Dynamic simulation of a triple-mode multi-generation system assisted by heat recovery and solar energy storage modules: Techno-economic optimization using machine learning approaches

Intelligent design and operation optimization allow energy systems to take advantage of the flexibility that multi-generation provides. This study proposes a basic solar-driven system integrated with thermal energy storage for round-the-clock energy harvesting. A modified configuration is then designed incorporating innovative multi-heat recovery approaches to increase the capacity and product diversity of the basic system. The modified system is able to cover vital urban utilities such as electricity, fresh water, cooling, and hydrogen throughout the day. To overcome the time-consuming procedure of dynamic techno-economic simulation as well as the limi-tation of commercial engineering equation solvers for tri-objective optimization, a deep learning approach is developed to reduce the computational complexity and improve the analysis accuracy. In this regard, the trained neural networks play an intermediary role in coupling the developed code with the MATLAB optimization toolbox. A comparison between the modified and conventional configurations indicates that implementing the multi-heat recovery approach results in a 29% increase in power generation while only increasing the overall system cost by 1.97%. From an economic perspective, the Sankey diagram depicts that the storage unit with a cost rate of 12.25 $/h accounts for 6.77% of the plant's cost rate, which enables the system to operate contin-uously. According to the sensitivity analysis and contour plots, the number of collectors significantly affects the total cost rate and fresh water production capacity while it has no tangible effect on the exergy efficiency.