Identification and Layout Optimization of Congestion Bottlenecks in Air-Cargo Hub Multimodal Regions Based on the Max–Min Traffic Allocation Model

Air-cargo hub cargo regions concentrate congestion in a few facilities competing for limited budget and land. We couple constraining-facility identification with capacity-oriented layout adjustment in one shortage-first Max–Min traffic-allocation workflow: path flows first minimize unmet demand and then peak utilization, and each facility is scored by the system relief a prespecified capacity step delivers after full reallocation, without bottleneck labels. On a synthetic eight-facility network whose demand intensity is calibrated on 61 days of desensitized terminal throughput records, ranking by post-reallocation relief rather than utilization separates the busiest facility from the most valuable expansion target. A frozen-forecast week authorizes construction only when forecast peak-utilization reduction clears a stated 4% threshold; a 10% forwarder-warehouse expansion then uniquely qualifies for the stated threshold and cuts modeled unmet demand by 43.7% and 18.5% at 1.5× and 2.0× pressure. In a random-holdout backtest, identification-driven rules lower unmet demand from 616.6 to 68.6, 48.4 and 0.0 ULD across three budget tiers under the same caps (25 re-draws; stability diagnostic p < 0.0001). Removing path substitution raises unmet demand from 616.6 to 1820.2 and 3409.8 ULD; an auxiliary negative-control classifier returning only the majority class provides no evidence of a learnable label shortcut under the tested features. The workflow screens candidates before capital is committed.