Abstract

This thesis is focused on the supply chain disruptions and it reviews cost-efficient risk mitigation strategies to sustain supply chain functionality when disruptions occur. In particular, we study the robust design of supply flow subject to minor operational risks and major disruptions. The contingent sourcing along with strategic stock is incorporated as risk management strategies. We consider a firm with two suppliers where the main supplier is cost-effective but prone to disruptions and the back-up supplier is reliable but expensive. The back-up supplier can scale up its capacity according to a speed related to its configuration in order to supply the required flow of material when the main supplier disrupts. When minor disruption occurs, the strategic stock can cover the losses. The design problem considered is to determine optimal strategic stock level and response speed of volume-flexible back-up supplier.
The back-up supplier might not provide the required supply level instantaneously due to non-steady production state and congestion during the response time. Therefore, there could be material shortages if the actual level of available capacity during the response time is ignored. The first chapter includes the incorporation of the clearing function into a contingency capacity planning model in order to represent the impact of congestion. The appropriate response speed is selected through a decision tree analysis considering different attitudes of the decision maker towards risk. The results show that considering congestion impact is especially critical for risk-neutral decision makers. The second chapter considers the randomness associated with the available capacity through a two-stage robust optimization model. The results show improvement in the quality of optimal solution by considering the randomness. The objective in the third chapter is to find an equitable solution which has an efficient performance with respect to all plausible scenarios. Therefore, the Ordered Weighted Averaging aggregation operator is incorporated in the objective function of a MIP robust model. In order to address the computational complexity associated with large set of scenarios, a novel clustering based scenario reduction model based on location covering model is proposed. The results show that the proposed methodology provide an accurate reduced scenario set within relatively short computational time.