Title: Repeated-value simplex designs - better rotations of the classical simplex experimental design

Authors: Joseph Colannino

Addresses: Faber Burner Company, 1000 Bald Eagle Street, Lock Haven, PA 17745, USA

Abstract: Classical simplex designs allow the assessment of n factors in n + 1 experimental runs. However, the classical design specifies levels, intervals, and coded values that differ from factor to factor, and this is experimentally inconvenient. This research note presents the means to rotate the simplex to orient all factors to the same three coded values { p, q, r}, or to ±1 if n = 2J - 1, where J is an integer. Such rotations render the simplex design much more convenient for experimental work. This note also presents a means for generating two-value simplex matrices that may be viewed as non-orthogonal fractions of two-level factorial designs.

Keywords: simplex; repeated-value simplex; RVS; simplex rotation; experimental design; linear regression.

DOI: 10.1504/IJEDPO.2021.123133

International Journal of Experimental Design and Process Optimisation, 2021 Vol.6 No.4, pp.349 - 359

Received: 18 Oct 2021
Accepted: 02 Mar 2022

Published online: 30 May 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article