An improved numerical model for early detection of bed agglomeration in fluidized bed combustion

Abstract

An improved predictive numerical index has been developed to predict the tendency of bed agglomeration in fluidized bed boilers. The index was developed based on the melt fraction resulting from the thermodynamic equilibrium model of fuel ash compositions together with SiO₂ as the bed material at temperatures ranging from 700 to 900 °C. The partial least squares regression (PLSR) coupled with the cross-validation technique is utilized to establish the correlation for the bed agglomeration index, I_a. The improved index, I_a has been validated by experimental observations found in various literature sources. The results obtained using the improved index, I_a demonstrated a significantly higher success rate in predicting the bed agglomeration tendency of biomass fuel ash compared to the other four conventional bed agglomeration indices. In addition, K₂O is the main element that accelerates the formation of bed agglomeration in the biomass firing while CaO was found to reduce the tendency of bed agglomeration in the fluidized bed combustion system.