A Practical Application of Stochastic Modeling Techniques for Turbidite Reservoirs

Abstract

A detailed study of the productivity and recovery potential for a turbidite reservoir has been completed using a practical application of stochastic modeling and reservoir simulation techniques. The objective of this study was to assess the impact of various reservoir characteristics on development considerations such as well spacing and expected production/recovery profiles. Distributions of typical sand body dimensions were estimated from a review of core, log, seismic, and analog data. These distributions were input to a stochastic modeling program to generate multiple realizations of reservoir descriptions for a range of net/gross ratios, facies distributions, and assumed sand body geometries. An interface program was developed to minimize the gridding problems associated with conversion of stochastic model output to reservoir simulation input data. Grid data generated from this program were input directly to practically sized reservoir simulation models. Results from the stochastic realizations and simulation models demonstrated the potential variability in connectivity and recovery profiles that may be expected for a turbidite reservoir. The reduced connectives in the stochastic model yield significantly different production profiles and lower recoveries than would be calculated with the assumption of continuous layer-cake type models. The approach developed for this study can be used to define and improve confidence limits for production and recovery profiles from typical turbidite reservoirs with only limited well information. Introduction Turbidite reservoirs are formed as a result of downslope movements of clasic sediments under the forces of gravity and fluid turbulence. These reservoirs are generally associated with deep sea submarine fans. Typical deposits consist of massive, structureless channel fill sands which pass laterally into and are overlain by progressively thinner, laminated intervals of sands and shales. The shifting, stacking, and erosion of these channel-levee systems results in a complex reservoir description with variable correlation of individual sands at the scale of typical well spacing. Interchannel areas, which are often dominated by muds (shales) and thin-bedded sands, can significantly reduce the connectivity of individual sand sequences. A schematic illustration of a typical facies distribution for this type of turbidite system is presented in Figure 1. An assessment of the impact of sand discontinuities or connectives in these reservoirs is required for realistic performance predictions and estimation of associated confidence limits. This is particularly important during the pre-development stage when major investment decisions, regarding well spacing and facility (platform) requirements, are being made on the basis of a limited number of exploration wells. Analytical calculations and conventional simulation approaches based on layer-cake type models will lead to optimistic results. Alternate model representations were proposed by Weber and van Geuns in which the architecture of middle and upper fan turbidites was described as either "jigsaw" or "labyrinth" distributions of sand bodies. P. 343^