Increasing mesh size from radioss model to optistruct model
Hi,
One more thing is that can a spring element be connected to the master node of RBE 2 as I am getting problems of loose spring nodes during radioss integration for NLGEOM analysis. In optistruct, the spring element is CBUSH
Answers
-
Hi Gopal,
Here is a test file where a CBUSH is attached to two RBEs and used to run a linear static run.
0 -
Hi,
I am running NLGEOM analysis and as during solving it converts into radioss engine and starter. On checking, radioss starter in hypercrash for any model error I am getting error of loose spring nodes, CBUSH got converted to TYPE 8 element.
Also,CBUSH is getting converted to type 8, with very low mass., whereas in optistruct I have not added any mass over, the reduced mass causes very low time step. How to avoid this?
On checking the same with the above model in NLGEOM, it solved properly but on checking the radioss starter with hypercrash model checker the same errors which came in my model came but still it solved, I think there might be something else which I might be missing.
0 -
Hi Gopal,
DO you see any other warning messages in HyperCrash?
Regarding mass, I will check and update to you soon.
0 -
-
-
Hi Gopal,
I will check and update to you soon.
0 -
0
-
Hi,
Please send me the link for the same.
0 -
Please Use the LINK to share the file
0 -
Hi,
The files have been sent.
0 -
Hi,
One more thing which I have seen in help content of NLGEOM (Radioss integration) , it says that the PBUSHT will not be translated fully while conversion from bulk data to block data format. I have this property in my model , does this causing a problem.
I have attached the image of the same. However, I thought this should create problem in starter but not engine and I am getting no errors in starter and segmentation violation during engine run.
0 -
Hi Gopal,
A CBush element will be translated to type8 general spring and some of the paramters like inertia, mass may not get translated. I am checking the same with experts, I will also share your comments. Will let you know soon.
0 -
Hi,
In CBUSH element , I think we do not have any option of assigning mass and inertia in optistruct, Thus on conversion to radioss i.e TYPE 8 spring, the small mass and inertia are being added by default to make sure it develops some time step for calculation.
0 -
Gopal,
You are right. Somehow on what parameters this mass and inertia is calculated is not known.
0 -
Hi Gopal,
Can you check if there are any modelling issues in the .rad file?
Try to clear any penetrations, temp nodes, etc... and check if that helps.
0 -
Hi,
I have gone through the .rad file and even starter .out has on error or warnings, also there are no penetrations, temp nodes.
One thing is that I am using the Radioss meshed model for simulation, thus no. of elements is high keeping 5mm minimum element size in radioss model. Does the model size has an effect on this. I was expecting that even with high element count, the model should have atleast initiated the run and could have taken a longer time for solving, but it is coming out as soon as starter gets read.
0 -
Hi Gopal,
Is the solver still terminating with segmentation violation error?.
Please share the model files through secure dropbox link.
0 -
Hi,
When I took a front section of the model, with same conditions the model is running but same is not happening with complete model. I will be trying the same with rear section and check whether it works independently or not.
I feel this is due to the large element count?
0 -
Hi Gopal,
This seems strange, Please update after the run.
0 -
Hi,
It seems that the rear section is having some issues, as time step is consistently decreasing and the solution not getting converged.I have sent the rear section file named test.hm.
0 -
Hi,
Please find the attached simplified model of rear section. it has only 1-d elements. In this, I have assigned a non-linear spring property. Running with non-linear spring , it gives error but when I run the same with a constant stiffness, the model runs. This is being to identify the problem coming in running the complete problem.
0 -
Hi george and prakash,
Any updates?
Thank you.
0 -
Hi george and prakash,
Have you gone through the files?Kindly update me for the same.
0 -
Hi Gopal,
We are trying to find the convergence issue in your model. And I think its related to the spring elements.
CBUSH were converted to Type 8 spring elements. If initial length is not equal to zero, the equilibrium is insured for forces. It is then recommended to use spring elements type 8 with a zero length. Can you recheck this in your model.
0 -
Hi,
Thank you for the reply.
Instead of going with NLGEOM analysis to avoid conversion issues between optistruct and radioss, I tried with NLSTAT analysis and the model is converging but I am not getting the expected results from the model i.e the spring forces are coming way too high and thus structure giving very high displacements.Have you gone through the table of the non-linear spring stiffness, is this correct way of entering the values. This is when I am working on complete model.
On more thing is about units I have used force and mm in table and for my model it is ton, mm, N ,sec.This seems correct to me but just wanted to confirm.
0 -
Hi,
In order to understand the problem at root level i was trying with one dof spring element and applying force on it with one constrained. On using linear spring stiffness model i.e only K value the run is fine, but when I assign a non-linear stiffness curve, it behaves abruptly and thus it seem that there is an issue with the table of non-linear spring properties. Kindly let me know about guidelines to be followed for assigning a curve to spring elements and also in case of any sample file available, kindly provide it.
0 -
Hi,
After working on individual spring element with the non-linear curve, it is found that there was an issue with the sign of the force and displacement used in the curve. In my case, the force being the gravity needs to be negative and also displacement is negative after changing the signs the solution is getting converged. This is with NLSTAT type of analysis.
Now I am trying on the full model.
Thank you for support.
0