Comparative evaluation of best-fit nesting heuristics for FDM 3D printing batch scheduling: a paired experimental study with sorting decomposition

Purpose This study aims to investigate the contribution of sorting pre-processing and placement criterion selection to fused deposition modeling (FDM) batch scheduling performance, addressing a fundamental attribution ambiguity in prior nesting literature where these two factors have never been empirically decomposed. Design/methodology/approach Five algorithms are compared: a Naive sequential baseline, a Sorted Naive intermediate baseline (enabling sorting-vs-placement decomposition) and three Best-Fit variants: Original (BF), height-penalized (BF-H) and volume-grouped (BF-G). Algorithms are evaluated on 210-part samples drawn from a 5,335-part preprocessing pipeline applied to the Thingi10K benchmark across 25 paired experimental runs. Performance is assessed on batch count, total processing time and platform utilization using Wilcoxon signed-rank tests, Repeated Measures ANOVA and pairwise post hoc comparisons. Findings Sorting alone significantly worsens batch count relative to the unsorted baseline (+2.84%, W = 52.0, p = 0.025), establishing placement logic as the primary performance driver. All three BF variants achieve 18.76–19.87% batch reductions versus the sorted baseline (p < 0.001). All BF variants are statistically equivalent to one another (all pairwise p > 0.05); BF-G achieves the lowest processing time (19,664 min, −7.09% vs Naive) and highest platform utilization (66.91%). Repeated Measures Analysis of Variance (ANOVA) confirms omnibus significance (F(4,96) = 244.04 and 171.42, p < 0.001). Originality/value This paper presents the first empirical sorting-vs-placement decomposition in FDM nesting optimization, introduces a preprocessing pipeline yielding 5,335 validated parts from Thingi10K and provides a statistically rigorous paired evaluation framework with effect-size reporting, a methodological standard absent from prior FDM nesting literature.