Robust Multi-objective Optimization of Peanut Shell Biochar Production under Uncertainty Using Gaussian Process and NSGA-II
Main Article Content
Abstract
As fossil fuels continue to dominate the global energy mix, the thermochemical conversion of biomass via pyrolysis offers a carbon-neutral pathway to produce high-value biochar. However, optimizing this highly nonlinear process remains challenging, especially when constrained by limited experimental data. This study investigates the pyrolysis of peanut shell—an abundant agricultural residue in ASEAN countries—aiming to simultaneously maximize biochar yield and fixed carbon content. To overcome the small dataset limitation (N = 9), a Gaussian Process Regression (GPR) model was developed. Incorporating Leave-One-Out Cross-Validation (LOOCV) and hyperparameter tuning, the GPR approach effectively mitigated overfitting while providing intrinsic uncertainty quantification—a distinct advantage over conventional machine learning methods. The surrogate model demonstrated high predictive accuracy (R2 = 0.9975 for yield and 0.9341 for fixed carbon). By coupling GPR with the NSGA-II algorithm, a robust multi-objective optimization framework was established. The generated Pareto front identified optimal operating conditions that secured high fixed carbon levels (e.g., > 75 wt%) while maintaining total predictive uncertainty below 2.5%. Ultimately, this data-efficient framework provides a highly reliable strategy for optimizing biomass pyrolysis and scaling sustainable bioenergy systems with minimal experimental overhead.
Keywords
Biomass pyrolysis, Biochar yield, Gaussian Process Regression, NSGA-II, Uncertainty quantification.
Article Details
This work is licensed under a Creative Commons Attribution-NoDerivatives 4.0 International License.
References
[2] Australia Awards Cambodia, Agricultural Residue Open Burning and Its Impacts in Southeast Asia, Australia Awards Cambodia, 2025.
[3] United Nations Centre for Regional Development (UNCRD), Agricultural and Livestock Waste in Southeast Asia, UNCRD, Nagoya, Japan, 2022.
[4] N. T. Vinh, L. N. Han, H. P. N. Thao, and H. T. T. Hien, Trial production of biochar pellets from de-oiled cashew nut shells, TNU Journal of Science and Technology, vol. 230, no. 2, pp. 38–46, 2025.
[5] A. Gorshkov, N. Berezikov, A. Kaltaev, S. Yankovsky, K. Slyusarsky, R. Tabakaev, and K. Larionov, Analysis of the physicochemical characteristics of biochar obtained by slow pyrolysis of nut shells in a nitrogen atmosphere, Energies, vol. 14, iss. 23, Dec. 2021, Art. no. 8075. https://doi.org/10.3390/en14238075
[6] Y. Li, R. Gupta, and S. You, Machine learning assisted prediction of biochar yield and composition via pyrolysis of biomass, Bioresource Technology, vol. 359, Sep. 2022, Art. no. 127511. https://doi.org/10.1016/j.biortech.2022.127511
[7] B. Liu, G. Zhang, S. Hong, and Z. Zhang, ANN assisted optimization of phosgene free hexamethylene diisocyanate production for economic and environmental benefits, Scientific Reports, vol. 15, Sep. 2025, Art. no. 33973. https://doi.org/10.1038/s41598-025-11730-3
[8] E. Yapıcı, H. Akgün, K. Özkan, Z. Günkaya, A. Özkan, and M. Banar, Prediction of gas product yield from packaging waste pyrolysis: Support vector and Gaussian process regression models, International Journal of Environmental Science and Technology, vol. 20, pp. 461–476, Feb. 2022. https://doi.org/10.1007/s13762-022-04013-1
[9] J. Wei, Y. Sun, D. Zhu, Q. Liu, X. Wang, and Z. Liu, PAL-GP: A parallel active learning framework based on Gaussian processes for efficient fatigue reliability assessment of structures, Structures, vol. 86, Apr. 2026, Art. no. 111294.
https://doi.org/10.1016/j.istruc.2026.111294
[10] W. Bi, Y. Wu, Y. Xu, and A. Zhang, Optimized deployment–monitoring synchronized array method for passive sonobuoys based on an improved NSGA-II algorithm, Applied Soft Computing, vol. 195, Jun. 2026, Art. no. 115065.
https://doi.org/10.1016/j.asoc.2026.115065
[11] K. Phuakpunk, B. Chalermsinsuwan, and S. Assabumrungrat, Comparative artificial neural network models for predicting kinetic parameters of biomass pyrolysis from the biomass characteristics, Results in Engineering, vol. 27, Sep. 2025, Art. no. 106701. https://doi.org/10.1016/j.rineng.2025.106701
[12] J. Xing, K. Luo, H. Wang, Z. Gao, and J. Fan, A comprehensive study on estimating higher heating value of biomass from proximate and ultimate analysis with machine learning approaches, Energy, vol. 188, Dec. 2019, Art. No. 116077.
https://doi.org/10.1016/j.energy.2019.116077
[13] P. J. García-Nieto, E. García-Gonzalo, J. P. Paredes-Sánchez, and L. A. Menéndez-García, Interpretable machine learning models for forecasting elemental composition of solid biomass fuels from proximate analyses in energy processes, Journal of the Energy Institute, vol. 123, Dec. 2025, Art. no. 102322. https://doi.org/10.1016/j.joei.2025.102322
[14] P. Talwar, M. A. Agudelo, and S. Nanda, Pyrolysis process, reactors, products, and applications: A review, Energies, vol. 18, iss. 11, Jun. 2025, Art. no. 2979. https://doi.org/10.3390/en18112979
[15] S. N. Mohan, N. N. Nair, P. V. S. Nair, K. V. Babu, and K. M. George, Biochar as an exceptional bioresource for energy, agronomy, carbon sequestration, activated carbon and specialty materials, Waste Biomass Valorization, vol. 7, no. 2, pp.201–235, Apr. 2016.
[16] A. K. Nanda, K. K. Berruti, A. Dalai, J. A. Kozinski, and S. C. Cameron, Advanced techniques in the production of biochar from lignocellulosic biomass and environmental applications, Biofuel Research Journal, vol. 3, no. 1, pp. 363–372, Mar. 2016.
[17] B. Zhao, D. O’Connor, J. Zhang, T. Peng, Z. Shen, D. C.W. Tsang, and D. Hou, Effect of pyrolysis temperature, heating rate, and residence time on rapeseed stem derived biochar, Journal of Cleaner Production, vol. 174, pp. 977–988, Feb. 2018.
https://doi.org/10.1016/j.jclepro.2017.11.013