Framing set up that works in compression but not in tension 2

[WARose]

I am modeling something where I cannot use tension/compression elements, t/c springs, t/c supports, etc to represent something that I need to offer (virtually) no resistance/stiffness in tension, but definitely some in compression.....so I have been brainstorming to figure out if there is a way to do that with a conventional element set up. Can't think of anything so far. It can be any combination of stick, plate, or whatever elements....just can't be what I talked about earlier with t/c only elements/supports. (They won't do for the type of analysis I am running.)

 

[Celt83]

Manually do the no-tension iterations that the analysis software does:

Run the model with the element in place
check that it is in compression
if the member is in tension delete it from the model and rerun
look at the nodal deflections where the member used to be, if based on the nodal deflections the member would be in compression
Add the member back but find a way to reduce the axial stiffness (mess with the section area to reduce AE/L to something between the real stiffness and 0)
Run the model again
if the member is now in tension, adjust the axial stiffness to be lower

Repeat the above for every element you need to be compression only

 

[Tomfh]

Uh, a compression only element?

 

[rb1957]

CBUSH ?

non-linear material curve ?

"Hoffen wir mal, dass alles gut geht !"
General Paulus, Nov 1942, outside Stalingrad after the launch of Operation Uranus.

 

[WARose]

[blue](Celt83)[/blue]

Manually do the no-tension iterations that the analysis software does:

Run the model with the element in place
check that it is in compression
if the member is in tension delete it from the model and rerun
look at the nodal deflections where the member used to be, if based on the nodal deflections the member would be in compression
Add the member back but find a way to reduce the axial stiffness (mess with the section area to reduce AE/L to something between the real stiffness and 0)
Run the model again
if the member is now in tension, adjust the axial stiffness to be lower

Interesting suggestion Celt83. may try it.

[blue](Tomfh)[/blue]
Uh, a compression only element?

Like I said: can't do it.

[blue](rb1957)[/blue]
CBUSH ?

non-linear material curve ?

Don't know what a "CBUSH" is....can't do the second.

 

[Just Some Nerd]

What would the software be out of curiosity?

----------------------------------------------------------------------

Why yes, I do in fact have no idea what I'm talking about

 

[3DDave]

Is this a one-way ratchet or just a gap that closes?

 

[JLNJ]

How about a stone arch?

 

[TLHS]

You can also do the 'delete the member' thing by assigning the element a very very soft material. If you do that, you still have the element there and can look at whether it's taking a very tiny compression to validate that you're right instead of trying to do math on nodes to figure out what direction things are moving.

Just make sure that none of your very soft elements are taking a bunch of load because they're in a load path that's necessary for stability and that you're not getting crazy overall structure deflections.

 

[TLHS]

You haven't lived until you've tried to approximate a post yield plate analysis by iteratively adjusting material stiffnesses by hand on different areas to eliminate local elastic stress hot spots to ensure you're okay justifying an existing condition.

 

[IFRs]

a boundary condition like a roller?
a contact element?

 

[Tomfh]

What is the “type of analysis” that prevents you from using non-linear elements and materials?

 

[canwesteng]

I would guess some kind of dynamics, lots of software won't allow for non linearity in steady state analysis. If that's the case, I would run it with and without the element, as that probably bounds the real solution.

 

[WARose]

I would guess some kind of dynamics, lots of software won't allow for non linearity in steady state analysis. If that's the case, I would run it with and without the element, as that probably bounds the real solution.

Exactly. And I think that is probably what I am going to be stuck with.