Abstract:The problem of hydrogen station location is often studied through understanding refueling behavior or reviewing the experience of gasoline stations. Driven by the notion "where you drive more is where you more likely need refueling", this paper develops a new approach where station siting is treated as a fuel-travel-back problem and the only required data is VMT distribution. Such a fuel-travel-back problem is a typical transportation problem and is solved as mix-integerprogramming model. When the total fuel-travel-back time is minimized, so is the average refueling travel time of a random motorist, for which theoretical deduction is provided. The model is applied to derive an optimal station roll-out scheme for Southern California. The results show that, if station size constraints are relaxed, only 18% of existing gas station number is needed to achieve the current fuel accessibility of gasoline in the region. Fewer stations lead to larger station size, suggesting a need to re-examine the current speculation on designs of hydrogen station and distribution system and to conduct more regional studies for discovery of optimistic and pessimistic regions for hydrogen. The results also indicate that early stations should be located strategically and even at low-demand locations, which is contradictory to existing proposition.