Quantifying the variance contribution of special cause mean shifts in reconfigurable batch manufacturing processes

Abstract

This paper presents a methodology for quantifying a portion of the special cause variation in reconfigurable batch manufacturing processes. The proportion of variability statistic measures the amount of instability in the process mean associated with inconsistencies in process set-up and triggered by changes in process settings or input materials. This statistic suggests the amount of special cause variation the manufacturer could eliminate by controlling process setup and production parameters. Two versions of the statistic are presented: measuring the variation associated with an inconsistent process set-up, and capturing variation related to both configurable process variables and parameters that may vary during a production run. The methodology also includes a process capability index that forecasts the process capability value the manufacturer could attain by eliminating the variance measured by the proportion of variation statistic. Manufacturers can use these quality measures as an input into developing a process improvement strategy. The measures will quantify the impact of better controls on special cause variation and the resultant process capability. Estimating the proportion of variation statistic and associated process capability index requires following a structured designed experiment with observational covariates. The methodology and associated statistics are demonstrated with applications to two automotive body stamping processes.