Performance Evaluation of Fire Damaged Reinforced Concrete Beams using Machine Learning
Abstract
This study aims to develop techniques to evaluate performance of structural damaged reinforced concrete (RC) beams using machine learning technique. Toward this goal, fire tests are conducted on RC beams having different fire... [ view full abstract ]
This study aims to develop techniques to evaluate performance of structural damaged reinforced concrete (RC) beams using machine learning technique. Toward this goal, fire tests are conducted on RC beams having different fire exposure time periods. For fire tests, RC beam specimens having compressive strength of 24 MPa are exposed to high temperatures according to ISO 834 standard time temperature curve and temperature distributions are measured during the fire tests by six thermocouples placed within the beam section. After fire tests, crack information on beam surfaces is obtained from the captured images taken by digital cameras and used as basic data for machine learning technique. The vertical displacements are measured using linear variable differential transformers (LVDT) during the fire test. For investigating structural performance, it is determined to use the cracks obtained after the fire tests, and the temperature distributions obtained during the fire tests. The length of surface cracks using AutoCAD is then included in the performance evaluation framework to estimate the heated temperatures of reinforcing bars. The machine learning technique is used for establishing a statistical model to estimate the degrees of actual fire damages from the images of beam surfaces. The images include significant cracks caused by fire damages as well as noises, and, therefore the machine learning technique is developed for efficiently discriminating the critical cracks in the region of interests. It is shown that the proposed technique can estimate the damage of fire damaged RC beams quickly and quantitatively. Further studies are also needed to increase the applicability of this method for wide ranges of cases such as the different materials, fire conditions, and member sizes.
Acknowledgments
This research was supported by a grant (17CTAP-C114986-02) from Technology Advancement Research Program (TARP) funded by Ministry of Land, Infrastructure and Transport of Korean government
Reference
Shin, Yeong-Soo, et al. "Experimental Studies on the Effect of Various Design Parameters on Thermal Behaviors of High Strength Concrete Columns under High Temperatures." Journal of the Korea Concrete Institute 23.3 (2011): 377-384
Michael O’Byrne et al. “Texture Analysis Based Damage Detection of Ageing Infrastructural Elements”, Computer-Aided Civil and Infrastructure Engineering 28 (2013) 162–177
Authors
-
Heesun Kim
(Ewha womans university)
-
Eunmi Ryu
(Ewha womans university)
-
Yoojin Lee
(Ewha womans university)
-
Jewon Kang
(Ewha womans university)
-
jieun lee
(Ewha womans university)
Topic Areas
Analysis, simulation and sensing , Big Data, data mining and machine learning , Other
Session
O4 » Concrete Structures (15:45 - Tuesday, 5th June, Sonaatti 2)
Paper
134.pdf
Presentation Files
The presenter has not uploaded any presentation files.