Integrating Stochastic and Deterministic Methodologies in Geological Reservoir Modeling

Taizhong Duan, Jeff Hamman, Don Caldwell, and Mark Petersen
Marathon Oil Company, Houston, TX

A workflow was implemented to build reservoir-scale geological models based on Roxar and Marathon proprietary software that takes advantage of both deterministic and stochastic approaches. Workflow includes two major sub-workflows, Roxar-based stochastic modeling, and multiple rock-type and Gassmann equation-based rock physics inversion. Input data includes well data, seismic acoustic impedance volume (AI) and their derived data types such as mapped sedimentary bodies, net-to-gross ratio or sand thickness map. Object and/or pixel-based stochastic modeling simulate facies, porosity, and permeability distributions in 3D and at reservoir scale. Simulated results are used as a pre-conditioned input or constraint for further processing. Rock physics inversion uses the pre-conditioned stochastically distributed porosity and facies to generate a synthetic AI volume at well-data scale. Simulated annealing algorithm is used to repeatedly update porosity, permeability, and water saturation until an upscaled effective medium AI response matches the seismic acoustic impedance. Output includes multiple realizations of facies, porosity, permeability, and water-saturation from multiple geological scenarios, which are consistent with seismic AI and can reasonably predict the same properties at well locations. Output results can be ranked based on their impact on reserves calculation and dynamic reservoir simulation, or analyzed stochastically if enough realizations implemented. This workflow captures the core value of a well-understood deterministic model (Gassmann rock physics) and also provides a tool to evaluate uncertainty of geological models. This uncertainty is inherited from poor data quality or data incompleteness, the non-unique solutions in inversion, and the lack of understanding in geological processes and their modeling.

AAPG Search and Discovery Article #90039©2005 AAPG Calgary, Alberta, June 16-19, 2005