Congestion-driven delivery delays can increase the risk of grocery stockouts on the Savannah–Atlanta replenishment lane, particularly during high-stress periods. This paper develops a stress-conditioned Monte Carlo inventory model that maps a public TEU-based congestion proxy into discrete stress regimes R∈{−1,0,1,2}. Each regime shifts the shipment delay distribution, including the probability of rare severe delays, and is used to estimate P(stockout∣R, B,s) and expected shortage across buffer coverage levels B∈{2,3,4,5,6} and policy scenarios s∈{Baseline, Safety Stock (+1 day), Tail Mitigation} Results show that stockout risk rises sharply under extreme stress when inventory buffers are thin. At R=2 and B=2, baseline stockout risk is 12.51%, compared with 1.83% under Safety Stock and 9.31% under Tail Mitigation. Expected shortage at the same setting falls from 6.370 at baseline to 0.849 under Safety Stock. At R=2 and B=3, the baseline stockout risk declines to 1.76%, and further to 0.24% with Safety Stock. Overall, increasing buffer coverage is the strongest lever for reducing both stockout frequency and shortage severity, while tail-focused mitigation is most effective when buffer levels are already moderate. The framework provides a practical way to translate public congestion signals into inventory-risk decisions for regional food supply chains.