Time-Aware Two-Dimensional Packing for Slicing-Aware 3D Printing Throughput Optimization

Abstract

Batching multiple parts onto a single fused-filament fabrication build plate improves throughput, but existing packing algorithms optimize for geometric density rather than print time. We introduce a slicing-aware cost model that estimates print time from part geometry and placement without invoking toolpath generation, achieving strong correlation with slicer-reported times (Pearson r = 0.90, Spearman ρ = 0.96). Evaluating packing algorithms on synthetic production builds, we find that greedy polygon-based packing matches or exceeds Large Neighborhood Search at three orders of magnitude lower compute cost—a negative result we attribute to high initial packing density leaving minimal room for iterative refinement. Against PrusaSlicer’s default auto-arrange, our method achieves a 5.7% throughput improvement (95% CI [4.7%,7.3%], N = 329), with median savings of 19.5 minutes per build.