Selection of representative geological models: multiobjective analysis

The decision-making process in oil field development and management involves analyzing production strategies. To define an efficient strategy, geological models composed by uncertain variables must be evaluated. However, combining these variables results in an exponential number of models. To make the definition of an appropriate production strategy feasible, it is essential to select a reduced subset of representative models (RMs). We propose a multiobjective approach that evaluates and selects a Pareto front of RM solutions. Two objective functions were defined and minimized by NSGA-III: the first evaluates the representativeness of the RMs in an ℝ⁵ space, while the second assesses their distribution across the percentiles of the production variables. The algorithm was applied to the UNISIM-I-D benchmark, and results were compared with the solution proposed by the RMFinder tool. The findings indicate the potential of the multiobjective approach for RMs selection and suggest further evaluation of the objective functions.