Disassembly sequence plan generation using a branch-and-bound algorithm

Abstract

This paper presents an approach to generate disassembly sequence plans (DSP) automatically for product recycling and remanufacturing. An algorithm is defined first that generates a geometrically based disassembly precedence matrix (DPM) from a CAD drawing of the product. The DPM is then used to generate a hierarchical disassembly tree (HDT) that represents the feasible DSP. Generation of the HDT, i.e. generation of all feasible DSP, is NP-complete. Thus, to keep the size of the HDT manageable, the branching and bounding processes are controlled using two user-defined variables. The first, w , controls the enumeration of the HDT, while the second, v, controls the bounding procedure in the HDT along with an evaluation function. The evaluation function incorporates tool changes, changes in direction of movement during disassembly along with individual part characteristics (e.g. high-valued parts, parts with hazardous content, etc.). The resulting HDT is called the reduced HDT (RHDT) since it only represents as many (near-) optimum DSP as the size of v. Experimental results demonstrate the applicability and effectiveness of the methodology.