Analytical studies of a large-scale laminar soil-box for experiments in soil-structure-interaction
Loading...
Authors
Bitsani, Anastasia
Issue Date
2017
Type
Thesis
Language
Keywords
biaxial shaking , earthquakes , nonlinear soil , nuclear facilities , soil-box , soil-structure-interaction
Alternative Title
Abstract
Nuclear facilities frequently have deep massive foundations, which are large enough to affect the response of neighboring soil and the nature of ground shaking these facilities have to withstand. Despite this well-recognized phenomenon, the ramifications of soil-structure interaction (SSI) are not completely understood due to the complexity of the mechanics involved. As a consequence, only simplified elastic models are currently used to study SSI for these and other facilities. To address this situation, the U.S. Department of Energy (DOE) has funded a multi-institutional project to investigate SSI effects in nuclear facilities. To this end, the research team at University of Nevada Reno (UNR) is fabricating a 400-ton, laminar, biaxial soil box and corresponding shake table, which will be used to (a) explore SSI phenomena at a scale not currently possible in the U.S., and (b) validate the ESSI nonlinear computational framework, developed by UC Davis. This thesis presents some of the numerical analyses that have been conducted in order to inform the design of the soil-box and shake-table, and to understand the (a) dynamic behavior of the soil-box, (b) the role of soil nonlinearity, (c) the fundamental interaction of the soil with the walls of the box, and (d) the effect of friction and gapping at the soil-wall interface. The preliminary design phase included the modelling of a 1D soil-column in DEEPSOIL and compared results from linear, equivalent linear and nonlinear analyses, for a suite of eight recorded ground motions obtained from the PEER database, and different scale factors, with scaled PGAs between 0.25g and 1.0g. Simulation of the nonlinear hysteretic soil-behavior was achieved via the use of the Pressure-Dependent Modified Kodner Zelesko model and the new General Quadratic Hyperbolic model. The effect of several parameters, such as the hysteretic soil material, the reference curve, the time-step and the time-scaling of the input motion, on the results of the nonlinear dynamic analyses was also evaluated. Furthermore, finite element modeling and nonlinear dynamic analyses of a 1D soil column and a more realistic 2D slice of the soil including the box walls were conducted in LS-DYNA using a nested surface plasticity model. Different mesh sizes, wall configurations, and contact conditions at the soil-wall interface, ranging from frictionless contact to perfect contact, were examined in order to decipher the role of sliding, friction and gapping on the behavior of the box. Wall configurations with and without vertical constraints, with linear axial springs, and with compression-only springs were investigated. The boundary effect close to the walls was also examined and the area of uniform soil stresses was identified for different design alternatives. The nonlinear dynamic analyses were used to quantify the base shear, overturning moment, pressures below the box, response spectra at different locations of the soil, forces in the walls, and the accelerations, displacements, strains and stresses of the soil and the box. The advanced numerical analyses presented in this thesis give an insight into the seismic behavior of the soil-box and are expected to be useful to other research teams designing their own soil-box. The numerical work demonstrated that:• Equivalent linear site response analyses give similar results with nonlinear analyses for small to moderate levels of shaking (PGA=0.5g), but they over-predict the base shear forces and under-predict the shear strains for higher levels of shaking.• The soil nonlinearity limits the increase of the base shear, offsets the fundamental period of the soil (from 0.13sec to about 0.5-0.6sec for input motions with PGA=1.04g), increases significantly the soil-strains (1-7% for aforementioned motions), and results in de-amplification of the input motion towards the surface.• It is important to use soil materials models (GQ/H) that can properly simulate the soil behavior at large-strains by reaching the correct shear strength especially at high levels of shaking, because such models can give a significantly different response of the soil column and reduce the base shear by 15% and increase the maximum shear strains by a factor of 2.• Laminar walls that are flexible in every direction (lateral and vertical) are witnessing vertical soil displacements in regions close to the walls, indicating that the soil is not in pure shear. For this case the stresses are not uniform along the whole length of a layer, with soil regions closer to the walls witnessing different stresses than the ones close to the center of the box, demonstrating the existence of a significant boundary effect caused by the walls.• Large overturning moment is generated at the bottom of the soil-box during strong lateral shaking, and this moment can introduce significant uplift in the walls, meaning that they should be designed not only for shear but also for tension.• To ensure that the soil-box will behave as realistically as possible, it is necessary to have walls with small lateral shear stiffness but very high axial and bending stiffness, together with a high-coefficient of friction at the soil-wall interface, which will transfer the complementary shear of the soils to the walls and ensure a minimal/negligible boundary effect.
Description
Citation
Publisher
License
In Copyright(All Rights Reserved)