Print this pageAn Empirically Calibrated Model for Sandstone Reservoir Quality Prediction
LANDER, R. H., and O. WALDERHAU
Presently available approaches to quantitative reservoir quality prediction tend to be based on empirical functions or thermodynamic/kinetic/fluid flow simulators. These approaches typically are limited in applicability to specific lithostratigraphic units, burial depth ranges, and geographic areas; require input data that are difficult or impossible to obtain; or demand specialized high-end computer hardware. We have developed a forward numerical model based on empirically calibrated models of compaction and quartz cementation in an attempt to provide a method better suited for sandstone porosity prediction in both mature and frontier basin settings. While the model in its present state is comparatively simple it provides accurate predictions for many quartz- rich sandstones using commonly available geologic data as input. In addition, the model incorporates risk analysis through the application of parameter optimization and Monte Carlo techniques.
The model simulates compaction and quartz cementation from the time of deposition to present and the frame of reference is designed to lead to predictions that can be readily compared with data obtained from petrographic thin sections. Input data required for a simulation include effective stress and temperature histories together with the depositional composition and grain size of the modeled sandstone. The model is able to reproduce observed correspondences between quartz cementation, thermal history, grain size, and grain coatings (including chlorite and microcrystalline quartz) as well as between compaction and the onset of fluid overpressure development.
The computational execution speeds of the model are fast (e.g., seconds) on desktop computers, making integration with parameter optimization and Monte Carlo simulation techniques practical. Routines are used to obtain distributions of optimized parameter values when petrographic and basin modeling data are available for calibration. These distributions provide the basis for rigorous evaluation of inherent model uncertainties in pre-drill reservoir quality predictions when they are incorporated into Monte Carlo simulations. Monte Carlo-based predictions also include uncertainties associated with the input parameters describing burial history and initial sandstone composition and texture making it possible to apply the model to frontier basin settings.