Gaussian Process Surrogates and Uncertainty Aware NSGA II for Optimizing Wind Turbine Blade Pyrolysis Parameters
Authors
Odysseas Dellas, Sotiris Kafatos, Thanasis Petrou, Achilleas KallisFiles
Abstract
Recovery of valuable products from used wind turbine blades requires operating conditions that balance condensable yield, energy demand and the preservation of reinforcing fibres. This study develops an uncertainty-aware optimization framework combining Gaussian process surrogates with the nondominated sorting genetic algorithm II. The case study considers temperature, heating rate, holding time and nitrogen flow within a bounded operating domain. Response definitions and uncertainty conventions are applied consistently throughout the analysis. Separate surrogate models predict liquid yield, specific energy demand and retained fibre strength, with conservative prediction bounds used for screening. The evaluation protocol specifies 120 development cases, 40 calibration cases, and 40 independent assessment cases; the displayed diagnostic subsets provide a numerical assessment of the decision rules. A typical compromise has a liquid yield of 27.4 wt%, an energy requirement of 6.20 MJ per kg of dry feed, and a retained strength of 76.0%. Compared with the designated reference setting, these values are an increase of 3.4 percentage points in yield and a reduction of 15.1% in energy. A conservative lower strength bound of 72.0% exceeds the acceptance threshold of 70.0%. Fifteen data panels show the variations in response, calibration requirements, trade-offs and operating sensitivity of the operating alternatives. The framework provides a reproducible specification linking process performance and surrogate uncertainty for subsequent engineering verification.
References
Meng, F., Hawkin, J. L., Gast, L., Job, S., Ó Brádaigh, C. M., Cullen, J. M., Leeke, G. A., McKechnie, J., Poulsen, A. K., Rothman, R., Ryan, A. J., Todd, I., Walton, A., Harper, G. D. J., Yang, Z., & Zhang, L. (2025). End-of-life management for wind turbines. Nature Reviews Clean Technology, 1(10), 677–698. https://doi.org/10.1038/s44359-025-00097-3 DOI: https://doi.org/10.1038/s44359-025-00097-3
Hu, Y., Zhang, Y., Li, Y., Wang, Y., Li, G., & Liu, X. (2024). Wind turbine blade recycling: A review of the recovery and high-value utilization of decommissioned wind turbine blades. Resources, Conservation and Recycling, 210, Article 107813. https://doi.org/10.1016/j.resconrec.2024.107813 DOI: https://doi.org/10.1016/j.resconrec.2024.107813
Royuela, D., Martínez, J. D., López, J. M., Callén, M. S., García, T., Verdejo, R., Murillo, R., & Veses, A. (2024). Pursuing the circularity of wind turbine blades: Thermochemical recycling by pyrolysis and recovery of valuable resources. Journal of Analytical and Applied Pyrolysis, 181, Article 106657. https://doi.org/10.1016/j.jaap.2024.106657 DOI: https://doi.org/10.1016/j.jaap.2024.106657
Lu, Q., Ji, H., Lu, Y., Yang, J., Chen, W., Li, J., Li, W., & Xu, M. (2025). Transition-metal-assisted pyrolysis to recover glass fibers from end-of-life wind turbine blades. Journal of Analytical and Applied Pyrolysis, 189, Article 107081. https://doi.org/10.1016/j.jaap.2025.107081 DOI: https://doi.org/10.1016/j.jaap.2025.107081
Zhang, B., Zhang, S., Yang, Z., Liu, W., Wu, B., Huang, M., & Liu, B. (2024). Pyrolysis process and products characteristics of glass fiber reinforced epoxy resin from waste wind turbine blades. Composites Part B: Engineering, 287, Article 111803. https://doi.org/10.1016/j.compositesb.2024.111803 DOI: https://doi.org/10.1016/j.compositesb.2024.111803
