search

underworldcode / underworld2

underworld2: A parallel, particle-in-cell, finite element code for Geodynamics.
http://www.underworldcode.org/
Other
170 stars 58 forks source link

simple example code for elasticity problem #188

Closed cenguo closed 7 years ago

cenguo commented 7 years ago

Hi Underworldians ,

I wish to use underworld2 for elastoplastic solid mechanics problems. The current examples for me to start with are all solving Stokes equations. Can anyone provide me a simple code solving quasi-static equilibrium equation with elastic material?

My another concern is the comparison of FEMLIP with MPM. It seems in order to obtain good integration result, ~50 particles were assigned to each element in the examples. However, only 4 material points are used for each grid in MPM (similar to FEM). Does that mean the efficiency of FEMLIP is much lower than MPM?

Many thanks, Ning

jmansour commented 7 years ago

Hi Ning. We will have an example with viscoelasticity in the next release (hopefully), which should be just around the corner.

In Underworld you have two options for integration: the voronoi lagrangian integration point scheme, or standard FEM gauss integration. For the latter, when you are using particles, currently it simply utilises a near neighbour approach to determine values at the gauss points.

Also, we currently default to using 9 (or 27 in 3d) particles per element.. this is mainly so that where we have lagrangian particles, we sufficiently sample their types. It is also required for higher order elements.

In terms of efficiency, the cost of integration is usually very small relative to the other costs (in particular the solve time), so it should usually not be a concern.

I'll leave this open as a reminder for us to add the elastoplastic example.

RebeccaFarrington commented 7 years ago

Little correction - there is currently a viscoelastic example committed, /docs/examples/2_11_ViscoelasticityInSimpleShear.ipynb Cheers, Rebecca

jmansour commented 7 years ago

Thanks Bec! I don't think we can give a useful timeline for a model also including plasticity, so I will close this ticket for now. Please check in with us again at a later date!