Case
Study, they illustrate the difficulties in building a  3-D description,
especially when one is attempting to integrate seismic information into
the model and present some solutions.
The chapter by T. Høye et
al. details the use of seismic reflectors (top, base, and intrareservoir
reflectors) into stochastic models to aid in constructing a structural
model of the reservoir. Their objective was to highlight a target well
location and assess uncertainty in the modeled results. Depth conversions
were aided by modeling surface velocity as a Gaussian random field
conditioned on the well locations. Follow their procedures in
Stochastic Modeling of Troll West with Special Emphasis on the Thin Oil
Zone.
S. A. McKenna and E. P.
Poeter illustrate the use of an indicator simulation technique to
incorporate soft (imprecise) data into a simulation of facies to help
reduce model uncertainty. The soft data used in their study are synthetic
data (real data plus noise). Their chapter, Simulating Geological
Uncertainty with Imprecise Data for Goundwater Flow and Advective
Transport Modeling, is applicable to the petroleum
industry.
In Fractal Methods for
Fracture Characterization, T. A. Hewett illustrates the concept of
fractals and self-similarity and scaling laws using fracture networks. He
points out that the most useful fractal sets for describing natural
property distributions are those displaying self-similarity.
Description of Reservoir
Properties Using Fractals, by M. Kelkar and S. Shibli, rounds out the
discussion of fractals with a case study. This work investigates the
utility of fractal geometry for describing the spatial correlation
structure of rock properties in carbonate and clastic rock
types.
A. S. Almeida and P. Frykman
obtained stochastic images of porosity and permeability in a Maastrichtian
North Sea chalk reservoir. In Geostatistical Modeling of Chalk
Reservoir Properties in the Dan Field, Danish North Sea, they use a
Gaussian collocated cosimulation algorithm, built on a Markov-type
hypothesis, to perform direct cosimulation of spatially interrelated
variables. Log-derived porosity was used as soft data during the
cosimulation.
Innovative modeling
techniques were combined to generate a realistic characterization of a
complex eolian depositional environment for a multicompany reservoir
management problem. D. L. Cox et al. first model the eolian bedding
geometries and dimensions of four different stratification types.
Permeability fields were generated within the conditional simulation of
reservoir model, scaled-up for flow simulation and compared to historical
field production data in Integrated Modeling for Optimum Management of
a Giant Gas Condensate Reservoir, Jurassic Eolian Nugget Sandstone,
Anschutz Ranch East Field, Utah Overthrust (USA).
C. J. Murray uses a variety
of methods, including cluster analysis, discriminant function analysis,
and sequential indicator simulation (SIS) to identify and model
petrophysical rock types. Simulated annealing was used to postprocess the
SIS images such that the images honor rock type transitional frequencies
in well data. Follow this process in Identification and 3-D Modeling of
Petrophysical Rock Types.
Uncertainty is a statistical
term used to describe what we do not know. The results of a geostatistical
model include a number of possible outcomes, each equally likely, and each
physically portraying, in a variety of ways, the portion of the model we
do not know. Thus, for example, a channel may vary in size, shape, and
location from outcome to outcome in an area where there is no hard data.
Tracking these variations and developing a sense of how variable the model
is can be very important. This section's final chapter, by R. M.
Srivastava, nicely illustrates the topic of visualizing uncertainty. In
The Visualization of Spatial Uncertainty, Srivastava demonstrates a
unique way of looking at the various possible outcomes of a given
geostatistical model. |
