Summary
Reservoir simulation is widely used to investigate the effects of geologic heterogeneity and engineering parameter variability on reservoir production behavior. Because many geologic and engineering factors interact to affect recovery behavior, an exhaustive examination of variations due to all possible parameter combinations is prohibitively expensive. The factors that most strongly influence production behavior must be identified to focus analyses and measurements. Experimental design varies multiple factors simultaneously and minimizes the number of simulations required to construct accurate response models. Response models can statistically test the relative importance of the factors examined in designed simulations.
In this paper, a method based on response surfaces, Monte Carlo simulation, and Bayes’ theorem (RSMCB) was used examine the effects of several types of geologic variability and different modeling methods. The method is demonstrated in an analysis of outcrop data from a heterolithic tide-influenced deltaic sandstone. The data set includes thousands of probe permeameter measurements, detailed bedding maps, shale diagrams, cement maps, facies maps, facies-specific rock properties, and variograms.
Because of the importance of geostatistical methods for reservoir model construction, designed simulations examined the effects of varying geostatistical parameters and compared geostatistical and quasideterministic models of geologic variability. Intrafacies variations have statistically significant effects at the bed scale, although shale resistance dominates these effects.
Improvements in reservoir model accuracy were quantified by computing the expected changes in factor uncertainties resulting from response measurements. Posterior distributions of geologic and geostatistical parameters were estimated from flow responses. Monte Carlo simulations with prior and posterior factor distributions quantify the uncertainty reduction caused by measurements. Uncertainty reduction can be compared to measurement cost, enabling better informed measurement decisions.
