The application of multivariate quality control in gas chromatography

Quality control is important in present laboratory activities. Nowadays, an analysis must satisfy high requirements, which can only be met under appropriate control. An example of a rather simple and often used tool for quality control is the Shewhart control chart. A disadvantage of the Shewhart chart is that it requires much effort with a risk for false alarms or mistakes, certainly when many control variables are involved. Multivariate control charts are an alternative to univariate Shewhart charts. In this paper, the results are presented for both univariate Shewhart control charts and multivariate control charts applied to the capillary gas chromatographic analysis of the fatty acid composition in BCR162 (soya-maize oil). Quite often one may observe a shift in the data when a process is monitored for a long period. Although such a shift may result in statistical out-of-control situations, the analytical method may still be capable to perform within the specifications. In this paper, we introduce an extension of the T2-chart for multivariate process control in order to handle statistically instable processes, with sufficient capability.