Explicit stochastic or robust treatment of fluctuating classification-yard sorting capacity in the integrated traffic routing and train formation plan problem

The integrated traffic routing and train formation plan problem (TR&TFP) jointly decides, for a rail freight network of classification yards, which train blocks to form between yards, the service frequencies, the physical paths of shipments, and the block sequences along each shipment's journey. The reference model of Lin et al. (2021) treats every yard's sorting (reclassification) capacity and its number of available classification tracks as fixed deterministic values, hedging capacity risk only through a constant proportional margin, even though the effective sorting capacity of a yard fluctuates with equipment and labor resources. The open problem, stated by the source paper as its key future direction, is to formulate and integrate these capacity fluctuations explicitly into the integrated TR&TFP, together with a tractable solution methodology, and to demonstrate improved real-world performance and resilience of the resulting train service plans. Published follow-ups have addressed demand uncertainty and a same-group arXiv preprint has given a fuzzy elastic-capacity formulation without solution or validation, but no published work treats fluctuating sorting capacity itself as an uncertain parameter inside the integrated model.

Background

In long-haul rail freight, shipments (groups of freight cars sharing an origin and a destination) rarely travel on one train from origin to destination. Cars are consolidated into blocks — groups of cars that stay together as a single unit between two yards — and are reclassified (sorted) at intermediate classification (marshalling) yards, where humps and switches sort inbound cars onto classification tracks to assemble outbound blocks. The train formation plan (TFP) is the tactical plan, typically revised every few years, deciding which blocks to form between which yards and at what train frequencies; traffic routing (TR) decides the physical path of each shipment over the network and hence the sequence of reclassifications it undergoes. Operating cost is dominated by car accumulation at yards (waiting for a block to fill), reclassification work, and travel distance, so TFP and TR are strongly coupled, and solving them sequentially (the historical standard practice) sacrifices global optimality.

Lin, Zhao, Lin and Liu (2021) therefore formulated the integrated TR&TFP as a non-linear binary program that simultaneously selects blocks, train frequencies, shipment paths, and block sequences, minimizing accumulation, reclassification, and travel costs subject to link capacities, yard reclassification capacities, and limits on the number of classification tracks. A simulated-annealing heuristic (a randomized local-search metaheuristic) with a penalty-function treatment of the hard capacity constraints solves it at national scale, validated on the east-west railway channel of China and a 235-node model of the China railway network. All capacities in this model are deterministic: the reclassification capacity $C^{R}_{k}$ of yard $k$ is scaled by a constant proportional factor $\theta_k$ that reserves a margin against overload, giving constraints of the form $x_{ij}\le\theta_k C^{R}_{k}$.

In reality the effective sorting capacity of a classification yard is not fixed: hump and yard equipment availability, maintenance, and staffing fluctuate from day to day, so the same tactical plan can be comfortably feasible in one week and strained in the next, and a constant $\theta_k$ margin is a crude, data-blind hedge. The source authors accordingly identify the explicit formulation and integration of capacity fluctuations — potentially through stochastic parameters or robust optimization — as a key direction for improving the real-world performance and resilience of the TR&TFP. The uncertainty-modeling machinery for this exists in neighboring literature: chance-constrained stochastic programming admits constraints that must hold with a prescribed probability under random parameters; robust optimization, notably the budgeted uncertainty model of Bertsimas and Sim, protects against coefficient deviations within a cardinality-limited set while often retaining mixed-integer tractability; and fuzzy elastic constraints soften a capacity limit through a membership function that scores how fully the limit is used.

Subsequent work has advanced the uncertainty agenda only on the demand side. Aghaee, Alinaghian and Aghaee (2024) formulated a robust counterpart of the integrated traffic routing and train formation model with yard capacity constraints and demand uncertainty using Bertsimas–Sim robustness; a series of papers treats train formation at technical stations under fluctuating wagon (demand) flow via chance constraints or robust optimization; and a dynamic train formation framework (Chi, He and Song 2025) integrates railcar-to-track assignment while reacting to demand fluctuations with deterministic capacities. The same research group as the source paper released an arXiv preprint (Lin, Wang and Zhao, 2021) reformulating the integrated model with elastic yard and link capacities expressed by fuzzy membership functions, but it presents the model only, with no solution algorithm or computational validation, and it has not appeared in a journal. No published study makes the fluctuating sorting capacity itself an uncertain parameter inside the integrated TR&TFP and quantifies the resulting benefit.

