determination of honeycomb sandwich Directional properties from test results

[trish129]

Hello,

I am performing a bending test as per ASTM D7250 of a honeycomb sandwich panel comprising the following :

Face Skin
Material: carbon fiber epoxyPREPREG
Uni directional
1 skin consists of

3 plies of 0 degree

3 plies of 45 degree

3 plies of -45 degree

3 plies of 90

i.e ,12 plies layed up in 1 skin.both skins of nearly same thicknes and same material


Quasi -isotropic configuration


Aluminium Core
thicknes: 1 inch

The tests would provide me the following properties:
Core Shear Modulus,G
Flexural Stiffness,D
Transverse Shear rigidity,U

panel orientation in test: flat-wise
panel length along the core-ribbon direction



1.what information do these three properties convey about a particular sandwich PREPARED by given fabrication process(oven,autoclaceve)?

2.how can these properties be incorporated in the finite element analysis of this particular configuration? that is,how can these properties be RESOLVED into the directional properties for inputs in the FEA program?

thank you,​

 

[RPstress]

The shear stiffness measured will be for just the core 'L' direction. Any use of shell or solid elements will probably need both L and T direction properties. For metallic honeycomb the L-direction shear modulus is between 2 and 3 times the T-direction.

If the skins are co-cured then they will be dimpled somewhat (also called telegraphing) even if a caul plate is used. Telegraghing will reduce the skin's in-plane modulus a bit compared with skins that are pre-cured (which should be pretty flat). This will slightly increase both in-plane and flexural deformation.

If skins are co-cured then the in-autoclave process will produce more dimpling than vacuum-only production. Vacuum-only will probably develop between 9 and 14 psi, and an autoclave pressure somewhere between 30 and 40 psi will probably be used, as long as the core is at least 3 lb/ft³ or 48 kg/m³; a lighter core will need a lower pressure to avoid risking core-crush during cure.

The core shear modulus can be used to calculate shear stiffness of plates in the core L-direction with different core thicknesses from the 1 inch of the test. Strictly, total shear deformation includes shear deformation of both the core and the skin, but shear deformation of the skin is small and often disregarded (the shear distribution goes to zero at the free surface and is pretty low at the skin-core interface).

The flexural stiffness is the bending stiffness which governs the curvature that a plate or beam takes up for a given moment. As a first approximation it depends only on in-plane skin stiffness and skin separation. It is proportional to skin separation squared and directly to skin thickness (strictly it is proportional to EI, which includes terms for both skin thickness cubed (skin bending) and skin separation squared, but skin thickness cubed makes only a small contribution).

The transverse shear rigidity is the amount of deformation in shear for a given transverse load. As such it is largely determined by the core shear properties, although both the skin shear and core shear should be included if high accuracy is needed.

Any finite element system will have facilities for defining skin stiffness separately from shear stiffness in arbitrary plate directions. The plate coordinate system should be chosen to have its shear directions in the core L and T directions. Skin X and Y directions should be selected to be in directions relevant to the skins (typically X is the direction of 0° skin plies). It is necessary to define skin X- and Y-directions separately from core L- and T-directions.

 

[trish129]

Many thanks Rpstress...quite an insight...

Given the fact that i am new to the world of composites and learning the ropes of the trade,especially its mechanics, i would like to know more...

1)curing pre-preg plies(12) in autoclave,making them into 2 skins and bonding them to the upper and lower surface of the honeycomb VIA an adhesive called cold bonding or co-bonding?

2)i fail to understand the FINE difference between core shear modulus and transverse shear rigidity for a given sandwich panel..what is the physical significance of these? do these hold for the skins or core or the whole fabricated panel itself?

3)in bending test,the applied load is out-of plane i.e transverse to the sandwich beam longitudinal axis(L core ribbon direction in my case).i intend to orient the panel FLATWISE.does it run counter to ASTM D7250?

