I have tried validating PLAXIS 2D for the very simple case of a block connected through a linear-elastic perfectly plastic Mohr Coulomb interface to an oscillating base. The model consists in a large soil cluster of linear elastic material (the oscillating base, or shaking table), a small soil cluster of linear elastic material (the block) and an interface in between which is assigned a Mohr-Coulomb material model with friction, zero cohesion, zero dilatancy, and linear elastic Ks and Kn shear and normal stiffness (to make it easier to check elastic movement). The maximum shear force that can be mobilised between block and base (the interface resistance) is easily computed as the self-weight of the block times tangent of the friction angle. If the amplitude of the acceleration wave applied to the oscillating base is greater than the acceleration inducing an inertial force in the block greater than the interface resistance, relative movement should occur between block and base. As the inertial force is greater than the resistance, the difference in force divided by the mass of the block should be the acceleration the block is subject to. Double integration of the acceleration over time should give the displacement of the block.
This does not happen in PLAXIS. Once the inertial force becomes greater than resistance, the software computes movement on the basis of the plastic multiplier rather than dynamics. If the given base acceleration is constant (resulting in a constant inertial force on the block), the relative movement at the interface computed by PLAXIS is that of a constant velocity, probably linked to a constant rate of plastic shear strain; and the computed velocity depends on the selected stiffness parameters for the interface material model. This goes against the laws of physics whereby the block should be subject to constant acceleration, independently by the elastic properties at the interface. PLAXIS appears to ignore dynamics completely and compute relative displacements on the basis of the plastic strain increment times the time step given.
PLAXIS therefore does not appear to be a viable finite element software to model dynamics problems in presence of frictional interfaces.
Has anyone ever tried a similar problem and has a solution for the issue raised.
Dear Francesco,
I have noticed that you have submitted a Service Request and a colleague of mine has already picked up your query.
The current status is under investigation so soon you will receive more information from my colleague.