In the July
2002 "Geophysical Corner" we described a new fracture detection
method incorporating a portable laserscan unit to completely image
analog outcrops in three dimensions. This month we are going to
explore how simple analysis of calibrated analog fracture models
can enhance exploration and production in fractured reservoirs.
After collecting
laserscan data from analog outcrops, semi-automatic processing extracts
important fracture data such as geometries, intersections, trace
lengths and orientation statistics. These statistical and spatial
properties are extrapolated in three dimensions and used to generate
synthetic fracture models at a scale consistent with existing or
planned wells.
The synthetic,
three-dimensional fracture networks have similar statistical and
topological characteristics of observed data, but offer distinct
advantages over their natural counterparts.
One important
benefit is the ability to quickly construct multiple realizations
of the observed fractures to test different hypotheses and perform
sensitivity testing of the input parameters. Results also can be
compared to well production volumes and modified to get good matches.
The new
laserscan technique provides robust statistical data on fracture
orientations, clustering and, to a lesser degree, fracture spacing.
Parameters such as fracture trace length and shape (i.e., aspect
ratio) also may be extracted from the laserscan data using techniques
such as trace analysis, but with lesser certainty.
Constructing
multiple fracture networks using different parameters and analyzing
them with a few simple tools may help to determine the relative
importance of these less-constrained parameters.
More importantly,
the simple analysis may indicate whether a more detailed fracture
investigation is truly necessary.
One technique
is three-dimensional connectivity analysis, which determines how
well connected or poorly connected fractures are within a natural
or synthetic network. Parameters such as fracture volume and fracture
area are extracted, and may be useful for fracture modeling or well
planning.
Often,
wells that intersect different components cannot communicate with
one another, so defining the likely extent and volumes of the components
is critical in making production estimates.
For example,
consider a simplified synthetic fracture network generated using
orientation and length data extracted from a laserscan of a ~500m2
outcrop face (figure 1).
- Orientation data
are extracted by orientation cluster analysis.
- The fracture length
distribution is estimated using two-dimensional trace analysis.
In this example, the length and aspect ratio (fracture length
divided by fracture height) of the fractures are less constrained
than other key parameters, reflecting a commonly encountered situation
in three-dimensional fracture modeling.
The resulting
scale model of the fracture network contains four fracture sets,
each with a similar statistical signature as those interpreted from
the outcrop laserscan data (figure 2).
A synthetic well is "drilled" into the model and used as the basis
for connectivity analysis.
The analysis
quickly reveals that the well intersects a number of fracture clusters
with relatively small drainage volumes (figure
3).
Since the
fracture lengths and aspect ratios are not uniquely defined, sensitivity
testing of this fracture model might include changing the aspect
ratio and lengths of one or more of the fracture sets, and reanalyzing
the model for connectivity.
To this
end, a new model is constructed using one fracture set with twice
the length and aspect ratio as in the first model. Connectivity
analysis shows that the model is indeed sensitive to these changes,
as reflected by the difference in drainage volumes intersected by
the synthetic well (figure 4).
Although
both fracture models contain three primary drainage volumes of roughly
equal size, the second model contains one volume (colored in yellow)
almost 500x larger than in the first model.
If the
actual production volume for the well were known, it would be essential
to compare these data with the new model to determine if a more
detailed assessment of fracture size is important.
In conclusion,
laserscanning can be used to generate calibrated fracture models
that enhance the understanding of a fractured reservoir. Models
can be built around existing or proposed wells, and analyzed with
simple techniques such as connectivity analysis.
Sensitivity
testing may help to provide additional confidence in well planning
and the fracture network interpretation.