New Fast, Stochastic, Boolean Simulation of Fluvial Deposits.

Abstract

Summary It is now a fairly common practice to characterize fluvial reservoirs via stochastic simulations of fluvial deposits. Three criteria are used to evaluate various simulation processes. They are:honoring input data (usually well data);defining fluvial heterogeneity boundaries; andspeed. There are two major branches in 3D stochastic simulations: pixel-based and object-based. The pixel-based method distributes property values rather than the heterogeneities. It is fast but the boundaries between heterogeneities are often blurred and do not reveal geological features. The object-based (or boolean) method distributes the heterogeneities directly and can therefore produce well-defined feature boundaries. However, object-based methods often involve a simulated annealing process which converges slowly. In this paper, we will present a new, straightforward object-based conditional simulation process that honors the input data, produces well-defined heterogeneites and works reasonably fast. The revolutionary step that makes this possible is a well data segmentation process and a channel generating process that we have developed. This processing enables us to rethink and redesign the entire object-based simulation process. As a result, we are able to, in a reasonable amount of time, produce multiple realistic realizations of a fluvial environment that honor well data and other geological information. P. 697