Mathematical models for Multi Container Loading Problems with practical constraints

• Published in: Computers & industrial engineering 2019-01, Vol.127, p.722-733
• Main Authors: Alonso, M.T., Alvarez-Valdes, R., Iori, M., Parreño, F.
• Format: Article
• Language: English
• Subjects: Container loading; Cutting stock problem; Integer programming; Optimization
• ISSN: 0360-8352; 1879-0550
• DOI: 10.1016/j.cie.2018.11.012

•We study the problem of putting products on pallets and loading pallets into trucks.•Models have been developed considering both static and dynamic cargo stability.•Optimal or quasi-optimal solutions are obtained on a large set of real instances.•We consider the extension of the model for heavy loads with a special configuration.•We also study how to serve products to meet delivery dates over a planning horizon. We address the multi container loading problem of a company that serves its customers’ orders by building pallets with the required products and loading them into trucks. The problem is solved by using integer linear models. To be useful in practice, our models consider three types of constraints: geometric constraints, so that pallets lie completely inside the trucks and do not overlap; weight constraints, defining the maximum weights supported by a truck and by each axle, as well as the position of the centre of gravity of the cargo; and dynamic stability constraints. These last constraints forbid empty spaces between pallets to avoid cargo displacement when the truck is moving, and limit differences between the heights of adjacent pallets to prevent tall pallets tipping over short ones. We also consider extensions of the models to the case of heavy loads, requiring a special configuration of the pallets in the truck, and to the case in which the orders must be served over a set of time periods to meet delivery dates. The computational study that we performed on a large number of real instances with up to 44 trucks shows that the proposed models are able to obtain optimal solutions in most cases and very small gaps when optimality could not be proven.