Optimizing Solar Chimney–Double-Skin Façade Integration in High-Rise Buildings: A Multi-Criteria CFD Parametric Study with Machine-Learning-Based Prediction

A building façade normally keeps the weather out, lets in daylight, allows ventilation, and shapes how a building looks from outside. This study asks whether it can also help generate electricity without giving any of that up. Solar chimney power plants (SCPPs) generate clean electricity from solar heat, but they have mostly been studied for open, land-abundant rural sites. Mounting one onto a façade instead risks the very things a façade is meant to protect: thermal comfort, natural ventilation, and architectural freedom. No prior study has looked at energy output, thermal behaviour, double-skin façade (DSF) operability, and architectural freedom together for solar chimneys integrated into high-rise buildings, and this is the gap this work addresses. We coupled solar chimney power plants with double-skin façades across five configurations, evaluated using Computational Fluid Dynamics (CFD) validated against the Manzanares pilot plant to achieve 3.3% for velocity and 3.0% for temperature. Using the DSF as both collector and absorber (Model 1) pushes power to its highest point, 78.7 kW, but it drives inner-façade air to 327.1 K and shuts off ventilation entirely. Confining the collector to the roof and upper chimney instead (Model 5) settles for a more modest 31.1 kW, but it preserves DSF ventilation over most of the façade height and keeps inner air below 305.8 K through the majority of that range, with only the uppermost portion becoming thermally unsuitable for natural ventilation. Weighing power, thermal load, ventilation, and façade freedom together in a composite score, Model 5 comes out as the preferred configuration under the adopted equal-weight multi-criteria assessment. A parametric study of height, irradiance, and ambient temperature for Model 5 produced design equations that, paired with a Random Forest climate forecast, benchmarked against three alternative algorithms and evaluated on a chronological hold-out set (city-level test R2 up to 0.83 for irradiance and 0.94 for temperature), power a predictive framework for hourly-to-annual energy output at any height and city, demonstrated here for seven cities across five continents.