Optimizing package express operations in China

Abstract

We explore a novel optimization model to support the planning and operation of an air service network at a package express carrier in China. The model simultaneously considers ground and air transportation, company-owned aircraft capacity and belly capacity purchased on commercial aircraft, multiple service classes for packages, and packages becoming available throughout the day. A column generation based algorithm is devised to solve real-life problems with more than 650,000 timed origin-destination demands (packages), a service network with 150 hubs at airports, and an air cargo fleet of 60 aircraft of varying capacities. An extensive computational study shows the efficacy of the model and algorithm, and provides managerial insight. In particular, we see that considering ground transportation when designing an air service network is critical to improving operational efficiency and company profits as well as to extending service coverage. Furthermore, we find that in an environment with ample and reasonably-priced capacity on commercial aircraft, an air service network design based on single-flight shipments can be more effective than the commonly used star network design with multi-flight shipments.