Liquefaction Modeling

Liquefaction is defined as the loss of shear strength of soil under monotonic or cyclic loading, arising from a tendency for loose soil to compact under shear loading. The term “liquefaction” was originally coined by Mogami and Kubo (1953). Note that this definition covers both static and dynamic liquefaction; the effective stress does not necessarily have to be zero for a soil to liquefy. In particular, when a saturated cohesionless soil is submitted to rapid static or cyclic loading, the tendency for the soil to densify causes the effective stress to decrease, and the process leads to soil liquefaction.

It is known that pore pressures may build up considerably in some sands during cyclic shear loading. Eventually, this process may lead to liquefaction when the effective stress decreases. Although excess pore pressure is generally associated with liquefaction, it is not the direct cause of liquefaction. In constant volume tests with no applied load, it is the decrease in contact forces between particles that is responsible for the decrease in effective stress. The process is documented by Dinesh et al. (2004), who modeled similar tests (in which no change in pore pressure occurs) using the distinct element method. Alternatively, in undrained, simple shear tests under normal pressure, it is the irrecoverable reduction of porosity during cyclic compaction that generates pore pressure and, consequently, a decrease in effective stress.

Dilation plays an important role in the liquefaction process. As soil densifies under repeated shear cycles, grain rearrangement may be inhibited. Soil grains may then be forced to move up against adjacent soil particles, causing dilation to occur, the effective stress to increase and the pore pressure to decrease. Thus, densification is a self-limiting process.

There are many different models that attempt to account for pore pressure build-up, but they often do it in an ill-defined manner because they refer to specific laboratory tests. In a computer simulation, there will be arbitrary stress and strain paths. Consequently, an adequate model must be robust and general, with a formulation that is not couched in terms that apply only to specific tests.

The following section describes two built-in models. The first one is the Finn model with with Martin/Byrne Formulations. It is simple but that accounts for the basic physical process. The second one is a practical three-dimensional model called P2PSand model that is an effective-stress sand plastic model based the bounding-surface theory. A brief review of more compressive models is followed.

Some User-Defined Models (UDMs) can be found from Itasca’s UDM website.

Several examples are presented to demonstrate simulating liquefaction and/or cyclic mobility.