New Petroleum Reservoir Modeling Techniques Improve Field Management and Optimize Recovery

Marjorie Levy
Chevron Energy Technology Company ([email protected])

Abstract

Because of the extremely high cost of developing a subsurface reservoir, commonly a billion dollars or more, it is critical to understand the volumes of hydrocarbon that are present within the reservoir and the amount that can be recovered. Each well is expensive, so we must make the most of the information collected from each well to constrain the uncertainty surrounding the architecture of the reservoir, its extent, and its internal heterogeneities, as well as the impact on recoverability. We approach this by constructing a geocellular model of the hydrocarbon accumulation that incorporates a reasonable range of possible reservoir characteristics, and then simulate the flow of fluids - hydrocarbons and water - throughout the life of the field. The results from any reservoir simulation are strongly dependent on the accuracy of the underlying geologic models. Until recently, it has not always been possible to build geocellular models that accurately portray the subsurface geology.

Over the past several years, Chevron has developed a new geologically-based modeling workflow, which combines Multiple Point Statistics (MPS) and Facies Distribution Modeling (FDM) to generate a 3D geologically-robust geocellular reservoir model. MPS is an innovative depositional facies modeling technique, developed by Chevron in collaboration with Stanford University, which incorporates 3D geological concepts in training images that more accurately integrate geological information into reservoir models. Training images allow MPS to retain complex spatial relationships among multiple facies and to model non-linear shapes such as sinuous channels or irregular bar forms that conventional variogram-based modeling techniques typically fail to reproduce. In addition, because MPS is pixel-based, not object-based, MPS models can be constrained by very large numbers of wells. FDM is a novel technique that is used to generate a facies probability cube to better constrain the facies spatial distribution in geostatistical models.

The MPS/FDM workflow above is preferred to variogram-based and object-based techniques to model important Chevron assets in both shallow-water and deepwater clastic reservoirs, and more recently, in carbonate reservoirs. Additionally, this workflow has been used in synthetic studies to explore the potential impact of architectural and textural parameters on flow behavior. Using experimental design methods, it is possible to determine the relative impact on production of a variety of field parameters. With this information, one can focus on better understanding the key subsurface parameters and gather new data to reduce their uncertainty. This work flow enables field management by lowering risk and optimizing production.

AAPG Search and Discovery Article #90086 © 2008 AAPG Foundation Distinguished Lecturer Series 2008-2009