Data-calibrated cost-aware freight routing in a mixed diesel-electric railway network: a multi-corridor study from Kazakhstan

Rail freight route choice often combines distance, train-formation and operating rules without making the cost effects of heterogeneous traction explicit. This study evaluates a transparent, data-calibrated route-cost framework for strategic planning on Kazakhstan’s mixed diesel-electric mainline network. Historical train, locomotive, energy and crew indicators are converted into segment-specific productivity parameters; shortest- and k-shortest-path logic generates candidates; and route costs combine energy, crew and locomotive maintenance, repair and overhaul. A 2019 planning workbook based on 2018 data contains 36 route records for 14 purposively selected origin–destination flows, reconciled into 34 cost profiles. Relative to current Train Formation Plan routes, the minimum-cost available candidates are cheaper for eight flows and reduce aggregate modelled variable cost by 12.64%. Seven of these eight candidates are longer. A minimum-distance candidate-set benchmark costs 8.39% more than current routing and 24.07% more than the minimum-cost candidates, confirming that distance is an incomplete cost proxy in the evaluated cases. Six attractive alternatives carry source notes requesting capacity or transit-time verification; excluding them through a conservative analyst-note screen reduces aggregate potential to 1.79%. Across 81 factorial scenarios varying diesel, electricity, crew and MRO costs by ± 20%, aggregate reductions range from 7.53% to 18.26% before the screen and from 1.43% to 2.53% after it. The results are internal model comparisons, not external validation or realized savings. The framework is therefore best used as an auditable economic layer for identifying corridors that warrant subsequent capacity-constrained analysis.