This paper presents a method to integrate a cost-based linear holding (LH) practice, with the commonly seen ground and airborne holding, for the network air traf¿c ¿ow management (ATFM) problem. In this method, the performance of dynamic ATFM delay assignment can be improved in a more cost-ef¿cient way. A brief comparison of the three holding manners is presented ¿rst, followed by illustrating one potential applicability of the proposed LH which inspires the motivation of this paper. Then,an etworkATFM model is built based on the classic Bertsimas and Stock Patterson model, with all the three types of holding accommodated. Under the model framework, a trajectory generation algorithm is introduced, aiming at computing, per ¿ight, the realizable linear holding(i.e., LHtimebound). This information is handled andsharedbythedifferent airlinesand it is then used as a constraint factor in the network ATFM model. The dynamic delay assignment problem is formulated as a mixed integer linear programming and solved with a commercially-of-the-shelfsolver. Computational experiments are givens howingthattheintegrationcontributestoa notable cost reduction when network capacity changes ahead of schedule, if compared to the case where only ground and/or airborne holding apply

Peer Reviewed