Risk quantification combining geostatistical realizations and discretized latin hypercube

Abstract

This work presents an alternative way to combine different types of uncertainty to quantify risk in petroleum field development. Risk quantification is key in decision analysis. Some areas need special attention, namely: (1) generating simulated scenarios compatible with geological models and (2) statistical techniques that address different types of uncertainty (especially continuous and discrete attributes, and realizations represented by geostatistical images) using the fewest possible simulation runs. Several statistical techniques address this, but most present significant drawbacks, potentially yielding incorrect risk quantification or demanding excessive time (for simulation runs) to reach good results. This simple, efficient methodology combines geostatistical realizations with other types of uncertainty (e.g., reservoir structure, fluid characterization and economic parameters) using a Discretized Latin Hypercube sampling. To verify the results, we applied the methodology to the UNISIM-I-D benchmark case showing that the method can be applied to a complex case yielding good results. We found the methodology to meet our initial objectives, to reliably and easily quantify risk within a minimal timeframe