Dynamic truck UAV location inventory routing for postdisaster humanitarian logistics

Purpose This study aims to develop and evaluate a dynamic, multi-item, multiperiod location–inventory–routing model for postdisaster relief that jointly determines day-by-day distribution center (DC) activation under an exogenous availability calendar, inventory buffering, truck line-haul replenishment, unmanned aerial vehicle (UAV) last-mile service and inter-DC UAV redeployment. This study also benchmarks flexible and restricted operating policies; examines sensitivity to demand surges, UAV speed, sortie capacity, holding cost and road detours; and evaluates rolling-horizon reoptimization under progressively revealed demand, facility and UAV-availability disruptions. Design/methodology/approach The author formulates a mixed-integer linear program on a time-expanded two-echelon truck–DC–UAV network. Using an earthquake-response benchmark, the baseline (BS) policy is compared with cross-dock-only, always-open, static-open and no-redeployment variants. Sensitivity analyses vary demand surges, UAV speed, daily sortie limits, inventory-holding costs and road-detour factors. A separate six-day rolling-horizon experiment compares ORACLE, ROLLING and nonadaptive execution across four structured disruption classes, with 30 replications per class, to evaluate adaptive performance under progressively released information. Findings The flexible BS consistently achieves the lowest cost; under demand surges from 2× to 10×, total cost rises from 8,233 to 18,541. Most speed-related savings occur between 40 and 60 km/h, while the main sortie-capacity gains occur around 4–6 sorties per UAV per day in this benchmark. BS remains least costly across road-detour factors from 1.0 to 4.0. In rolling-horizon tests, ROLLING closely matches ORACLE’s total cost and planned-network/emergency-supply split, although both rely substantially on emergency/external supply under deliberately severe disruptions. Originality/value This study integrates four features rarely considered together in humanitarian logistics: time-varying DC usability, an explicit multiperiod horizon, hybrid truck–UAV two-echelon distribution and interperiod UAV redeployment while also incorporating multi-item inventory carryover. Beyond static policy comparison, it shows how these flexibilities perform under demand surges, road inefficiency and progressively revealed disruptions. The framework therefore links facility operability, inventory positioning, line-haul replenishment, UAV dispatch and adaptive reoptimization within one decision-support model, providing a structured basis for resilient postdisaster truck–DC–UAV planning.