An improved Taguchi's method in design of experiments for milling CFRP composite

Abstract

The Taguchi's method of design of experiments has been applied in industry for many years. The greatest advantage of this method is to save the effort in conducting experiments: to save the experimental time, to reduce the cost, and to find out significant factors fast. Because the original data for ANOVA is assumed to be normally distributed with uniform variances, Taguchi does not consider these assumptions. Instead, he suggests use of extra experiments to confirm data. These extra experiments raise the cost accordingly, and thus violate his original purpose. In our research, we propose two methods to avoid extra experiments: (1) confirming data and (2) screening effects. To verify the proposed methods, we conduct experiments in milling CFRP composite as examples. From our analysis, we obtain some significant effects that are originally regarded as errors in Taguchi's method, and the best operating conditions thus obtained are more accurate, while the extra experiments are no longer required.