Multi-scale analysis of weather data for building performance assessment in Brazil
| Ano de defesa: | 2024 |
|---|---|
| Autor(a) principal: | |
| Orientador(a): | |
| Banca de defesa: | |
| Tipo de documento: | Tese |
| Tipo de acesso: | Acesso aberto |
| Idioma: | eng |
| Instituição de defesa: |
Universidade Federal de Viçosa
|
| Programa de Pós-Graduação: |
Não Informado pela instituição
|
| Departamento: |
Não Informado pela instituição
|
| País: |
Não Informado pela instituição
|
| Palavras-chave em Português: | |
| Link de acesso: | https://locus.ufv.br/handle/123456789/33575 https://doi.org/10.47328/ufvbbt.2025.024 |
Resumo: | The choice of weather data is fundamental for assessing building performance in an accurate and representative way. Typical weather files usually apply a statistical approach to select months representative of current climatic conditions, but they do not encompass site-specific characteristics for different locations worldwide. This work analyses the potential of a multi-scale analysis for building performance assessment from different resolutions of weather data, in a comprehensive geographical territory, given the size of Brazil. It presents an overview of weather data and its application on building performance analysis and a general procedure to retrieve and process weather data, and different approaches to compile weather files for building performance assessment. The study also provides an extensive analysis of the Brazilian territory, presenting a climatic profile and trends for the entire territory. The analysis focuses on a climatic and bioclimatic summary, and on building performance simulations for representative cities according to the Brazilian bioclimatic zoning. Then, it compares the records from ERA5-Land and INMET to quantify the differences and present the impact on building performance analysis. The study proposes a new weather file compilation method for Brazil and applies statistical tests to determine whether the new approach delivers better results than the existing TMY methods. The procedure encompasses correlation and sensitivity analysis based on machine learning models to propose a performance-based method. The initial analysis of the Brazilian territory showed predominantly a temperature increase based on the 2008-2022 records, with some locations reaching more than 1 °C. However, the bioclimatic approach based on Givoni’s chart showed that ventilation strategies are still the most effective approach instead of HVAC systems. Following the comparison between high resolution spatial data and weather stations from Brazil, some locations present insufficient years for a multi-year analysis and some municipalities show a significant variation not only of weather data, but also of the building performance results. Finally, the analysis of new weather files for Brazil allowed concluding that creating typical year weather files based on a performance approach delivers the best outcomes, since they are closer to the historical records. Keywords: climate analysis; building performance simulation; weather files; machine learning |