--> Abstract: Object Based Modeling and Stochastic Simulation of Fluvial Reservoirs Geometry, by C. A. Poletto, W. L. Lanzarini, and G. Tavares; #90933 (1998).

Datapages, Inc.Print this page

Abstract: Object Based Modeling and Stochastic Simulation of Fluvial Reservoirs Geometry

Poletto, Carlos A. - Petrobras/E&P; Lanzarini, Wilson L. - Petrobras/Cenpes; Tavares, Geovan - Pontifícia Universidade Católica do Rio de Janeiro

Reliable 3-D simulations of fluvial reservoirs were performed using an object-based method through a stochastic modeling technique of geometric objects. The basic unit of a reservoir is represented in cross-section by the geometry of a semi-elliptical surface. The use of parametric and implicit representations of semi-elliptical surfaces makes the algorithm efficient to generate and visualize objects.

The functions which define a fluvial channel into a 3-D domain were constructed using reservoir data from the Açu Formation (Potiguar Basin, Brazil), well known not only in many oil fields, as well as in near surface outcrops. The main parameters were thickness (h), width (w), sinuosity (s) and preferential direction (d). The thickness was taken from single reservoirs, defined as fining-upward successions in cores and geologic cross-sections. Sinuosity and channels preferential direction were measured on geologic maps. In addition, a proportion curve controls the reservoir distribution within the simulated domain.

The simulation process of each object starts from a random germ into a defined 3-D domain, with random values for the respective functions and ends when all conditioning wells are honored and the reservoir proportion is reached. The accepted objects are built by discrete geometric modeling of the surfaces. The equiprobable architecture of the reservoir in three dimensions is the result of erosion and stacking pattern of the generated objects. (Fig. 1).

Simulations of reservoir geometry and architecture were taken into volume of a developed field, having 1632 x 1632 x 34 meters, totaling 106 cells, and using real data from 30 wells. The realizations are equiprobable and suitable to all conditioning data. Only the reservoir distributions were simulated. The voids were assumed as non-reservoir facies. (Fig. 2).

AAPG Search and Discovery Article #90933©1998 ABGP/AAPG International Conference and Exhibition, Rio de Janeiro, Brazil