I am a researcher working on modelling soil liquefaction due to induced seismic excitation. The FEM code I am using is Plaxis2D.
Plaxis2D can model the onset of liquefaction using advanced constitutive models such as UBC-SAND. I am however interested in doing this for more basic material models, specifically Linear Elastic and Mohr Coulomb. Of course, neither of these models can simulate post liquefaction behavior as the soil becomes a fluid. I am interested to find out if they can model the onset of liquefaction.
In addition to this I would like to figure out how to detect this onset. Research has shown to me that liquefaction due to cyclic loading, or "cyclic mobility" occurs when there is a buildup of pore water pressure over time from this cyclic loading. After a while the buildup may overcome the effective stress (since total stress = effective stress + pore pressure) and the model will perhaps fail to converge as this will effectively be the liquefaction point.
Does anyone know what I'm talking about? Is it possible to model liquefaction onset using Mohr Coulomb or Linear Elastic constitutive models?
Hi Amine,
To capture the liquefaction behavior, the simulation of excess pore water pressure is a prerequisite for your work. Hence, the contraction of pore space must be captured by your model. If you just use M-C model (a shear failure yield surface), you can just get a shear-induced expansion when stress approaches and remains on the yield surface. In addition, it is usually assumed that the solid phase is incompressible. As a result, the plastic volume change calculated from M-C model will increase the porosity, and thus decrease the PWP. So if you use M-C model, the resultant plastic strain can never contribute to the liquefaction.
If you would like to model the excess pore water pressure, you should incorporate a shear-induced contraction yield surface into your model (such as cap yield surface, Cam Clay model (or Modified Cam Clay)).
For the cyclic loading, it is another topic. You should try to incorporate the viscoelasticity or elasto-viscoplasticity into your model to model the corresponding dynamic behaviors of soil (such as rate dependence, creep and relaxation) (e.g., Perzyna model).
Good luck,
Liang
I second Liang Cui's post. Mohr-Coulomb is a complete dead-end for modeling liquefaction. They developed those complex constitutive models for a reason.
Hlw Amine,
Mohr coulumb can help you to get an idea about the liquid motion and possibilities of failure. You can use UU (Uniconsolidated undrained), CU (consolidated undrained) and CD (consolidated drained) methods.
Warm wishes,
Sharmin
As others explained, you need a more advanced model in order to capture volumetric change after failure. For example, the Hypo-plastic model for sand can also help.
- Badges
- Science topic
- Similar topics
- Engineering
- Modeling