Air refueling extends aircraft range and endurance, but its operational value hinges on where the refueling airspace is placed and how tanker missions are sequenced. This paper addresses the joint optimization of refueling airspace planning and tanker scheduling, in which each receiver selects a refueling point from a continuous feasible interval along a fixed route. The upper level determines refueling point locations (continuous variables), while the lower level schedules multiple heterogeneous tankers (discrete combinatorial variables); the two levels are tightly coupled through spatiotemporal constraints and fuel propagation. We propose a bottleneck-driven decoupled update (BDDU) strategy built on the Whale Optimization Algorithm (WOA). BDDU extracts bottleneck states from lower-level scheduling feedback and applies per-dimension step-size control to damp the coupling amplification effect inherent in bi-level optimization. Across three scenarios of varying coupling intensities and scales, BDDU-WOA raises the feasibility rate from 50% (WOA baseline) to 90% (+40 percentage points; $p<0.05$, Fisher’s exact test). The gain stems from a bottleneck-aware, dimension-wise step-size control mechanism with an adaptive, parameter-free classification threshold and only two tunable parameters, adding roughly 10% computational overhead. The method is intended for pre-mission planning of large-scale air refueling operations.

In multi-mission air operations, area-patrol missions and long-range missions typically share a limited tanker fleet while imposing different constraint structures: patrol refueling is bounded by hard time windows, whereas long-range missions are governed by restricted-zone avoidance, detour tolerances, and multi-segment fuel verification. Conventional scenario-wise independent planning splits resources and wastes cross-region ferry mileage. This paper adapts the established paradigms of the location–routing problem (LRP) and the vehicle routing problem with time windows (VRPTW) to this joint refueling scenario: a joint planning model prioritizes the number of tanker sorties over total system flight distance, and a decoder-coupled adaptive large neighborhood search (ALNS) integrates airspace selection, task assignment, tanker routing, and dual-timeline rendezvous decoding, with all mission hard constraints embedded in a deterministic, reproducible evaluator that adjudicates feasibility at every search iteration. Experiments at three scales (17, 42, and 100 tasks) show 100% mission coverage and 100% patrol time-window satisfaction: relative to scenario-wise independent planning, tanker sorties decrease by 16.2–19.7% and tanker flight distance by 14.6–15.5% (significant after Bonferroni correction on 90 paired replicates per scale); against genetic algorithm (GA) and ant colony optimization (ACO) baselines—and against a route-encoding GA under an equal solution-space representation—the method is superior in solution quality and runtime ($p<0.001$), and the separation persists when the baselines receive a 25-fold evaluation budget. Monte Carlo simulations characterize how plan feasibility degrades under execution-time disturbances. Within the studied instance families, the framework yields executable joint refueling plans within operational runtimes.