Enhanced near zero-energy building performance through intelligent hydrogen storage management across diverse climates

This study explores the implementation of solar energy systems to achieve near-zero energy performance in buildings across diverse climates. The energy systems were modeled using TRNSYS, a powerful transient simulation tool that, while versatile, lacks built-in optimization capabilities. To address this, an optimization framework combining neural networks and genetic algorithms was introduced. This system modeled in this study includes photovoltaic (PV) panels and hydrogen storage. Results reveal that PV panels alone can generate between 27 % and 45 % of the required electricity across these cities. By integrating hydrogen storage, the system's energy contribution rises significantly, fulfilling between 54 % and 84 % of the building's total energy needs, depending on the specific climate. Hydrogen storage enhances system reliability, reducing dependence on external power sources. In the Gansu case study, the neural network and genetic algorithm-based optimization found the best mix of PV panel array, fuel cell capacity, and electrolyzer power to lower the need for electricity from the grid, CO2 emissions, and project costs. At the optimal configuration, the system achieved annual CO2 emissions of 125 tons, a loss of power supply probability (LPSP) of 0.28, and an hourly operational cost of $1.37.