Insight into textile sludge combustion behavior: Kinetic study by thermal analysis and advanced machine learning modeling

Abstract

In this study, the combustion potential of textile sludge is evaluated through thermogravimetric analysis. The analysis is conducted in an oxidative atmosphere with heating rates ranging from 5, 10, 15 and 20 °C/min and temperatures ranging from ambient to 900 °C. Kinetic analysis was complemented using the isoconversational model-free approach, including the differential and integral methods (Friedman, Ozawa-Flynn-Wall and Kissinger-Akahira-Sunose). The average activation energies (E_a) calculated by these methods were about 296.73 kJ/mol, 324.97 kJ/mol, and 318.75 kJ/mol, respectively. The reaction mechanism was derived from the combined kinetic analysis, which showed a high R² value of 0.99507, indicating a strong correlation between the experimental data and the kinetic analysis results. The analysis of the activation energy distribution was performed by utilizing four pseudo-components (PC1-PC4). Furthermore, Artificial Neural Networks (ANN), Classification and Regression Trees (C&RT), Boosted Regression Trees (BRT), and Multivariate Adaptive Regression Splines (MARS) were employed to predict the E_a for textile sludge combustion. This detailed exploration of kinetics and the development of innovative predictive modeling techniques like ANN, C&RT, BRT, and MARS establish a new standard for creating customized models for the thermochemical conversion of textile sludge.