This study develops an interpretable machine learning (ML) surrogate to predict hourly indoor air temperature and discomfort indicators for a representative Mexican social-housing prototype in San Luis Potosí (cold semi-arid, Köppen–Geiger BSk). A four-zone EnergyPlus model with constant window opening (50%) and no internal gains was used to generate a parametric dataset spanning 24 building orientations, seven roof solar absorptance levels, and two neighborhood configurations (surrounded vs. corner). Zone-specific bagged-tree regression models were trained in MATLAB using weather predictors, temporal indicators, and weather-memory features (including outdoor temperature lags and rolling averages). Orientation and roof absorptance were included as explicit design predictors, enabling the surrogate model to generalize across the full combinatorial design space rather than requiring a separate model for each configuration. Interpretability was assessed with SHAP values. Evaluated on orientation–absorptance combinations deliberately held out during training, the surrogate achieved high accuracy across zones of the house (R² = 0.98–0.99; RMSE = 0.31–0.67 °C) with stable, near-zero-centered residuals. When propagated into adaptive-comfort metrics computed directly relative to the monthly neutral temperature $T_n$, ML predictions preserved the main cold and hot discomfort degree-hour patterns across the full design space. The proposed surrogate enables rapid, physically consistent comfort-oriented screening of roof finishes and orientation choices in naturally ventilated social housing.

The variability of solar energy limits its reliability as a thermal resource, motivating the use of thermal energy storage (TES) to extend heat availability beyond periods of direct irradiance. This study numerically compares latent and sensible TES integrated into a solar dish system from a resource-oriented perspective across representative Mexican climates. Rather than focusing only on stored energy, the analysis evaluates how each storage strategy affects the temporal availability and post-irradiation persistence of usable thermal energy over 24 h charge–discharge cycles. A salt-based PCM (58.1LiNO₃–41.9KCl) was assessed against steel-based sensible storage under identical operating conditions. Under average-day forcing, the minimum PCM mass required to effectively utilize latent heat while sustaining a 320 W test load was found to be 13 kg. Under these conditions, the PCM case showed smoother thermal transients and longer post-irradiation energy availability, enabling nocturnal operation. In contrast, a mass-matched 13 kg steel store showed negligible post-irradiation availability, while a volume-matched 55 kg steel configuration achieved similar nocturnal operation only by substantially increasing mass, with limited improvement in accumulated energy. Hot-day forcing extended the operating window, whereas cold-day forcing yielded negligible charging so that operation could not be sustained within a single daily cycle.