Probabilistic Site Response Analysis and Soil-Structure Interaction: A Data Science and Statistical Investigation Utilizing a Comprehensive Dataset in Japan and the US
Loading...
Authors
Boushehri, Reza
Issue Date
2023
Type
Dissertation
Language
Keywords
Alternative Title
Abstract
Site response analysis (SRA) plays a crucial role in the performance-based design ofstructures subjected to earthquakes. It allows for the prediction of ground motion and the
subsequent effects on structures. While the accuracy of current SRA methodologies has
been the focus of much research, there are several sources of uncertainty in soil media that
should be accounted for in design practice. These include variations in soil properties,
which can affect the response of the ground to earthquakes. In addition to ground motion,
the effects of soil-structure interaction (SSI) must also be taken into consideration, as they
can significantly influence the response of a structure to an earthquake. Therefore, the
incorporation of these factors is essential in the design process to ensure the safe and
efficient performance of structures during earthquakes. There has been recent progress in
developing methods to understand how uncertainties in soil parameters can affect the
results of nonlinear (NL) SRA. However, it is not yet clear how reliable these methods are,
and further testing using real-life data is needed to determine their strengths and
weaknesses. Additionally, soil-structure interaction (SSI) analysis using the "direct"
method, which involves analyzing the entire soil-structure system at the same time under
strong shaking scenarios, is becoming more common in engineering practice. However,
the accuracy of this method is uncertain due to the complex nonlinear behavior of soil,
structures, and foundations. Similarly, further testing using real-life data is needed to
understand the strengths and weaknesses of this approach. Proposing new methodologies,
and validating those against actual recordings for a large inventory of sites this report provides some recommendations for design purposes. The work done is presented in three
parts.
In the first part, a new methodology is proposed for quantifying the effects of epistemic
uncertainty in soil parameters on NL SRA results. The methodology is validated using data
from a well-instrumented geotechnical downhole array located in Japan, and the results of
46,200 NL finite element (FE) analyses are presented. The findings demonstrate that
increasing the number of soft realizations and implementing higher levels of earthquake
intensity leads to higher dispersion in ground motion. In contrast to previous studies, the
proposed methodology has no significant effect on the predicted median surface response
spectra and amplification factors for this case study. The findings of this part were
published in the Earthquake Spectra in 2022.
The second part aims to benchmark a three-dimensional (3D) Nonlinear (NL) SoilStructure Interaction (SSI) analysis methodology in the time-domain code LS-DYNA for
a mid-rise well-instrumented building in Japan. The findings of this study indicate that the
response of a structure is significantly influenced by the accuracy of the predicted free-field ground surface response, particularly at the structural fundamental period rather than
the site period. Additionally, the results show that the responses of fixed-base and flexible-base structural models are significantly different, with the main contributing factor being
SSI effects that cannot be accounted for in the fixed-base model. A discussion on the
minimum amount of data needed to accurately capture nonlinear structural response using
the proposed SSI methodology is also included in the study. The findings of this part were
submitted to Soil Dynamics and Earthquake Engineering. The third part utilizes data from 14 well-instrumented downhole arrays in the United States
and Japan to evaluate the reliability of the proposed methodology presented in part 1 to
examine the individual and combined effects of uncertainties in soil parameters, such as
the shear-wave velocity profile and shear strength, on NL SRA results for sites with
different characteristics to provide some recommendations for design purposes. Results
from 182,400 NL FE analyses are presented. The results of this study demonstrate that the
applied methodology effectively quantifies the effects of soil parameter uncertainty on the
response, without resulting in any undesired reductions in the predicted median surface
response spectra, unlike some previous studies. Additionally, the results indicate that the
impact of soil parameter uncertainty on the response strongly depends on the site
conditions and the intensity of the seismic motion. The manuscript for this portion of the
study is currently in preparation and will be submitted to a peer-reviewed journal in the
near future.