On the robustness of central composite designs when missing two observations

Abstract

AbstractQuite often, many situations arise in experimental work where some experimental runs that may not be feasible because of certain circumstances or existing constraints in the system under investigation. A lot of work has been done in studying classical designs, such as the central composite designs (CCDs), with missing observations. In this paper our interest is to research the losses that may occur when one or two observations are missing. The general formulas for missing two observations are presented in a closed form as a function of α, where α is the value used in the axial part in the CCDs. These are calculated explicitly for CCDs up to factors, for the first time in this paper, and are displayed in tables for practitioners to use. The losses for any value of α, in the axial points of the design, can be derived by simply replacing α with the desirable value in the provided formulas. A detailed comparison between our findings and results already known in the literature also been made.