Weather files for building performance simulation in Brazil: typical, performance-based, and extreme years

This repository contains a large collection of weather files for building performance simulation in Brazil, covering 480 locations distributed across the country. The files were created from ERA5-Land hourly weather data for the period 2008–2022 and are provided in EPW format, which can be used in EnergyPlus and other simulation tools that accept this format. The dataset includes files representing different ways of describing local climate conditions for simulation, including conventional typical meteorological years (TMY), performance-based typical years (BTMY), and extreme reference years (ERY). Altogether, the repository provides more than 3000 weather files for applications related to thermal comfort, energy needs, and building performance assessment in Brazil. These files were developed within a national benchmark study on weather-file representativeness for Brazilian building simulation. In addition to traditional TMY methods, the dataset includes performance-based files obtained from a workflow that links weather data, building simulation outputs, and machine learning to better reflect the climatic variables that most affect building performance. The repository is intended to support both research and practice, helping users select and apply weather data that are more consistent with Brazilian climatic diversity and with the objectives of building simulation studies. Corrections or suggestions are welcome at [e-mail] ----- If you use the weather files (EPW) in analysis, simulation, or derived datasets, please cite the Zenodo dataset concept DOI below: Zenodo dataset: 10.5281/zenodo.10546589 If you use or discuss the dataset-generation workflow, methodology, validation, or benchmark study results, please also cite the associated journal paper: Building Simulation: 10.1007/s12273-026-1416-1