Effect of damping on dynamic soil reaction curves used in 1D spring models to consider soil-structure interaction.

In the last decades, several damages of pile-supported structures due to earthquakes have been reported. Soil-Structure Interaction (SSI) is an important factor to estimate the seismic loading on pile accurately and hence this is a vital question for engineers, how to model the SSI appropriately. The finite element method (FEM) can be used to estimate the seismic loading and predict the behavior of pile-supported structures during seismic excitation accurately. However, FEM is very complex and time inefficient. Researchers have proposed numerous alternative 1D spring models to consider the interaction between pile and soil, which are applicable for the beam on nonlinear Winkler foundation (BNWF) method. Among those spring models, PySimple1 is widely used and easily accessible by an open-source program - OpenSees. However, the shortcomings of PySimple1 is that it adopts the soil reaction curves (p-y curve) which were developed based on the static pushover tests or slow cyclic load tests on pile head. Recently, a dynamic p-y curve has been proposed as a hyperbolic function based on a series of centrifuged tests performed by Yoo et. al. 2013. This dynamic p-y curve is significantly different from the conventional static p-y curves in three aspects: initial stiffness of soil, ultimate bearing capacity of the soil, and overall shape of the curve as mentioned by Choi et. al. 2015. The proposed hyperbolic function is well adopted by soil-structures interaction research community for the development of dynamic nonlinear spring models, such as PySimple3, PySimple5, however the explanation for why the dynamic soil reaction curve is significantly different, is not properly explained. In this paper, PySimple1 and PySimple5 are considered for a detailed investigation of the difference between static and dynamic P-y curves. Results of a series of centrifuged test performed by Wilson in 1998 is adopted for comparison of each model. BNWF model of the experiment is prepared using PySimple1 and PySimple5. For PySimple1, an additional dashpot is defined, and by performing a parametric study, necessary damping values are quantified to obtain comparable results with experimental data. While the definition of additional dashpot is not needed for the model prepared with PySimple5. Equivalent linear models are also prepared and analyzed to point out the effect of the damping force on seismic loading on the pile during seismic excitation. It can be concluded that in seismic loading conditions the soil is in excited condition and hence it exhibits damping force on pile foundation. This damping force causes a significant difference between the static and dynamic soil reaction curves while preparing those soil reaction curves from the back-calculation of measured bending moment during the experiment. [ABSTRACT FROM AUTHOR]