In this work, we present a surrogate-based modeling and superstructure optimization framework for ammonia production, with particular interest in exploring the application potential of novel microwave-assisted ammonia synthesis reactors. The superstructure considers the flowsheet from hydrogen and nitrogen acquisition, ammonia synthesis, to downstream product purification. Representative commercialized and emerging modular/intensified technologies are examined such as Haber-Bosch reactor versus microwave reactor, steam methane reforming versus electrolysis, etc. We showcase the use of SMOTE-integrated neural network modeling to incorporate microwave reactor as emerging process technology, in system-level analysis based on experimental data. For the other process technologies with available mechanistic simulations, surrogate models are developed to address the computational complexity of large-scale superstructure optimization. These neural network-based models are embedded into a single mixed-integer nonlinear programming formulation. The following application scenarios are investigated to: (i) maximize ammonia production, (ii) maximize ammonia production per unit of energy consumption, (iii) minimize equivalent annualized operating cost, and (iv) minimize the environmental impact quantified via Eco-Indicator 99. Multi-objective optimization is also performed using the -constrained method for minimizing environmental impact and total expenses. This work delivers a rigorous comparative assessment of current and next-generation ammonia production technologies, yielding insights into the optimal technology selections under varying decision-making priorities and evaluation criteria.
扫码关注我们
求助内容:
应助结果提醒方式:
