Hello,
This message is for Prof. Jeremic:
Thank you for all the good work you have done with the geotechnical part of OpenSees. I have a question:
are there any plans to include in the near future 2D plane strain and axisymmetric plasticity with the soil models you have developed, including reduced integration elements? From the user viewpoint the first attempt to solve a geotechnical problem is using a 2D model and then attempt a 3D solution, so this would be invaluable.
Thank you in advance for your answer.
Best regards
2D Plane Strain Plasticity
Moderators: silvia, selimgunay, Moderators
Hello Perou,
Thanks for your kind words!
We have done mostly 3D models. There are some 2D and 1D models that we use to verify
our formulations and implementations, and those were done using 3D elements as well. In those cases (2D and 1D) we make appropriate boundary condition changes and still use 3D elements. You can find some of those models in this document:
http://sokocalo.engr.ucdavis.edu/~jerem ... es_2up.pdf
there are examples (pages round 19 9 and around 215, dealing with verification examples for u-p-U formulation, and also for single phase elastic-plasticmodels look arond page 87...).
So yes it would be useful to have some 2D implementations in, there but we do well without them too, by using 3D models with appropriate bondary conditions. The computational effort is slightly biger but the development and implementation effort is much smaller. We also work on mesh genrators that will generate these 2D models without showing too many details of the (actuall) 3D models and present it to OpenSees for computations...
As for reduced order integrations, I am advocating against it, but our new element implementation (not yet submitted to OpenSees repository but soon to be available through my web site at UCD) wll have option for different order and type of integration methods (Gauss, Lobatto...).
So if you think that it is OK to do under-integrate you element stiffness, go for it, but again, one can show that for plasticity you are making a large error, and I would argue that one needs to increase integration order above what is suggested for (perfect) elastic elements. This problem of integration order is one of homework problems in the course I curretly teach (see HW#2 at:
http://sokocalo.engr.ucdavis.edu/~jeremic/CG/
Best regards, Boris
Thanks for your kind words!
We have done mostly 3D models. There are some 2D and 1D models that we use to verify
our formulations and implementations, and those were done using 3D elements as well. In those cases (2D and 1D) we make appropriate boundary condition changes and still use 3D elements. You can find some of those models in this document:
http://sokocalo.engr.ucdavis.edu/~jerem ... es_2up.pdf
there are examples (pages round 19 9 and around 215, dealing with verification examples for u-p-U formulation, and also for single phase elastic-plasticmodels look arond page 87...).
So yes it would be useful to have some 2D implementations in, there but we do well without them too, by using 3D models with appropriate bondary conditions. The computational effort is slightly biger but the development and implementation effort is much smaller. We also work on mesh genrators that will generate these 2D models without showing too many details of the (actuall) 3D models and present it to OpenSees for computations...
As for reduced order integrations, I am advocating against it, but our new element implementation (not yet submitted to OpenSees repository but soon to be available through my web site at UCD) wll have option for different order and type of integration methods (Gauss, Lobatto...).
So if you think that it is OK to do under-integrate you element stiffness, go for it, but again, one can show that for plasticity you are making a large error, and I would argue that one needs to increase integration order above what is suggested for (perfect) elastic elements. This problem of integration order is one of homework problems in the course I curretly teach (see HW#2 at:
http://sokocalo.engr.ucdavis.edu/~jeremic/CG/
Best regards, Boris