Published 1994 as part of Computer Applications 3
Copyright © 1994 The American Association of Petroleum Geologists.  All Rights Reserved.
 

Because the reservoir properties of various rock types can differ considerably, it is best to simulate the distribution of reservoir properties in two steps: simulate the rock types, then simulate the reservoir properties conditional on the rock types; however, identifying the rock types themselves can be difficult, due to the interaction of depositional processes and diagenesis, and the variable amount and quality of data available from well to well. This chapter presents a methodology using cluster analysis for identifying petrophysical rock types in a training data set of cored wells. A case study from the Cretaceous Muddy Sandstone of the Powder River basin illustrates the methodology. After identifying the rock types, analyzing the core properties provided geologic information concerning each rock type. Discriminant function analysis extended the rock type classification to wells without core data. Sequential indicator simulation was used to model and simulate the spatial distribution of the rock types in three dimensions. An annealing method was used to postprocess the simulations so that they honor the rock type transition frequencies seen in the well data.