Regression modeling to predict ultrafine particles emission in a mineral plant combining meteorological and process variables

Abstract

Ultrafine particles are object of main health concern, but its concentration is challenging to be continuous monitored in mineral and metallurgical industrial processes. This paper shows the development of an empirical regression model correlating the ultrafine particles concentration measured by two continuous analyzers, electrodynamic (EDA) and optical scatter (OSA) with meteorological and process parameters. The analyzers were installed at stack of an industrial mineral fertilizer plant over 4 seasons. The results showed that EDA have poor correlation with process or meteorological parameters (r-squared less than 10%) what can be caused by particles not being charged evenly on the stream as its better accuracy for particles over 10µm, as previous studies had suggested. The OSA ultrafine particles concentration model showed r-squared of 45% correlation with meteorological parameters and raw material feed. The model presented and standard error of 0.21 mg/Nm 3 which is considered adequate for industry compliance purposes. OSA shows promising application if meteorological parameters are included, as already in practice for ultrafine particles monitoring outdoors.