Dimensional management in metal assembly process validation: A multivariate approach.

Managing dimensional issues in the sheet metal assembly process during the initial stages of the production launch involve a series of complex decisions. This dissertation provides a series of methodologies that address the multivariability of components and assemblies and their relationship. In the first paper, a statistical process control methodology for the stamping process is presented. The methodology uses a combination of Hotelling T2 multivariate charts and Individual/Moving Range univariate charts. The multivariate charts are used to monitor the stability of groups of checkpoints simultaneously and to identify those particular ones that have a critical contribution to out of control signals. The Individual charts are then used to follow up and monitor those critical checkpoints more closely. The second paper, addresses the issue of modeling the relationships between stamping components and their subsequent assemblies using a slow build approach to plan and carry out dimensional studies. The proposed data analysis methodology is based on the partial least squares regression methodology to identify significant characteristics in a highly collinear and multivariate setting with a relatively small number of parts. A methodology for tolerance revision and adjustments is presented in the third paper. This is a very important outcome of the process validation in which the process variation is estimated using a limited number of die sets and parts. The power of the methodology is compared to that of other common practices.PhDApplied SciencesIndustrial engineeringUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/125267/2/3016846.pd