Alireza hajizadeh - Department of chemical and petroleum engineering, Sharif University of Technology
Farhad A. Farhadpour
Saeid Jmashidi

In building a geological facies model of a reservoir, honoring the large scale heterogeneities and long range connectivity is of crucial importance since it strongly controls the flow paths. Traditional variogram-based geostatistical methods are inadequate in this regard because they employ two point correlations of data which is too limiting for capturing long range geological heterogeneity. For example variogram based methods have been shown to fail to reproduce long range and/or complex structures such as channels. A relatively new field, termed multiple-point geostatistics which does not rely on variogram models allows capturing structure from so-called training images. The basic logic relies on the fact that geological structures need to be defined using models which use correlation between more than two points at a time. Multiple-point geostatistics borrows multiple-point patterns from a training image which only reflects a prior structural concept and does not need to include any accurate data. It then anchors such patterns to subsurface well-log, seismic and production data thereby capturing the long range connectivity while honoring the hard data. The application a particular multiple point statistics method, the single normal equation simulation (SNESIM) to a fluvial reservoir is demonstrated in this article

Geological modeling, Multiple- point statistics, facies, training image