Spatial variations in the distribution of trace ionic impurities in the water-steam cycle in a thermal power plant based on a multivariate statistical approach

Abstract

AbstractAbstract In this study, a multivariate statistical approach was used to identify the key variables responsible for process water quality in a power plant. The ion species that could cause corrosion in one of the major thermal power plants (TPP) in Serbia were monitored. A suppressed ion chromatographic (IC) method for the determination of the target anions and cations at trace levels was applied. In addition, some metals important for corrosion, i.e., copper and iron, were also analysed by the graphite furnace atomic absorption spectrophotometric (GFAAS) method. The control parameters, i.e., pH, dissolved oxygen and silica, were measured on-line. The analysis of a series of representative samples from the TPP Nikola Tesla, collected in different plant operation modes, was performed. Every day laboratory and on-line analysis provides a large number of data in relation to the quality of water in the water-steam cycle (WSC) which should be evaluated and processed. The goal of this investigation was to apply multivariate statistical techniques and choose the most applicable technique for this case. Factor analysis (FA), especially principal component analysis (PCA) and cluster analysis (CA) were investigated. These methods were applied for the evaluation of the spatial/temporal variations of process water and for the estimation of 13 quality parameters which were monitored at 11 locations in the WSC in different working conditions during a twelve month period. It was concluded that PCA was the most useful method for identifying functional relations between the elements. After data reduction, four main factors controlling the variability were identified. Hierarchical cluster analysis (HCA) was applied for sample differentiation according to the sample location and working mode of the TPP. On the basis of this research, the new design of an optimal monitoring strategy for future analysis was proposed with a reduced number of measured parameters and with reduced frequency of their measurements. Graphical abstract