Physical and computational modeling of biaxial base excitation of sand deposits

Natural soil is often loosely deposited, with the granular particles locked together by friction. During seismic events, shear waves overcome this frictional confinement. The force exerted by the overburden pressure causes the particles to reorient into a higher density configuration. As the solid particles of the soil densify, the water between these particles is forced out, and has to travel to the surface. This pore water motion is resisted by the permeability limitations of the soil resulting in increased pore water pressure. Pore water pressure reduces the strength of the soil by limiting the contact forces between the soil grains. Once the hydrostatic force is sufficient to carry the applied load, the inter-granular frictional forces will approach zero, and the soil will behave like liquid. This liquefaction behavior causes structures to sink or topple in the soil, creating an incredible risk to human life, and loss of property. The effects of earthquake loading on site response are complex and three dimensional. The ability to understand and predict soil response during earthquakes is an important aspect in the design and management of soil-structure systems. Numerous studies were conducted over the last four decades to analyze and quantify the response of soil deposits subjected to base excitations. Most of the previous studies, especially those dealing with physical modeling were done using uniaxial base excitation. The simulation of base excitation progressed from single frequency sine excitation, to more complex motions with variable amplitude and finally multi-frequency content. The vast majority of research is conducted using uniaxial excitation, and only few used simple biaxial shaking.;The work presented herein is a centrifuge study conducted at the Center for Earthquake Engineering Simulation (CEES) of Rensselaer Polytechnic Institute (RPI) to assess the dynamic response characteristics of level deposits and SSI under multidirectional shaking. Synthetic sinusoidal waves were used as base excitations to test loose and dense models under biaxial and uniaxial shaking. Dense arrays of accelerometers were used to monitor the deposit response along with pore water pressure transducers. Three primary and calibration tests were conducted in order to assess the behavior of the RPI 2D shaker and 2D laminar box. Followed by three uniaxial tests and three bi-axial shaking tests conducted on soil models to study the impact of multidirectional shaking on the soil liquefaction. The three uniaxial shaking tests consisted of: i) Two tests with input energy similar to that of the bi-axial input shaking energy and ii) Test with 10% increase in one of the components of the biaxial shake amplitude, as commonly done in practice for uniaxial simulation of multidirectional field shaking. For biaxial tests two studied free field and one studied SSI. The observed acceleration and pore pressure are used along with non-parametric identification procedures to estimate the corresponding dynamic shear stress-strain histories. The measured results along with the obtained stress and strain histories are used to shed the light on the mechanisms of liquefaction occurring through the stratum, excess pore pressure buildup, soil contraction and the difference in soil behavior when it is subjected to biaxial shaking. This difference is evident in the strain energy generated in the biaxial test compared to that of the equivalent and traditional uniaxial tests, and the non-proportional response of the soil under biaxial shaking.