Non-Linear VS P-Delta

[Anonymous]

Hi!

In order to study a trestle tower behavior, I need to evaluate it with an appropriate 2nd order analysis.

Some problems appear during the analysis, that let me apprehensive.

1: A instability (type 3) warning is always shown at beginning of calculations. I cannot resolve this instability, even removing all bar releases. So I question if it has some relevant influence on final results;

2: I perform a non-linear P-Delta analysis, and a non-linear analysis. As you can see on the models attached, the results are wrong for P-Delta, as the deformation on combination C110 is less than on static-linear combination C100.

With the simple non-linear analysis model, deformation results seems more trusty. Is the simple non-linear analysis enough?Does it count geometrical non-linearity of the deformation pattern?

3: I ask you to explain the main difference between these analysis that cause such different results. And, in which cases I should evaluate structural behavior with one or another.

Thanks,

[Anonymous · Accepted solution]

The P-Delta model in following:

[Artur.Kosakowski]

1. I think that the reported instability is caused by large difference in stiffness of the elements of the model (rigid links). If you delete rigid links (together with the bar 164  they support) the instability is no longer reported. In such case it can be ignored.

2. P-Delta analysis does not converge therefore its results are incorrect (please be careful when you press Esc to avoid display of messages and check the calculation report before exploring the results)

Case 110 :

ELU NL PD

Analysis type:

Nonlin. Combination

　　

Non-linear process: divergent.

Maximum value of process parameter when convergence is obtained

: 0.000

Maximum value of process parameter when convergence is not obtained :

0.006

3. Please check the attachment.

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]

Thanks,

If I understand it correctly, for the case exposed the main part of non-linear effects results from b/ (second order geometrical non-linear effects), because deflection of each element should not be relevant on global behavior of the Tower.

So, if adequate initial imperfection is induced on the model, a non-linear analysis (b/) must reproduce the main non-linear effects of the geometric non-linearity, correct?

And are the instability warnings the reason for divergence on the Non-Linear P-delta analysis?

[Artur.Kosakowski · Accepted solution]

---

If I understand it correctly, for the case exposed the main part of non-linear effects results from b/ (second order geometrical non-linear effects), because deflection of each element should not be relevant on global behavior of the Tower.

I would say so.

So, if adequate initial imperfection is induced on the model, a non-linear analysis (b/) must reproduce the main non-linear effects of the geometric non-linearity, correct?

I'm not sure if I understand you correctly. I would assume that the imperfection is added to simulate the second order analysis when the linear static is used whereas non-linear analysis takes into account increase of deflection due to vertical load by its principle.

And are the instability warnings the reason for divergence on the Non-Linear P-delta analysis?

The reason is the use of RLink elements to model rigid links (they have to be used for Sparse solver which is the default one). If you use 'standard' rigid links (you have to use Skyline solver) then you can converge P-Delta too.

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]

Ok. Your explanation was very helpful.

I refered to a normative initial imperfection for global analysis in frame strucutures, as provide in eurocode EC3 (5.3.2 - EN1993-1-1), that can reproduce an eventual constrution imperfection. Than, with a non-linear analysis I could assess second order effects of load cases.

Thank you

[skilak]

Hi,

I am trying to implement the initial imperfection as per EC3 5.3.2 for global analysis (I assume that the local will be taken into account while doing the members checks e.g. buckling curves). The initial imperfection can be implemented in the analysis parameters/buckling deformation if I select a case (buckling one) there e.g. LC 1: Dead. For that case I will ask for 10 buckling modes. Question, for the initial imperfection I have to select the mode and a coefficient, is it always the first mode (the lowest critical load factor) or, for example, when with Wind X+ , I have to find the relevant mode in the wind direction say Mode 5. Shall I understand that the Modes in the Tab - Buckling Deformation are the modes in the chosen load case? Or, for a new load combination e.g.. Dead + wind I have to perform new buckling analysis? I have a shell-thin metal on the roof of a steel structure then is there a chance to turn off the shell from buckling analysis as I get plenty of buckling modes of the shell instead of the bars what quite obvious? Thank you very much in advance for your kind help.

Best regards,

Tom

[Artur.Kosakowski]

The buckling deformation 'deforms' the structure to the shape that corresponds to the selected mode of the selected buckling analysis case. Then calculations (for all defined load cases and combinations) are performed for model with 'deformed' geometry. IF you want to have 'different deformations' for different load cases I'd rather  use the notional loads instead adding them to the combinations of other loads.

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.

[skilak]

Hi,

Does Robot generate the Equivalent Notional Forces or I have to do it by hand? On an odd structure it might be a bit time consuming.

Best regards,

[Artur.Kosakowski]

Please check http://forums.autodesk.com/t5/Autodesk-Robot-Structural/Notional-load-definition/m-p/4369111

[Anonymous]

Sorry for diging out an old topic but I am a bit confused about non-linear and P-Delta analysis and I have not found satisfactory explenation where difference lies between these two methods.

I have created simple cantilever loaded at the tip with horizontal and vertical force. 

Please find my findings below (non-linear analysis only and P-Delta). Could you provide me with some information where the difference in the results comes from? Or provide with some technical note which explain the differences between that two methods.

Thank you

 

[skilak]

Hi,

In brief, non-linear is to be used when you have a physical non-linearity of material e.g. plastic hinges, tension only elements or plastic properties of cross section. P-delta (vertical force and horizontal displacement) is for unbalancing the structure - the analysis will attempt to unbalance the whole structure there, where the stiffness will allow (perfect symmetrical structures may not be able to be unbalanced unless you introduce a small imperfection of any sort).

I hope it helps a bit.

Best regards,

[Pawel.Pulak]

The names "Non-linear" and "P-delta" used in the current version of Robot are somehow misleading and they are incoherent with the effects considered when activating them.

It is discussed and explained in for instance this forum topic (the second post of it redirects also to some explanations on AUGI forum)

---------------------------------------------
If this post answers your question please click the "Accept as Solution" button. It will help everyone to find answer more quickly!

Regards,