An Application of Hotelling's T2 Decomposition in a Multivariate Statistical Process Control

ABSTRACT Multivariate Statistical Process Control (MSPC) is the most acceptable monitoring tool for several variables, it is advantageous when compared to the simultaneous use of univariate scheme. However, one of the challenges in this scheme is the identification of influential variable(s). Any signal by this scheme implies one or more variables in the process are out-of-control. Therefore, identification of such variable(s) correctly is very important. The aim of this research is to determine the variable(s) that cause out-of-control signal in cement production process of Cement Company of Northern Nigeria (CCNN), Sokoto. The data used for this research were percentages of clinker minerals (Alite, Bilite, Aluminate and Ferrite), obtained from the Cement Company of Northern Nigeria (CCNN), Sokoto. We achieved our aim through the construction of Hotelling’s T2 phase I and II control charts, presentation of MYT decomposition models and computation of the T2 decomposition terms. Observations composing of 100 samples regarding cement clinker variables in Phase I were acquired in order to form the reference data set in the application. It was found that 82 of these samples formed the reference data set as a result of 8 steps analyses. In Phase II, T2 chart was constructed for 30 new observations and it was detected that 7 samples were out-of-control. The MYT decomposition revealed the T2 values and their corresponding critical values, the T2 values that were higher than their corresponding critical values were responsible for the out of control signal. We conclude that variable 4 (Ferrite) is the influential variable.