Site response analysis and liquefaction evaluation

When an earthquake occurs, the seismic waves propagate from the source till the ground surface, causing ground shaking. The effects of an earthquake can be different, such as structural damages, landslides and soil liquefaction. In order to identify and mitigate the seismic hazards, appropriate earthquake engineering studies which involve different technical fields, such as geology, geotechnical and structural engineering, seismology are required.

One of the aspects that needs to be taken into consideration is the modification of the earthquake characteristics when seismic waves travel through the soil deposit, that acts as a filter.

Liquefaction

The term liquefaction is used to describe a variety of phenomena that occurs in saturated cohesionless soils under undrained conditions. Under static and cyclic loading, dry cohesionless soils tend to densify. If these soils are saturated and the applied load acts in a short time, as in the case of an earthquake, the tendency to densify causes an increase in excess pore pressures that cannot be rapidly dissipated and consequently a decrease in the effective stresses occurs. When this happens, the soil behaves as a fluid.

To establish if liquefaction will occur in a specific site subjected to a selected earthquake semi-empirical procedures or dynamic methods can be used. The semi-empirical procedures consist in the evaluation of a safety factor as the ratio of the cyclic shear stress required to cause liquefaction and the equivalent cyclic shear stress induced by the earthquake. The dynamic method is based on a one-dimensional wave propagation analysis in terms of effective stresses, which gives the possibility to calculate the pore pressure ratio at any depth.

In the attached document an example is given for a dynamics analysis performed with the PLAXIS 2D finite element code, aimed at modelling the onset of liquefaction in loose cohesionless soils.
Two different approaches, commonly used in engineering practice, are compared. First, the simplified procedure introduced by Seed & Idriss (1971) and updated by Idriss & Boulanger (2014) is carried out. The onset of liquefaction is determined by a curve which separates a liquefiable state from a non liquefiable state. This curve is built on the basis of a large number of case-histories. This approach is based on a series of coefficients that allow to "scale" the seismic event and the in situ conditions to a standard situation.

The second approach consists of a fully dynamic analysis by means of the finite element code PLAXIS 2D. In this case, it is important to select the appropriate dynamic boundary conditions and constitutive models to reproduce the behaviour of saturated soils under cyclic loads. The results of the PLAXIS calculation are in good agreement with the results of the simplified procedure, since the onset of liquefaction is successfully modelled in all the five sand layers.