Buckling analysis - visual results look not correct.

[Anonymous]

Hello, support.

See attached picture. Please, gan you give any comments?

[Artur.Kosakowski]

Check if you displayed deformation or shape of buckling mode please. If this does not answer your question attach the model please. 

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.

[legar]

You should simply read software messages and turn on suggested algorithm DSC... 🙂

[Anonymous]

In my other topic Rafal said you need ignore warning, so it is quite difficult to understand which warnings you need to ignore and which not. Also, program gives only suggestion. So, as I understood, difference in this model between using DSC algorithm and a simle algorithm is only not correct visual results?

Other thing, I remember I had the model with different springs and offsets. And i was not able to use DSC algorithm, because program gave me a warning(i don't remember which exactly) so, in what cases I can't use this algorithm? (or there are no any limits for this algorithm?

[legar]

It is hard to say in few words, when you should use which method (algorithm) for calculations.

At first read carefully ARSA help adequate chapters - there are some useful suggestions.

At second... after calculations you can check results precision and come to a conclusion about results credibility.

As an example... if you calculate your file without algorithm DSC and with algorithm DSC and compare critical coefficient you can find that modes 1-3 and 5 are practically the same and in mode 4 the difference in critical coefficient is ca 5%.

Generally if you not turn on algorithm DSC then results will be a little strange near bar releases but overall results could be (not always) practically the same.

BTW

Maybe RG or AK (ARSA support) will describe it better ...

JerzyCP 

[Artur.Kosakowski · Accepted solution]

The DSC algorithm has been introduced as it is more precise for structures with nonlinear, buckling and dynamics analysis however in many of the typical situations the results with DSC on and off are very similar or the same. DSC cannot be used on bars with offsets.

The warnings are introduced because they indicate something that is (or at least my be) important and should not be ignored without looking at them at all  

@legar

Thank you  for your help.

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.

[Anonymous]

Thank you for answers!

[Pawel.Pulak]

May be a little bit too late - some additional explanations related to activating or not DSC for releases.

When DSC is inactive the degrees of freedom corresponding to released directions are (simplifying) ignored in appropriate members with releases. That is why it can be used only for "full" releases and not for "elastic" or "non-linear"ones.
When DSC is active additional calculation nodes are generated in releases and discontinuity elements (with zero length) corresponding to releases are generated between user nodes and these calculation nodes. It can be even seen that the number of equations reported by the solver is bigger in such case comparing to models without activating DSC.

In case of static linear analysis both approaches give the same results.

In case of dynamic, nonlinear or buckling analysis DSC gives more precise results. Sometimes these differences are only quantitative but sometimes thay can be qualitative, i.e. not considering some effects without DSC which would be considered with DSC.

Below is the example of very simple frame with one member with pinned releases.
When running buckling analysis of this frame with DSC active the first buckling mode is the local buckling of member with releases - see below: 

Then running the same analysis with DSC inactive the first buckling mode is the global buckling of whole frame - which in the previous case corresponded to the second buckling mode:

Without DSC the local buckling of the released bar iss not found at all. It is so because this buckling shape is in such model (without any internal nodes inside the released bar) related only to rotations of the ends and they are ignored without DSC.
Of course adding internal node in released bar will result in finding local buckling of it also without DSC but the precision will be worse than in analogous model with DSC (qualitative difference will change into quantitative one)

I hope it slightly explains the observed effects and the messages displayed by Robot.

[Anonymous]

Thanks for informative answer! I already wanted to ask about elements local buckling. In help i found information about how progam makes analysis of buckling, but i didn't find anything if prgram check the local stability of elements (I don't know hot to say in english, in russian it would be 'явные и скрытые формы потери устойчивости'. So, as I understood DSC algorythm also helps to determine 'скрытые формы потери устойчивости'

[Pawel.Pulak]

I am not sure what you mean by  'скрытые формы потери устойчивости' (hidden forms of loss of stability?).

If you mean only standard Euler buckling then Yes - using DSC will enable to find such local buckling during buckling analysis (another thing is precision of such results in relation to Euler formula - see other topics on this forum mentioning necessity to divide members in smaller parts to increase accuracy)

If you mean the lateral buckling, i.e. loss of stability of top or bottom flanges of I sections etc. then No - it is not a case, it will not be found. Such buckling can be found only modling web and flanges as shell structure using surface finite elements.

[Romanich]

I think "скрытые формы потери устойчивости" it is buckling modes that realized in limits of one FE and does not cause rotation and displacement for nodes.

[Pawel.Pulak]

Thanks Roman for the explanations:)

If  "скрытые формы потери устойчивости" are buckling modes that are realized in limits of one FE and do not cause rotation and displacement for nodes then activating DSC does not always help to find them.

It concerns situations when such modes are related not to rotations but to displacements.

See for instance the screen capture below made for slight modification of the very simple test model used before. This time the model does not contain any releases and one of columns (bar 101) is made of significantly weaker section than other ones to show the effect.

The local buckling of this column would not be found (the activation of DSC unimportant here) if the additional node 4 would not be defined along it.

And some remark to to prevent possible concerns related to such effects - such local stability is checked in design modules (steel design or RC design) when considering the slendernes (the buckling length) of members.

Regards,

[Anonymous]

Yes, Romanich translation is correct. 

---

@pp2008 wrote:

And some remark to to prevent possible concerns related to such effects - such local stability is checked in design modules (steel design or RC design) when considering the slendernes (the buckling length) of members.

---

But in design modulus we need to define buckling length, which we can take from buckling analysis for most critical member. In this case I think, we need to check all types of buckling(global and local). 

[Pawel.Pulak]

Usually the buckling length in design modules is not taken from buckling analysis but from individual buckling conditions of each member.

See the message 7 in this forum post:

http://forums.autodesk.com/t5/Autodesk-Robot-Structural/Buckling-Analysis/m-p/3436847#M4527

Regards,

[Anonymous]

Absolutely disagree with message 7 from your link. Anyway, thank you for explanation.