Modelling and analysis of integrated machine-level planning problems for automated manufacturing.

The wide application of NC technologies and the advance in integrated manufacturing have significantly improved productivity. This however also leads to complexity in shop floor planning, particularly in machine-level planing due to the increased flexibility in the selection of machining parameters, tools, and tool paths. Part sequencing, tool replacement and machining speed selection for metal cutting in general and tool set selection, machining speed specifications as well as path sequencing for hole making in particular have a direct impact on manufacturing economics and hence have drawn much attention of many researchers. However, these problems have been often solved in isolation to each other, thereby causing inconsistent and conflicting planning actions on the shop floor. As a result, the solutions obtained in such a way can not be used for real shop floor planning. At the best a lengthy process is needed to resolve the conflict between the separately obtained solutions. Such a decision process obviously does not meet the need of modern manufacturing environment where quick and consistent planning decisions are imperative. The main purpose of this study is to model and solve several combined planning problems facing today's manufacturing industry. These include (a) part sequencing and tool replacement with sequence-dependent setup times and probabilistic tool life; (b) Just-In-Time (JIT) part scheduling with variable processing times and sequence dependent setups; and (c) tool set selection, machining speed specification, operation sequencing and path selection for hole making operations. These problems are combinatorial in nature and are often classified as NP-complete. Consequently, optimal solutions may not be obtained within polynomial times. In this dissertation, tabu search technique has been employed to solve the above combined planning problems. The computational results have shown that these problems can be efficiently solved and consistent decisions can be made based on the solutions. The effects of some important parameters such as initial solutions, move selection, termination criteria, and tabu list size on the search performance have also been examined