Manfredi, P. (2024). Probabilistic uncertainty propagation using Gaussian process surrogates. International Journal for Uncertainty Quantification, 14(6), 71–104. https://doi.org/10.1615/int.j.uncertaintyquantification.2024052162 DOI: https://doi.org/10.1615/Int.J.UncertaintyQuantification.2024052162
Zheng, W., & Doerr, B. (2025). Approximation guarantees for the nondominated sorting genetic algorithm II (NSGA-II). IEEE Transactions on Evolutionary Computation, 29(4), 891–905. https://doi.org/10.1109/tevc.2024.3402996 DOI: https://doi.org/10.1109/TEVC.2024.3402996
Kim, H., & Sanz-Alonso, D. (2025). Enhancing Gaussian process surrogates for optimization and posterior approximation via random exploration. SIAM/ASA Journal on Uncertainty Quantification, 13(3), 1054–1084. https://doi.org/10.1137/24m1677009 DOI: https://doi.org/10.1137/24M1677009
Xu, M.-X., Ji, H.-W., Wu, Y.-C., Meng, X.-X., Di, J.-Y., Yang, J., & Lu, Q. (2024). Recovering glass fibers from waste wind turbine blades: Recycling methods, fiber properties, and potential utilization. Renewable and Sustainable Energy Reviews, 202, Article 114690. https://doi.org/10.1016/j.rser.2024.114690 DOI: https://doi.org/10.1016/j.rser.2024.114690
Xu, M., Yang, J., Ji, H., Wu, Y., Li, J., Di, J., Meng, X., Jiang, H., & Lu, Q. (2024). Recovering high-quality glass fibers from end-of-life wind turbine blades through swelling-assisted low-temperature pyrolysis. Waste Management, 187, 179–187. https://doi.org/10.1016/j.wasman.2024.07.020 DOI: https://doi.org/10.1016/j.wasman.2024.07.020
Liang, T., Wang, F., Wang, S., Li, K., Ma, X., & Mo, X. (2024). Towards unified aleatory and epistemic uncertainty quantification for machinery health prognostic through sequential heteroscedastic Gaussian process regression. Advanced Engineering Informatics, 62, Article 102719. https://doi.org/10.1016/j.aei.2024.102719 DOI: https://doi.org/10.1016/j.aei.2024.102719
Cabral, R., Bolin, D., & Rue, H. (2025). Robustness, model checking, and hierarchical models. Journal of the Royal Statistical Society Series B: Statistical Methodology, 87(3), 632–652. https://doi.org/10.1093/jrsssb/qkae107 DOI: https://doi.org/10.1093/jrsssb/qkae107
Papadopoulos, H. (2024). Guaranteed coverage prediction intervals with Gaussian process regression. IEEE Transactions on Pattern Analysis and Machine Intelligence, 46(12), 9072–9083. https://doi.org/10.1109/tpami.2024.3418214 DOI: https://doi.org/10.1109/TPAMI.2024.3418214
Deb, K., Pratap, A., Agarwal, S., & Meyarivan, T. (2002). A fast and elitist multiobjective genetic algorithm: NSGA-II. IEEE Transactions on Evolutionary Computation, 6(2), 182–197. https://doi.org/10.1109/4235.996017 DOI: https://doi.org/10.1109/4235.996017
Aghaei pour, P., Hakanen, J., & Miettinen, K. (2024). A surrogate-assisted a priori multiobjective evolutionary algorithm for constrained multiobjective optimization problems. Journal of Global Optimization, 90(2), 459–485. https://doi.org/10.1007/s10898-024-01387-z DOI: https://doi.org/10.1007/s10898-024-01387-z
Yousef, S., Eimontas, J., Zakarauskas, K., Stasiulaitiene, I., Striūgas, N., & Tuckute, S. (2025). Catalytic pyrolysis of wind turbine blades waste for plasticizers recovery and its life cycle assessment. Journal of Environmental Management, 395, Article 127690. https://doi.org/10.1016/j.jenvman.2025.127690 DOI: https://doi.org/10.1016/j.jenvman.2025.127690
Ren, K., Liu, S., Kong, S., Ma, S., & Feng, R. (2026). Thermal recycling of wind turbine blade composites: Optimizing pyrolysis and oxidation for fiber recovery. Composites Part A: Applied Science and Manufacturing, 202, Article 109460. https://doi.org/10.1016/j.compositesa.2025.109460 DOI: https://doi.org/10.1016/j.compositesa.2025.109460
