聚焦复杂不确定环境下应急物资灾前布局和灾后运输分配问题，考虑承载应急物资运输的道路因交通拥堵或车流量限制而导致交通中断限制因素，构建了一种新的两阶段应急物资储运调度决策优化方法。在调研国内外相关文献的基础上，综合考虑应急物资“储备-运输”过程中的不确定性因素和道路拥堵造成的交通中断或车流量限制因素，采用均值-CVaR风险测度方法构建一类风险规避型两阶段应急物资“储备-运输”调度随机规划模型，进而合理规划运输路线，优化“储备-运输”网络布局，实现调度总成本最小化目标。同时，给出了两类随机信息价值分析指标——完备信息的风险规避价值RAVPI和随机解的风险规避价值RAVSS，探究随机分布信息和随机规划方法在不同拥堵程度下对应急物资调度策略的差异化影响。研究发现：①承载应急物资运输的道路出现交通中断或车流量限制情形将显著增加应急物资调度总成本，并使灾前物资储备布局、灾后运输线路选择的最优调度策略发生明显改变；②在道路交通中断或车流量限制程度显著条件下，采用随机规划方法开展应急物资调度决策可使调度总成本降低10%，有效发挥随机分布信息的额外价值。

This study focuses on the pre-disaster allocation and post-disaster transportation of emergency supplies in complex and uncertain environments, and proposes a two-stage decision-making optimization methodology for emergency supply inventory-transportation scheduling operation, considering road traffic disruptions due to traffic congestion and flow restrictions. Based on a review of relevant literature, this study comprehensively considers the uncertainty and the disruptions or restrictions caused by road congestion among the supplies inventory-transportation scheduling operation process, and a risk-averse two-stage stochastic programming model is developed using the mean-CVaR risk measure, which aims to schedule transportation routes and optimize the inventory-transportation network configuration while minimizing total scheduling costs. Two indices of stochastic information value are introduced: risk-averse value of perfect information (RAVPI) and risk-averse value of the stochastic solution (RAVSS), which are used to assess the differential impact of stochastic distribution information and stochastic programming methods on emergency supply scheduling operation strategies under varying congestion levels. Key findings include: ①Traffic disruption or restriction substantially increase total scheduling costs, altering the optimal pre-disaster inventory layouts and post-disaster transport routes; ②Under the roads with significant degree of traffic disruption or restriction, the operation strategy from stochastic programming methodology reduces total scheduling costs by up to 10%, which can leverage the additional value of stochastic distributional information effectively.