Disassembly Sequence Planning (DSP) for multi-material building components requires methods that encode not only topological adjacency between parts but also the material composition and interface conditions that determine which removal sequences preserve material recovery value best. While material-oriented objectives have appeared in DSP for manufactured products, no graph-based approach has embedded material- and interface-sensitive weighting directly into sequence generation for components, nor operationalised detachability and structural complexity jointly as primary edge weights in adjacency graph traversal. This study proposes and validates a graph-based DSP workflow for aluminium-framed windows, operationalising the Detachability Index and Structural Complexity Index as edge weights within adjacency graph representations, generating disassembly sequences using greedy traversal heuristic algorithm under unweighted and weighted conditions within parametric modelling environment. We validated workflow on nineteen aluminium-framed window assemblies spanning fixed and operable configurations, part counts from 8 to 48, and four manufacturers through three complementary metrics: Spearman’s Footrule Distance for Index Shift Change at the part-displacement level, Normalised Spearman Footrule Distance for Sequence Shift Change at the global reordering level, and Precision@k for the early-sequence material prioritisation. Results show that weighting altered disassembly sequences in 17 of 19 assemblies. Precision@0.20n values fell between 0.75 and 1.00, confirming consistent aluminium-first prioritisation at early disassembly stages. Index Shift Change scaled with assembly size, with weighting inducing approximately six positional changes per part count, confirming material distribution and interface density govern global reordering more than assembly scale alone. The workflow establishes a replicable, material-aware DSP methodology, supporting material recovery and circular economy objectives.
Aktas et al. (Fri,) studied this question.