Optimization of Tactical Freight Rail Export Planning in Minas Gerais: A Service Network Design Approach with Selective Backhaul

The tactical planning of freight rail operations in export corridors presents challenges arising from rugged topography, infrastructure constraints, and the need to meet contractual time windows. This work proposes a mathematical formulation based on the Service Network Design (SND) problem, extended to encompass a heterogeneous fleet, selective backhaul routing, and dynamic CO₂ emissions quantification. The model endogenizes the decision to capture return loads by evaluating the balance between revenue and traction and shunting costs. The Mixed-Integer Programming (MIP) formulation introduces continuous temporal propagation via arc sequencing that eliminates infeasible solutions. Experiments were conducted with real ANTT data for January 2024 on the EFVM corridor, yielding a 30.8% profit gain and a 56.3% reduction in CO₂ emissions relative to a no-backhaul baseline.