Optimizing power-to-ammonia plant: Sizing, operation, and production forecasting using deep learning approach

Hydrogen and ammonia hold substantial potential as energy carriers. This study investigates the optimal sizing of a renewable power plant for sustainable hydrogen and ammonia production within a hot, semi-arid climate in Morocco. A dynamic energy model is proposed in this work using Limited-memory Broyden–Fletcher–Goldfarb–Shanno with Bounds (LM-BF-GS-B) optimization algorithm. Moreover, a deep learning-enhanced approach was adopted to forecast power and ammonia production. The developed models assess the plant's performance based on criteria including affordability, sustainability, and acceptability while exploring the technical, economic, social, and political dimensions. The obtained findings demonstrate that a hybrid renewable energy system consisting of an optimized combination of wind and solar with energy storage integration is the most cost-effective configuration. This can reduce the amount of storage required. The investigated power plant has the capacity to supply an average daily amount of 438 t/day of ammonia. In addition, highly accurate results were achieved using the long short-term memory (LSTM) framework. The obtained MSE is in the range of 561.4, 5.86, 0.15 for PV, wind, and ammonia production outputs, respectively. Furthermore, an attractive levelized cost of electricity (LCOE) of approximately 0.018 $/kWh is achieved. The levelized costs of hydrogen and ammonia is equal to 2.59 $/kg and 623.9 $/t, respectively.