Stochastic reservoir description techniques are combined with a three-dimensional streamline simulator to optimize volumetric sweep efficiency in a mature, West Texas waterflood. Results from the streamline model were validated using an IMPES, finite-difference scheme. It is shown that the combination of multiple-realization property distribution with an efficient streamline model is an attractive alternative to the traditional, finite-difference approach. Performance predictions are made for the placement of new infill wells (vertical and horizontal) and for pattern modification by selectively shutting in existing injectors.

Geostatistical techniques including kriging and cokriging are used to generate realizations of property distributions. Fundamental to the geostatistical techniques described is the honoring of hard data and we illustrate the importance of integrating various data sources to constrain these techniques. Specifically, production, core, log, pressure transient, geological and 3D seismic data were combined to estimate near-well properties and to place realistic constraints on distributing porosity, permeability and net-to-gross thickness.

The advantages and the applicability of the streamline approach are outlined as well as the caveats. Primarily, it was found that streamline simulation is much more efficient in solving two-phase, immiscible problems and produces results equivalent to finite-difference methods.

You can access this article if you purchase or spend a download.