(451e) Machine Learning-Based Modeling and Dynamic Optimization of Microwave-Enhanced Methane Coupling with Programmable Heating

Conventional reaction processes operate at steady state with continuous thermal heat input. Given the possible time-varying uncertainties and ongoing trend toward electrification [1-2], this may not provide the optimal strategy in response to the increasingly dynamic energy supplies, volatile market demands, etc [3-4]. Microwave (MW) heating has emerged as a promising approach as it enables targeted localized heating which notably improves energy efficiency and reaction selectivity [5-6]. Non-equilibrium dynamic MW operations can further leverage programmable heating to optimize energy use in real time and enable heat integration between endothermic and exothermic reactions [7-8]. Despite the potential, even the steady-state modeling of MW-assisted catalytic reactions remain a formidable challenge due to the sophisticated electromagnetic field and its selective heating impact on catalyst particles [9-10].

In this work, we present a machine learning (ML)-aided approach for the dynamic modeling and optimization of a MW-assisted non-oxidative methane coupling process. Heating is applied to facilitate this endothermic reaction following by cooling to enable coupling with potential exothermic reactions in a cyclic manner. Via ML-aided data-driven modeling, we study the following key research tasks: (i) dynamic modeling based on time-series experimental data, (ii) predictive modeling under various reaction conditions, (iii) optimizing the heating and cooling profiles to minimize energy use and prevent temperature overshoot. Artificial neural networks are adopted as the foundational ML technique. To address the challenge of limited experimental data in the context of ML for big data analytics, we employ weighted sampling to prioritize learning in regions of low data density during the heating process. Transfer learning is also explored to use a similar endothermic reaction process with well-established mechanistic model to transfer the knowledge of heat management to this novel microwave reaction process. This is instrumental to enhance sampling efficiency and modeling accuracy given the available experimental data. Dynamic optimization is finally applied to optimize the cyclic temperature profiles using the resulting data-driven surrogate model. This approach provides a systematic methodology to optimize the dynamic operations of emerging technologies based on experimental data.

References:

[1] Mallapragada, D. S., Dvorkin, Y., Modestino, M. A., Esposito, D. V., Smith, W. A., Hodge, B.-M., Harold, M. P., Donnelly, V. M., Nuz, A., Bloomquist, C., Baker, K., Grabow, L. C., Yan, Y., Rajput, N. N., Hartman, R. L., Biddinger, E. J., Aydil, E. S., & Taylor, A. D. (2023). Decarbonization of the chemical industry through electrification: Barriers and opportunities. Joule, 7(1), 23–41.

[2] Meloni, E. (2022). Electrification of Chemical Engineering: A New Way to Intensify Chemical Processes. Energies, 15(15), Article 15.

[3] Li, C., & Grossmann, I. E. (2021). A Review of Stochastic Programming Methods for Optimization of Process Systems Under Uncertainty. Frontiers in Chemical Engineering, 2.

[4] Sakki, G. K., Tsoukalas, I., Kossieris, P., Makropoulos, C., & Efstratiadis, A. (2022). Stochastic simulation-optimization framework for the design and assessment of renewable energy systems under uncertainty. Renewable and Sustainable Energy Reviews, 168, 112886.

[5] Goyal, H., Chen, T.-Y., Chen, W., & Vlachos, D. G. (2022). A review of microwave-assisted process intensified multiphase reactors. Chemical Engineering Journal, 430, 133183.

[6] Chen, W., Malhotra, A., Yu, K., Zheng, W., Plaza-Gonzalez, P. J., Catala-Civera, J. M., Santamaria, J., & Vlachos, D. G. (2021). Intensified microwave-assisted heterogeneous catalytic reactors for sustainable chemical manufacturing. Chemical Engineering Journal, 420, 130476.

[7] de la Fuente, J. F., Moreno, S. H., Stankiewicz, A. I., & Stefanidis, G. D. (2016). Reduction of CO2 with hydrogen in a non-equilibrium microwave plasma reactor. International Journal of Hydrogen Energy, 41(46), 21067–21077.

[8] Fu, B. A., Chen, M. Q., Li, Q. H., & Song, J. J. (2018). Non-equilibrium thermodynamics approach for the coupled heat and mass transfer in microwave drying of compressed lignite sphere. Applied Thermal Engineering, 133, 237–247.

[9] Yang, R., & Chen, J. (2021). Mechanistic and Machine Learning Modeling of Microwave Heating Process in Domestic Ovens: A Review. Foods, 10(9), Article 9.

[10] Masud, M. A. A., Araia, A., Wang, Y., Hu, J., & Tian, Y. (2025). Machine learning-aided process design using limited experimental data: A microwave-assisted ammonia synthesis case study. AIChE Journal, 71(1), e18621.