Rafael Medeiros
Did you find this post helpful? Feel free to Like this post.
Did your question get successfully answered? Then click on the ACCEPT SOLUTION button.
Either using a non-linear release or a non-linear support. Mind to set the parameters of the non-linear analysis as shown on the attached picture.
If you find your post answered press the Accept as Solution button please. This will help other users to find solutions much faster. Thank you.
Hi Artur
That´s exactly what I tried to do except for marking " matrix update after each iteration" . But also doing that I keep getting a zero reaction value where I get 100tf when the UZ direction is fixed
Rafael Medeiros
Did you find this post helpful? Feel free to Like this post.
Did your question get successfully answered? Then click on the ACCEPT SOLUTION button.
Please send the model and indicate the node number and load case to look at.
Artur,
I solved the problem using Bilinear model with a very high K (10000000)for the 1st spring, zero K for the 2nd spring and a corresponding D1 for the force I wanted to limit.It worked great!
But now I´m having a weird problem when running the model.I have like 120 combinations and ,it takes like 20secs to solve each one .But after something like 25 combinations ,it starts to not converge. So I stopped the calculations ,deleted most of combinations and only left some that robot reported as non convergent.Ran it again without saving the model and to my surprise they converged also in 20secseach and without using " matrix update after each iteration" !!!
Nodes with non linear supports are 10 ,21 ,32 ,33 . Combinations are 41,73,75,89
Model is attached. Rename it to .rar
Rafael Medeiros
Did you find this post helpful? Feel free to Like this post.
Did your question get successfully answered? Then click on the ACCEPT SOLUTION button.
Try to use the parameters from the attached picture. I have calculated the model and all combinations converge.
Case 132
: 1,35xPP+1,05xSC+0,84xVTsuc+0,84xVT270-1,2xTPgrad
Analysis type:
Nonlin. Combination
Non-linear process: convergent.
Maximum value of process parameter when convergence is obtained
: 1.000
Maximum value of process parameter when convergence is not obtained :
1.000
If you find your post answered press the Accept as Solution button please. This will help other users to find solutions much faster. Thank you.
Thx a lot Artur,
It worked.
But What I don´t understand is why a combination that doesn´t converge in the first run (without marking "matrix updating after each iteration"), converges if I use calculation restart without modifying anything?
Why the calculation process behaves differently??
The problem is solved, but I´d like to understand that.
thx
Rafael Medeiros
Did you find this post helpful? Feel free to Like this post.
Did your question get successfully answered? Then click on the ACCEPT SOLUTION button.
I have to admit that I do not fully understand the sequence of actions you made and the idea behind them. Definitely deleting some of the load cases/combinations while the analysis was terminated and then restart of the analysis is not the thing I personally do. The intention of this option was to change the parameters of the analysis when you can see that the process has got little chance to converge with the current settings. I can try to investigate this situation providing you give me the detailed (step by step) description to follow. As you could see all I did was to use the settings I suggested before
BTW: Have you checked in the calculation report that they actually converged?
Artur,
For just an example to show you ,I limited last load case to be number 41.Combinations 39,40 and 41 don´t converge in the 1st run .After I get the no convergence warning for load case 41 ,I just click OK and do an analysis restart for just case 41 without changing any settings.The result is completely different and now the combination converges
Watch it in this video.
http://screencast.com/t/jbHGlJc2z
Rafael Medeiros
Did you find this post helpful? Feel free to Like this post.
Did your question get successfully answered? Then click on the ACCEPT SOLUTION button.
As far as I can see Robot starts to have difficulty to converge the last of many non-linear cases for the BFGS method and the multi-threaded solver (when you restart calculations or delete some of the previous cases then divergent cases will converge). Thank you for pointing this situation out - we will check what exactly happens.
The solution is to either use the Full Newton-Raphson method for the multi-threaded solver (the settings I used) or use the Automatic solver (Sparse_M in this case) for the BFGS method.
Ok Artur,
Thanks!
Rafael Medeiros
Did you find this post helpful? Feel free to Like this post.
Did your question get successfully answered? Then click on the ACCEPT SOLUTION button.
As far as I can see Robot starts to have difficulty to converge the last of many non-linear cases for the BFGS method and the multi-threaded solver (when you restart calculations or delete some of the previous cases then divergent cases will converge). Thank you for pointing this situation out - we will check what exactly happens.
Corrected in Robot 2015.
If you find your post answered press the Accept as Solution button please. This will help other users to find solutions much faster. Thank you.
Rafael Medeiros
Did you find this post helpful? Feel free to Like this post.
Did your question get successfully answered? Then click on the ACCEPT SOLUTION button.
Rafael Medeiros
Did you find this post helpful? Feel free to Like this post.
Did your question get successfully answered? Then click on the ACCEPT SOLUTION button.
I think ( still testing) I found a way to do it using non-linear bar releases .
Structure to the left - panels are normally connected
Structure to the right - Exact same structure ,(with same load ,10tf/m2 on the top slab area) except that the vertical panel is 2cm shorter . Mesh is 1 mt . Created very stiff bars (2cm) connecting the panels at each node. Then I created a non-linear release limiting the moment at 25tf.m(12.5tf.m on the sides) using " rigid with linear hardening" type. Seems to be working nicely. For a load that does not reach 25tf.m on that edge , both structures show the same bending moment. Is there anything else that I should be aware?
Rafael Medeiros
Did you find this post helpful? Feel free to Like this post.
Did your question get successfully answered? Then click on the ACCEPT SOLUTION button.
Rafael Medeiros
Did you find this post helpful? Feel free to Like this post.
Did your question get successfully answered? Then click on the ACCEPT SOLUTION button.
I thought about compatible nodes but I think they can only be used with bars.
It is the only way you can define (or shall i wrote control to which object they are assigned) them but you may achieve this goal in the following way:
1. Define compatible nodes (with the intended nonlinear function) at the intersections of bars created at the intended location of a common edges of panels.
2. Delete bars
3. Move one of the set of nodes either vertically or horizontally
4. Define panels with (temporarily separated) edges at the lines of these nodes and mesh them (make sure the generated meshes match number of nodes along the edges)
5. Freeze meshes
6. Move the 'shifted' nodes (as well as the panel's edge if you want) to their 'original' locations