Table 6 ANOVA results for the first-order regression model

significantly  estimates  the  shrinkage.  The  P  value indicates

the significance of process parameters on the shrinkage. If the P values are less than 0.05 (i.e., a = 0.05 or 95%  confi- dence), the process parameters in the regression model are significant. As can be clearly seen in Table 5, Tmelt, Itime, and Ctime are statistically the most significant process  parameters for the shrinkage. The other process  parameters  (Tmold  and Pinj) are not significant because their P value is bigger than

0.05.  Table 6  reveals  the  results  of  the  ANOVA     analysis

showing that the process parameters on the shrinkage are sta- tistically significant. The F value of 98.72 in Table 6 implies that the first-order regression model is significant. The bold value in Table 5 and 7 shows whether the process parameters are important or not.

The second-order regression model, in terms of coded values of process parameters, can be found as  below:

Source DF SSE MSE F P

Regression 5 5.3364 1.0673 98.72 0.000

Residual error 21 0.2270 0.0108

Total 26 5.5634

Table 7    The results of the second-order regression  model

Predictor Coefficient SE coefficient T P

I2 0.02506 0.04224 0.59 0.559

Table 5    The results of the first-order regression   model

Predictor Coefficient SE coefficient T P

S = 0.1035 R-Sq(pred) = 93.90%

PRESS = 0.339428 R-Sq = 95.9%

R-Sq(adj) = 94.9%

Constant 1.0120 0.6120 1.65 0.113

Tmold 0.001267 0.001225 1.03 0.313 Table 8    ANOVA results for the second-order  regression

Tmelt 0.037461 0.002451 15.29 0.000 model

Pinj —0.00030 0.001225 —0.24 0.809

Itime —0.38861