Development and testing of advanced methods for the screening of Enhanced-Oil-Recovery techniques

Enhanced Oil Recovery (EOR) techniques must undergo preliminary laboratory and pilot testing before implementation to field-wide scale, and the whole evaluation process requires heavy investments. Hence forecasting EOR potential is a key decision-making element. A critical difference amongst EOR techniques resides in the oil-displacement mechanism upon which they are based. The effectiveness of these mechanisms depends on oil and reservoir properties. As such, similar EOR techniques are typically successful in fields sharing similar features. Here we implement and test a screening method aimed at estimating the optimal EOR technique for a target reservoir. Our approach relies on the information content tied to an exhaustive set of EOR field experiences. The basic screening criterion is the analogy with known reservoir settings in terms of oil and formation properties. Analogy is assessed by grouping fields into clusters: we rely on a Bayesian hierarchical clustering algorithm, whose main advantage is that the number of clusters is not set a priori but stems from data statistics. As a test bed, we perform a blind test of our screening approach by considering 2 fields operated by eni. Our predictions for analogy assessment are in agreement with the EOR techniques applied or planned in these fields.