LAUSR

Abstract The aim of this study was to develop a data analysis method which could provide operational insights and guidelines waste-to-energy (WtE) plant operators. A method to filter outliers with changing properties was combined with a cross-correlation analysis method that can capture nonlinearity and quantify time lags between variables. The method was applied to a dataset obtained from a commercial WtE plant. The method was able to detect already established correlations such as the influence of combustion conditions on NOx and CO emissions, which both had positive correlation with O2 concentration in the flue gas, while the effect of combustion conditions was unnoticeable for HCl emissions. Furthermore, the method could detect that NOx and SO2 emissions exhibited positive correlations with the furnace temperature. Time lags provided additional information about the sensor locations and plant dynamics. This methodology can be used especially when process data is available while good process models are not immediately accessible for determining non-obvious process phenomena not only for WtE sector but also for process industry in general.