Examination of membrane performance with multivariate methods: A case study within a pulp and paper mill filtration application

The membrane permeability is affected by the characteristics of the membrane, chemical surroundings of the membrane and the process conditions in the filtration application. In pulp and paper mill applications the process waters contain various solutes of different chemical nature and also the process chemistry might vary. Thus, the reasons causing the changes in permeability might be very complicated. Therefore, optimisation of the membrane permeability requires a considerable number of experiments, which produces a vast amount of experimental data. Due to the multivariate nature of filtration phenomena, univariate examination cannot show effectively the correlation structure between the variables in the filtration system. Usually, only univariate monitoring and interpretation is typically used to analyse filtration data. However, in this study the experimental results from a pilot-scale pulp and paper mill filtration application were examined with multivariate methods principal component analysis (PCA), dynamic partial least squares (PLS) regression and parallel factor analysis (PARAFAC) in order to find the most significant factors affecting the membrane permeability and to extend the knowledge obtained from univariate examination. The multivariate methods applied revealed crucial information about the influence of the quality of the filtrated process water on the permeability. Moreover, the tested methods were shown to be very effective tools in the examination of membrane filtration results.