I would like to have your opinion on the interpretation of the results on Simulation.
My question concerns a place where the equivalent stress increases very sharply. In the image, the probed locations are spread over less than 8mm. My pieces: a tube diam. 450mm, 500mm in height, and a 20mm thick gusset. What projection should I make to know my maximum constraint? Linearize the curve (i.e. max stress at 120MPa)? Or not consider the last points (i.e. max stress at 169 MPa? or 135 MPa?)?
From the little I see of your play, it seems normal to me that you have a concentration of stress in this place.
It should be noted that this is an artificial concentration that is not significantly to be retained.
You have two ways to do it and several ways to deal with the problem.
1°) the 2 ways: slightly modify the part or not modify it. If you have the possibility to modify make a radius between the two perpendicular faces.
2° the means allowed by a better parameterization of the simulation. All of which relate to the way in which the mesh is made. 2a) Proceed to a refinement of the mesh locally so as not to multiply the number of degrees of freedom to be calculated on the whole part. To do this, it is necessary to set a local mesh control using a ratio (select in the mesh folder) which amounts to creating finer meshes over a very limited area. 2b) If you have opted higher for a radius instead of the sharp edge, you can (must) also control the transition to the edge formed by the right face and the radius.
I assumed your piece was in one piece. On the other hand, if it were two pieces (two volumes) that were interconnected, it would also be necessary to act on the compatible or incompatible mesh.
Depending on the results obtained, the adaptive mesh can also be used with the H or P method. But that's another story.
NOTE: Von mises is relative to the limit, but in the results you have a tool that allows you to see more or less the areas under constraint. This last tool makes it easier to distinguish very local constraints as in your case.
As I say every time ;-) Without seeing the play, it is difficult to say more.
Kind regards
PS: Please post your coin with an ang go pack including the result of your simulation if you want to know more
(Sorry for the lack of info, it's tricky to disseminate certain details).
It is indeed 2 different parts, a through gusset (orange), welded on a split tube (transparent blue). The mesh has already been refined in the area (on the surface that bears the most load). The weld bead is not modeled, I'll try to do that today.
I don't want to change the shape of the parts, or artificially (they are sheet metal parts) by adding a very small fillet?
I also tested a simulation with shells but it is not conclusive, the stress concentrations seem even stronger and less well localized.
Have you looked in the results the choice offered to look at the stress concentrations. This is enough to see that the concentrations are not too high inside and that they are grouped together. (Von mises and Ures only show the outer (surface) nodes, which is not necessarily meaning, it's rather NS.
If you have a concentration of constraints only on certain nodes, you shouldn't be afraid with that. You have to look at the values of the adjacent nodes, as you did in the first image.
Rather than simulating welds that will not bring anything AMHA, especially on small thicknesses (unless you artificially create the radius that will normally reduce the point stresses). It would be better to move on to frequency analysis. This is how we see whether this excess of very localized stresses remain stable or grow and in how long.
From what I see in the last simulation, there is nothing to whip a cat about. Especially since we don't know the talent of the welder or the frequency load of the whole.
Kind regards
PS: note that I don't have access to all the settings you used, although it's probably very good :-) for I do not know you.
Do not use Hulls which is only used for sheet metal or very thin parts
For the analysis of singular constraints by Simulation, my tests were not at all conclusive, but I don't know if I was the one who misused it. I didn't take the time to go into it.
To look at the inner nodes, I have the impression that the iso visualization works well and quickly!?
I'm going to look at what I can do with frequency analysis.
I will come back to testify as soon as I move forward with these tests.