuthal anisotropy,
particularly from stages 3 to 12. So, azimuthal anisotropy played a role in the
perforation localization mismatch, but it,
alone, did not provide a definitive conclusion as to the root cause of the localization bias. Consequently, we would
argue that an orthorhombic anisotropy is
the minimum level of velocity model
complexity to resolve this bias.
Reducing Localization Error
With such a narrow horizontal subtended angle from the ray paths developed
from the aggregation of preformation
shots, it was determined that it would be
beneficial to leverage the larger vertical
aperture and attempt to revise the initial
isotropic velocity model. Both 1-D and
3-D elliptically anisotropic VTI models
were developed by inverting the measured
velocities in a single model from all imaged perforation shots.
Note that as a result of enhanced adaptive signal processing, the total number
of imaged perforation shot localizations
increased from 56 to 59.
Figure 2 is an example of this effort.
It shows the final VTI velocity profiles
in depth as an extraction at the x-y location
of the treatment well. The curves represent
sonic velocity (black), horizontal velocity
(green), and vertical velocity (red). The
1-D VTI profile is displayed at left and
the 3-D VTI profile is at right.
A key quality assurance step in constructing a velocity model for microseismic
localization is maintaining consistent flatness of the travel-time corrected moveout for both arrays, and for all recorded
perforation shots. Figure 3 shows ray
paths and 3-D VTI model move-out corrected gathers of a perforation shot from
stage 9 in the midlateral section of the
well.

Note the contribution of the two large
vertical aperture arrays and the flatness
of the perforation shot. The deepest receivers of the left observation well flattened
as a result of slight lateral velocity variations. The global root mean square error
across both gathers was 0.506 milliseconds.
Implementing a single (unified) 1-D
elliptically anisotropic velocity model reduced localization error from a median
of ≈75 to ≈32 feet, compared with a 1-D
isotropic velocity model. Minimizing localization error ultimately provides a
more robust platform on which to perform
a reliable and competent engineering interpretation. In particular, the toe stages
no longer had localization ambiguity and
could be interpreted with confidence.
Events were detected using a technique
that seeks a parametrically weighted
agreement or weighted semblance between
the modeled versus observed move-out
of compressional and shear waves, respectively. The algorithm can trigger on
compressional, and both horizontal and
vertical shear-wave modes.
There is noted sensitivity between the
number of valid event detections and the
minimization of the residual error in a
velocity model. This observation was evident in the data. The 1-D anisotropic
model permitted the population of valid
event detections and localizations to in-

crease from 1,675 to 2,344.
Key Improvements
The key improvements to processing
this microseismic dataset with an
anisotropic velocity model include:
* Reduced perforation localization
error in all three dimensions;
* Detection of more valid events;
* More robust interpretation of microseismic azimuths by stage;
* More complete and reliable fluid
system analysis with better containment;
* More complete and reliable distribution by formation analysis with evidence
of less downward growth; and
* Extraction of geologically relevant
rock properties from near- and far-field
microseismic events.
Figures 4A and 4B show the perforation
localizations within the isotropic velocity
model and the reprocessed perforation
localizations mapped within the anisotropic
1-D VTI velocity model. Figure 5 shows
the perforation localizations contrasted
again within the anisotropic 3-D VTI velocity model.
Utilizing the 3-D VTI inverted velocity
volume, the effective p-wave and s-wave
velocities were extracted at each perforation location along the treatment lateral.
A computation of vertical Poisson's ratio
at each perforation interval was performed
and plotted with respect to measured

FIGURE 5
Perforation Localizations in Anisotropic 3-D VTI Velocity Model

FEBRUARY 2017 47

American Oil and Gas Reporter - February 2017

Table of Contents for the Digital Edition of American Oil and Gas Reporter - February 2017