(711a) Time-Series Integrated Surrogate Modeling for Control of Ammonia Synthesis and Adsorption Processes

As the second most produced chemical compound, ammonia is a critical component in the fertilizer industry. It is also utilized in plastic production, pesticides, dyes, and textiles [1, 2]. Moreover, recent applications of ammonia, such as its potential as a carbon-free hydrogen carrier and fuel source, indicate a growing demand for its production in the future [3]. The Haber- Bosch process has been employed for industrial-scale ammonia production for over a century. This process combines nitrogen and hydrogen gases over a metal-based catalyst within a reactor operating under high pressure and temperature [4]. Subsequently, an adsorption process separates the produced ammonia from the unreacted gas mixture. A model-based dynamic optimization and control approach can achieve optimal synthesis and storage with minimum energy consumption.

Developing such a framework relies on an integrated multi-scale model that precisely describes the reaction kinetics and absorption dynamics. The computational challenges associated with this model, especially for integrated process systems, render it impractical for real-time decision-making and control applications. In our previous work [5], we tackled the high computational demands of a single multi-scale model for ammonia synthesis in a packed-bed reactor (PBR) by utilizing long short-term memory (LSTM) surrogate models [6, 7, 8]. The effectiveness of the LSTM surrogate model was demonstrated through an integrated state estimation and model predictive control (MPC) approach [9], enabling dynamic regulation of the outlet concentration and hotspot temperature. However, while effective for single reactor operation, this framework faces limitations in extending further to integrate subsequent processes such as absorption.

In this work, we develop an LSTM-based MPC framework for integrated ammonia synthesis and absorption processes. Creating a single surrogate model for this integrated process poses substantial computational challenges during training. It often leads to poor predictive capabilities due to the distinct reactor and adsorbed bed dynamics. Therefore, a novel integrated surrogate modeling approach is introduced that leverages offline simulations of the integrated multi-scale models for rapid and accurate prediction of the system’s outputs. This approach focuses on developing separate encoder-decoder LSTM models for the ammonia synthesis reactor and adsorption bed based on the offline open-loop simulations of individual processes using COMSOL Multiphysics. Each LSTM surrogate model is designated to capture each process’s temporal dynamics, considering their integrated dynamics. Furthermore, the absorption surrogate model incorporates the outputs of the reactor’s surrogate model, allowing it to harmonize its outputs with the reactor’s dynamics. This comprehensive surrogate model is tailored to grasp the kinetics and thermodynamic effects of the synthesis reaction on the absorption process. The integrated LSTM-based model is then employed as the basis for MPC design. The resulting closed-loop performance in regulating the ammonia synthesis reactor’s hotspot temperature and outlet concentration, and adsorption capacity is evaluated under various operating conditions.

References:
[1] FAO. World fertilizer trends and outlook to 2022. https://doi.org/10.4060/ca6746en, 2020.

[2] D. R. MacFarlane, P. V. Cherepanov, J. Choi, B. H. R. Suryanto, R. Y. Hodgetts, J. M. Bakker, F. M. Ferrero Vallana, and A. N. Simonov. A roadmap to the ammonia economy. Joule, 4(6):1186–1205, 2020.

[3] A. M. Elbaz, S. Wang, T. F. Guiberti, and W. L. Roberts. Review on the recent advances on ammonia combustion from the fundamentals to the applications. Fuel Communications, 10:100053, 2022.

[4] J. Humphreys, R. Lan, and S. Tao. Development and recent progress on ammonia synthesis catalysts for haber–bosch process. Adv. Energy Sustainability Res., 2:2000043, 2021.

[5] T. O. Cabral, A. Bagheri, and D. B. Pourkargar. Learning-based model predictive control of an ammonia synthesis reactor. In Proceedings of the American Control Conference. IEEE, 2024.

[6] M. Aparicio and J. A. Dumesic. Ammonia synthesis kinetics: surface chemistry, rate expressions and kinetic analysis. Top. Catal., 1:233–252, 1994.

[7] R. Schl ̈ogl. Catalytic synthesis of ammonia - A “never-ending story”? Angew. Chemie - Int. Ed., 42(18):2004–2008, 2003.

[8] Q. Qian, A. An, A. Fortunelli, R. J. Nielsen, and W. A. Goddard III. Reaction mechanism and kinetics for ammonia synthesis on the Fe(111) surface. J. Am. Chem. Soc., 140:6288–6297, 2018.

[9] J. B. Rawlings, D. Q. Mayne, M. Diehl, et al. Model predictive control: theory, computation, and design, volume 2. Nob Hill Publishing Madison, WI, 2017