The cardinal objective of response surface methodology is reliability and quality improvement through well designed experiments. The Central Composite Design (CCD) is the most popular and most used response surface design. Extensive attempts are made in this work to improve the prediction performances of the CCD and its variants by augmenting the design with partially replicated components. Substantial improvements on the prediction characteristics of the designs are well documented. The results contained therein will be very useful to practitioners in various fields of research who earnestly desire quality and reliability through statistical process control.