aircraft panel: problem and advice

[volaff]

kind users, I have been finding myself for some time studying an aircraft panel made of composite material (fig.1).

of such panel the model fem (fig4) was created in which the various are connected to the main skins (red and blue) trami8te riveting, modeled by beam elements.

in particular the panel has been subjected to a static traction test.
the problem that I cannot solve is linked to the bound parts by fig3 rivets.

by launching static analysis, in fact, there is a discontinuity in red (fig 2 and 5) that we cannot explain (although the "effect" has been multiplied by a factor 20).

the panel has been binding correction for which that draft for us is "anomala".

We initially thought it was a mesh-related problem.
The problem is even more regular and we cannot understand what is due.

have any of you studied similar panels or had similar problems? !

every board is well accepted.

I've been trying for more than five months now, and I've been trying all the time without being able to handle it.

Thank you for your attention.
Good Sunday.

fly

 

[Er Presidente]

kind users, I have been finding myself for some time studying an aircraft panel made of composite material (fig.1).

of such panel the model fem (fig4) was created in which the various are connected to the main skins (red and blue) trami8te riveting, modeled by beam elements.

in particular the panel has been subjected to a static traction test.
the problem that I cannot solve is linked to the bound parts by fig3 rivets.

by launching static analysis, in fact, there is a discontinuity in red (fig 2 and 5) that we cannot explain (although the "effect" has been multiplied by a factor 20).

the panel has been binding correction for which that draft for us is "anomala".

We initially thought it was a mesh-related problem.
The problem is even more regular and we cannot understand what is due.

have any of you studied similar panels or had similar problems? !

every board is well accepted.

I've been trying for more than five months now, and I've been trying all the time without being able to handle it.

Thank you for your attention.
Good Sunday.

fly

Why is the reinforced panel around the windows interrupting before the pillars?

 

[volaff]

in practice does not interrupt.
I'll explain.
If you do it, there's some kind of "frame" all around the panel: That's the part of the test machine that transfers the load to the panel.

the two skins are "under" to this frame, in the sense there is but you do not see

I hope I have not misinterpreted your question

the problem in the overlap zone of the two main skins I have the problem mentioned above.

 

[Er Presidente]

in practice does not interrupt.
I'll explain.
If you do it, there's some kind of "frame" all around the panel: That's the part of the test machine that transfers the load to the panel.

the two skins are "under" to this frame, in the sense there is but you do not see

I hope I have not misinterpreted your question

the problem in the overlap zone of the two main skins I have the problem mentioned above.

I'm late for the area.

 

[volaff]

Actually, there's no interruption in that area.

riding between the upper and lower riveted green part there is another element called stringer which is glued and riveted on the upper skin.

in practice that horizontal gray "striscia" you see is another linking element on the skin.

Thank you for your interest!

 

[Matteo]

hello, would you be able to put some image with higher resolution, possibly with the edges of the visible elements? In particular, I refer to image n°2.

 

[volaff]

Is that okay?

 

[Matteo]

It's better. the ideal would also see the real mesh.

in which direction is static traction applied?

 

[volaff]

Here we go.

Sundayg.png


the traction load is applied along y sil side purple and celestial "horizontal" in fig.4 of the first post.

Actually, it is a force per unit of length.

thank you very much for the interest

 

[Matteo]

I'm sorry but this evening I'm a little "tard"... is the load parallel to the red arrow or the blue arrow? is the load applied on which edges (better if it has an explanatory image)? do you have an image of the movements (not too amplified, otherwise it deforms everything and becomes incomprehensible)?
the "problem" (as long as it is such) is repeated even if you replace the property of the material from laminated to homogeneous (e.g. aluminum alloy? ).

if needed I can evaluate the simulation at equal geometry and loads, but with another solutor (I don't guarantee immediate response, but I can try).

 

[volaff]

hello thanks for the interest.

in practice the situation in question is the following (blue arrow parallel load) attached.

I do not think the problem is the solutor (at least I hope).

Would you please explain what makes you think the problem is related to the solutor type?

