Search for a command to run...
Under conditions of a full-scale war, urban logistics has undergone a fundamental transformation, shifting from a commercial function to a critical element of sustaining urban agglomerations. The operating environment of frontline cities, in particular, Kharkiv, is characterized by complex uncertainty: stochastic demand, destruction of transport infrastructure, and direct threats to participants in the freight transportation process. This necessitates a revision of classical deterministic supply chain management approaches in favor of adaptive strategies. The purpose of the article is to formalize and investigate the urban freight routing problem in a war- induced environment by developing a two-stage stochastic model for a heterogeneous vehicle fleet, which accounts for stochastic demand and military risks as sources of disruptions to transport execution and enables minimization of expected total costs considering possible failures, idle time, and forced route replanning. The methodological basis of the study is a simulation–heuristic approach that makes it possible to overcome the computational complexity of the problem caused by the high dimensionality of the scenario space. The proposed algorithm combines a metaheuristic (genetic algorithm) to generate routing solutions with Monte Carlo simulation to evaluate their robustness under uncertainty (air-raid alerts and blockages of network segments). A comparative analysis of the baseline (risk-ignoring) and the developed (risk-adaptive) models on a test case revealed fundamental differences in routing strategies. It was found that accounting for military risks leads to the formation of routes that increase planned mileage by 11,8 % and operating costs by 11,8 % in order to avoid hazardous areas. As a result, it was shown that the proposed approach reduced total expected costs by 13,5 %, confirming the priority of reliability and safety over distance minimization under martial law.