An end-to-end hydrogen supply chain framework with a two-stage model

Abstract

Byproduct hydrogen offers a cost-effective entry into hydrogen energy. However, the inherent complexity of hydrogen supply chain (HSC) presents hurdles in production, transport, and utilization, leading to low utilization and increased costs. Hydrogen supply chain network design (HSCND) and hydrogen fueling station planning (HFSP) both aim to optimize HSC from different angles, but integrating these two into a single framework poses challenges. Furthermore, HSC design directly influences hydrogen fuel cell vehicle (HFCV) adoption, influenced by factors like vehicle registration, driving tolerance range, traveling range, and more. The aim of this study is to develop an end-to-end supply chain framework that simultaneously considers HSCND and HFSP, encompassing all segments of the HSC. We design a two-stage model that aligns with each stage of HSC. From the end-user’s (station) perspective, we find optimal locations of Hydrogen Refueling Stations (HRSs) on expressways by means of a capacitated flow-refueling location model (CFRLM) combined with a hydrogen vehicle refueling station logic. This model considers government policy and consumer adoption factors with the goal of striking a balance between consumer interests and HRS decision-makers. For transporting byproduct production, we use a Voronoi Diagram (VD)-based capacitated vehicle routing model (CVRP) to deliver byproduct hydrogen from industrial areas to stations. The VD is employed to determine the assignment of HRSs in different target periods. Finally, we propose a modified branch-and-cut (BC) algorithm to solve the VD-based CVRP, which significantly outperforms commercial solvers in both speed and solution quality. Through sensitivity analysis, we identify critical factors impacting overall cost and hydrogen utilization. Tested on real and benchmark datasets, our algorithm increases byproduct hydrogen utilization by over 10% overall, reaching up to 70% in key industrial areas, while also achieving maximum path and flow coverage.