I am willing to listen to every advice or suggestion

I did not try to replace the properties of the material from laminated to homogeneous since I never thought about it before sincerely

ps for the pictures of the movements I refer to in the morning, sleep is prevailing on my memebra

Good night and thank you again.

 

[Matteo]

hi, I didn't mean that the problem is the solutor, indeed the opposite: solving with another solutor (I have no ansys available) nothing changes.

back to the problem: I think it's normal that there's a stressful area like that: the connection between red panel and purple panel at that point undergoes a stress of bending and transmits a pair on the red panel.

speculating that the vertical load is applied on the edges in both sides and at the bottom (horizontal): did you correctly bind the vertical edges?

 

[volaff]

Hello Matteo, thank you for the answer in the meantime.

Honestly I think I have correctly bound the vertical and horizontal edges but a control never hurts.

Given the test condition t3 (uni-axial load test along the y direction) it is necessary to block the translation along the z axis and rotation in the plane. the first condition also occurs in reality and translates, in modeling,
with the application of a constraints (carrello) in the z direction to all knots belonging to the outer perimeter of the dummy (purple and celestial zone).

rotation, instead, should not occur as the load is self-balanced, and also seen the symmetric geometry of both the panel and the loads.
to avoid a slight dissimmetry entails undesirable movements that falsify the results adopts a constraint along the y direction, on two knots located in halfway and on the frame of the dummy, and a bond along the x direction to the upper end (or to the lower one) in correspondence of the dummy halfway.

You'd explain why, "I think it's normal that there's a stressful area like that: the connection between red panel and purple panel at that point undergoes a stress of bending and transmits a pair on the red panel"? ? ?

In fact, even in the static test the panel tends to flench but so (fem analysis) seems exaggerated

Thank you so much again for the disposition!

 

[Matteo]

Wait, I'm not saying there's a mistake in your modeling. . I was only looking for possible differences between real model (I believe it is that of the photo) and mathematical model.

first you should confirm my csys:
- normal z to the panel
- y vertical (compared to photos)
- x horizontal (compared to photos)

 

[volaff]

That's the "global."

local references were used to highlight tensions on shell elements (the two skins that make up the panel).

we have searched and searched but from the point of view of the modeling (wins, loads, links) everything seems fine.

I press that the model already had it and I had to change it in some parts but, at a certain point, as you can desuem from the photos (ansys) the two panels seem almost "staccarsi" and the red one breaks (not little).

According to you, could an idea be to increase the stiffness of the connection rivets between the two subpanels (skin)? ? ?

In fact the rigidity of the same was calculated using an experimental report called mc Donnell and douglas: in this way the cutting factor to be inserted between the parameters of the beam4 elements has been calculated.

Thanks for the kindness.

See you soon!

 

[Matteo]

Okay, now I have a clear picture of load and bond conditions.

then if the panels break off as in the image treated by ansys with amplified sports, it is correct that you read that stress.

I wonder: is it so (i.e. the two panels break off) even in the real model or should there be continuity at that point?

from the photo we see that the external metal support is interrupted: is it so? the interruption also concerns the actual panel?

 

[volaff]

Please attach me a photo to see what you're referring to?

Maybe this way?

 

[Matteo]

Please attach me a photo to see what you're referring to?

Maybe this way?

it seems that the red and blue panel are not in continuity of mesh (or in congruence of movements) on the yellow line... that they are not completely.

 

[volaff]

in practice the incongruence of the movements occurs right in that area and until here we are both.

a panel is superimposed (subsequent) compared to the other as you can see from images 1 and 2 attached.

a panel (blue) is found at a height z = 0 while the red one at z=1.4.

in the overlap zone between red and blue there is another element called stringer (in heavenly): I took care to make sure that the area of the stringer falling on the red skin is meshata equal to that of the red skin itself.

such stringer is either riveted or glued on the red skin.
(the elements in violet are rivets and glue elements).

Thank you again!

 

[Matteo]

Could it not be that the red and blue surface is too distant glue elements? It almost seems that the most extreme part of the contact is not "glued" completely. try to put "neck" elements even on the extreme part of the overlap?