Thermal performance in naturally ventilated buildings based on future weather data and multiple climates

The increasing energy demand for cooling in buildings, driven by urbanization and rising global temperatures, has highlighted the need for sustainable design solutions. Natural ventilation (NV) in buildings offers a promising alternative to reduce energy consumption while maintaining occupant comfort. However, NV has been largely overlooked in modern building design, partly because of the advent of artificial cooling systems, but also due to the complexity of assessing its thermal performance. Building performance simulation (BPS) tools are essential for analyzing NV designs, yet they face significant challenges. These include accurately modeling wind-related variables and addressing the limitations particular to passive-cooled buildings. Surrogate models present a practical alternative to computationally intensive BPS, offering faster results while maintaining acceptable accuracy. They are particularly useful for exploring diverse climate scenarios and future weather projections, which are crucial for assessing building performance under climate change. This study developed a surrogate model to estimate the thermal performance of naturally ventilated cellular office buildings across different climates, including future scenarios. Sensitivity analysis was employed to identify key input variables and sources of error of single-zone models compared to multi-zone models, providing insights into the relative influence of parameters such as wall exposure, transmittance, and ground contact. The results showed that NV performance differs significantly between offices within the same building, necessitating individualized outputs. Additionally, while wind variables had limited impact due to simplified NV modeling, associating wind speed with direction improved accuracy. Dew-point temperatures and other climate variables were found to play a more significant role in predicting performance. The surrogate model demonstrated satisfactory performance despite the complexity of NV systems and achieved applicability across various climates worldwide. This work underscores the potential of surrogate models in shifting design paradigms towards NV, supporting sustainable building practices and reducing reliance on energy-intensive cooling systems.