A Framework for Transportation Network Optimization for Real-World Logistics: Extensions and Applications

Le résumé fourni par la source

Transportation network optimization is traditionally stated as a classical transportation problem in which supply is allocated to demand at minimum cost. While elegant in an analytical sense, the underlying formulation lacks aspects that are now dominating the real world of logistics: time dependence, binding capacities, multimodal transfers and uncertainty. This paper proposes an integrated framework which retains the interpretability of minimum-cost flow and adds extra dimensions to it along four realism axes. In the baseline, explicit penalty terms for surplus and unmet demand are included to reflect pragmatic trade-offs when exact balance is infeasible. First, spatial routing is linked with intertemporal decisions under inventory carryover and storage costs in a transshipment model. Second, capacity constraints are placed on arcs and facilities so as to obviate a bounded operationally credible feasible set. Third, multimodal routing is added using mode-specific arcs and transfer rules that represent physical compatibilities and service frictions. Fourth, uncertainty in costs, capacities and demands is factored in by separating anticipative decision from adaptive recourse. A dual-price-based diagnostic layer, parametric ranges and scenario testing are used to transform the results of the optimization into management actions such as expanding capacity, mode shifting and buffer-stock. The proposed formulation is a trade-off between tractability and operational fidelity, and provides the building block for scalable solution strategies (exact when possible and heuristic when necessary) that are appropriate for large-scale, practical logistics systems.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.