The Manufacturer's Pallet Loading Problem (MPLP) consists in arranging, orthogonally and without overlapping, the maximum number of rectangular and identical pieces of sizes (l, w) onto a pallet (L, W). The MPLP has been successfully handled by block heuristics, which generate loading patterns with one or more blocks where the pieces have the same orientation. The resulting patterns produced by these methods are limited to the so-called 1st order non-guillotine patterns. In this work we present a tabu search implementation based on the algorithm proposed by Dowsland to solve the MPLP; differently than block heuristics, the solutions are not limited to particular loading patterns. Computational experiments with a set of 34 instances extracted from the literature as well as from practical contexts indicate that this approach is capable to solve most of the examples. For the unsolved instances, we propose a simple additional procedure, which resulted in optimal patterns.
Pallet loading problem; tabu search; combinatorial optimization
