Solving the box filling problem for fresh produce with simulation

Horticultural products are unique in the food industry due to their physical variability in size and shape. A resulting key challenge is engineering an efficient supply chain, where package design for loose packed product enables consistent fill weight and volumetric efficiency along the supply chain. Predetermining how much fruit of a weight range can fit into a box of a certain size is challenging, as the natural differences in size and shape of individual fruit, coupled with random stacking patterns, produces an irregular bulk shape that has a unique weight on every occasion. Consequently, deciding on a package design that enables efficient packing for all count sizes and generates the maximum possible volumetric efficiency is therefore difficult without extensive iterative experimental investigation. Alternatively, with accurate descriptions of fruit shape and size, Monte Carlo style computer simulations of the fruit stacking process can be used to optimize box designs to achieve near equal fill and overall volumetric efficiency. This paper demonstrates the value of such simulations in informing package design. Using predetermined product size ranges, populations of randomly sized fruit within each size range are stacked in a digital box of a specified footprint in Blender. Gravity, friction and collision forces are used to calculate realistic motion paths as the fruit fall and stack naturally into the box. Any over fills of a box design are removed, and the remaining fruit are counted for number and net weight to define fill outcome. After repeated simulations with new random population distributions, count and weight predictions are generated, which can be utilised to inform consequences of packaging design decisions, and lead to decisions that enable economic and environmental efficiency.