buckling analysis on roof bracing

[achillesgr]

hello,

i am a bit confused about buckling analysis on roof bracings.

i found some exambles but are with oposite methods. some make analysis and some dont.

please someone help with that.

the bracings are X with L60x60x5mm and between two purlins.

1. should i declare these members as tension only?

2.as members with tension only, should i run buckling analysis on them?

(i am located in Europe and we must conform with Eurocodes).

Thanks .

[Artur.Kosakowski · Accepted solution]

1. should i declare these members as tension only?

If they are intended to work in compression then you should not. If not, then yes.

2.as members with tension only, should i run buckling analysis on them?

(i am located in Europe and we must conform with Eurocodes).

No, as there is going to be no compression in the tension only bars 

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.

[achillesgr]

Artur,

there is not a mechanism that prevents the L 60*60*5  to take compresion, that is simply prevented from the lalge members length comparing to the section.

so that is considered as "tension only"?

i undersrtand that if there was a larger section assuming a IPE300 then there will be a real and significant compresion, but for an L60*60*5 with 5.5m long witch with the slightest force it bends and is usless??should i declare it in analysis as tension only?(so all the horizontal force goes to the other one witch is real scenario)

i hope to understand what i mean.

Also, what the X means at the buckilg diagrams dialogue?that the member will not be checked?

there is also options for "sway " or "non sway"

if i dont want to check the member should i check the X?

what about the "sway " or "non sway"?swould i check the "sway"?(because it is a sway member?) or it doesn t matter if the X is choosed?

can you please clear these?

thanks!

[Artur.Kosakowski]

If you have a long bar with small cross section the small value of the compression will cause this element to buckle so that it will not be able to transfer this compression force to other elements of a model. This is when you set is as tension only.

If you mark X then it means that the bar cannot buckle (e.g. is welded to a thick steel plate). In such case bucking check is not done. The sway or not sway setting is in such case irrelevant.

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.

[teixeiranh]

What RSAP is going to do is to shut down the element "tension only" if it is in compression. Then it will refresh the stiffness matrix of the structure and perform the next step in the incremental method of solving.

You may choose such elements to work as tension only if they have a high slenderness ratio. If they don't, however, you shouldn't.

I wonder if there is a way to turn that "tension only" element in a "limited compression" element...it would be very cool. 

[Artur.Kosakowski]

Via nonlinear releases.

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]

Robot runs correctly the buckling calculation set when I have "truss tension only" bars? I do not think so...

Look at the next example--- truss tension bars will act only one at each time, when I put both bars at the model in "tension only mood", the buckling length of the column comes incorrect. It seems like the truss tension X bars can't avoid the sway movement of the structure!! the buckling length becomes 2 x length of the columns, although I expect to achieve buckling length equal to the length of rhe column.

Usually, in buckling calculations, I only put one of the bars in the model : if the bar has compression force, I assume that the value of its axial force is only mathematical and that it is the other bar that will be at work, with a tension force of the same magnitude of the compression force given in the model. It is a accurate simplification to analyse models concerning sway movements/ horizontal load cases and buckling and modal analysis.

What do you think about that?

[Artur.Kosakowski]

I believe it has already been several times that we wrote on this forum that buckling analysis is not intended for finding buckling lengths of elements of a model 

In fact it is only intended to find the level of load that would cause the model to collapse. The buckling length is then calculated for each element based on Euler formula and axial load obtained in a given element for a given mode. As you can see from it this may be correct for selected mode only for a single element of a model provided it is the local lack of stability rather than a global one e.g. entire truss moving sideways.

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]

I agree with all that you wrote. But what I mentioned do not collide with these facts that you exposed.

• In fact, buckling factor represents the level of load at which the structure experiences his colapse (in other words, it is the mathematical multiplicative value of the load that makes the determinant of the stiffness matrix (with its geometrical part included) becomes zero).
• The buckling length, for each specific mode, is only real and meaningful for bars that have macroscopic deformation in this mode (menu diagrams, deformation topic activated). I agree with that topic. It can also happen that some bars, because they have less load than their neighbour bars, may have an increased buckling length in “helping” the other bars (those bars with less load wil have then a lesser buckling length than the buckling length corresponding to an equal load level in all columns). In a frame with plenty columns, if the two exterior columns contain axial loads such that their buckling load is not reached, when the interior columns reach their independent buckling loads, shear resistance will be developed in those exterior columns which counteracts the sidesway tendency of the global frame. If all columns are samely loaded and they want to buckle simultaneously, there will be no shear resistance available. However, with different load levels at interior and exterior columns, the exterior columns will "brace" the interior ones. The critical load for the interior columns is increased and their effective length is decreased. The stabilizing effect can be such that the effective length of some of the columns could be reduced to 1.0, even though there is no apparent bracing system.
• If it is true that a particular mode is correct only for the few elements of a model that experience deformation in this mode, it is also true that their buckling mode could concern only local lack of stability, rather than a global one. But in the particular example that I attached the columns have the same load applied, so they will buckle in union. No bar will help the other bar. And the frame is so simple that only “global modes” (sway and non-sway) appear - for the first buckling modes all the bars of a specific frame buckle simultaneously. So, as you can see, the first mode, valid only for the columns of the frame with “tension truss” bars (only these bars have deformation when you activate “deformation” in diagrams menu), implies a buckling length of 11.68m (about 2.33 L), and the second mode, valid only for the columns of the frame with one “truss” bar, implies a buckling length of 4.95 (almost L), what is a correct value, because the truss bar will avoid sway movement and the rigid connection between the nodes of the frame will help columns buckling less than L, activating the flexural restriction between columns and beam nodes.
• As the columns of my example do not have different level of initial loads, and as all the bars at each frame buckle together at their particular modes,  this example, in my opinion, proves that the use of tension bars at buckling analysis is not correct. Because we should achieve a "non sway" buckling length in both frames, instead of the erroneous  2.33 L.
• So the buckling length at my example is an important and meaningful value. It does not concern local modes and we should reflect about it carefully... because it apparently proves that robot does not process the non linear condition of the truss tension bars correctly. If it does, we must have a buckling length less than L - it's the ultimate goal of putting ties in frames after all.