Filter choppy crimpy buckling modes with MaxMin(rotations)

[Wrench88]

I work on both solid laminate and cored panels and when I am running sol105 buckling runs, I will get hundreds to thousands of the short choppy crimping modes that are a result of low G13,G23 stiffness of the foam core with real modes interspersed throughout. I typically look for at least 4 nodes per 1/2 sinewave for it to be real vs choppy modes and by choppy I mean where each adjacent node is the max/min peak of the sinewave. Clicking through each mode is very laborious.

I had the idea of using total strain energy output to f06 to see if there was a pattern to which real modes vs crimpy modes could be filtered. This worked reasonably well but not well enough (pic included which shows the first 200 modes on x axis, y is total strain energy, red circles are crimping modes, green are real modes). real modes typically appear with low strain energy, however this is not a bullet proof method. I had the idea to output the max nodal magnitude rotations for the top 10 nodes per mode. My thinking is the short crimpy modes should have very very high nodal rotations as each adjacent node will be the peak and minimum of the sinewave. Whereas with a real buckling mode, there should be smooth waves with reasonably small nodal rotations.

My problem is I can only get MaxMin to output to the .f06 for the static case of the 105 solution, not for each mode. These are the two lines I added to the Global Requests and conditions start text (case control) in FEMAP.

"ESE(PRINT, THREASH=100.) = ALL" <--- this works and outputs total element strain energy for each mode
"MAXMIN(VMAG=10,CID=0,DISP)=ALL" <-- I can only get this to outputs for the static

Anyone know why I can't get MAXMIN() to output displacements for each mode? ORRRR have a different way to solve my overall problem of finding the real modes easy and efficiently

note: I am covering the crimping modes with hand calcs.

 

[SWComposites]

Maybe the solution is to use a coarser mesh to try to suppress the shear crimping modes.
Can you show the model?
How are you modelling the sandwich panel? shell elements for entire sandwich? shells for the facesheets and solids for the core?

 

[Wrench88]

Can't post the model, but lets say it's a ~75" radius barrel, where there are 6 locations on the barrel where 180 klbf of load in introduced at each of the 6 location in direction of the barrel centerline. So load is very concentrated before it fans out into the rest of the structure. I'm using CQUAD4 elements with laminate definitions so facesheets and core are all defined for each element. As it is, I have demeshed into as course of elements as I think I can get away with. That did help a significant amount to reduce those modes but the overall problem still remains.

 

[SWComposites]

then try increasing the transverse shear stiffness values a bunch for the buckling analysis to suppress the shear crimping modes. should not have much effect on the overall flexure stiffness. (you could verify this with a sol 101 run).

 

[tim1958]

The problem might be numerical, but since shear crimping is a real failure mode, it is possible that the core you are using has insufficient stiffness for the application. As SW said, try increasing the value of the core shear stiffness until these modes disappear.

It sounds like the problem might be the concentrated load points however. The waves might start there are spread over the entire shell. This would be an indication that you need to introduce the loads into the shell more gradually. Either a discrete fitting or a local beef up of the facesheets and/or locally replacing the foam core with a plug of higher density material might solve the problem. As a quick check, replace the core with facesheet material all the way through keeping the same total thickness. Perhaps over a rectangular patch (say 2" wide x 5" long) at each load point.