Quality improvement using taguchi method in a manufacturing company

Abstract

Quality improvement is crucial for manufacturing companies to survive in today’s marketplace. The main purpose of this study was to improve the quality of one selected product produced by a manufacturing company using the Taguchi Method. The product, a plastic injection moulding is studied as it showed the highest reject quantity. The type of defects found in the product includes bubble, short mould, scratches and over pack. However due to company’s request and limitations, only two defects, short mould and over pack, were further investigated. Experimentation using Taguchi method was decided as the approach to reduce the occurrence of the defects. To that end, four main factors affecting the surface defects were identified. They are injection pressure (A), injection speed (B), melting temperature (C) and holding pressure (D). Taguchi method was chosen since it provides fast and lower costs for conducting the experiments as well as in determining optimum parameter. Qualitek-4 software was used to facilitate the signal-to-noise (S/N) ratio analysis in the Taguchi Method and to predict the results at optimal parameters setting. The results from the two levels of experiments (L8) suggest that all parameters and 2-factor interactions investigated were significant for short mould and all the factors should be set at high level to achieve optimum condition. In the case of the over pack problem, two significant factors and one significant interaction factor which are injection speed (B), melting temperature (C) and interaction AB (injection pressure and melting temperature) were found to be significant and should be set at low level. Confirmation run was conducted for over pack problem and the recommended optimal settings are injection pressure at high level, injection speed at low level, melting temperature at low level and holding pressure at high level (A2B1C1D2). Future studies could look at introducing others factors that may be involved in the process and to possibly use conventional design of experiment method