Sustainable Freight Modal Allocation Using an Integrated Modelling and CRITIC–SREM Framework

Abstract

Freight modal allocation requires balancing economic efficiency, operational performance, environmental impacts, reliability, and infrastructure capacity. This study develops an integrated ANN–CRITIC–SREM framework for evaluating alternative freight allocations among road, rail, and inland waterway transport. The framework is demonstrated through a case study of the Prahovo–Šabac corridor in Serbia. Latin hypercube sampling was used to generate representative system states for training and validating a multi-output feed-forward artificial neural network employed as a surrogate model for simultaneously estimating six decision criteria. The CRITIC method was used to determine objective criterion weights, while SREM was applied to rank the alternatives. The ANN showed close agreement with engineering-generated reference values on an independent testing subset within the defined parameter ranges. Under the adopted baseline assumptions, the alternative comprising 60% road and 40% rail transport achieved the highest overall score. Sensitivity analysis showed that this alternative remained preferred for most analysed values of the SREM balancing coefficient, whereas urgent-delivery, environmental-priority, and capacity-priority scenarios favoured different modal structures. The results indicate that the integrated framework provides a transparent and flexible decision-support basis for evaluating freight modal allocation under changing economic, operational, environmental, and capacity-related priorities.