Date of Award

12-2013

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Legacy Department

Civil Engineering

Advisor

Dr. Ronald D. Andrus

Committee Member

Dr. Nadarajah Ravichandran

Committee Member

Dr. C. Hsein Juang

Committee Member

Dr. WeiChiang Pang

Abstract

A new seismic site coefficient model is developed from the results of over 60,000 total stress, one-dimensional equivalent ground response simulations assuming conditions in South Carolina. Computed site coefficients ( F ) are plotted versus average shear wave velocity in the top 30 m (VS30 ) and grouped by location, spectral acceleration (Soutcrop ) and spectral period. Locations considered in the Coastal Plain include Aiken, Charleston, Columbia, Florence, Lake Marion, Myrtle Beach, and the South Carolina side of Savannah. Locations considered in the Piedmont include Columbia, Greenville, Greenwood, and Rock Hill. In all the plots of VS30 versus F , the following three distinct trends can be seen--(1) an increasing trend in F as VS30 increases from a low value; (2) a zone of peak values of F , depending on (S outcrop ); and (3) a decreasing trend in F as VS30 increases beyond the zone of peak F values. Development of the mathematical site coefficient model begins by estimating the peak coefficient (FP ) and the corresponding average shear wave velocity (VS30P ) for each VS30 -F plot. Next, the values of FP and VS30P are studied to determine the most significant influencing variables. Variables found to be most influential are Soutcrop , mean predominant period of the outcrop ground motion (Tm ), average shear wave velocity in the top 100 m (VS100 ), and depth to top of soft rock (HB-C ) or hard rock (HHR ). Then, regression analysis is applied to the values of FP and VS30P . Finally, assuming the best-fit values of FP and VS30P , median relationships for the plotted site coefficients are expressed by a linear relationship for lower values of V S30 and a linear or exponential relationship for higher values of VS30 . The amount of variability within the plotted site coefficients is characterized by 95% upper bound and 5% lower bound relationships. The 95% upper bounds are, on average, 42% higher than the median relationships; and the 5% low bounds are, on average, 36% lower than the median relationships. Computed site coefficients for the Coastal Plain are found to be greater in Myrtle Beach, followed by Savannah, Charleston, Florence, Columbia, Lake Marion and Aiken. More closely matching values of Tm and T100 may explain the higher site coefficients in Myrtle Beach and Savannah. Computed site coefficients for periods of 0.0, 0.2 and 1.0 s (designated as FPGA , Fa , and Fv , respectively) are compared with the 1994 National Hazard Reduction Program (NEHRP) Fa and Fv values, which are commonly assumed in current seismic design codes. Significant differences are found between the computed site coefficients and the NERHP values, particularly for Site Class D and E, and where the top of rock is at shallow depths. The computed FPGA , Fa and Fv median relationships are recommended for South Carolina because they are: (1) based on regional conditions; (2) continuous with VS30 , (3) considers depth to rock, and (4) consider the frequency (or period) content of the outcrop motion. If it is desired to design with more conservatism than the median relationships provide, the median coefficients can be increased by 40% to obtain values corresponding to the 95% upper bound. Because the proposed seismic site coefficient model is based on a very broad range of soil/rock conditions, much of it can be directly applied to other areas of the world. Specific variables needed to apply the model are: VS30 , VS100 , HB-C or HHR , Soutcrop , and Tm . It is important to remember that the soft- or hard-rock site coefficients selected should correspond to the Soutcrop values available for the area. A relationship to estimate Tm based on HHR and site-to-source distance is suggested for areas influenced by the Charleston Seismic Hazard Zone. This Tm relationship may not be applicable for other areas. Finally, the simplified procedure for constructing acceleration design response spectrum (ADRS), called the three-point ADRS method, is shown to be adequate when VS30 > 200 m/s. When VS30 ≤ 200 m/s, significant spectral peaks may occur at periods greater than 1.0 s. The objective of the multi-point ADRS is not to replace the building code philosophy, but to present an option for the designer to make sure that longer period accelerations are not under-predicted by the three-point ADRS.

Share

COinS