Automated data-processing technique: 2D Map for identifying the distribution of the U-value in building elements by quantitative internal thermography

Computing a 2D colour map of average U-values pixel-by-pixel could become a challenging task in terms of complexity and time, especially for entire façades under the influence of anomalies. In a quantitative IRT test, a thermal image with a resolution of 320 × 240 pixels involves 76,800 elements wit...

ver descrição completa

Detalhes bibliográficos
Autores: Tejedor Herrán, Blanca|||0000-0002-2064-0617, Barreira, Eva, Almeida, Ricardo M. S. F., Casals Casanova, Miquel|||0000-0001-5379-894X
Tipo de documento: artigo
Data de publicação:2020
País:España
Recursos:Universitat Politècnica de Catalunya (UPC)
Repositório:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglês
OAI Identifier:oai:upcommons.upc.edu:2117/334178
Acesso em linha:https://hdl.handle.net/2117/334178
https://dx.doi.org/10.1016/j.autcon.2020.103478
Access Level:Acceso aberto
Palavra-chave:Facades
Heat -- Transmission
Thermography
Buildings -- Energy conservation
Quantitative infrared thermography
Automated data-processing technique
Measured U-value
Processed image
2D map
2D correlation coefficient
Quality image
MATLAB
SURFER
Façanes
Calor -- Transmissió
Termografia
Edificis -- Estalvi d'energia
Àrees temàtiques de la UPC::Edificació
Descrição
Resumo:Computing a 2D colour map of average U-values pixel-by-pixel could become a challenging task in terms of complexity and time, especially for entire façades under the influence of anomalies. In a quantitative IRT test, a thermal image with a resolution of 320 × 240 pixels involves 76,800 elements with different TWALL for each instant “t”. This research aims to create a thermographic 2D U-value map for the characterization of heavy walls in a stationary regime. The method was divided into three steps: (i) metrology; (ii) assessment of how mesh discretization affects the image quality by MATLAB; (iii) development of a 2D map by SURFER. The results demonstrated that all 2D maps were a great reproduction of the original image, considering as optimum a TWALL mesh comprised of 1600 elements of 8 × 6 pixels. The automated data-processing method only took 20 min and image quality losses were estimated at 6.65%