Intelligent decision modeling for optimizing railway cold chain service networks under uncertainty

Abstract

Railway cold chain service network design (RCC-SND) aims to optimize the utilization of stations and lines as well as train allocations in a manner that minimizes costs while satisfying the service requirements of shippers. Furthermore, the uncertainties associated with freight demand, transportation costs, quality loss, station handling capacity, and arc capacity make the RCC-SND a complex decision-making problem. To tackle this challenge, we first formulate a Mixed-Integer Nonlinear Programming (MINLP) model to determine hub locations, freight wagon flows, and service frequency. To cope with uncertain parameters with varying degrees of uncertainty incorporated in the model, we extend the problem using fuzzy programming and further convert it to its crisp counterpart. A real-world cases study in Southwest China is performed to validate the proposed model, whose results provide different strategies for decision-makers with varying preferences. There are some main findings: As the number of hubs increases from 5 to 6, a maximum total cost savings of 1.99% can be achieved. Railway operators may opt for different decision preferences, for decisions prioritizing economic efficiency, the cost can decrease by 2.69% compared to deterministic optimization.