Thesis Type: Postgraduate
Institution Of The Thesis: Middle East Technical University, Faculty of Engineering, Department of Civil Engineering, Turkey
Approval Date: 2019
Student: Eray Altıncı
Supervisor: KEMAL ÖNDER ÇETİNAbstract:
Assessment of cyclic (seismic) response of soils is considered to be one of the most interesting research areas of geotechnical engineering. Cyclic loading-induced stresses cause accumulation of shear strains and generation of excess pore water pressure, which are accompanied by reduction in stiffness. In the most extreme cases, this process leads to soil liquefaction. As opposed to the random nature of earthquake waves, it is generally preferred to apply uniform loading cycles to undisturbed or reconstituted soil samples in the laboratory. The method proposed by Seed et al. (1975) can be used to convert the acceleration time history to number of equivalent uniform stress cycles. This approach uses a weighting factor, which varies as a function of selected failure criterion (Cetin and Bilge, 2012), and allows direct calculation. The aim of this study is to develop a new semi-empirical framework to predict number of equivalent stress cycles. For this purpose, more than 4000 acceleration time histories from PEER database with different magnitudes, site-source distance and site types have been assessed. These records were filtered and re-sampled to a common time step, and then the number of uniform cycles was calculated for 17 different weighting factors ranging from 0.2 to 1.0 to form a database. This database was assessed by regression methods to develop a probability-based model for prediction the number of equivalent uniform stress cycles for a given earthquake magnitude, site-source distance, soil properties and selected cyclic target strain performance criterion.