Combined modelling methodology for optimisation of box design based on hybrid genetic algorithm

Abstract

The hybrid genetic algorithm (HGA) was used to optimise box design to maximise cooling performance, mechanical performance, pallet footprint, and container packing efficiency and to minimise cardboard usage. These factors are normally investigated independently, but the industry requires combined functionality. Here, we present a case study, for optimisation of regular slotted carton boxes filled with wrapped beef mince. After packing, individual boxes are chilled to a desired storage temperature and then palletised for shipping. Four models were developed to predict design performance, including cooling rate, mechanical performance, cardboard usage, and box stacking on pallets. The combination of the model results was used to score the average performance of box designs. The models were solved by Comsol Multiphysics, Ansys APDL, and Cape Pack. The overall design generation and optimisation were developed with Matlab that controls all these software packages, evaluates the interactions between results, and runs the HGA for box optimisation. The HGA was conducted for 10 generations each with a population of 100 individuals. The optimisation routine successfully found optimum dimensions for the box for the defined conditions with relative short simulation times (about 3 hours per generation). This paper demonstrates how overall optimisation of packaging can be achieved through combining the strengths of multiple simulation software packages.