Data-driven distributionally robust generation of time-varying flow corridor networks under demand uncertainty

Abstract

Flow corridors are novel long tube-shaped, high-density airspace structure (like freeways in sky) which could achieve a very high throughput, while allowing traffic to flexible deployment and shift as necessary. In current research, the design of flow corridor networks cannot capture either the dynamic nature of traffic or the uncertainty in demand variations, which may fail to ensure satisfactory efficiency and reliability. In order to propose more efficient and reliable flow corridor networks for practice operations, this paper is devoted to propose a data-driven framework for the robust generation of time-varying flow corridor networks under demand uncertainty. Specifically, a delay-based method is proposed firstly for optimal design of a static flow corridors network which could be more effective in absorbing frequent flight delays from today’s air transportation system. Next, a multi-objective combinational optimization model is presented with its fast approximate evolutionary algorithm for generating time-varying flow corridor networks. Finally, to handle uncertainties in traffic operations over time, the data-driven Distributionally Robust Optimization (DRO) approach is employed to ensure the efficiency and reliability of the proposed networks. The framework is applied to the Chinese airspace to design a robust national-wide time-varying flow corridor network for numerical test. The numerical test results confirm that the proposed time-varying networks outperform previous designs in the average alleviated delays, average occupancy, and activation time with only a small trade-off in the number of served flights.