Multilevel multifactor designs for multivariate calibration

Multilevel experiments are common in chemistry, especially in calibration and mixture problems. This paper presents designs for l = 2–5 different concentration levels and l2 corresponding experiments. The importance of orthogonality between successive factors is discussed. It is shown that up to 12 mutually orthogonal factors can be generated for a five-level design. Methods of generating designs are generalised. The designs are restricted to first-order (linear) models, typical of most instrumental calibration experiments.