4)ASTM D7250 suggests two methods(1) when the facing modulus is known (2) when the facing modulus is NOT known...i intend to go with the second one,though i can test the CURED face skins YET i turn a blind eye to it and want to go on with the second one so that i test the panel independent of the tests used to determine facing properties.and for this ,the standard D7250 suggests performing a short span test(c393) and a long span test(d7249).for the long span the d7249 recommends a specimen size of 24 inch length and 3 in width and a 22 in support span.and for the short beam test it recommends 8 inch specimn length and 6 inch support span ..

i am going by this.just wanted to know IF,by mistake,i happen to cut a panel of length 540 mm for the long beam test.how would it affect my test results?

5)if it is known that the major loading on the structure is going to be that of bending loads,what would these values(D,U,G),flexural stiffness G in particular,tell for panels processed by different methods(out of autoclave,autoclave and cold bonding)? what would the comparison of these values suggest about these processed?

6)i go back to my main question..after getting the values for D,U,G from the bending test D7250...how do i get the values for input in FEA program? because the properties i get would be REPRESENTATIVE of the whole sandwich panel is there any way to resolve them into the directional properties for the FEA?

 

[RPstress]

1) Cocuring vs. cobonding vs. secondary bonding. CMH-17 vol. 3 chapter 5.7.8 (it was 2.7.8 in MIL-HDBK-17) says quite clearly that a cobond is when adhesive is present and one of the substrates is also being cured. The language is not as clear as it might be but it also says that in a cobond one of the substrates is precured or metallic.

I thought that curing the prepreg at the same time as effecting the bond to a core could be termed cocuring or cobonding but CMH-17 disagrees. ASM handbook 21 isn't much better but strongly implies that a cobond has adhesive as well as using the matrix of one or both of the parent materials to make the bond. One or both of the parent substrates must be cured at the same time because if just the adhesive is being cured it's a secondary bond.

I think it is best to be clear about what parts are being cured and whether adhesive is present even if it takes a few more words. Annoyingly I've been inclined to use the term cobond if one parent is being cured and this is being used to bond it to something else without any extra glue in the joint, but CMH-17 and ASM HDBK 21 would disagree with that usage. I should probably change my habits.

Cocure seems to be when more than one substrate is being cured at once. Cobond is when there's adhesive in there and one of the substrates being bonded is also being cured. I may not like it but that seems to be what our most authoritative references say.

Aside: if a prepreg has enough resin in it to bond to a honeycomb (without addition of any extra glue) then it is sometimes termed 'self-adhesive.'

A cold-bond is not a term I use in my industry (almost all bonds are elevated temp), but might be when no extra heat is used. We would usually term this a 'room temperature' bond and more often a 'room temperature-set bond' or 'room temperature cure'. It would be independent of any cocuring or cobonding of the materials concerned.

2) According to ASTM D7250 the transverse shear rigidity is the amount of deformation for a given transverse load and is measured by force times lever arm (force times span) divided by deflection. As such it is not a material property but a property of the structure being examined (the beam). It is measured as a force (N or lb), since it is a moment divided by a deflection. The core shear modulus is N/mm² or psi and is the usual shear modulus material property.

3)"in bending test,the applied load is out-of plane i.e transverse to the sandwich beam longitudinal axis(L core ribbon direction in my case).i intend to orient the panel FLATWISE.does it run counter to ASTM D7250?"

I find that I do not know what you mean by orienting the panel flatwise. Are you intending to apply load in the plane of the skins? See attached sketch query.

There still three queries you have but I must put off attempting to address these for a while. This may complicate matters... Maybe others can help you out (maybe they already have).

 

[RPstress]

D'oh! The 'query' sketch attached should have been a .jpg.

 

[trish129]

thank you Rpstres for your replies..

i am taking the x and y axes as the IN-PLANE axis..i.e the horizontal plane of the sandwich panel..and Z as the transversal(along the cell axis) the vertical direction,the core thickness..

taking the orthogonal material properties approach in defining the properties in ANSYS APDL.

and what exactly will be the properties?the values?

