A heuristic algorithm for the Air Transport Unit Consolidation Problem

TL;DR

This paper introduces a heuristic algorithm for the air transport unit consolidation problem, aiming to minimize costs by optimally packing packages into transport units considering air transport-specific constraints.

Contribution

It presents a novel heuristic and local search approach tailored for the air transport unit consolidation problem, incorporating cost-based objectives and real-world constraints.

Findings

The algorithm produces high-quality solutions efficiently.

It outperforms existing practical solutions on real data.

Computational tests validate its effectiveness and efficiency.

Abstract

Consolidation of loose packages into transport units is a fundamental activity offered by logistics service-providers. Moving the transport units instead of loose packages is faster (with one movement only, multiple packages are loaded instead of having one load operation for each package), safer (chances of damage and loss is reduced) and cheaper. One of the typical objective of consolidation problems is the minimization of the number of transport units used, e.g. containers. In air transportation, however, transport units have multiple aspects which concur in the calculation of the cost and thus optimization in the number and characteristics of the transport units is required. In this paper, we present the air transport unit consolidation problem where the aim is to determine how to consolidate loose packages in transport units with the goal of minimizing the corresponding cost. The problem is a variant of the three-dimensional bin packing problem where the objective function is formulated according to the way costs are calculated in the air transport business. In addition, side constraints are included to take into account specific requirements. We propose a heuristic algorithm which construct an initial feasible solution and then improves it through a local search algorithm. Computational tests on randomly generated instances show that the algorithm provides high quality solutions in a reasonable computing time. In addition, tests on real data show that it improves solutions found in practice by the company providing the data.