Renewable ammonia synthesis via experiments and optimization using periodic operation and machine learning integrated approaches

초록

The global demand for ammonia necessitates environmentally friendly production methods, as traditional synthesis emits about 0.5 gigatons of CO2 per year and requires high pressures and temperatures, highlighting the urgency of exploring alternative synthesis processes to address this challenge. Our study introduces the development and optimization of a novel, renewable ammonia synthesis methodology, leveraging low-pressure renewable hydrogen and advanced machine learning techniques. We employed a Al2O3-supported ruthenium catalyst to synthesize ammonia at low pressure, which is favorable for ammonia production under mild reaction conditions. The catalyst's efficiency was notably enhanced by integrating a periodic operation system, which modulated pressure between high and low levels during the reaction. This method significantly improved ammonia yield, even at reduced pressure settings. Experimental data were analyzed using an artificial neural networks (ANN) model to predict the synthesized ammonia volume and yield. This model underwent optimization with genetic algorithms (GA), considering experimental parameters such as the partial pressures of nitrogen (N2) and hydrogen (H2), reaction time, and reaction temperature. The feed gas's H2/N2 ratio was calculated based on these partial pressures. This optimization culminated in achieving an ammonia production rate of 3.298 cc/gcatmin and a yield of 78.13 % without catalyst promoter in a 1.07:1.93 Imbalanced mixture of H2 and N2 predicted by the machine learning techniques, illustrating the potential of this approach for enhancing ammonia production efficiency under environmentally favorable conditions.

키워드