Problem Statement

As called for by Lin et al. (2021), make the fluctuating sorting capacity of classification yards an explicit modeled quantity in the integrated traffic routing and train formation plan problem: reformulate the integrated TR&TFP so that yard reclassification capacities and/or the number of available classification tracks are no longer fixed deterministic values scaled by a constant margin $\theta_k$, but instead appear as stochastic or uncertain parameters, or through an equivalent explicit treatment of capacity fluctuation driven by yard equipment and labor resources; develop a solution methodology able to solve the resulting model on realistic railway networks; and demonstrate computationally, against the deterministic margin-based model, whether and by how much the resulting train service plans improve real-world performance and resilience under capacity fluctuations.

References

1. Lin, Boliang; Zhao, Yinan; Lin, Ruixi; Liu, Chang (2021). Integrating traffic routing optimization and train formation plan using simulated annealing algorithm. Applied Mathematical Modelling 93:811-830. https://doi.org/10.1016/j.apm.2020.12.031
2. Lin, Boliang (2017). Integrating car path optimization with train formation plan: a non-linear binary programming model and simulated annealing based heuristics. arXiv:1707.08326. https://arxiv.org/abs/1707.08326
3. Lin, Boliang; Wang, Zhenyu; Zhao, Yinan (2021). A Train Formation Plan with Elastic Capacity for Large-Scale Rail Networks. arXiv:2111.03473. https://arxiv.org/abs/2111.03473
4. Aghaee, Shima; Alinaghian, Mehdi; Aghaee, Mohammad (2024). Robust Integrated Model for Traffic Routing Optimization and Train Formation Plan with Yard Capacity Constraints and Demand Uncertainty. International Journal of Intelligent Transportation Systems Research 22(2):407-415. https://doi.org/10.1007/s13177-024-00406-3
5. Bertsimas, Dimitris; Sim, Melvyn (2004). The Price of Robustness. Operations Research 52(1):35-53. https://doi.org/10.1287/opre.1030.0065
6. Li, Bing; Jiang, Shangtao; Zhou, Yanjie; Xuan, Hua (2023). Optimization of train formation plan based on technical station under railcar demand fluctuation. Journal of Industrial and Production Engineering 40(6):448-463. https://doi.org/10.1080/21681015.2023.2221699
7. Li, Bing; Bian, Menglin; Zhou, Yanjie; Xuan, Hua (2026). Hybrid train formation plan integrating one-block and two-block trains from a technical station under wagon demand fluctuation. Engineering Optimization 58(2):585-617. https://doi.org/10.1080/0305215X.2025.2475381
8. Li, Bing; Hua, Zhaoyan; Xue, Yang; Zhang, Jiabei; Xuan, Hua (2026). Robust optimization of the formation plan of a direct train originating at a technical station under uncertain wagon flow conditions. Engineering Optimization, online first:1-23. https://doi.org/10.1080/0305215X.2026.2693731
9. Chi, Jushang; He, Shiwei; Song, Rui (2025). Optimization Framework of Dynamic Train Formation Planning in Combination With a Railcar-to-Track Assignment. IEEE Transactions on Intelligent Transportation Systems 26(10):16017-16031. https://doi.org/10.1109/TITS.2025.3584535
10. Delbart, Thibault; Braekers, Kris; Caris, An (2026). Three-stage service network design in rail-road networks with demand and capacity uncertainty. Flexible Services and Manufacturing Journal 38(2):763-804. https://doi.org/10.1007/s10696-025-09603-y
11. Lin, Boliang; Zhao, Yinan; Lin, Ruixi; Liu, Chang (2022). Shipment Path Planning for Rail Networks Considering Elastic Capacity. IEEE Intelligent Transportation Systems Magazine 14(3):160-174. https://doi.org/10.1109/MITS.2020.3014109
12. Arsalani, Pouria; Reisi-Nafchi, Mohammad; Dardashti, Vahid; Moslehi, Ghasem (2023). Two new mixed-integer programming models for the integrated train formation and shipment path optimization problem. Networks 81(3):359-377. https://doi.org/10.1002/net.22133