Alternative-fuel station network design under impact of station failures

In this paper, we have formulated a mixed-integer non-linear programming model for alternative-fuel station location problem in which each station can fail with a site-specific probability. The model aims to maximise the total expected traffic volume that can be refuelled by the unreliable alternative-fuel stations. Based on the linearisation techniques, i.e., probability chains and piecewise-linear functions, we linearise the non-linearity of compound probability terms in the non-linear model to solve this problem efficiently. An efficient Tabu search algorithm is also developed to solve the large-size instances. In addition, we extend the model to deal with reliable multi-period alternative-fuel station network design. Computational experiments, carried out on the well-known benchmark instances where the probability of station failures is uniformly generated, show that the proposed models and algorithm can obtain the optimal solutions within a reasonable computation time. Compared to a standard station location model that disregards the potential for station failures, our model designs more reliable alternative-fuel station network under risk of station failures. A sensitivity analysis of failure probabilities in the station network design is investigated to demonstrate the robustness of our model and study how variability in the probability of station failure affects solution robustness.