Non-linear disassembly planning for increasing the end-of-life value of products

Abstract

Disassembly of products at their end-of-life is an important topic, since the general objective of product recycling is to increase the amount of parts recovered and reused while minimizing the amount of disposal. In addition, materials have to be recovered and recycled, subsequently. In general, disassembly can be defined as a systematic method for non-destructively separating a product into its constituent parts, subassemblies or other groupings. In general, disassembly is a prerequisite for recycling or remanufacturing, since most products should be disassembled before being recycled or remanufactured as secondary parts or materials. Among the various decision problems in a disassembly process, this thesis focuses on the disassembly process planning, which is the problem of determining the sequence of disassembly operations for a given product and disassembly tools under certain objective function such that the resulting sequence satisfies the constraints established by the product. A deterministic disassembly sequencing method is developed using two kinds of graph based disassembly sequence representations, i.e. extended process graph and modified directed graph of assembly states; the problem is solved using integer and linear programming formulation, respectively. The graphs represent not only the operation sequences, but also sequence-dependent operation times, which should be considered as direct labor costs, since most disassembly operations are done manually and processing times are affected by the order of operations due to the tool changing. The last part of the thesis deals with the uncertainty issues, e.g. defective parts or joints in an incoming product, disassembly damage, and imprecise net profits and costs. Among those, imprecise net profits and costs are mainly of interest in this thesis. Practically, it is quite often that the estimated disassembly operation time and the expected value of a component or material can only be obtained in a rough range. In other words, the profit values are given in the form of intervals that correspond to ranges in which the true coefficient values lie. The problem considered in this thesis can be transformed into a linear programming problem with interval objective function coefficients. Then, in order to solve the problem, the minmax regret approach is applied, which transforms the multi-objective decision making problem into a problem with a single objective, and a dynamic programming algorithm is developed to compute the minimum worst case deviation (robust deviation) between alternative solution paths as an exact solution approach.