ABSTRACT

Predrill estimates of reservoir quality (porosity, permeability, and net thickness) aid prospect risk assessment. Four approaches can be used to predict sandstone and carbonate reservoir quality: seismic detection, analog/statistical prediction, petrological analysis, and numerical modeling.

In favorable settings, seismic records can detect lateral variations in reservoir quality. Because lithology variation causes a response similar to porosity variation, local calibration is necessary for successful application.

Statistical prediction methods are based on databases that incorporate reservoir-quality uncertainty with predictive variables such as depth or burial history. Porosity is usually the reservoir quality of interest, but permeability may be predicted from porosity and fabric where properly calibrated. Reservoir quality risk is estimated from cumulative probability curves. Analog/statistical predictions are only as good as the analogs; where analogs are poorly chosen, results are poor.

Petrological approaches characterize diagenetic patterns and aid prediction of permeability from porosity. Specific depositional and diagenetic controls on reservoir quality can only be identified by this approach. Once reservoir-quality controls are identified, process and statistical models can extrapolate these results to other locations.

Numerical reservoir-quality process modeling is used where no good analogs are available, such as deep, rank-wildcat wells. The most successful models are based on quartz cementation in quartzose sandstone. The burial diagenetic models available for predicting average carbonate porosity are not yet reliable. Available numerical models for permeability prediction are complex and unreliable, reflecting the many controls on permeability evolution.

The real strength of reservoir-quality predictive technology comes from convergence of different approaches to the same answer. Two examples (a Cretaceous limestone and a Miocene shelf sandstone) are examined to illustrate how these reservoir-quality prediction techniques can be used.