Prediction of gas product yield from packaging waste pyrolysis: support vector and Gaussian process regression models

Yapıcı E., Akgün H., ÖZKAN K., Günkaya Z., Özkan A., Banar M.

International Journal of Environmental Science and Technology, cilt.20, sa.1, ss.461-476, 2023 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 20 Sayı: 1
  • Basım Tarihi: 2023
  • Doi Numarası: 10.1007/s13762-022-04013-1
  • Dergi Adı: International Journal of Environmental Science and Technology
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Agricultural & Environmental Science Database, Aqualine, Aquatic Science & Fisheries Abstracts (ASFA), Biotechnology Research Abstracts, CAB Abstracts, Compendex, Environment Index, Geobase, INSPEC, Pollution Abstracts, Veterinary Science Database
  • Sayfa Sayıları: ss.461-476
  • Anahtar Kelimeler: C, LDPE, Gaussian process regression, LDPE, Pyrolysis, Support vector regression, CATALYTIC PYROLYSIS, DENSITY POLYETHYLENE, POWER-GENERATION, PLASTIC WASTES, NEURAL-NETWORK, H-BETA, RECOVERY, DEGRADATION, ALUMINUM, LDPE
  • Eskişehir Osmangazi Üniversitesi Adresli: Evet

Özet

© 2022, The Author(s) under exclusive licence to Iranian Society of Environmentalists (IRSEN) and Science and Research Branch, Islamic Azad University.The pyrolysis process enables the transformation of plastic waste into products such as oil, solid residue, and gas at temperatures of around 300–900 °C by thermal decomposition. Conversion of such waste into valuable products depends on various factors, such as raw material composition, temperature, heating rate, residence time, and catalyst. From this point of view, in this study, predictions of gas product yield based on different pyrolysis conditions including waste types (LDPE–C/LDPE), temperature (400–600–800 °C), heating rate (5–10–20 °C/min), type of catalyst (zeolite-clay-sludge) and amount of catalyst (5%, 10%, 15%, by weight) were carried out with support from the vector regression (SVR) and the Gaussian process (GPR) models using the results of experimental studies performed under various conditions. Different kernel functions were used for SVR (Linear, Quadratic, Cubic, Gaussian) and GPR (Squared Exponential, Matern 5/2, Exponential, Rational Quadratic). The Gaussian Kernel Function presented a good prediction performance (89% R2 and 0.0011 RMSE) for SVR while the Exponential Kernel Function was the most appropriate for GPR (93% R2 and 0.0011 RMSE). On the other hand, the deviations in the SVR model with linear Kernel change over a wide range of 0.25–80.85%, and the GPR model with exponential kernel show deviations close to each other in the range of 0.06–3.91%. The present study provides new information for future studies by understanding the pyrolysis process of plastic waste and predicting product yield.