Journal of the Nigerian Statistical Association, Vol. 30, 2018 Ukaegbu & Chigbu 30 Comparison of the central composite design with augmented alternative designs for response surface exploration Eugene C. Ukaegbu 1 and Polycarp E. Chigbu Department of Statistics, University of Nigeria, Nsukka, Nigeria The cardinal objective of response surface methodology is reliability and quality improvement through well designed experiments. The choice of response surface design that satisfies this objective has been the subject of many studies on response surface explorations. The standard central composite design (CCD) remains popular choice among practitioners but offers high number of experimental runs with increase in number of factors. In this study, we propose augmented Minimum-run resolution V CCDs (MinResV CCD) as useful alternatives to the standard CCD in response surface exploration. The standard CCD and the augmented MinResV CCD are evaluated and compared with some alphabetic and graphical criteria in spherical and cuboidal design regions. Some augmented versions of the MinResV CCDs displayed better potentials for quality improvement and with smaller design runs in most cases than the standard CCD. Keywords: Design augmentation, Design size, Fraction of design space graph, Prediction variance, Spherical region 1. Introduction The central composite design (CCD) of Box and Wilson (1951) was developed for exploration of response surfaces with curvature and the design exists for 2  k factors, where k is a fixed positive integer for both spherical and cuboidal response surfaces. The structure of the central composite design which consists of the cube (factorial) portion, the star (axial) portion and the centre point makes it flexible and adaptable. The number of runs, N, of the CCD increases rapidly as k increases, especially for 6  k . This has serious implication on the cost of experimentation, as some practitioners cannot afford to use the design for response surface exploration. For this reason, the small central composite design (SCD) by Hartley (1959) and minimum-run resolution V (MinResV) design by Oehlert and Whitcomb (2002) were developed as smaller alternatives to the CCD. The SCD and MinResV designs are readily available for 2  k like the CCD but are more useful for higher number of factors, especially for 6  k . Both designs also exist for the cuboidal and spherical regions. However, unlike the CCD, the two smaller designs do not possess the orthogonality and rotatability properties. The three designs, CCD, SCD and MinResV CCD, are made up of the same three components: the cube, the star and the centre point. The difference between them lies in the structure of their cubes. The cube component of the CCD is at least of resolution V. To obtain the SCD, the cube component of the CCD is replaced with a resolution III factorial design. The minimum-run resolution V CCD is obtained by replacing the resolution V factorial component of the CCD with a minimum-run resolution V design which gives fewer runs than the standard CCD (see, for example, 1 Corresponding Author. E-mail: eugene.ukaegbu@unn.edu.ng, Phone : +2348033728139