Use Non-Structural Core In HyperSizer

[Jason_the_CA_surfer]

Does anyone know how to use nonstructural core in HyperSizer. I am trying to analyze a stuffed hat (monolithic panel with strips of sandwich panel stiffeners) in hypersizer. The intent is that the core is non structural, it is only there for manufacturing purposes and cannot be considered for damage tolerance reasons.

What I have done thus far is create a core material that has orders of magnitude less stiffness for a nastran run, generated pcomps that reference this material, and run with a model that uses shell elements that reference this pcomp. Nastran seems to run fine with this core. I then import it into HyperSizer.

In HyperSizer I created a new material for this non structural core; I did this by copying a traditional aluminum honeycomb core and then dividing all stiffness and stress allowables by 100. When I imported the model imported fine, it picked up the new core material. The problem comes when I run the MS check in HyperSizer. HyperSizer is applying load to the core and giving very negative margins. Somehow load is getting appllied to the core and the very low stress allowables are being referenced, causing these negative margins.

What I want is for all the shear load to be analyzed as transverse shear in the top and bottom facesheets. How can this be acheived.

Thanks,

Jason J
Goodrich Stress Engineer

[Jason_the_CA_surfer]

I have found the following on core imports in HyperSizer. In order for hypersizer to recognize a material at import as a core it must satisfy the following:

HyperSizer looks at the thickness and the properties of the ply to determine if it should interpret the ply as a honeycomb.  The rules that HyperSizer uses are:

1.    This is not the last ply in the PCOMP

2.    The thickness of this ply is greater than 10 times the thickness of the previous ply (i.e. if the ply is ply 5, then it says if thickness_5 > 10*thickness_4, then this is true)

3.    The transverse shear stiffness (G1z) of the material is greater than the in-plane stiffness, E1 of the material.  This should never happen for a composite ply material but is very common for a honeycomb which has very little in-plane stiffness.

Jason J
Goodrich A350 Stress

[Phil]

Does anyone know how to use nonstructural core in HyperSizer. I am trying to analyze a stuffed hat (monolithic panel with strips of sandwich panel stiffeners) in hypersizer. The intent is that the core is non structural, it is only there for manufacturing purposes and cannot be considered for damage tolerance reasons.

What I have done thus far is create a core material that has orders of magnitude less stiffness for a nastran run, generated pcomps that reference this material, and run with a model that uses shell elements that reference this pcomp. Nastran seems to run fine with this core. I then import it into HyperSizer.

In HyperSizer I created a new material for this non structural core; I did this by copying a traditional aluminum honeycomb core and then dividing all stiffness and stress allowables by 100. When I imported the model imported fine, it picked up the new core material. The problem comes when I run the MS check in HyperSizer. HyperSizer is applying load to the core and giving very negative margins. Somehow load is getting appllied to the core and the very low stress allowables are being referenced, causing these negative margins.

What I want is for all the shear load to be analyzed as transverse shear in the top and bottom facesheets. How can this be acheived.

Thanks,

Jason J
Goodrich Stress Engineer

Jason,

Is this truly a hat concept (with a web and a crown?)  Or is this a honeycomb where there are facesheet separators in for manufacturing but don't actual contribute to the structural response?  If I understand correctly,  you are looking for a capability where the facesheets are uncoupled and the core material has no strength.

There are two ways I can think of doing this:
1) you can turn off all core analyses on the failure tab.  This way, HyperSizer will ignore the core completely when doing failure analysis.    The core will not contribute stiffness for in-plane loads at all.   It will only contribute to bending stiffness in that it keeps the facesheets separated, therefore increasing the D matrix contribution  of the facesheets.

There is no physical way to keep the core from taking transverse shear load.  If the sandwich is loaded in transverse shear, then the transverse shear stress in the core must be the peak transverse shear stress.  This stress value is approximately Qx / thickness and is independent of the core material.  It must be this way to transfer shear from the bottom to the top facesheet.

2) If the core is truly non-structural and cannot take load at all, you could model this in the FEA with two separate CQUAD elements and then in HyperSizer just treat these two CQUADs as unstiffened laminates.  This means that these two elements are completely uncoupled both in HyperSizer and in the FEA.  In this case, any transverse shear is picked up in the two facesheets and none is transferred across the core.

It sounds like to me option two is what you are looking for.

Please let me know if this sounds correct.

[Phil]

I have found the following on core imports in HyperSizer. In order for hypersizer to recognize a material at import as a core it must satisfy the following:

HyperSizer looks at the thickness and the properties of the ply to determine if it should interpret the ply as a honeycomb.  The rules that HyperSizer uses are:

1.    This is not the last ply in the PCOMP

2.    The thickness of this ply is greater than 10 times the thickness of the previous ply (i.e. if the ply is ply 5, then it says if thickness_5 > 10*thickness_4, then this is true)

3.    The transverse shear stiffness (G1z) of the material is greater than the in-plane stiffness, E1 of the material.  This should never happen for a composite ply material but is very common for a honeycomb which has very little in-plane stiffness.

Jason J
Goodrich A350 Stress

Jason, these rules are exactly correct.

Phil

[Jason_the_CA_surfer]

Hi Phil,

Thank you for the response. The physical part I am analyzing is a stuffed hat panel (but for a damage tolerant requirement the core must not be considered structural, so for analytically I am analyzing a "hollow hat" or some call it a "skin stringer" configuration).

Ideally I would like to build a new model with the sidwalls of the hats modeled and top of the hat as there own quads; unfortunately all I have at this time is a sandwich panel model (quads on a midsurface) because the design change actually came during my PDR stress analysis. So I was looking for a quick and dirty way to transfer all the transverse shear than normally would be applied to the core to the facesheets. For now I have just turned off the core check, which will suffice for PDR analysis. I can just do a hand calc to superimpose the shear to the interlaminar shear M.S. check for now.

Thanks again,

Jason Joel

[Phil]

We have skin-stringer concepts in HyperSizer natively.  You can size and analyze a hat very effectively in HyperSizer.  We don't have a filled hat, but since your core is non-structural, that would not be a concern.  You could use the same shell model to size the hats, you don't need to model the stiffeners discretly.  Go to the Basic user manual and look at the first tutorial example for how to build a hat section in HyperSizer.   

One major concern I have is that the load paths for a skin-stringer structure will be very different than they will be for a honeycomb sandwich.  The A11/A22 ratio will be high, but the D11/D22 ratio will be very high for the skin-stringer, maybe as much as 20-1.  The panel essentially has no bending stiffness in the transverse direction.  If you use FEA loads that were developed using honeycomb panels which are roughly the same stiffness in the two directions, then the hat would have to size up tremendously to take the high transverse loads.

In order to size a skin-stringer properly, you really need to iterate the FEM and put the skin-stringer ABD matrix into the FEM so that the load paths re-distribute.  Basically, your Y direction loads, Ny, My and Qy should drop way off when you go to a skin-stringer design and all loads will be primarly carried in the X direction.

Phil