Peak Load of an Assembly?

[metalman8357]

Hi all,

I'd like to use FEA as a way to predict the maximum download a structural assembly can take before failure. In our lab, we build setups that involve steel beams, connectors, and bolts. We apply a download to an area of the assembly, and we monitor load vs. crosshead displacement. Eventually, the assembly will reach some peak load and buckle and begin to lose load. Is there a way that I can use FEA to predict this peak load? For example, if we test an assembly and the peak load is 50 kips, I'd like to build this model in CAD and test in FEA to determine the theoretical 'peak load'.

I'm currently running Autodesk Simulation Mechanical 2014 through the non-linear mechanical event solver.

Any tricks or ideas to determine this?

Thanks,
M

 

[structSU10]

generally you need to know the buckling mode of the member. once you do that (through an elastic buckling analysis) you can apply an 'initiator' force near the area that buckles, like 10 lb, and run a NL analysis - this should show you the ultimate capacity, roughly.

This is an intensive process, requiring a strong computer and days to weeks of analysis depending on debugging of the model and such. Simple hand calculations may serve a good purpose to get an idea.

 

[metalman8357]

StructSU10,
While that would probably give me an extremely accurate result, I was hoping for something a little quicker in terms of analysis time.

My question was actually a little more basic as I am relatively new to FEA. For example, when I run my assembly now all I'm getting is displacement, von mises stress/strain, or max principal strain in the results. If I apply a load that is 10x what the assembly should take to fail (compared to test results), it will just show an extremely high stress value like 1 million ksi Von Mises stress, which isn't practical.

If I just take a standard steel bar that is 0.25" in diameter and constrain both ends and apply a download (to simulate a 3 point bend test), the bar should bend until some peak load is reached and then it will to continue to yield and not hold this peak load any longer. Is there a way to determine this peak load?

 

[inline6]

you need to use a elasto plastic non linear material. a linear buckling analysis also goes hand in hand and is a good start point since the structure might elastically buckle before it forms plastic hinges or grossly yields/collapses.

 

[metalman8357]

When we test these assemblies in the lab, the steel studs buckle suddenly and the assembly loses load. If I model this situation in FEA, when I apply the corresponding peak load for the test, it shows that the studs have not even started to yield. I think I need to run a linear buckling analysis to modify my simulation, like users have stated. Can someone explain briefly how this works? Autodesk simulation has a linear buckling analysis module and if I input the area of interest into this module, it shows 5 different modes for buckling. What is the difference between these modes? It also gives a buckling factor. Is this what I would use in my regular non-linear model and instead of applying my original load, I multiply this load by the buckling factor? Clearly, I'm new to all of this but I'm a very quick learner and anything helps!

 

[structSU10]

The linear buckling module gives you the eigen-values corresponding to the first 5 elastic buckling modes of the structure. The load factor it gives you is what you multiply the applied load by to achieve the onset of that buckling mode. The first mode you see will happen first - if elastic buckling is pertinent. Note if a member is not in the slenderness range where elastic buckling controls, inelastic buckling may control, so the load factor given by the program may be very high.

The way you might get what you want is to figure out from the buckling mode shape where to apply an initiator load on a non-linear analysis, which will induce a failure when appropriate. This is not a simple type of analysis, and there is no way that I know that will make it fast or easy to do.