Abstract
The collapse of building structures during recent earthquakes, particularly in the countries around the Kurdistan region (KR), including Turkey (2011 Van earthquake) and Iran (2003 Bam earthquake), has raised many questions about the safety of existing buildings in the region and structures that are going to be constructed in the future. The KR, which is located in northern and northeastern Iraq, is also considered to be the most hazardous region of Iraq. However, many buildings in the region, especially unreinforced masonry (URM) buildings, were not engineered to withstand seismic loads. The seismic vulnerability assessment of these types of buildings in this region is a necessary step towards the development of regional seismic retrofitting and pre-disaster mitigation plans. Fragility analysis is often used for this purpose and to graphically represent a structure’s seismic vulnerability in terms of fragility curves.Considering that there are several important uncertainties involved in such an analysis and after developing and proposing seismic zonation maps, response spectra and the seismic zone factor Z for the KR, the results of analytical fragility analyses of URM buildings in the KR indicated that the correct selection of a ground-motion intensity measure (IM) is the most important variability involved in assessing the fragility of URM buildings. The results suggest that the variability in the mechanical parameters of materials can be neglected because the variability of ground motion is considerably more substantial. Furthermore, the use of incremental dynamic analysis (IDA) and a well-selected IM allows fragility curves to be derived with only a few records (i.e., a minimum of 7 records) with the same performance as for with numerous records (i.e., 60 records). Moreover, the pronounced difference in the results when using only one IM vs. using two IMs cannot be ignored; hence, fragility surfaces are preferred over the more commonly used fragility curves.
Furthermore, a framework based on using machine learning models (i.e., a wrapper-based approach) for the optimal selection of an IM for developing fragility curves is proposed in this study. The feasibility of the wrapper method for selecting the best IM is compared to a statistical regression (log-logistic regression) used to develop fragility curves and the results were encouraging.
Finally, the outcomes of the study indicate that the seismic safety of the investigated low-rise buildings in the region is questionable and that these structures must be strengthened to prevent failure.
| Date of Award | Oct 2015 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Nikos Nanos (Supervisor) & David Begg (Supervisor) |