A Study On Robust Decision Making Under Uncertainty

The main contribution of this thesis is to develop and apply robust optimization approaches to facilitate better decisions in an environment with uncertainty. From the aspect of theoretical contributions, we build upon recent advances in the domain of robust optimization techniques to develop models dealing with uncertainty. From the aspect of practical application contributions, we provide a framework of applying robust optimization approaches to supply chain management and revenue management problems.More specifically, we investigate three classes of robust optimization approaches.First, the proposed robust optimization approach integrates the concept of goal programming and the scenario-based description of unknown data, the philosophy of which is built upon the trade-off between solution robustness and model robustness. We apply this class of robust optimization approach to study the supplier selection and allocation in outsourcing problem and the bottleneck generalized assignment problem under uncertainty. Second, a novel robust optimization approach handling linear programming with right-hand-side uncertainty is developed by incorporating new parameters: uncertainty level, infeasibility tolerance and reliability level. Two types of uncertainty, namely, bounded uncertainty and symmetric uncertainty are considered,respectively. We apply the proposed robust optimization models to address the container slot allocation problem with Minimum Quantity Commitment under uncertain demand, which is faced by some companies in the international trade industry.Third, a computationally tractable robust optimization approach avoiding the worst case is presented by modelling the uncertain parameters in a polyhedral set, which does not require exact information of the underlying probability distributions. Robust capacity planning policy is determined within this framework.