Zhang, Y., Zhang, Z., Ma, W., Chen, P., Bai, B., Li, L., Lai, Y., & Wang, X. (2025). Pyrolysis and oxidation characteristics and energy self-sustaining process design of retired wind turbine blades. Journal of Analytical and Applied Pyrolysis, 185, Article 106872. https://doi.org/10.1016/j.jaap.2024.106872 DOI: https://doi.org/10.1016/j.jaap.2024.106872
Wang, Y., & Politis, D. N. (2026). Model-free bootstrap and conformal prediction in regression: Conditionality, conjecture testing, and pertinent prediction intervals. Journal of Nonparametric Statistics, 38(1), 311–345. https://doi.org/10.1080/10485252.2025.2603398 DOI: https://doi.org/10.1080/10485252.2025.2603398
Hajibabaee, P., Pourkamali-Anaraki, F., & Hariri-Ardebili, M. A. (2024). Adaptive conformal prediction intervals using data-dependent weights with application to seismic response prediction. IEEE Access, 12, 53579–53597. https://doi.org/10.1109/access.2024.3387858 DOI: https://doi.org/10.1109/ACCESS.2024.3387858
Ginsbourger, D., & Schärer, C. (2025). Fast calculation of Gaussian process multiple-fold cross-validation residuals and their covariances. Journal of Computational and Graphical Statistics, 34(1), 1–14. https://doi.org/10.1080/10618600.2024.2353633 DOI: https://doi.org/10.1080/10618600.2024.2353633
Trottner, L. (2024). Covariate shift in nonparametric regression with Markovian design. Information and Inference: A Journal of the IMA, 13(2), Article iaae011. https://doi.org/10.1093/imaiai/iaae011 DOI: https://doi.org/10.1093/imaiai/iaae011
Pajasmaa, J., Miettinen, K., & Silvennoinen, J. (2025). Group decision making in multiobjective optimization: A systematic literature review. Group Decision and Negotiation, 34(2), 329–371. https://doi.org/10.1007/s10726-024-09915-8 DOI: https://doi.org/10.1007/s10726-024-09915-8
Zhang, H., Yi, W., Li, H., & Wang, S. (2025). Performance metrics evaluation of multi-objective optimization for transportation. Electronic Research Archive, 33(6), 3968–3988. https://doi.org/10.3934/era.2025176 DOI: https://doi.org/10.3934/era.2025176
Cheng, X., Yang, Y., Zhu, C., Kong, X., & Liu, J. (2026). A constraint violation-based adaptive switching surrogate-assisted evolutionary algorithm for expensive constrained multiobjective optimization. Swarm and Evolutionary Computation, 104, Article 102372. https://doi.org/10.1016/j.swevo.2026.102372 DOI: https://doi.org/10.1016/j.swevo.2026.102372
Feng, Z., Wang, J., & Huang, W. (2026). An active learning multi-objective robust optimization framework via Gaussian process models under uncertainty. Engineering Optimization, 58(4), 976–999. https://doi.org/10.1080/0305215x.2025.2495800 DOI: https://doi.org/10.1080/0305215X.2025.2495800
Dayimu, A., Simidjievski, N., Demiris, N., & Abraham, J. (2024). Sample size determination for prediction models via learning-type curves. Statistics in Medicine, 43(16), 3062–3072. https://doi.org/10.1002/sim.10121 DOI: https://doi.org/10.1002/sim.10121
Li, Y., Yorke-Smith, N., & Keviczky, T. (2024). Machine learning enabled uncertainty set for data-driven robust optimization. Journal of Process Control, 144, Article 103339. https://doi.org/10.1016/j.jprocont.2024.103339 DOI: https://doi.org/10.1016/j.jprocont.2024.103339
Malluhi, B., Fezai, R., Kravaris, C., Nounou, H., Al-Rawashdeh, M., & Nounou, M. (2024). Guided experimental design for static nonparametric modeling. Chemical Engineering Science, 298, Article 120327. https://doi.org/10.1016/j.ces.2024.120327 DOI: https://doi.org/10.1016/j.ces.2024.120327