Translating third-order data analysis methods to chemical batch processes

Measurements collected from batch processes naturally produce a third-order or three-dimensional data form. The same structure also results when multi ple samples are measured using hyphenated analysis techniques such as liqui d chromatography with diode array detection. Analysis of third-order data b y principal components analysis (PCA) is achieved by a nonunique rearrangem ent that produces a two-dimensional array. This preferentially models only one of the three orders present. In contrast, methods such as parallel fact or analysis (PARAFAC) apply a particular decomposition that accounts for al l three orders explicitly. The results from either approach should be relat ed if data are to be interpreted reliably for applications to batch process es such as on-line monitoring and control. This work compares these two app roaches from an applied point of view. To accomplish this objective, exempl ary methods are selected from each type of analysis, parallel factor analys is (PARAFAC) and multiway principal components analysis (MPCA). These are e mployed to analyze data obtained during the manufacture of a condensation p olymer in an industrial batch reactor. (C) 1999 Elsevier Science B.V. All r ights reserved.