(598b) Process Synthesis Optimization for Microwave-Assisted Ammonia Production with Data-Driven Modeling

The pursuit of energy efficiency and environmental sustainability has stimulated the development of innovative process intensification technologies in chemical process industry. The use of microwave heating for chemical reaction synthesis presents a promising technique which can result in higher selectivity at milder reaction conditions benefited from direct and selective volumetric heating [1]. Despite the recent impetus [2-4], the modeling and simulation of microwave-assisted reaction systems remain an open challenge due to the sophisticated multi-phase (e.g., gas-solid), multi-physics (e.g., reaction, electromagnetism), and multi-scale (e.g., catalyst particle in millimeters to reactor in meters) phenomena. This also hinders to further evaluating the systems integration of these emerging technologies to retrofit existing process infrastructure [5].

To address these challenges, this work aims to develop a process synthesis optimization framework with application to microwave-assisted ammonia production. The framework features: (i) Machine learning-aided modeling of microwave reaction using experimental data, and (ii) System-level superstructure optimization of various ammonia production routes with cost and sustainability considerations. Specifically, a set of 46 data is first collected from experimental results examining the impact of reaction temperature, pressure, flowrate, and hydrogen/nitrogen ratio on ammonia concentration [6]. A neural network (NN) model is then developed using the original experimental data and the synthetic data generated via Synthetic Minority Oversampling Technique (SMOT) [7-8]. The prediction accuracy and optimization validity of the SMOTE-integrated NN model are verified against the experimental data and expert knowledge. Utilizing this data-driven model for microwave-assisted reactor, we proceed to construct a comprehensive superstructure network encompassing different commercial or emerging technology alternatives to design the overall ammonia production flowsheet. The superstructure includes options for hydrogen production through steam methane reforming and water electrolysis, as well as nitrogen extraction via air separation unit. These streams are sent to Haber-Bosch reactor and/or the microwave reactor for ammonia synthesis. The effluents from these reactors undergo separation processes, such as distillation and membrane separation, to obtain the desired ammonia product. Data-driven surrogate models are developed from first-principles simulation (e.g., Aspen Plus) using neural networks to reduce modeling complexity while ensuring sufficient physical accuracy [9]. We demonstrate the integration of these surrogate models into mixed-integer nonlinear programming for superstructure optimization, thereby systematically identifying the optimal production routes under various flowsheet design objectives (e.g., productivity, cost, sustainability).

References

[1] Hu, J., & Reddy, B. M. (Eds.). (2023). Advances in Microwave-assisted Heterogeneous Catalysis (No. 45). Royal Society of Chemistry.

[2] Goyal, H.; Chen, T. Y.; Chen, W.; Vlachos, D. G. A Review of Microwave-Assisted Process Intensified Multiphase Reactors. Chem. Eng. J. 2022, 430, 133183.

[3] Lakerveld, R.; Sturm, G.; Stankiewicz, A.; Stefanidis, G. Integrated Design of Microwave and Photocatalytic Reactors. Where Are We Now? Curr. Opin. Chem. Eng. 2014, 5, 37–41.

[4] Mevawala, C.; Bai, X.; Bhattacharyya, D.; Hu, J. Dynamic Data Reconciliation, Parameter Estimation, and Multi-Scale, Multi-Physics Modeling of the Microwave-Assisted Methane Dehydroaromatization Process. Chem. Eng. Sci. 2021, 239, 116624.

[5] Industrial Decarbonization Roadmap. Department of Energy. September 2022.

[6] Wang, Y., Khan, T. S., Wildfire, C., Shekhawat, D., & Hu, J. (2021). Microwave-enhanced catalytic ammonia synthesis under moderate pressure and temperature. Catalysis Communications, 159, 106344.

[7] Liu, X., Xu, Y., Li, J., Ong, X., Ibrahim, S. A., Buonassisi, T., & Wang, X. (2021). A robust low data solution: Dimension prediction of semiconductor nanorods. Computers & Chemical Engineering, 150, 107315.

[8] Masud, M. A. A.; Araia, A.; Wang, Y.; Hu, J.; Tian, Y., Machine Learning-Aided Process Design for Microwave-Assisted Ammonia Production. Computer-Aided Chemical Engineering. Under review

[9] Henao, C. A., & Maravelias, C. T. (2011). Surrogate‐based superstructure optimization framework. AIChE Journal, 57(5), 1216-1232.