ABSTRACT
The optimum disassembly sequence is the best out of the many possible sequences that may exist in fulfilling specific criteria. If these criteria are quantifiable, often methods can be identified for automatic selection of such sequence. On the other hand, if such methods require too much effort or if no linear models can be formulated, then one could use metaheuristic and heuristic methods, many of which are available. However, they often return suboptimal solutions. Methods for determining the maximum possible number of disassembly sequences and, consequently, the upper limit of the size of the search space, were discussed in chapter 4. It was observed there, that this number tends to increase exponentially with the number of components
N
, as long as no additional constraints are present. However, in real world products, a set of both topological and geometric constraints is present, which considerably reduces the maximum size of the problem. Even so, the search space remains computationally large, even if it is reduced by a couple of orders of magnitude. The semiautomated incorporation of these constraints in the model was discussed in chapter 5. We discussed three basic representation methods: the state diagram, the disassembly AND/OR graph, and the disassembly precedence graph. Of these, the disassembly AND/OR graph appear to be the most appropriate tool for representing all the possible sequences. It can also be conveniently transformed into a mathematical model. We, therefore, will focus on this representation methodology. We will mainly discuss exact methods as they are considered the benchmarks for evaluating heuristic methods. Heuristic methods can be designed to study arbitrary complex products, mainly by simplifying the problem by introducing additional assumptions and rules, thus returning possibly good, but suboptimal solutions.
