We deal here with the routing of vehicles in charge of performing internal logistics tasks inside some protected area. Those vehicles are provided in energy by a local solar hydrogen production facility, with limited storage and time-dependent production capacities. One wants to synchronize energy production and consumption in order to minimize both production and routing costs. Because of the complexity of resulting bi-level model, we deal with it by shortcutting the production scheduling level with the help of surrogate estimators, whose values are estimated through fast dynamic programming algorithms and through machine learning.