At what thickness do you transition from shear panel element to a plate element? 1

[Kristina Sornikova]

Dear All,

I try to find at what skin or frame or bulkhead thickness (typical CSHEAR thickness 0.016") do you start transitioning from a CSHEAR to CQUAD4? Is there a rule of thumb or formula?

 

[rb1957]

in day's world, why would anyone use CSHEAR ? except in situation where a plate element is clearly inappropriate.

sure you can use CSHEAR for beam webs and lump axial capability in the caps, but this is very olde fashioned (and I should know !).

Sure you can use CSHEAR for light, unpressurised skins, and lump the axial capability with the stringers, but why ?

another day in paradise, or is paradise one day closer ?

 

[Kristina Sornikova]

Thank you much for reply, so is it OK to use CQUAD4 instead of CSHEAR? It is for a global loads model.

 

[rb1957]

yes

another day in paradise, or is paradise one day closer ?

 

[Kristina Sornikova]

Thank you again, but is it not important to use MID2 and MID3 to none?

 

[rb1957]

yes …
"The most common use of the PSHELL entry is to model an isotropic thin plate. The preferred method to define an isotropic plate is to enter the same MAT1 ID for the membrane properties (MID1) and bending properties (MID2) only and leave the other fields blank. For a thick plate, you may also wish to enter an MAT1 ID for the transverse shear (MID3). You can also use PSHELL to model anisotropic plates."

another day in paradise, or is paradise one day closer ?

 

[Kristina Sornikova]

Thank you, but I ask to model CSHEAR type behavior with PSHELL by not using MID2 and MID3? Will it make fatal error?

 

[rb1957]

? I see nothing gained by modelling with CQUADs only to trick them to be CSHEAR by their property and I don't see how to do this. If you want shear only behaviour, use CSHEAR.

The difference between CQUAD and CSHEAR is that CQUAD reacts shear and membrane (and plate bending) loads; whereas a CSHEAR reacts only shear loads. Ie a CSHEAR is a CQUAD with the in-plane loads removed (and only the shear stiffness remaining. Now CSHEARs require 4 axial elements along the perimeter, in order for them to work; so you add the axial stiffness of the panel (usually taken as effective width = 30t) to whatever structure members bound the panel (like stringers).

Modelling this way is very olde school.

So returning to your original question, there is no thickness reason to transition from CSHEAR to CQUAD, it is a modelling approach (to choose to use CSHEAR in the first place). I would recommend using CQUAD everywhere, but this probably is a decision above your paygrade and my opinion is of little importance. I suspect you're working to some modelling guidelines that say "use CSHEAR for webs", and if the guidelines don't limit the thickness (I suspect they don't, otherwise you wouldn't ask the question !) then use CSHEAR for webs !?.

Possibly a logical transition is when the panel is fully effective, W = 2*30t … a 3" wide panel becomes fully effective at 0.05"; here W is the short side of the panel. Transitioning between CSHEAR and CQUAD is tricky in that you have to change the effective area of the perimeter members (dropping the web effective area where you use CQUAD).

another day in paradise, or is paradise one day closer ?

 

[Kristina Sornikova]

Thank you so much for detail response. So if I model with CQUAD4, is there a good way to model stiffener and joint without using fastener elements like CBUSH or CBAR etc.?

 

[rb1957]

my apologies (I'm too old and too blunt) but good grief ! I would Never model rivets or fasteners connecting pieces (like webs and stringers) together. It's probably a sign of the times … that model size is now "limitless" so "infinite" detail is modelled. The reale world permits load redistribution along a line of rivets so that the average load is an adequate representation of the load applied … F is the load sheared by n rivets … rivet load is F/n.

The pieces of the structure share common nodes (yes, infinitely stiff connections). I won't model rivets with co-incident nodes and CBUSH or other. But you may have to follow whatever modelling guidelines you've been given.

another day in paradise, or is paradise one day closer ?

 

[Kristina Sornikova]

Thank you so much for responses... so the key is common nodes.... I understand now...

 

[rb1957]

NP.

I'd recommend getting some texts on FEA, possibly some practical guides on building models.

another day in paradise, or is paradise one day closer ?

 

[Kristina Sornikova]

Thank you much, I have good experience with FEA, but can always improve.

So with common nodes, you model offset of the stiffener cross section CG connected to skin? or you make stringer inline with skin element edge?

 

[rb1957]

you can use element offsets, but in a fuselage, placing the stringer in the skin plane (using the skin nodes without offsets) isn't a significant "thing". The bigger discussion point is whether to model the stringers with CROD or CBEAM elements.

In a wing, with the much smaller section, it is much more important to model the stringer CG correctly (ie to use offsets).

another day in paradise, or is paradise one day closer ?

 

[Kristina Sornikova]

I understand better now, thank you very much..

 

[Kristina Sornikova]

One more question, how you extract shear flows from all this PSHELL model on a curved skin panel?

 

[rb1957]

there are many ways to post-process the FEM output.

"Most" elements give you shear flow directly.

You can get there by difference in Grid Point Force Balance, comparing the two nodes of the CQUAD along the edge you're interested in.

Your office should have their own method for doing this, or a method that your supervisor approves of.

another day in paradise, or is paradise one day closer ?