A probability-based multi-path alternative fueling station location model

TL;DR

This paper introduces a probability-based multi-path model for optimally locating alternative fueling stations on transportation networks, accounting for demand uncertainty and maximizing expected coverage.

Contribution

It presents a novel probabilistic multi-path location model for AFS deployment, incorporating demand uncertainty and using a genetic algorithm heuristic for complex optimization.

Findings

Model effectively accounts for demand probability distribution.

Genetic algorithm provides feasible solutions for NP-hard problem.

Numerical experiments validate model applicability and heuristic efficiency.

Abstract

We develop a probability-based multi-path location model for an optimal deployment of alternative fueling stations (AFSs) on a transportation network. Distinct from prior research efforts in AFS problems, in which all demands are deemed as given and fixed, this study takes into account that not every node on the network will be equally probable as a demand node. We explicitly integrates the probability into the model based on the multi-path refueling location model to determine the optimal station locations with the goal to maximize the expected total coverage of all demand nodes on a system level. The resulting mixed integer linear pro- gram (MILP) is NP-hard. A heuristic based on genetic algorithm is developed to overcome the computational challenges. We have conducted extensive numerical experiments based on the benchmark Sioux Falls network to justify the applicability of the proposed model and heuristic solution methods.