CAD-Based Product Partitioning for Automated Disassembly Sequence Planning with Community Detection

Remanufacturing valuable components, like lithium-ion batteries from an automotive battery pack, is crucial in turning our society more sustainable. For scalable remanufacturing, disassembly planning with a minimum number of assembly steps is required. CAD-based assembly-by-disassembly method can be utilized to generate disassembly sequence plans automatically. However, this approach is computationally expensive due to graphical collision analysis. For assembly step minimization, the computing time can be reduced by identifying dismantlable subassemblies. Therefore, we propose an automated partitioning of assemblies into feasible subassemblies, which is then fed into the final step of the assembly-by-disassembly method to produce a precedence graph. A comparison of the two popular community detection algorithms - the Louvain algorithm and the Girvan–Newman algorithm - have been applied to exemplary CAD models. The Louvain algorithm showed the best results with a distance-weighted contact matrix. The results show that subassemblies can be proposed for disassembly planning. Nevertheless, the uncertainty in clustering quality is still too high to utilize the approach in a fully automated planning pipeline.