GEEL, CORNELIS R., Delft University of Technology, Delft, Netherlands

ABSTRACT: Stochastic Modeling of a Deep-Water Sandstone Reservoir Analog

In order to assess the variability of deep-water sandstone reservoirs, an object-based stochastic modeling technique was developed using geological rules, geological objects, and simple statistics.

A sedimentological study was done on inner fan turbidites of the Eocene Hecho Group, south-central Pyrenees, Spain. A 1200-m-wide, 50-m-high exposure was selected, which contains a series of laterally stacked turbidite channels. A clear trend in channel geometry and channel fill was observed, which formed the basis for a set of geological rules that relate geological events, channel geometry, and channel fill.

Based on bed geometry, grain size, shale content, and degree of amalgamation, three objects (bed types) were defined: (1) thick sandstone beds, (2) thin sandstone beds, and (3) thin shale beds. Bed geometries were stored by recording for each individual bed the width and three bed-thickness measurements at regular distances. For each bed type and each channel, multidimensional frequency charts were prepared.

The generation of synthetic cross sections was done in two stages. (1) Channel outlines were generated according to the geological rules. (2) Each channel was filled from base to top by randomly drawn objects whose geometry was determined by the frequency charts and subsequent cubic spline interpolation. The generated cross sections were transformed into high-resolution permeability grids, which were reduced to low-resolution grids by using a special upscaling technique. These grids formed the basis for a Monte Carlo series of waterflood simulations, which eventually validated the stochastic modeling technique.

AAPG Search and Discovery Article #90990©1993 AAPG International Conference and Exhibition, The Hague, Netherlands, October 17-20, 1993.