Application of multivariate statistical methods to the modelling of a flue gas treatment stage in a waste-to-energy plant

Among all the flue gas components produced in waste-to-energy plants, acid airborne pollutants such as SO2 and HCl have the most rigorous emission standards provided by the European Parliament. Their removal is thus a key step of the flue gas treatment which is mainly achieved with the Dry Treatment Systems (DTS), technologies based on the direct injection of dry solid sorbents which is capable to subtract the acid from the gas stream with several important advantages and high removal efficiencies. However, the substantial lack of a deeper industrial knowledge makes difficult to determine accurately an optimal operating zone which should be required for the design and operation of these systems. The aim of this study has been therefore the exploration, while basing on an essential engineering expertise, of some of the possible solutions which the application of multivariate statistical methods on process data obtained from real plants can give in order to identify all those phenomena which rule dry treatment systems. In particular, a key task of this work has been the seeking for a general procedure which can be possibly applied for the characterization of any type of DTS system, regardless of the specific duty range or design configuration. This required to overcome the simple mechanical application of the available techniques and made necessary to tailor and even redefine some of the available standard procedures in order to guarantee specific and objective results for the studied case. Specifically, in this so called chemometric analysis, after a pre-treatment and quality assessment, the process data obtained from a real working plant was analyzed with basic and advanced techniques in order to characterize the relations among all the available variables. Then, starting from the results of the data analysis, a linear model has been produced in order to be employed to predict with a certain grade of accuracy the operating conditions of the system.