I AM STUCK UP HERE.how do the obtained properties from bending test help in inputting data for the FEA of the same sandwich panel?
For e.g. i need the values of elastic modulus in Ex,Ey,Ez directions,values of Poission ratio Pxy,Pyz,Pxz,values of angular deformations Gxy,Gyz,Gxz for the CURED pre-preg face skins and the aluminium core for FEA of the sandwich panel in ANSYS APDL...how do i get these values?

the pressing issue at this point is to Obtain the 9 properties for the orthogonal material analysis in ansys apdl...9 properties for the cured pre-preg face skins and 9 For the core...how do i get these values?..i have the core manufacturer data sheet in which it called out the coupon test results and the carbon fiber data manufacturer data sheet and epoxy resin data sheet..how do i correlate the data given in these sheets with the required 9 properties for ansys apdl..?

 

[RPstress]

Since you have the core manufacturer data you have two values of shear modulus and a compression modulus (and you should also have the corresponding strengths). On the face of it you shouldn't need to do the testing you've been suggesting. Are you planning on using orthotropic (or maybe anisotropic) 3D elements for the core and orthotropic 2D elements for the skins? If so then the core L and T shear moduli and through-thickness E are all you need for 3D elements. If they are anisotropic elements the other moduli can be made zero (it may pay to put in a very small value for them, about 0.1 MPa or 10 psi or similar). The Poisson's values for the core should be zero and it should be possible to use zero (if not then try very small values). The shell elements will need some test data for Ex, Ey and Gxy or a manufacturer's data sheet should be good enough if it's available.

There is no simple way to model a sandwich panel with just 3D elements and I wouldn't advise trying. (I'm not familiar with any facilities in Ansys that might make it easier than I think. I last used Ansys fairly briefly almost twenty years ago; I think it was version 5.2.) Using 3D elements for the core and 2D for the skins should give you all the stresses you need, including pretty comprehensive through-thickness stresses (variation in through-thickness through the skins won't be available).

A 2D element can be easily used to represent a sandwich. Such shell elements should have options for modeling each skin and the core as layers (and you could model the layers in each skin for laminated skins—a lot depends on what exactly data you have for the skins). You cannot model any through-thickness properties with shell elements (though usually there is a facility to enter through-thickness shear stiffnesses; what you won't get is any indication of through thickness tension/compression in curved areas).

Unless you are using 3D elements for the skins you do not need Ez for a skin. However, if you do then it is usual to find an Ey value for unidirectional material and use that across-the-fibres value as Ez. It is always very close and is frequently used because it is quite hard to measure E3 and Ez experimentally. (If you have only got woven material data then find data for a similar but unidirectional material—as long as it's documented and a plan is made to acquire material and do a test on the correct configuration it is acceptable to do this.) Similarly Poisson's values can be approximated if basic skin material properties are known (nu21 = nu12*E2/E1, nu13 = nu12, nu23 = nu12, nu31 = nu32 = nu21). For shears take G12 and say G13 = G23 = G12. There should be micromechanics approximations available which can be used too. Look for micromechanics at

http://www.sti.nasa.gov/.

These sorts of approximation should be readily available in referenceable sources such as Barbero, Kaw, Jones and quite possibly CMH-17 (I'm not entirely sure what's in CMH-17, which I really should be—mea culpa).

If you told us a bit about what you're actually trying to do it might make it easier to help.

 

[trish129]

thank you for your replies Rpstress...

i yearn to have your email address..can i have it..

 

[RPstress]

You're welcome to contact me directly if there's anything you don't want in Eng-Tips (but there may be considerable delay in replying, of course).

Eng-Tips moderators disapprove of e-mail addresses in 'clear' (they have explained in the past that trolls and recruiters pester people too much). This is a simple encoding using rot13() that they have approved of in the past. The website

http://online-calculators.appspot.com/rot13/

seems to work ok for rot13 (it seems to be faulty for rot47 though, as 'encrypting' then 'decrypting' doesn't give the same as the original).

My simple rot13'd e-address is "evpuneq.creel@txanrebfcnpr.pbz" (strip off the inverted commas). rot13 doesn't conceal the .s and @ which rot47 would but the encryption is so light that it makes little difference (it feels a bit stupid using the term encryption and it can only defeat simple automated trolling but I doubt if I'll be irritated by hordes of recruiters).