Passive solar shading is an effective strategy for reducing building energy demand, but its performance varies with climate, façade orientation, and thermal inertia. This study develops a sequentially coupled framework that links geometric shading calculation, anisotropic window heat gain prediction, and indoor thermal balance analysis across low- and high-latitude scenarios. For the low-latitude case, the model identifies a stable engineering overhang depth of about 1.85 m under the reference design space and weather inputs, while preserving winter solar admission. When compared with an unshaded baseline case with the same envelope, glazing, weather file, and internal gain assumptions, the optimized dynamic shading configuration reduces annual cooling load by more than 42% in the Guangzhou case study. For the high-latitude case, coupling shading with thermal mass parameters improves annual energy performance, and the best tested configuration achieves an energy-saving efficiency of 37.83% with an annual heating load of 96.14 MWh in the Stockholm scenario. The uncertainty and sensitivity analysis reports deterministic quantitative ranges and representative cases: the low-latitude recommended depth remains within the 1.85–1.864 m engineering neighborhood, while the Stockholm sensitivity sweeps show heating-load reductions of approximately 32.2–34.1% and indoor temperature variation reductions of up to 60.5–78.3% across the tested thermal mass parameter ranges. The discussion also clarifies the influence pathways of literature-sourced PCM and thermal property parameters, especially latent heat, thermal conductivity, and effective heat capacity. The quantitative validation boundary analysis distinguishes internal verification, controlled baseline benchmarking, and the external EnergyPlus/IDA ICE or measurement comparison still required for calibrated prediction. The results support the framework as a model-development tool for comparing passive design strategies under clearly defined assumptions, validation boundaries, practical engineering limits, and deterministic sensitivity ranges.
Chen et al. (Sun,) studied this question.