World Journal of Engineering Research and Technology

Aguh Patrick Sunday* and Ejikeme Ifeanyi Romanus

ABSTRACT

In this study, we investigated the adequacy of approximation of fitted model to the real system. To meet this objective, the numerical optimization method was applied as an alternative to the conventional method by Suresh (2014) for the validation of optimized performance characteristic (response) model. Box ? Behnken design, a spherical design, with all factorial and axial points at the same distance from the center of the design was used to build a model for predicting and optimizing the flexural strength of WARPP. The mathematical model equations (response surface models) were derived from the response surface methodology (RSM) optimization process using Minitab 16 software. This response surface model is representable with nonlinear power equation and second order polynomial of the four control factors. In the statistical analysis, the analysis of variance (ANOVA) table, the coefficient of determination, R2 and the adjusted coefficient of determination, R2adj values and p values indicated that the model was adequate for representing the experimental data. The p values show that the linear effects of volume fraction (B), injection force (C), operating temperature (D) and quadratic effects of volume fraction (B^2), injection force (C^2) and the interaction effects of volume fraction and injection force (B*C) and volume fraction and operating temperature (B*D) were highly significant. To gain a better understanding of the control variables for optimal flexural performance, the graphical models were presented as 3 ? D response surface and 2 ? D contour graphs. The optimal setting for the particle size (A), volume fraction (B), injection force (C) and operating temperature (D) were found to be 1.40mm, 5%, 200tons and 215oC respectively at the highest flexural strength value of 50.0933MPa. The predicted R2 value of 99.97% and adjusted R2 value of 99.99% are within 0.20 of each other indicating that the predicted response model and experimental data are in reasonable agreement (Mourabet et. al., 2013).

[Full Text Article] [Download Certificate]