Electric Vehicle Charging Network Design under Congestion
Antoine Deza, Kai Huang, Carlos Aníbal Suárez
TL;DR
This paper extends a multistage stochastic integer model for EV charging network design by incorporating congestion constraints through a queuing-based formulation under Poisson arrivals. It develops an approximation method, a refined heuristic, and a branch-and-price algorithm (via Dantzig-Wolfe decomposition) to solve the enlarged model, and demonstrates that congestion-aware planning can yield different, potentially more efficient deployment patterns for medium-sized networks. The approach balances capital and operating costs with congestion performance, and numerical experiments show that the congestion constraints significantly affect deployment decisions and queue metrics while providing scalable solution options. Overall, the work advances practical infrastructure planning by integrating queueing dynamics into stochastic capacity-expansion for EV charging networks.
Abstract
This paper presents an extension of a recently introduced multistage stochastic integer model designed for optimizing the deployment of charging stations under uncertainty. A key contribution of this work is incorporating additional constraints accounting for congestion management at charging stations. The solution approach combines a greedy heuristic with a branch-and-price algorithm, enabling the efficient handling of medium instances. In the branch-and-price algorithm, when the solution to the restricted master problem is not integer, a greedy heuristic and a local search procedure are conducted to obtain feasible solutions. Computational experiments illustrate the effectiveness of the proposed framework.