Modeling of an automotive body assembly system for dimensional control

Quality and productivity are the two major areas of today's automotive manufacturers' improvement efforts. Correct geometry of the final product is one of the key factors for quality and productivity in the automotive body assembly process. Dimensional control during production and variation management during the design stage are two important tasks that heavily impact these manufacturing goals. This thesis presents a generic modeling methodology for dimensional control of the automotive body assembly that integrates critical characteristics of product and process. This modeling technique can be used during design as well as manufacturing stage of a new product development. The developed modeling of sheet metal assembly is divided into two parts: (1) modeling for rigid parts assembly and (2) beam-based modeling for flexible parts assembly. These modeling methods include both process characteristics, such as locating schemes for single and multi-stations, jointing methods, and assembly joint interactions, and product characteristics, such as sheet metal flexibility, joint geometry, and fabrication errors of the sheet metal assembly process caused by tooling and part variabilities. The presented modeling methodology for dimensional control can be implemented to solve manufacturing problems (diagnostics of multiple dimensional faults) as well as to solve design problems (dimensional tolerance analysis and allocation). The diagnosis of dimensional faults is based on the developed correlation maps approach which integrates correlation analysis and graphical representation. The major advantages is the enhancement of fault diagnosisability which enables identification of multiple dimensional faults in multi-station assembly environment. The novel flexible beam-based tolerance analysis and allocation was developed. This method include part-to-part interaction conditions by including part fabrication error. Additionally, tolerance allocation algorithm is developed using non-linear programming methodology in combination with projection theory which provide a direct approach in solving tolerance allocation problem. Two industrial cases have verified the proposed modeling methodology for identification of multiple faults in multi-station sheet metal assembly systems. Tolerance analysis and tolerance allocation schemes are proven by experimental results.