We are collecting some Requirements Data.
Please reply to this post with a list of what you, as an OpenSees USER IN SOIL MODELING, require of OpenSees.
Thank you,
The OpenSees Development Team
We are collecting OPENSEES User-Requirements Data
Moderator: Moderators
We are collecting OPENSEES User-Requirements Data
Silvia Mazzoni, PhD
Structural Consultant
Degenkolb Engineers
235 Montgomery Street, Suite 500
San Francisco, CA. 94104
Structural Consultant
Degenkolb Engineers
235 Montgomery Street, Suite 500
San Francisco, CA. 94104
requirements USER IN SOIL MODELING,
For bricks:
output Gauss point's material stiffness tensor
output data in VTK format
output Gauss point's material stiffness tensor
output data in VTK format
I'd like to see a standard set of tests that each element and/or constitutive model are put through for verification purposes. Each version of OpenSees.exe should run these tests and be compared against analytical solutions where applicable. The results should also be posted on the OpenSees website for users to have access to.
I'd also like to see axisymmetric quad and axisymmetric line elements (with and without pore pressure DOFs).
I'd also like to see axisymmetric quad and axisymmetric line elements (with and without pore pressure DOFs).
I think OpenSees is perfectly capable of doing both static and dynamic analyses and should be use for both.
Analytically, there should be the capability of using Mohr Coulomb, Modified Cam Clay ,and the hardening soil model (very common in plaxis). The first two exist but their implementation inside the NewTemplate3Dep is too complicated, and, to my opinion, wastes time in too many function calls. The hardening soil model is very useful and should be implemented.
Have a procedure to initialize the state of the model. For example, in the mutliyieldpressure dependent I have to use the elastic mode of the model run one analysis and then the elastoplastic. This is not sufficient since the initial stresses are not represented exactly. In the DM soil model I have to run an independent elastic analysis, store the data in an array and use this as initial conditions withouth running an elastic step. As you can see there is not a unified way to apply the initial geostatic conditions. VERY IMPORTANT PROBLEM.
The DM soil model as implemented does NOT work for low stresses (even after a very important bug i fixed and send you last year). I have a version that uses various algorithms to improve the integration of the model that works for my models, which I could share at some point, but this is definitely an issure!
In soils it is common that some region is saturated (u-p elements 3DOF) and some region is not satureated (normal quad 2dof/node elements). So in a model that i have both elements i have to create double the nodes in the interface and then connect them with equalDOF objects! Very time consuming. I would add a command in OS that would be something like:
map quad quadup 1 2 1 2
which would say that the dof 1 and 2 at every node on the quad element correspond to dof 1 and 2 on the quad up element. This is way the elements will share nodes and OS will know how to create the stiffness matrix.
Partially saturated soils! It would be great to see some sort of model being able to predict the mechanical behavior of these soils.
Fluid solver, to be able to do e.g. dynamic embankement analysis next to the sea level! Very important. If this is too big of a task we can get some people to try and work on an ABAQUS element to connect OS with ABAQUS fluid solver with OpenFresco (I had some correspondence with Andreas some time ago).
Large displacement u-p elements for plane strain. Speaks for itself, very much needed for 2d liquefaction simulation.
Last, a way to automate free field absorbing boundaries similar to Flac. Today this can only be done either with extensive manual input (which I am doing), or with rough simplifications.
Thank you very much,
Please do not hesitate to contact me for more details, I've working with OS for many years and I would like to see the geotechnical modules improve and become useful in so many more cases!
Analytically, there should be the capability of using Mohr Coulomb, Modified Cam Clay ,and the hardening soil model (very common in plaxis). The first two exist but their implementation inside the NewTemplate3Dep is too complicated, and, to my opinion, wastes time in too many function calls. The hardening soil model is very useful and should be implemented.
Have a procedure to initialize the state of the model. For example, in the mutliyieldpressure dependent I have to use the elastic mode of the model run one analysis and then the elastoplastic. This is not sufficient since the initial stresses are not represented exactly. In the DM soil model I have to run an independent elastic analysis, store the data in an array and use this as initial conditions withouth running an elastic step. As you can see there is not a unified way to apply the initial geostatic conditions. VERY IMPORTANT PROBLEM.
The DM soil model as implemented does NOT work for low stresses (even after a very important bug i fixed and send you last year). I have a version that uses various algorithms to improve the integration of the model that works for my models, which I could share at some point, but this is definitely an issure!
In soils it is common that some region is saturated (u-p elements 3DOF) and some region is not satureated (normal quad 2dof/node elements). So in a model that i have both elements i have to create double the nodes in the interface and then connect them with equalDOF objects! Very time consuming. I would add a command in OS that would be something like:
map quad quadup 1 2 1 2
which would say that the dof 1 and 2 at every node on the quad element correspond to dof 1 and 2 on the quad up element. This is way the elements will share nodes and OS will know how to create the stiffness matrix.
Partially saturated soils! It would be great to see some sort of model being able to predict the mechanical behavior of these soils.
Fluid solver, to be able to do e.g. dynamic embankement analysis next to the sea level! Very important. If this is too big of a task we can get some people to try and work on an ABAQUS element to connect OS with ABAQUS fluid solver with OpenFresco (I had some correspondence with Andreas some time ago).
Large displacement u-p elements for plane strain. Speaks for itself, very much needed for 2d liquefaction simulation.
Last, a way to automate free field absorbing boundaries similar to Flac. Today this can only be done either with extensive manual input (which I am doing), or with rough simplifications.
Thank you very much,
Please do not hesitate to contact me for more details, I've working with OS for many years and I would like to see the geotechnical modules improve and become useful in so many more cases!
CapabilityRequests on Soil Modeling
1.Soil Reinforcement may be implemented, e.g. geogrids, geofabrics etc.
2.Reg DSSI, Scaled Boundary FEM (Wolf & Song, 1996) , if implemented, would be great capability.
3. In one of the earlier posts, Boris talked about limitations of current quad elements in plane-strain analysis. These may be removed.
4. Some GUI, like building Tcl for Soil Modelling will be another great addition.
2.Reg DSSI, Scaled Boundary FEM (Wolf & Song, 1996) , if implemented, would be great capability.
3. In one of the earlier posts, Boris talked about limitations of current quad elements in plane-strain analysis. These may be removed.
4. Some GUI, like building Tcl for Soil Modelling will be another great addition.
-
khosravifar
- Posts: 9
- Joined: Fri Nov 03, 2006 3:15 pm
- Location: Fugro Consultants Inc.
1. Being able to assign different convergence criteria to different regions (i.e. soil and structure).
2. It would also be nice to record the residuals (in "test" command) in different elements to get a sense of which part of the model is problematic. This is helpful when the model consists of more than one type of elements, such as SSI.
3. Some references on how to choose system command options, which I guess requires having more access to the stiffness matrix!
4. Recording stress and stress at nodes and gauss points (I guess right now it only records at gauss points.)
2. It would also be nice to record the residuals (in "test" command) in different elements to get a sense of which part of the model is problematic. This is helpful when the model consists of more than one type of elements, such as SSI.
3. Some references on how to choose system command options, which I guess requires having more access to the stiffness matrix!
4. Recording stress and stress at nodes and gauss points (I guess right now it only records at gauss points.)