Honeycomb Sandwich Close Outs Modeling (FEA/Nastran) 1

[Burner2k]

Hi folks,
I am trying to model a honeycomb panel of a fuel tank subjected to internal pressure. We use tapered or ramp close outs for Honeycomb sandwiches and had a few questions regarding the same. I am using Patran/Nastran version 2010.

To start with, I did represent the end doublers followed by ramp and panel section with max core thickness. When I did a simple 2D panel under the influence of uniform pressure with all sides simply supported, the results are way off from hand calculated values.

To provide some more info, I used 2D representation of Honeycomb Panel. The panel I am trying to model has Aluminum face sheets (2024) with Aluminum Core (5052).

When I modeled the above without the closeouts in FEA, the results are matching very closely to hand calc values.

So, is it acceptable to not include close outs in FEA model? If its necessary, any inputs on how better I can represent the same in FEA apart from the above procedure?

My employer is urging me to find out the best practices regarding the above. I am receiving very little help here.

Feel free to request further info if the above does not make sense.

Thanks a lot in advance folks,
- B

 

[KSshopper]

I don't think there is any industry best practice. I've worked for three separate companies and seen it done three different ways:

1. Model the entire composite bondment with shell elements (PCOMPS). Model the core ramp area with average (1/2) height core. Use a hand analysis to calculate:
a. strain in the ramp ply,
b. flatwise tension and compression at the ramp transitions
c. shear at the height (thickest portion) of the core. Note that core shear is not checked at the bottom of the ramp near the laminate. Otherwise you would never show it good.

2. Model the core with solid elements and the facesheets with shell elements. The core stresses and ply strains can be read directly from the model. NASTRAN is pretty good for this since the solid elements have features that avoid shear locking in thin solids. This method is great unless you need to make design changes that blows away the geometry on your carefully crafted FEM.

3. Model with shell elements and ignore stresses and strains in the transition. If the core shears and laminate strains are acceptable at the boundaries of the transition, they will be acceptable in the ramp as long as you respect the design rules (ply drops, core ramp angles, etc).

I like item 3. Items 1 and 2 tend to show low margins at the ramps despite never seeing a panel fail like that. It's a big paradigm shift for me since core ramps were big deals at the last place I worked.

 

[SWComposites]

But the hand calc values are probably for a uniform thickness plate - so why would you expect correlation vs a ramped panel?

 

[Burner2k]

KSshopper,
Many thanks for the reply.

I did try method 1 using PCOMP (using an average height for the ramp) along with representing the end doublers as well as the laminate with max core thickness for a simply supported panel with uniform pressure. The peak deflection values were off from hand calculated one. Same case for face sheets stresses.

>>3. Model with shell elements and ignore stresses and strains in the transition. If the core shears and laminate strains are acceptable at the boundaries of the transition, they will be acceptable in the ramp as long as you respect the design rules (ply drops, core ramp angles, etc).

If I am interpreting the above correctly, you are saying not to include ramp (& doublers) in the FE model. Is this correct? Also, do you know of any source where I can find design rules for transition & end doublers? Niu has given some information in Chapter 7 in his Stress Sizing book, but the design curves he has presented are for Composite face sheets & core. In my case, both are metallic. I can adopt the same procedure to verify the design, but wanted to find out if a source is available for metallic as well.

Thanks & regards,
- Vishak

 

[KSshopper]

Clarification on item 3, you represent the ramp in the model as you would in item 1, just evaluate the stresses at the boundary of the edgeband and full height core. It's simple, but you should have some test data backing it up.

Strange that your hand calcs differ. That's definitely something to resolve first. I would work a simple problem first and slowly make it more complex. This is pretty easy to do with PCOMPs.

I would stay away from those design curves. They are specific to a manufacturer's material system and construction. You can use the results of your analysis to size the core height, edgeband, and facesheet thickness. Niu's example is pretty good, but I would use the equations from ASM Handbook Volume 21 - composites to justify the ramp (what he calls tapered area) instead of the design chart.

 

[Burner2k]

KSshopper,
I used the Honeycomb Sandwich Design Technology document from Hexcel (as well as equations from a leading Aero OEM design manual) to calculate deflections, face sheet stresses etc.

The Hexcel document is available here: Honeycomb Sandwich Design Technology (pdf)

I realize that in the above document, the laminate is considered to have uniform thickness through out i.e. right from support to support. I will check ASM Vol. 21 and find out what equations are presented for ramp design. Thanks for pointing that source to me.

Just to make sure that my FEA representation of ramps is correct, I will post screenshots of my PCOMP once I get to work.

Thanks again for your response. Appreciate that.
Regards,
- B

 

[Burner2k]

KSshopper,
Please find the attached pdf file which summarizes the simple analysis I am doing to verify my FEA representation.

I would appreciate your thoughts on how I can improve (or correct) my modeling to get more accurate results.

Do you know where I can find design rules for edgeband (thickness & length determination) and ramp (length)?

Thanks & regards,
-B

 

[Burner2k]

I think we are just going to model the entire honeycomb panel without the ramps for now and use hand calculation to design the edgeband & ramps (check for delamination & core crush depending on the face skin loading).

Kinda not making progress on getting an accurate enough representation of the panel with closeouts in FEA. Employer is not happy

Thanks all for your inputs. May be in future, we will revisit our modeling techniques & hopefully will get better.

Regards,
- B