Assessing the influence of the production strategy on the selection process of representative models

Abstract

In reservoir engineering studies related to production optimization under uncertainty, the reservoir is often represented by a large set of reservoir models. Each model represents one reservoir scenario, i.e., one combination of all reservoir uncertainties. Because reservoir simulation is time-consuming, representative models (RM) can be used to increase the computational efficiency of processes related to production optimization and decision-making (Schiozer et al., 2019). The RM can be selected using the proposal by Meira et al. (2017), which combines a mathematical function that captures the representativeness of a set of models with a metaheuristic optimization algorithm, to ensure full representation of the probability distribution of uncertain attributes and the variability of production, injection, and economic forecasts. Previous applications have used, as input for RM selection, the production forecasts obtained from an initial production strategy optimized for the base case. In most of these applications, the RM were used for production optimization (either specialized or robust optimization), meaning that the set of RM was assumed to adequately represent the uncertain system for optimization purposes. However, this representativeness has not yet been demonstrated, especially when different production strategies are under consideration for field development. As a result, this study was formulated to investigate the intrinsic relation between production strategy and production forecasts, and its effects on the representativeness of the RM. The main objective of this proposal is to assess the influence of the production strategy on the selection process of representative models. Specific objectives of this study include: · Assess if the RM selected for one production strategy are representative for other strategies; · Assess how the objective functions are influenced by the production strategy used in the RM selection; · Quantify the loss of representativeness of the RM when the production strategy is modified. Finally, this work proposal aims to identify ways to minimize such influence and increase RM representativeness, so that reliable decision-analysis processes are ensured in the 12-step decision structure of Schiozer et al. (2019). (AU)