Max Principal Stress / Vector / Bending / Torsion

[sartorbjk]

Hello,
I have an orthotropic fiber reinforced CMC. I have designed a T shaped sample in order to see if I can predict material failure under complex loading conditions (since the actual part which is in service is subjected to complex loading conditions). As can be seen from the overall part of the picture, I am applying a force to the sample.

I am totally aware of the fact that it will fail at the notch (and I already did the tests and it failed due to the notch, the actual geometry was not like as the attached picture, however, during production of the samples, some mistakes were done, so I had to change the geometry again, there supposed to be no notch).

Now, ignoring the notch,

I am looking at my maximum principle stress(MPS) and its vector.
For instance, for this analysis, my vector information is (at MPS location)
Node:43949 Stress:167.11
Components:
x:0.99031
y:0.13888
z:0.00044
My first question is:
1)
I am trying to get maximum tensile (x) and maximum shear(MS)(xy) possible at the same location in order to form the complex load. Maximum shear is concentrated on another place. If I look at the node where MS is, maximum principle stress is lower, however, y component of the MPS is higher (like x:0.6 and y=0.3). Does that mean, y component of the MPS can be related to shear stress? As it increases, the contribution of the shear stress will increase (which will lead to a failure).

My second question is:
2)Taking normal x(which makes a sample to fail under normal bending) and shear xy stresses into account, is it a good way to make a ratio from them to determine a complex loading state?

I hope I made my question clear.
I am looking forward to hearing from you.

Best Regards,
Sartor

 

[rb1957]

how are you accounting for the fibers in your material ?

is the resin brittle ? (ie, are linear FEA results valid ?)

if you've tested this, how did it break ? (the resin yielded, and eventually the fibers snapped)

if you've got "complex" loading why not apply it to the FEA ?

finally, it's principAL (and yes, i am an old foogie)

 

[corus]

rb1957, it's actually an old fogey, not an old foogie

http://en.wiktionary.org/wiki/old_fogey

and yes, it takes one to know one.

Tata

 

[EdR]

It seems clear to me that you do not understand what you are doing for the following reasons:

1. Stress is NOT a vector but rather is a tensor. Stresses cannot be manipulated using the rules of vector math....

2. Maximum principal stresses occur at points of zero shear so trying to combine them with maximum shear stresses directly makes no sense. You cannot relate the components of principal stress to the shear stress directly. (must use various stress relationships).

3. You cannot use a ratio of normal and shear stresses to determine a complex loading state.

Suggest you review a mechanics of material text to get a better idea of what you should be doing....

Ed.R.

 

[sartorbjk]

@rb1957:
I have carbon fibers 0°/90°(x/y) and the composite behavior is non-linear.

Fiber breakage/pullout was seen after breakage.
according to linear fea, i could say that the failure might occur at some particular regions. And at those regions, I am looking for Normal X and XY shear. at the same time, I am wondering if I can relate maximum principal stress vector to conclude about normal x and shear xy stresses?

@edr:
1) the direction of the maximum principal stress can be seen as a vector and my question was exactly this: what does this vector coordinates actually mean?

2) I never said i am comparing MPS to MS to find a complex loading case. I am just wondering, what does this vector representation mean and how it is related to shear/normal stresses at a particular point?

3) What I would use was a test/sample, where I could apply considerable amount of torsion to my sample, and than simply make a tensile test. However, this is not possible due to some other problems.

Best Regards,
Sartor

 

[rb1957]

ok ... do you think the FEA knows your composite construction and failure modes ?

if you're using a composite pre-processor to create the material matrix, then that should enable you to decompose the FE results into fiber and resin stresses.

 

[ESPcomposites]

I have to agree with edr in this one, based on this thread and the other. There are many basic things you should review before you can get meaningful help here. I think people here can help when you get stuck on a problem, but it is too difficult if a foundation is not properly set. I think there is also a good recent thread about the use of FEM without a strong foundation. Good luck, you seem interested in understanding it properly.

Brian

http://www.espcomposites.com

 

[sartorbjk]

@rb1957:
at the moment, it does not know!
The failure model that i would like to use has no connection with microscopic details. I need to see if I can already use the implemented failure criteria whether they can represent my material's behavior, or not.
If not, I need to go more for more advanced things to predict material behavior.
The theoretical problem is: behavior of interface between fiber and two types of matrix is not very well understood. I am just wondering if I can skip this part to model material behavior. So far, it seems like I can. Hill's criteria looks like working. and it is already stated that hill does not say anything about failure modes.


Best Regards,
Sartor

 

[rb1957]

ohhh that's a worrying post ...

 

[sartorbjk]

@rb1957:
I totally understand your concern, however, I do not agree. We know our material. Before deciding if we need a code which describes our material, we would like to see what are the capabilities of the things that are already implemented in commercial FE-codes.

Best Regards,
Sartor

 

[rb1957]

you may well know your material, but you have to tell the computer what it is.

if i'm using Al, it is easy to specify ... several built-in models or i can specify the few properties myself (E, mu, ...).

If you're working in composites you have to tell the computer the fiber construction. typically it's a ply lay-up for a face of a surface/panel. typically there's a pre-processor that takes your fiber properties, and orientations and combines them together for the inplane stiffness.

but you're not doing this ?? if not , how are you defining the material property ??

 

[sartorbjk]

I totally agree with your general idea. That is how it is supposed to be done.

"If you're working in composites you have to tell the computer the fiber construction. typically it's a ply lay-up for a face of a surface/panel. typically there's a pre-processor that takes your fiber properties, and orientations and combines them together for the inplane stiffness."
I do not do this.

I am giving orthotropic material properties. I am adding Hill potential and I define a hardening rule (MISO).
and all in-plane properties are very well estimated.

In-plane does not mean everything, though.

Best Regards,
Sartor

 

[rb1957]

i'm just surprised that you have a solid mesh ... i'd've thought that you needed to model through the thickness, in order to model your fiber layup ??

and i don't think you can apply von mises or principal stress failure criteria to composites ... i thought they failed in many different and mysterious ways ...

 

[sartorbjk]

We were also surprised but that was exactly what we wanted to see at the beginning before investing money into some more complicated things.
FE-code does not know anything about our fiber layup.

We are not applying any failure criteria at the moment, although it seems like tsai-wu will work for in-plane failure (again, nothing about failure mechanism but the values are in good agreement). von mises and others, they will not work, for sure.


Best Regards,
Sartor

 

[rb1957]

sorry but what material definition are you usng ??

and why chase max principal stresses if not to use them ??

 

[sartorbjk]

Linear elastic orthotropic, Hills potential, Multilinear isotropic hardening

well, max principal stress, we used to estimate failure location.

Best Regards,
Sartor