mesh density and simulation results

[Damianes]

its about static simulation,

I knew cad simulation is not a real calculation but i not knew it's totally not useful in this way. What i mean :
i have simple bar 60x25 x 300, did three simualltions just with changed mesh density
when mesh max element size is 32 max stress is 40
on max 7mm result is 43 - ok some difference ... better mesh better results i think but:
on 3.8mm i have max stress 49.4

so if i set max element size for lets say 0.1mm my result will be 10000 ?
so the simulation is useful only to see what places of part are the most sensitive ?
is it any possibility to use like estimation of normal paper/excel calculation ?

its something i do not understand here, my knowledge is little here can you explain me what i do not know, what should i learn ?
and can you say more about what is going on here please ?

[bnemec]

CAD FEA is real calculation, just a calculation of a theoretical model. Also, not sure what you mean by "in this way", but if you mean "solve the problem all by itself without needing my understanding or interpretation" then you are correct, it's totally not useful in that usage. Simulation is just a tool

What's going on at the max stress area? What you describe is common with welds. It's just how FEA works. The stress (or whatever is being solved) is typically varying continuously throughout the part so each element is an average of the boundary conditions. Some conditions like welds or other things that cause sudden changes will always have a high stress, make the element smaller and, in my layman understanding, the average stress in that element increases. This is where the theoretical model diverges from physical reality. It can tell you something about it, but not everything, and you must understand material properties and other principles to know how to interpret the sim data. There are other things going on in that region of the physical part that the particular FEA solver is not taking into consideration.

[TienDoVan]

I think you should use global mesh for the entire model and use control mesh for areas where you need to accurately calculate stresses (areas where stress changes rapidly). For solid Element, it is also possible to use H/P - adaptive to find convergence of results through calculation loops.
​

[dave.laban]

Where you are listing max stress, I'm going to go out on a limb and guess this max stress is found on one of the nodes that's associated with a fully rigid fixed restraint at one end of your beam. Which I'm also guessing is perhaps not the most representative of load cases. What are the bulk stress values, and how do they compare to your hand calcs for your simple beam model?

[Frederick_Law]

All tools used by untrained person are not useful.
The more powerful the tool is, greater the damage.

FEA use a mesh to estimate the geometry. Bigger mesh, less accurate. Small mesh, more accurate.
Mesh also means no smooth surface. Unless the surface is flat.
FEA work perfect on a square.
Not as much on a sphere.
The biggest mesh for the sphere is a cube. So it got 8 corners for stress concentration.
Use smaller mesh, it become more corners. More points for concentration.

"so if i set max element size for lets say 0.1mm my result will be 10000 ?"
Why not do a test with a few more smaller mesh and see?
That's the great thing about sim, you can run and try as many time with as many different setting you want.
You can only test a physical part so many times.