Joint optimization of train blocking and shipment path:An integrated model and a sequential algorithm

Chongshuang Chen,Jun Zhao
DOI: https://doi.org/10.48550/arXiv.1909.11310
2019-09-29
Abstract:The INFORMS RAS 2019 Problem Solving Competition is focused on the integrated train blocking and shipment path (TBSP) optimization for tonnage-based operating railways. In nature, the TBSP problem could be viewed as a multi-commodity network design problem with a double-layer network structure. By introducing a directed physical railway network and a directed train services (blocks) network, we formulate completely the TBSP problem as a mixed integer linear programming (MILP) model that incorporates all decisions, objectives and constraints especially the merge flow (called intree rule here) in an integrated manner. The scale of the MILP model can be reduced efficiently if we only enumerate the arc selection and block sequence variables for each shipment on the legal paths from its origin to its destination satisfying the given detour ratio. We further develop a sequential algorithm that decomposes the TBSP problem into the shipment path subproblem and train blocking subproblem which are solved sequentially. Computational tests on the three given data sets show that the reduced MILP model can solve DataSet\_1 to optimality in 8.48 seconds and DataSet\_2 with a gap of 0.16\% in 6 hours on a GPU workstation. The reduced model also can provide strong lower bounds for DataSet\_2 and DataSet\_3. The sequential algorithm can find a high quality solution with 0.04\% gap within 0.26 seconds for DataSet\_1, 0.42\% gap within 4.53 seconds for DataSet\_2 and 1.54\% gap within 0.58 hours for DataSet\_3 respectively on a Thinkpad laptop.
Optimization and Control,Networking and Internet Architecture
What problem does this paper attempt to address?