Investigation of Co-Combustion Characteristics of Waste Biomasses and Anthracite by TG-DTG/DSC and Machine Learning Method

ABSTRACTThe co-combustion characteristics and thermodynamics of waste biomasses, namely coconut shell and corncob, with anthracite were investigated using TG-DTG/DSC analysis and artificial neural networks. TG-DTG experiments were conducted at different ratios of CS-AN and CC-AN, and the samples were heated from 293 K to 1273 K at heating rates of 10, 15, 20, 25, and 30 K·min−1. The results indicated three main stages in the combustion process of the mixtures and the combustion profiles shifted to higher temperatures with increasing heating rates. As the mass fraction of biomass in the mixtures increased from 10 wt.% to 90 wt.%, the residual mass of the co-combustion mixture decreased by 19.50 wt.% and 21.78 wt.% for CS-AN and CC-AN, respectively. The comprehensive combustion index of CS-AN increased from 24.645%2·min−2·K−3 to 167.274%2·min−2·K−3, while that of CC-AN increased from 26.262%2·min−2·K−3 to 102.654%2·min−2·K−3. The DSC results showed a peak of heat absorption followed by an exotherm in the co-combustion process. As the mass fraction of anthracite in the mixtures increased, the curve shifted toward the high temperature region and the exothermic time increased. The apparent activation energies (E) for CS, CC, and AN were 75.36, 112.08, and 90.90 kJ·mol−1, respectively. Additionally, 50 wt.% CS and 50 wt.% CC were more conducive to CS-AN and CC-AN co-combustion processes. 63 ANN models were utilized to predict the post-combustion residual weight of CS5-AN5. ANN56 was identified as the best performer, with error results (RSME = 1.482, MBE = 1.120, R2 = 0.9982) and a topology of (3 × 15 * 5 × 1), which proved to be the most suitable model for the co-combustion characteristics of this mixture. This study provides theoretical and practical guidance for solving engineering problems related to the energy utilization of waste biomasses (coconut shell and corncob) with anthracite.KEYWORDS: Co-combustionwaste biomassesanthracitethermo–kineticsTG-DTG/DSC analysisartificial neural network AcknowledgementsThe authors gratefully acknowledge the financial support from the Demonstration and Guidance Project of Science and Technology Benefiting People in Qingdao (No. 21-1-4-sf-15-nsh), High quality courses of postgraduate education, Ocean University of China (No. HDYK21007), Support plan for youth entrepreneurship and technology of colleges and universities in Shandong Province (No. 2019KJB013), Postgraduate Education Joint Cultivation Base Construction Projects, Ocean University of China (No. HDYJ23005).Disclosure statementNo potential conflict of interest was reported by the author(s).Additional informationFundingThis work was supported by Demonstration and Guidance Project of Science and Technology Benefiting People in Qingdao (No. 21-1-4-sf-15-nsh), High quality courses of postgraduate education, Ocean University of China (No. HDYK21007), Support plan for youth entrepreneurship and technology of colleges and universities in Shandong Province (No. 2019KJB013), Postgraduate Education Joint Cultivation Base Construction Projects, Ocean University of China (No. HDYJ23005).