Optimal parameters and tolerances in concurrent product and process design using simulation and fuzzy goal programming

Abstract

Purpose The purpose of this paper is to propose an approach to determine the optimal parameters and tolerances in concurrent product and process design in the early design stages utilizing fuzzy goal programming. A wheelchair design is provided for illustration. Design/methodology/approach The product design is developed on the basis of both customer and functionality requirements. The critical product components are then determined. The design and analysis of experiments are performed by using simulation, and then the probability distributions are adopted to determine the values of desired responses under each combination of critical product parameters and tolerances. Regression nonlinear models are then developed and inserted as constraints in the complete optimization model. Preferences on product specifications and process settings, as well as process capability index ranges, are also set as model constraints. The combined objective functions are finally formulated to minimize the sum of positive and negative deviations from desired targets and maximize process capability. The optimization model is applied to determine the optimal wheelchair design. Findings The results showed that the proposed approach is effective in determining the optimal values of the design parameters and tolerances of the critical components of the wheelchair with their related process means and standard deviations that enhance desired multiple quality responses under uncertainty. Practical implications This work provides a general methodology that can be applied for concurrent optimization of product design and process design in a wide range of business applications. Moreover, the methodology is beneficial when uncertainty exists in quality responses and the parameters and tolerances of product design and its critical processes. Originality/value The fuzziness is rarely considered in research and development stage. This research considers membership functions for parameters and tolerances of a product and its related processes rather than crisp values. Moreover, presented optimization model considers multiple objective functions, sum of deviations and process capability. Finally, the indirect quality responses are calculated from the best-fit probability distributions rather than assuming a normal distribution.