A methodology for optimal door fit in automobile body manufacturing.

Abstract

The dimensional quality of an automobile door directly affects the quality of door fit on the car body. An inadequate door fit will not only affect the aesthetic value of the vehicle, but will also cause functional problems such as wind noise, water leakage, and difficulty in door closing. The automation of door fitting also requires the dimensional variations of both doors and body side openings to be controlled. The objective of this thesis is to develop a systematic approach for the reduction of the dimensional variation of automobile doors on the body side opening, through effective utilization of measurement data for process control. Three state-of-the-art dimensional measurement systems, hard gauge, Coordinate Measuring Machine (CMM), and in-line Optical CMM (OCMM) are compared and evaluated from the view point of, (1) process parameter identification, and (2) process monitoring, so that they can be effectively applied for process control. The door fitting process is formulated as a general optimization problem. Three quantitative indices are defined to evaluate the dimensional quality of door fit. They are: (1) gap width deviation, (2) gap parallelism, and (3) car-to-car variation. Based on these indices, simulations are subsequently conducted to provide guidelines for door gap design, and criteria for upstream process variation reduction. A systematic computer-aided fixture adjustment scheme is then proposed, and the adequacy of a feature modeling is verified by experiment. Variation of the door dimension is attributed to the variations of the assembly and the stamping processes. Variation reduction of the assembly process is approached through principal component factor analysis and solved case by case. A two-level classification scheme is proposed for efficient process diagnosis, and is demonstrated to be effective for process fault classification using in-line OCMM data. Variation of the stamping process is broken down into run-to-run variation and within-run variation, and is quantified using CMM data. The relationship between the two types of variation is analyzed. A successful case study is presented for within-run variation reduction using experimental parameter design.