Using the Taguchi design and Central Composite Design to increase the robustness of a process from its raw material variability

This research presents an experimental study where four controllable and one noise factor are analyzed using the Taguchi approach and the Central Composite Design (CCD) in order to induce robustness in a process that is being designed to treat residual grease from small restaurants. Three response variables are analyzed separately to establish the parameters conditions to obtain the most robust process output. These suggested conditions are validated conducting confirmation runs that are compared statistically to the predictions from the Taguchi prediction model for the Taguchi approach, and the Response optimizer for the CCD. The accuracy of each method is measured as a percentage of predictions that can not be declared statistically different from the confirmation runs samples. The data analysis indicates that the Taguchi approach has a 100% accuracy level to predict robust responses while the Central Composite Design reports a 25% of accuracy.