Bioenergy potential of Saccharum bengalense through pyrolysis, reaction kinetics, TG-FTIR-GCMS analysis of pyrolysis products, and validation of the pyrolysis data through machine learning

• Published in: Chemical engineering journal (Lausanne, Switzerland : 1996) Switzerland : 1996), 2023-06, Vol.465, p.142930, Article 142930
• Main Authors: Asghar, Azeem, Liu, Chen-Guang, Ali, Imtiaz, Khan, Aqib Zafar, Zhu, Hui, Wang, Ning, Nawaz, Muhammad, Tabish, Tanveer A., Mehmood, Muhammad Aamer, Rasool, Raqiqa Tur
• Format: Article
• Language: English
• Subjects: ANN; Green energy and chemicals; M-DAEM; Net-zero carbon emissions; Pyrolysis; TG-FTIR-GCMS
• ISSN: 1385-8947; 1873-3212
• DOI: 10.1016/j.cej.2023.142930

[Display omitted] •Bioenergy potential was Saccharum bengalense was explored through pyrolysis.•Pyrolytic products were studied using coupled TG-FTIR-GCMS.•ANN, M−DAEM machine learning was applied to elucidate the pyrolysis process.•Evolved gases contained green chemicals for industrial and energy applications.•MLP-based ANN regression model demonstrated perfection in training, test, and validation. Saccharum bengalense grows on non-arable lands throughout the world. Its oven-dried biomass showed 18.05 ± 0.37 (MJ kg−1) of High Heating Value (HHV), 76.47 ± 0.58 % of volatile matter, and very low content of sulfur (0.46 ± 0.04 %). Here, its oven-dried biomass was pyrolyzed at four heating rates (β: 10, 20, 40, 80 ℃min−1) in a thermogravimetric analyzer. The thermogravimetric data were subjected to isoconversional models including Friedman, KAS, and FWO to understand the pyrolysis process. Pyrolysis reaction occurred in three consecutive stages where main pyrolysis reaction happened during the second stage. The average values of activation energy were shown to be 186.01 kJ mol−1, 169.56 kJ mol−1, and 170.99 kJ mol−1 as assessed through Friedman, KAS, and FWO methods, respectively, which indicated the suitability of the biomass for pyrolysis. The TG-FTIR-GCMS analysis showed that pyrolytic gases contained acetone, aromatic compounds, hydrocarbons, phenols, ketones, alcohols, esters, and aldehydes. The MLP 3–6-1 network model was shown to be the best which was used to understand the conversion process. Activation energies were accurately determined using the MLP-based ANN regression model, where values of R2 (0.999) indicated perfection in training, test, and validation. The data from the ANN model were consistent with the TG-based interpretation of the pyrolysis and represented the first-order thermal degradation. The M−DAEM model had a good fit with the experimental data because the dα/dt and R2 of α were higher than 99.98 % and 99.99 %, respectively. The pyrolytic products were copious because of the high percentage of carbon atoms. The data demonstrated that S. bengalense biomass has promising potential to become a biorefinery feedstock to produce chemicals and energy from renewable biomass while keeping the energy-water-environment nexus sustainable.