Effect of Reservoir Heterogeneities on Flow Simulation Requirements

Abstract

Abstract Most enhanced oil recovery and waterflood production forecasts are based on numerical simulation models. In the past, these models have been shown to be overly simplistic in modelling geophysical heterogeneity. Overly simplistic models have resulted in unrealistic estimates for ultimate recovery for many secondary and tertiary recovery schemes. Therefore, the heterogeneity, or variance of the rock properties, is a key factor in predicting waterflood or EOR recovery. It is not possible to make accurate performance predictions for EOR or waterflood schemes without adequate reservoir characterization. Then the question arises: what level of geological detail is required to accurately characterize a reservoir. Another related question that follows is what are the number of layers that are needed in the flow simulation. To partially answer these questions, reservoir flow simulation studies were done to examine the effect of permeability heterogeneity on predicted waterflood/EOR performance. The level of geological detail and layering, to accurately characterize a reservoir, was found to strongly depend on the:The size, type and scale of heterogeneity.The type of process being modeled. Typically, logs and core data are used to determine flow units. The resolution of the data needed for engineering calculations directly impacts the amount of data that is needed from logs and core data. A history match/simulation study of a field undergoing tertiary solvent injection was done to confirm the techniques proposed. The study presents various dimensionless groups and their criteria in order to determine if permeability variations can be averaged together, or if they must be directly modeled. Introduction Most Enhanced Oil Recovery (EOR) and waterflood production forecasts are based on numerical simulation models. In the past these models have been poor at predicting future performance even when history matching was done. According to Archer the success of a numerical model depends on two particular conditions:–the ability of the equations to represent the physics of flow and equilibrium in the reservoir/well system.–the ability of the cell or node properties to represent the three dimensional reservoir description. At the present time the authors believe the ability to approach the first condition is further advanced than the second, at least for non EOR, non-compositional formulations. This paper concentrates on the ability of cell properties to represent the reservoir description. In the past, numerical models have been shown to be overly simplistic in modeling geological heterogeneity. Overly simplistic models have resulted in unrealistic estimates for ultimate recovery for many secondary and tertiary recovery schemes. The heterogeneity or variance of the rock is a key factor in predicting waterflood or EOR recovery. It is not possible to make accurate performance predictions of EOR schemes unless reservoir characterization is adequate. The goal of reservoir characterization is to place more emphasis on the actual geology of the reservoir so it can be more accurately modeled. Reservoir numerical simulation studies implicitly assume that the reservoir rock properties are homogeneous on the scale of the grid block used. Or alternatively they implicitly assume that heterogeneities on smaller scales are not worth modeling. Detailed geological work has shown that heterogeneity is usually present at all scale levels and therefore it is impossible in practical simulations to use a grid which is fine enough to resolve all scales of heterogeneity. P. 131