Ganasan, Reventheran (2021) Damage intensity class for reinforced concrete beam-column joint subjected to lateral cyclic loading. Doctoral thesis, University Of Malaya.
![[img]](http://eprints.uthm.edu.my/style/images/fileicons/text.png) |
Text
24p REVENTHERAN GANASAN.pdf Download (10MB) |
Abstract
The 2015 moderate earthquake that occurred in Ranau, Sabah, Malaysia had damaged more than 60 numbers of non-seismic designed reinforced concrete (RC) buildings from minor to severe damages. These damages were caused by lack of shear links confinement and insufficient anchorage length, particularly on the RC beam-column joint (BCJ) of the buildings. Apart from high-rise buildings, almost all of the existing buildings in Malaysia are non-seismically designed, thus the seismic performance of the buildings in resisting future earthquakes are of concern by the government/public, building owners and local researchers after the destructive earthquake in Ranau. Therefore, an appropriate assessment method is needed to predict the damage level of the existing and new RC structures, thus it may increase the preparedness level for future earthquakes. The damage intensity (DI) class is one of the useful structural assessment methods in evaluating the damage level of the structure based on its structural design under certain loadings. However, the currently available DI class from the literature are limited to vertical static loading only. Thus, there is a need to develop a DI class for lateral cyclic loading as simulating the earthquake loading. Hence, this thesis aims to develop an ANN chart model of DI class for RC beam-column joints, with and without seismic consideration based on lateral cyclic loading and to compare this with the existing DI chart for static load from the Acoustic Emission (AE) method. In phase 1, a total of 7 specimens were designed, prepared and tested to investigate the effects of shear links and anchorage length towards the structural performance and cracking behaviour of RC BCJ under lateral cyclic loading. The AE monitoring was carried out simultaneously during the lateral cyclic loading test to assess the AE signal such as AE hits, AE amplitude, rise time, counts and duration during the crack formation in every drift ratio (DR) level. In phase 2, iv the specimens design parameters, lateral cyclic loading results and the subsequent AE signals were used as the input parameters for the development of the ANN crack width prediction models and DI class. The observation shows that the additional shear link provision has shown a significant effect on the strength up to 55% compared to the application of anchorage length, hence became the main factor in reducing crack width. In the AE analysis, the higher numbers of shear link provision had indicated that the development of the crack within the specimen is slower compared to other specimens and it does not fail at the maximum DR level. From the ANN results, the Support-Vector Machine (SVM) – dot with root mean square error (RMSE) index performance crack width prediction model has shown the lowest error between the experimental and predicted data up to 5% thus being selected among other prediction models. The DI class has been developed in associated to damage index and response index for every specimen. Naïve Bayes (NB) Kernel classifier model was selected as it shows the highest accuracy and precision up to 98% which is compatible with the experimental results compared to deep learning (DL) and SVM classification model. Keywords: RC beam-column joint; Acoustic emission; Lateral cyclic loading; Shear link aspect; Anchorage length aspect; ANN prediction models
| Item Type: | Thesis (Doctoral) |
|---|---|
| Subjects: | T Technology > T Technology (General) |
| Depositing User: | Mrs. Sabarina Che Mat |
| Date Deposited: | 13 May 2024 06:52 |
| Last Modified: | 13 May 2024 06:52 |
| URI: | http://eprints.uthm.edu.my/id/eprint/10781 |
Actions (login required)
 |
View Item |