Constrained Response Surface Optimisation for Precisely Atomising Spraying Process

Abstract

This paper presents a development of a design of experiment technique for quality improvement in automotive manufacturing industrial. The quality of interest is the colour shade, one of the key feature and exterior appearance for the vehicles. With low percentage of first time quality, the manufacturer has spent a lot of cost for repairing work as well as the longer production time. To permanently dissolve such problem, the precisely spraying condition should be optimised. Therefore, this work applied the multiple regression and response surface methods or RSM to investigate significant factors and to determine the optimum factor level in order to improve the quality of paint shop. Firstly, 2 k full factorial was employed to study the effect of five factors including the paint flow rate at robot setting, the paint levelling agent, the paint pigment, the additive slow solvent, and non volatile solid at spraying of atomising spraying machine. The response value of colour shade at 15 and 45 degree are measured using spectrophotometer. Then the regression models of colour shade at both degrees were developed from the significant factors affecting each response. Consequently, both regression models were placed into the form of linear programming to maximise the colour shade subjected to 3 main factors including the pigment, the additive solvent and the paint flow rate. This led to the determination of new levels of decision variables and brought 70 % reduction on paint repairs cost and improve first time quality from 70% to 88% for the production of interest