Box-Behnken design and artificial neural network guided insights on viscosity reduction and maximum swollen volume of heat-treated sugarcane bagasse based kraft black liquor

Abstract

Heat treatment is an efficient technology for reducing the viscosity and improving the combustion behaviour of non-wood black liquors for chemical recovery in pulp mills. However, the interactive effects of process parameters and predictive modelling of this technology are little reported in the literature. This research presents Box-Behnken Design (BBD) and Artificial Neural Network (ANN) guided insights on a heat treatment process used for addressing the high viscosity and lower swelling problems associated with kraft black liquor of sugarcane bagasse. Here, the heat treatment of sugarcane bagasse based semi-concentrated kraft black liquor (SB_SCKBL) was performed considering the temperature (175–185 °C), time (10–20 min), and residual active alkali (RAA) as Na₂O (4.5–6.5 %) as the process inputs while the viscosity reduction and maximum swollen volume (MSV) as the outputs. The viscosity reduction and MSV quadratic models revealed R² of 1.0 and 0.9869, respectively, with p_value < 0.05, signal: noise >4.0 and non-significant lack of fits. A significant interaction between temperature–time, temperature-RAA, and RAA-time was noted towards viscosity reduction and MSV. A viscosity reduction of 75.24–76.96 % and MSV of 10.3-11.0 ml/g have been obtained at the optimum process conditions with an overall desirability of 0.66. ANN models developed for viscosity reduction and MSV using a back-propagation algorithm gave the overall R-values of 0.999 and 0.832, respectively. The original and optimally heat-treated SB_SCKBL was studied for physicochemical characteristics, SEM, FTIR, XRD, TGA and DSC. This study might be beneficial for designing black liquor heat treatment plants in pulp mills operating on agro-residues.