Minimizing Manufacturing and Quality Costs in Multiresponse Optimization

Most industrial processes and products are evaluated by more than one quality characteristic. To select the best design and operating control factors it is necessary to take into account all measures of quality simultaneously, in what is known as multiresponse optimization. Determination of the best operational settings for products or processes is usually accomplished through single-objective function optimization routines. Here we propose a new objective function for multiresponse optimization, based on Taguchi’s loss function. Our function innovates in taking into account manufacturing and quality-related costs in the optimization, in addition to the usual criteria of minimum response variance and distance-to-target, and maximum robustness. We also propose a simplified procedure for modeling the variance of responses from non-replicated factorial designs. The paper contains a case study from the rubber industry.