PROBABILISTIC AND HAZARD ANALYSIS FOR PORE PRESSURE INCREASE IN SOILS DUE TO SEISMIC LOADING

The build-up of excess pore pressure in a layer of cohesionless soil during an earthquake can lead to ground movements which damage structures and cause loss of life. The extreme case of pore pressure increase up to the zero effective stress condition is often called liquefaction and there are hundreds of recent cases of ground failures and damages to structures due to liquefaction.

This study represents an initial part of a project undertaken at Stanford to assess the damage potential associated with pore pressure increases in saturated cohesionless soils under seismic loading. The specific objectives of this project are: (1) Developing procedures to evaluate the likelihood of pore pressure build-up during an earthquake; (2) Assessing the probability of ground deformation and damage caused by the pore pressure build-up; and, (3) Evaluating the resulting potential for economic loss. This dissertation is oriented towards casting the phenomenon of pore pressure build-up into a probabilistic and hazard analysis framework.

To achieve these objectives, the behavior of cohesionless soils under uniform and non-uniform cyclic loading is documented and investigated. Two probabilistic models are proposed to study the development of pore pressure in sands under random loading. The first of these models is based on laboratory data, while the second uses an analytically based effective stress technique. Both models incorporate uncertainties in soil parameters, laboratory data, and earthquake loading parameters, and compute the cumulative distribution function of pore pressure at the end of any cycle (zero-crossing) of loading. The two models are cast into a methodology which evaluates the seismic hazard at a given site in terms of pore pressure build-up.

The probabilistic models correctly predict the behavior of sites for which the field behavior during the Niigata earthquake is known. The pore pressure probability curves computed with these models reflect the effect of earthquake duration, showing that for the same PGA or root mean square of acceleration substantially different levels of pore pressure are likely to develop. The effect of uncertainty in soil parameters is shown to be more important for moderate intensity loading and short duration events than for high intensity and long duration events.

The probabilistic pore pressure models developed in this dissertation are advantageous over conventional methods for several reasons. First, they are not restricted to the question of liquefaction per se, but provide a complete description of pore pressure, for any pore pressure ratio between 0. and 1.0. Second, they include uncertainty in both soil and loading conditions. And, third, they make a distinction between cases where liquefaction is certain and cases where it is not 100 percent probable.

The hazard methodology demonstrates that the hazard at a given site can be described in terms of a soil related parameter, the level of pore pressure. This methodology is well suited to compare the hazard potential of sites with different soil resistance and seismic environment.