Scholarlink

<p>Weather files strongly influence building performance simulation, especially in climatically diverse countries such as Brazil. Yet, the representativeness of typical meteorological years (TMY) for Brazilian applications remains insufficiently assessed. This study aims to evaluate four established TMY approaches and to introduce a performance-based alternative (Brazilian typical meteorological year, BTMY) that incorporates building-specific climatic sensitivity. Using ERA5-Land data for 480 locations, we compiled 15 actual meteorological years (2008–2022) per site and assessed their impact on operative temperature and energy needs in representative residential models. Spearman filtering removed collinear meteorological parameters, and optimized XGBoost and ε-SVR models predicted daily operative temperature with <i>R</i><sup>2</sup> = 0.91 ± 0.03 (RMSE = 0.64 ± 0.19 °C) and daily energy needs with <i>R</i><sup>2</sup> = 0.87 ± 0.05 (RMSE = 0.05 ± 0.01 kWh m<sup>−2</sup> yr<sup>−1</sup>), identifying dry-bulb temperature as the dominant meteorological parameter (75% importance for operative temperature; 59% for energy needs). Against 15-year simulations, minimum Finkelstein–Schafer and Best Rank I achieved the lowest mean operative-temperature bias (0.15 °C, SD 0.09-0.10 °C), ISO was similar (0.16 ± 0.13 °C), while Pissimanis performed worst (0.23 ± 0.27 °C). BTMY weather files matched the TMY methods and clarified key climatic drivers. Results showed that ERA5-Land data source mattered more than the reference-year method. Finally, the study delivered more than 3000 new weather files to support simulation practice and policy development.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Weather files for building simulation in Brazil: A national benchmark

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月1日

查看全文
  • Mario Alves da Silva,
  • Giovanni Pernigotto,
  • Alessandro Prada,
  • Andrea Gasparella,
  • Joyce Correna Carlo

摘要

Weather files strongly influence building performance simulation, especially in climatically diverse countries such as Brazil. Yet, the representativeness of typical meteorological years (TMY) for Brazilian applications remains insufficiently assessed. This study aims to evaluate four established TMY approaches and to introduce a performance-based alternative (Brazilian typical meteorological year, BTMY) that incorporates building-specific climatic sensitivity. Using ERA5-Land data for 480 locations, we compiled 15 actual meteorological years (2008–2022) per site and assessed their impact on operative temperature and energy needs in representative residential models. Spearman filtering removed collinear meteorological parameters, and optimized XGBoost and ε-SVR models predicted daily operative temperature with R2 = 0.91 ± 0.03 (RMSE = 0.64 ± 0.19 °C) and daily energy needs with R2 = 0.87 ± 0.05 (RMSE = 0.05 ± 0.01 kWh m−2 yr−1), identifying dry-bulb temperature as the dominant meteorological parameter (75% importance for operative temperature; 59% for energy needs). Against 15-year simulations, minimum Finkelstein–Schafer and Best Rank I achieved the lowest mean operative-temperature bias (0.15 °C, SD 0.09-0.10 °C), ISO was similar (0.16 ± 0.13 °C), while Pissimanis performed worst (0.23 ± 0.27 °C). BTMY weather files matched the TMY methods and clarified key climatic drivers. Results showed that ERA5-Land data source mattered more than the reference-year method. Finally, the study delivered more than 3000 new weather files to support simulation practice and policy development.