Configuration Design of a High Performance and Responsive Manufacturing System : Modeling and Evaluation

Sammanfattning: Configuring and reconfiguring a manufacturing system is presented as an issue with increasing importance due to higher frequency of system configuration or major reconfigurations to accommodate new set of requirements and/or the need to configure the system to make it usable across generations of products or product families. This research has focused in the modeling, evaluation and selection decisions which involves multiple, incommensurate and conflicting objectives. Which renders configuration a multi criteria decision making or multi objective optimization problem. A manufacturing system configuration design is strategic, i.e., the effects are long term and determines the competitiveness of manufacturing. A case study in one of Swedish large discrete part manufacturing which produces variants of products for two different market segments is conducted to verify the fit between the manufacturing strategy and the existing system configuration. The relevance of aggregate modeling is discussed and it’s argued that system dynamics has the advantage over analytical methods in its capability to capture the complexity, its capability to evolve into more rigorous and detailed model, and the lesser time needed for the assessment especially when there are a number of alternatives. Circumstances are when a cost model may suffice for certain comparative analysis. The challenge with cost models is the difficulty in projecting intangible factors in terms of cost. However, approximation to some important factors can be made that may give insights in the comparative performances of alternatives. In line with this view a cost model that comprises the investment and operation costs, quality and reliability is proposed. Application of AHP and ANP for preference weight (subjective) elicitation and qualitative performance evaluation, entropy for objective weights calculation that may help to evaluate the discriminating ability of a criteria, Pareto frontier Analysis particularly Data Envelopment Analysis for selection and ranking of alternatives are shown to be relevant and applicable in configuration design. A comprehensive design decision matrix called House of Assessment is proposed that captures the dependency among the criteria and evaluation objectives weights of the criteria using entropy to determine the discriminating ability of the criteria whenever appropriate.