Understanding the trends and uncertainties in Building Energy Simulation (BES) performance indicators under future climate conditions is crucial for mitigating issues such as overheating and power outages. To address this, we generated a set of weather files for all 27 state capitals in Brazil, considering six climate model projections (three General Circulation Models as driving models and two nested Regional Climate Models) and two distinct emission scenarios from the CORDEX project. We analyzed the variability in climatic variables and subsequently performed BES on a representative Brazilian social housing unit to evaluate its impact on the performance indicators outcomes. Consistent with previous studies, a substantial increase in cooling-related demands was observed in the more pessimistic scenario (RCP8.5) and mild increases in the more optimistic scenario (RCP2.6), with a trend toward stabilization after 2050. Regarding uncertainties, we found higher Relative Standard Deviation (RSD) values for the cooling degree hours indicator. The capitals in the Central-West, Southeast, and South regions exhibited greater uncertainty regarding temperature indicators, whereas the irradiation parameters displayed higher uncertainties in the Northeast region. For the BES outcomes, RSD values as high as 19.9% were found for cooling load values. It was also demonstrated that locations, periods, and scenarios exhibit different extreme climate model projections. Ideally, employing an ensemble of weather files developed from other models would help assess associated uncertainties in the building performance indicators.

中文翻译：

---

用于评估巴西建筑热性能的多个区域气候模型预测：了解不确定性

了解未来气候条件下建筑能源模拟 (BES) 性能指标的趋势和不确定性对于缓解过热和断电等问题至关重要。为了解决这个问题，我们为巴西所有 27 个州首府生成了一组天气文件，考虑了六个气候模型预测（三个大气环流模型作为驱动模型和两个嵌套区域气候模型）以及 CORDEX 项目中的两个不同的排放情景。我们分析了气候变量的变化，随后对巴西代表性的社会住房单元进行了 BES，以评估其对绩效指标结果的影响。与之前的研究一致，在更悲观的情景（RCP8.5）中观察到与制冷相关的需求大幅增加，在更乐观的情景（RCP2.6）中温和增加，并在2050年后趋于稳定。关于不确定性，我们发现冷却度小时指标的相对标准偏差 (RSD) 值较高。中西部、东南部、华南地区首府气温指标不确定性较大，东北地区辐照参数不确定性较高。对于 BES 结果，发现冷负荷值的 RSD 值高达 19.9%。研究还表明，地点、时期和情景表现出不同的极端气候模型预测。理想情况下，采用从其他模型开发的天气文件集合将有助于评估建筑性能指标中的相关不确定性。