Sandwich Yield Strength Modelling

[flyforever85]

Hi all,
thanks for the previous answers!

I have a new proble mand I'm not sure the approach is correct. I have a sandwich panel layed up as follows:

Inconel 625 (E=29700 ksi, yield 66 ksi thickness 0.013 in)
Steel Honeycomb
Inconel 718 (E=29000 ksi, yield 160 ksi thickess 0.010 in)

(I modelled this as a composite - I use Patran and Nastran)

I was asked to find the what would be the thickness of 625 at the bottom (replacing 718) to obtain the same strength.

This was my approach: I calculated the load until I reached yield (3500 lbf). Then, I kept the load constant and I started increasing the thickness of the bottom layer until I reached 66 ksi (yield of 625). In this way though, I got a thickness of 0.3 in! I don't think this is the right approach.

Thoughts??

Thanks!

 

[FEA way]

So your failure criterion is skin yielding, right ? And you only check stresses in the skins, not in the core ? Then for the first part you should compare stresses with strength of weaker skin (625) because it will fail first. Then if you replace stronger skin (718) with 625, nothing will change because that unchanged (top) 625 skin will still fail first. Assuming similar stress levels in both skins, of course. What are your loading conditions ?

 

[flyforever85]

Thanks for you reply, FEA way. I have a concentrated load at the center of the plate, and the plate is simply supported. In real scenario, the load would be dynamic (a weight dropped) but I'm modelling only the static case for simplicity. I started recently this job and they have a sample of this structure that failed at the bottom (the BC though were way different in the real case). They want to change the thickness and material at the bottom for other reasons.

 

[FEA way]

So it’s typical bending test. Stress willbe highest at the top and at the bottom in this case. Thus both skins will experience very similar level of stress and my considerations from previous post should apply. Focus on the weaker skin or choose another failure criterion.

 

[flyforever85]

Well, the bottom skin will fail first because there is a stretching component that top top skin doesn't have (both skins are glued to the honeycomb).

I tried with only the panel (without honeycomb) but that problem gets more complicated because with such a low thickness, there is a large deformation of the plate and the problem becomes non-linear.

Also, Nastran doesn't distinguish what happens to the top and bottom skin since it's a 2D surface.

Thoughts?

 

[FEA way]

Yes, adhesive bond strength is another factor to consider but in order to evaluate its failure you would have to perform an analysis using cohesive zone approach.

Could you show how your model looks like ? Can you replace that honeycomb with homogenized solid equivalent ?

 

[flyforever85]

As I said, the model is very simple, the sandwich is modeled with a 2D surface only. The honeycomb is actually a material with the same E11,22 and 33 of the honeycomb

 

[flyforever85]

pic1

 

[flyforever85]

pic 2

 

[FEA way]

Ok, so it's actually a composite section shell with point load. I suggest trying other methods of modeling sandwich composites to verify results such as continuum shell (solid shell) and, especially, regular solid elements (be careful with point load applied on solid elements - stress singularity may influence your results). Traditional shell elements often produce inaccurate output.

 

[flyforever85]

Ok, I will try different option. Would my approach in terms of finding the right thickness be valid? I mean, would you change the thickness until you find the yield strength of 625?

Thanks for all the feedback!

 

[FEA way]

Yes, just make sure that you read stresses from appropriate layer.

 

[SWComposites]

Why do you need a FEM to do this calc?

 

[flyforever85]

I've never dealt with honeycomb until now and I thought it was easier via FE modelling. If you have equations that can help me, please share!

 

[SWComposites]

See Composite Materials Handbook-17 Volume 6 and/or ASTM D7249.