A new robust counterpart model for uncertain linear programming problems

Many practical decision-making problems involve a significant level of data uncertainty. In such a case, modeling the uncertainty involved is critical to making informed decisions. The set-based robust optimization approach is one of the most efficient techniques for finding optimal decisions in problems involving uncertain data. The main concern with this technique is over-conservatism. This drawback has been widely investigated, and several robust formulations have been developed in the literature to deal with it. However, research is still ongoing to obtain effective formulations to handle uncertainty. In this study, we derive a robust counterpart formulation for an uncertain linear programming problem under a new uncertainty set that is defined based on a pairwise comparison of perturbation variables. The performance of the proposed robust formulation is evaluated using numerical studies and in terms of different performance metrics. For this purpose, robust counterpart models corresponding to the production-mix sample problems are solved at different protection levels. Then, for each solution obtained, violation probability is calculated using a Monte-Carlo simulation approach. The results revealed that the proposed method outperforms the existing ones.