Pareto ANOVA tables

Table 2.17 Pareto ANOVA table to assess the s/n ratio in the worked examp le. ...

Taguchi also suggested the use of Pareto s ANOVA [12]. This technique does not require any statistical assumption so a statistical analysis of the responses cannot be performed. Figure 2.7 shows a Pareto s ANOVA table. 

Figure 2.7 Pareto s ANOVA table to assess the statistical significance of design factors according to Taguchi.

As our objective is to maximise the s/n ratio, we select factor A at level 1 or 2 indistinctly, factor F at level 2 or 3, factor C at level 1 and factor D at level 2. With this selection we are sure to obtain minimal variability. However, we also need an average equal to 3.5. Table 2.18 summarises the Pareto ANOVA for the average and Figure 2.9 displays the contribution of each term when the average response of the experiments in the experimental matrix is considered. 

Table 2.18 Pareto ANOVA for the average values of the worked example.

Pareto ANOVA analysis is an excellent tool for determining the contribution of each input parameter and their interactions with the output parameters (surface roughness). It is a simplified ANOVA analysis method that does not require an ANOVA table. Further details on Pareto ANOVA can be found in Park [12]. 

The Pareto ANOVA analysis for aluminum (AISI 6061) is given in Table 32.4. It shows that feed rate (B) has the most significant effect on surface roughness with a contribution ratio (P = 65%), followed by cutting speed (A) P = 9%), and depth of cut (C) (P S 3%). The interactions between cutting speed and feed rate (A x B) and feed rate and depth of cut (B x C) also played roles with contributing ratios P S 13%) and (P = 9%), respectively. It is worth pointing out that the total... 

Table 32.4 Pareto ANOVA analysis for aluminum (AISI 6061)...

Table 32.5 Pareto ANOVA analysis for mild steel (AISI 1030)...

Table 32.6 Pareto ANOVA analysis for alloy steel (AISI 4340)...

Variations in surface roughness for input parameters cutting speed and feed rate are illustrated in Fig. 32.5. Effect of depth of cut is omitted because it demonstrated negligible percent contribution in Pareto ANOVA analyses (Tables 32.4-32.6). 

---