An analysis of robustness and flexibility in freight transportation systems

Abstract

Freight transportation is a complex large scale system that operates under a highly dynamic and uncertain environment. Due to the scale and complexity of the system, a highly interdependent set of decisions are made across multiple planning levels. The interaction between tactical level decisions and execution level decisions largely determines the overall effectiveness of the system. This thesis aims to provide an analysis of how the interaction between the degree of robustness of a tactical plan and the flexibility level of an execution policy affects the performance of freight transportation systems in a dynamic and uncertain environment. Such analysis is conducted under two types of transportation systems, including a generalized distribution system and a military logistic system, that operate under demand uncertainty. Execution policies with different levels of flexibility - defined as the degree of freedom to which decisions can be adjusted at the executional level - are obtained by controlling the stickiness level of the different decisions as the decision maker transitions from the tactical plan to the executional plan. Robustness is used as a metric to measure the ability of a system to withstand random changes. Tactical plans with various degrees of robustness are obtained through the use of robust optimization. The performances of tactical plans with various degrees of robustness and execution policies with different levels of flexibility are evaluated through simulation. Results from the analysis on the distribution system show that the optimal level of robustness required from a tactical plan to achieve the lowest expected total cost decreases as the flexibility level of the system increases. Finally, the analysis on the military logistics system shows that the effect of increasing the flexibility level of the execution policy on the performance of the system depends on the uncertainty level of the demand.