The use of statistical quality improvement methods in automotive body manufacturing: Three research topics.

The significance of quality improvements for automotive body manufacturing is profound. Manufacturers may invest half the cost for developing a new car trying to meet quality expectations for stamping and body assembly. Thus, the objective of this dissertation is to provide quality improvement methods for three important topics: part and assembly design, metal forming in stamping, and variation control in assembly. Each of these is critical to reduce costs and enhance quality. The first research topic examines metal forming. Process control of steel forming is complex because it involves both steel properties and press conditions. A logistic regression model is used to relate a common formability problem (split/non-split) and critical process and material parameters. By modeling the split probability, the significant parameters and their appropriate control ranges are identified. The second research topic explores variation sources and clustering characteristics for automotive doors using statistical quality improvement methods. These methods such as seven tools, regression, and design of experiments are used to identify variation sources in the assembly line. Cluster analysis and principal component analysis are also used to solve process problems. The third topic examines the use of simulation for body assembly. It extends the applications of tolerance design and tolerance synthesis. By using production data for the inputs to the design intent simulation model and then comparing the results with actual assembly outputs, model accuracy is significantly improved. The assumptions used in the body simulation are also thoroughly examined by multivariate techniques.PhDIndustrial and Operations EngineeringUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/104561/1/9527762.pdfDescription of 9527762.pdf : Restricted to UM users only