A Rolling Horizon Framework for Intermodal Loading Assignment to Improve Fuel Efficiency

Aerodynamic efficiency of intermodal freight trains can be significantly improved by minimizing the adjusted gap lengths between adjacent loads. This paper first develops a static model to optimize load placement on a sequence of intermodal trains that have scheduled departure times. This model applies when full information on all trains and loads is available. Then, a dynamic model is developed to account for the realistic situation where information on future trains and incoming loads may not be completely available. This paper seeks ways to balance between: (i) the advantage from optimizing multiple trains together; and (ii) the risk of making suboptimal decisions due to incomplete future information. We propose a rolling horizon scheme to address this challenge, where exponentially decreasing weights are assigned to the objective functions of future trains. Numerical results based on empirical data show significant aerodynamic efficiency benefits from these optimization models.