Non-linear analysis

[brhm]

Hi,
I have tried to make a non-linear analysis (geometric deformations) for a "large" structure (20772 nodes with bars, plates etc) and the results are suspicious.

I used first the incremental algorithm and I got minor bending moments than the linear analysis..
I changed to the Arc-length (poor documented...) the results were better or more logic.

I used first 5 load increment steeps. The results: Internal forces 20% higher than the lineal analysis but also the applied loads were 20% higher (from the reactions table results).
Where do this loads come from???? I have used the same load all the time!

Then I used 10 load steeps and got 10% higher internal forces and applied loads.
30 steeps; 3% higher
50 steeps; 2%
100 steeps almost the same than the lineal.

If you don't have notice the ROBOT algorithm for 2nd order analysis seems to be:
Results_2nd_order = Results_1s_order x [1 + 1/ (load steeps)]

I have even read in a doctoral thesis that "ROBOT cannot perform analysis of second order att all".

Some body know better than me about this subject?

[Tomasz.Fudala]

Please contact technical support through Subscription Center.
Don't forget to attach a file with your structure.

[pascal.bogdan]

Hi,
Where did you found some documentation about non-linear analysis with Robot Structural Analysis?

[brhm]

Hi Bogdan,
That is the problem the lack of documentation...there are several members of this forum tryng to perform this kind of analysis...
I send my model to the support team and I'am waiting for comments.

That I found was a paper where the autor compares the results of second order analysis from several programs againts the method of amplification of moments included in the AISC.
It is known that the amplification of moments approach is a conservative and approximately method and the results of this comparation could be expected.

Any way here is the link:
http://scholar.lib.vt.edu/theses/ava.../AMSThesis.pdf

[smm.144]

brhm thanks for the information. The document is about Robot Millennium.(I have not completely read it yet..)
But Robot algorithm since Millenium to RSAP has not changed?
About the first post are you able to model the structure in another software to compare results?
Please let us know if any new information you have..

[brhm]

I got help from a team from Autodesk support net. Here comes some tips:

1) “Method of solving the system of equations”
How many of you did know about this possibility?
When I set this option to “Skyline” the results were correct.
To set this option, choose from the menu: "Tools", "Job preferens", "Structure analysis"
-But Autodesk: with method will work better for a certain structre?

2) ”Calculation report”
Did you know that you can see how “much correct” is your analysis?
Bring up the report from you analysis (after calcultation of course); from the menu choose: “Analysis”, “Calculation report”, “Full note”
Look at Precision value; large number here (>1) indicates a good precision, = 1 is not acceptable; for low precision you get instability of type 3 and when you accept that you must assure that the results are according to your expectation

3) I performed a static non-linear analysis (because of geometrical deformations and linear materials) I performed the incremental method. Just with 5 load increments (“Load increment number”) and it worked.
I got help from a team from Autodesk support net.

1) “Method of solving the system of equations”
How many of you did know about this possibility?
Where when I set this option to “Skyline” the results were correct.
To set this option, choose from the menu: "Tools", "Job preferens", "Structure analysis"

2) ”Calculation report”
Did you know that you can see how “much correct” is your analysis?
Bring up the report from you analysis (after calcultation of course); from the menu choose: “Analysis”, “Calculation report”, “Full note”
Look at Precision value; large number here (>1) indicates a good precision, = 1 is not acceptable; for low precision you get instability of type 3 and when you accept that you must assure that the results are according to your expectation

3) I performed a static non-linear analysis (because of geometrical deformations and linear materials) I performed the incremental method. My structue with 5323 nodes and bars, shells, etc. worked with just 5 load increments (“Load increment number”).

I attach the in a PDF respective screen shots.

Autodesk: Don’t you think it’s time to give us a real user manual, please?

[diming]

brhm thanks for your post
2 years have past since your post, is there any example of non-linear analysis out from autodesk?

[enggrazi823687]

hello all, it sounds like all of us are suffering from the same problem on daily basis. As someone has mentioned that this is the time for Autodesk to undertake the vaildity of results of Robot Structural seriously and look in possible ways of correcting it ASAP. We do understand that all structural analysis softwares have evolved in their current state over the years and it requires patience as well. However, since Autodesk has taken up this software on board, I am sure that the outcome will be good but we all have to wait and see. Keep calm and carry on

[rigid_joint]

autodesk never really loved robot

[kmarsh]

Robot documentation is quite thorough and detailed with respect to the methods of analysis. Here are some selected articles:

Theoretical Basis for Modal Analyses:
http://docs.autodesk.com/RSA/2013/EN...F04CCC4D5.htm?
Non-linear vs. P-Delta Analysis
http://docs.autodesk.com/RSA/2013/EN...84F87843A.htm?
Theoretical basis of methods used during structure dynamic analysis
http://docs.autodesk.com/RSA/2013/EN...F04CCC4D5.htm?
Defining criteria for convergence:
http://docs.autodesk.com/RSA/2013/EN...3128665253.htm

As far as validity of the results, I just recently discussed the above mentioned thesis with our support folks and it has been 10 years since the paper mentioned was written. Not only have we confirmed that, as far back as we can easily go (release-wise), Robot is, in fact, able to provide non-linear analysis for the structures presented in the above referenced paper but Robot has also improved significantly since then. We quickly re-did the experiement presented in the paper and found that Robot's results were very close to other published results.
Below are forces and moments obtained for Frame 4 using linear analysis (case 4), 2nd order (case 5) and 3rd order (case 6) nonlinear analysis. You will find that Robot calculated moments are in the same range with other analysis results provided.

non-linCase4.jpg

I would encourage folks to follow the setup in the article and run the experiements for yourself. They're fairly straightforward and I think that going through the exercise yourself is worth much more.

Please don't hesitate to ask more questions, Robot is a very powerful analysis and design program and we want you to be successful with the software. There are also a lot of other users who have been working with it and can help answer questions.

Thanks again,
-Ken