Spatial Truss - Instability

[t.sautierr]

When I change the structure from 2D frame to 3D I get instabilities and Ican't find why .... please help.

I must be a silly thing but, I don't find ...

[t.sautierr]

As far as I remember, this not clearly stated in the help ? I mean for example that the self weight is vonverted into point loads at the ends of the bars.

I assume that it is the same for user defined loads if thay are not point loads on nodes?

I'll make some tests.

[Artur.Kosakowski · Accepted solution]

I assume that it is the same for user defined loads if thay are not point loads on nodes?

You can't define bar loads (forces) for trusses.

[k.budzinski]

Jestem w trakcie modelowania trójwymiarowej konstrukcji z ryglami w postaci kratownic (praca dyplomowa). W trakcie przeliczania natknąłem się na problem w postaci niestabilności, które dotyczą kratownicy. Pszeszukałem fora i natknąłem się na Pana posty. Czy byłby Pan tak miły i postarał się mi pomóc z wybrnięciem z sytuacji.

Co do konstrukcji całość modeluję jako rama przestrzenna (potrzebne mi do zadania obciążeń powierzchniowych za pomocą okładzin) i tu zaczynają się probelmy. Pręty kratownicy zamodelowałem jako kratowe (charakterystyki zaawansowane prętów - pręty ściskane lub rozciągane) i po przeliczeniu wyskakują niestabilności typu 1, 2 czasem 3. Z ciekawości gdy usunałem charakterystyki zaawansowane i całość była tak naprawdę wielką ramą kratownicowopodobną wszystko było ok - zero niestabilności.

Domyślam się, że chodzi o stopnie swobody ale niestety nie potrafię sobie z tym poradzić.

Byłbym bardzo wdzięczny za wszelką pomoc

[Artur.Kosakowski]

Hi Krzysztof,

As it is the English forum I'd like to ask you to posts in this language so that others could understand and find the solution for a similar situation of just be able to help. Thank you.

What you described suggests to many releases and usually it is because there are no bracing in the out of the plane of the truss direction. Try to use "standard" bars for posts and bracings with partial releases first. 

http://forums.autodesk.com/t5/robot-structural-analysis/steel-design/m-p/3573540

Then check values of moments and use "full" releases (keep RX at both ends as fixed)  or truss bars where they are necessary due to "excesssive" values being obtained. 

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.

[k.budzinski]

"As it is the English forum I'd like to ask you to posts in this language so that others could understand and find the solution for a similar situation of just be able to help. Thank you."

Of course I'll try my best and describe my problems in english. Sorry for writing first post in polish.

"What you described suggests to many releases and usually it is because there are no bracing in the out of the plane of the truss direction. Try to use "standard" bars for posts and bracings with partial releases first. 

http://forums.autodesk.com/t5/robot-structural-analysis/steel-design/m-p/3573540

Then check values of moments and use "full" releases (keep RX at both ends as fixed)  or truss bars where they are necessary due to "excesssive" values being obtained."

I tried what you suggested by still no results. The problem is that I want on bars of truss only axial/normal forces (no shear force nor bending moments). Releases help me only with the nodes, there are no bending moments but they appear on the bar (with shear forces as well). Other thing it's not possible to put relases to truss bars and when I try to model my truss with some bars with releases and some as truss bars I get instabilities again even having bracings.

Can you explain how truss bars work? Is it possible to design spatial truss using truss bars without getting instabilities?

Thank you in advance for answer.

[Artur.Kosakowski]

"What you described suggests to many releases and usually it is because there are no bracing in the out of the plane of the truss direction. Try to use "standard" bars for posts and bracings with partial releases first. 

http://forums.autodesk.com/t5/robot-structural-analysis/steel-design/m-p/3573540

Then check values of moments and use "full" releases (keep RX at both ends as fixed)  or truss bars where they are necessary due to "excesssive" values being obtained."

I tried what you suggested by still no results. The problem is that I want on bars of truss only axial/normal forces (no shear force nor bending moments). Releases help me only with the nodes, there are no bending moments but they appear on the bar (with shear forces as well).

In reality they will be there due to self-weight. A truss model is just the assumption that they are small enough to be neglected. 

Other thing it's not possible to put relases to truss bars

Therefore I asked you to remove this attribute from them first (Try to use "standard" bars)  

and when I try to model my truss with some bars with releases and some as truss bars I get instabilities again even having bracings.

If you have instabilities with partial releases used instead of truss bars then check the obtained shape of deformation as well as values of displacements and rotations in nodes. If they are as "expected" select whole model (CTRL+A) > run intersection command > open Bars table on the values tab and sort it by length. Check if you have very small values there.

Can you explain how truss bars work?

In the similar way as bars with pinned releases at both ends.

Is it possible to design spatial truss using truss bars without getting instabilities?

Yes, for a model with correct geometry and support conditions 

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.

[Tuctas]

Keep in mind that if in all the members of a node all moment releases are assigned, then surely instabilities will arise.  

[k.budzinski]

First of all thank you very much for help and priceless advice on my problem. Finally I managed to solve my truss (in spatial frame type of construction).

I got no warnings nor instabilities and deformation of whole construction seems to be as I suspect it to be.

I try to explain what I did to get proper results:

1. First of all I gave up the idea of modelling bars of truss as "truss bars" (advances characteristics), instead I used releases. Here I should mention that if you want to have pinned connection you should remember that at least one bar must be with fixed connection (otherwise instabilities will occur). Like on picuture below, bottom right bar has fixed connection in node 22.

2. When I put all releases on my truss bars (using pi-pin releases) I got a lot of warnings and instabilities again and deformation shape of the whole structure didn't look like I had expected. 

3. So next step was to change the definition of pin-pin release (I unchecked field with rotation about Z axis at end of a bar) like on the picture below:

4. And this time I had no problems at all with warnings and instabilities and shape of deformation od whole structure looke much better (as I expected it to look like):

5. That's all, hope that my case will help others with the same problems like mine.

PS. To compare similarities and differences of bars with pin-pin releases and "truss bars", I modelled another structure. This time simpler structure. First I designed everything in RM-Win [B)] (different software) and then in Robot Structural but using releases on bars [C)] and using truss bars [D)]. Loads are as on the picture below [A)] + self-load as well. And here are the results:

Results are almost the same in all cases but in my opinion when you want model truss bars you can get closer to reality if you model them using pin-pin releases (because truss bars also carry self-load so there should be some bending moments on bars).

[Mirko.Jurcevic]

If I may add to this thread (maybe Artur already mentioned it, and I assume that you know this, but however), you must be careful with pinned releases from many sides.

In general, if you have two bars connected in a node, you must not release it from both sides because you'll make that node "free in space".

Releases on one side is the right thing to do (and enough).

Examples below:

This is OK:

This is not OK: