A Sequential Framework for Reinforcement Bar Cutting with Reusable Leftover Management and Computational Traceability

Reinforcement cutting is commonly evaluated through stock-bar consumption, although bar counts alone do not describe material-source histories. This study compares Sequential Greedy (SG), Exhaustive-Window Greedy (EWG), Hybrid Operational Rolling Exhaustive-Window Greedy (HOREWG), First-Fit Decreasing (FFD), Best-Fit Decreasing (BFD), and arc-flow mixed-integer linear programming (MILP). Two real-project datasets and one dependent expanded dataset are analysed using 12,000 mm stock bars, three diameters, zero kerf, and common precedence, source-access, recording, and validation rules. MILP achieves the lowest stock-bar consumption, and its plans are replayed without replacement stock. Sequential-method rankings vary across cases, while wider local search does not consistently reduce stock consumption. Under floor-element precedence for 16 mm reinforcement in the expanded dataset, HOREWG, FFD, and BFD each use 868 bars but record 171, 649, and 646 source activations, respectively; MILP uses 836 bars with activations ranging from 903 to 1022 across the tested replays. Linked cut, source, and residual records enable reconstruction of material-use histories for heuristic and MILP-derived plans. The findings distinguish material efficiency from source-activation behaviour, including differences between plans with identical bar counts. Activations represent computational events; physical handling, fabrication, and economic benefits were not evaluated.