RELIABLE PLANNING OF HINTERLAND-PORT FREIGHT NETWORK AGAINST TRANSFER DISRUPTION RISKS

Abstract

Many previous cases have shown that port operations are susceptible to disruptive events. This paper proposes 2-stage Stochastic Programming (SP) for port users to reliably plan the hinterland-port intermodal freight network with consideration of risk aversion in cost. Probabilistic disruptions of intermodal terminals are considered as scenario-specific. In the 1st stage, intermodal paths are selected to obtain proper network capacities. In the 2nd stage, cargo flows are assigned for each disruption scenario on the planed network. The 2-stage model is firstly formulated in a risk-neutral environment to achieve the minimum expectation of total cost. Then, the Mean-Risk (MR) framework is adopted by incorporating a risk measure tool called Conditional Value-at-Risk (CVaR) into the expectation model, so as to reduce the cost of worst-case disruption scenarios. Benders’ Decomposition (BD) is introduced to efficiently solve the exponential many problem. Some numerical experiments are performed under different risk aversion parameters. With this study, network planners can decide network capacities with reasonable redundancies to improve the freight reliability in a cost-effective way. The proposed method provides a simple approach for the planners to quantify their risk appetites in cost and to impose them in the planning process, hence to trade-off the Expected Cost (EC) and the worst-case cost.