Accurate is a relative statement. It is relative to the assumptions made.
There are less assumptions made in solid elements assuming you have a good mesh. But beams should be similar.
But relative to the real world, my guess is you have more assumptions than just that.
My recommendation would be to break out some equations and show yourself the software solves problems with beams and shells and solids in a way that match your hand calcs. Then start with solid elements in your assy and change elements to beams. You will find that eventually one connection or part is changing your answer. And you will know what idealizion assumption you need to be aware of when comparing to real life.
Another test now. Make a change to both models. See if the difference between solutions holds.
But overall, I wouldn't be concerned unless you want high correlation to real life. In that case, you have a lot of work to do to remove assumptions.
I assume that you are using beam elements for the weldment and solid elements for the extruded model. If that is the case, I suspect that your solid model has much too crude a mesh to get accurate results. I also suspect that it will either take too long or not solve at all as you get the mesh fine enough to get reasonable answers. I think that you would be better off with shell elements if the beam elements don't get you the answers that you want. This will make setting up your solution harder, as you have to make sure that your contacts are all good.
Kudos for looking at this. It's really important to look at how choices like this affect your results. I would add a couple of things:
Axial stress is not the same as von Mises, axial stress is one component of von Mises. I believe SolidWorks beam elements are incapable of resolving von Mises stress, which may be higher or lower than axial stress. There is no expectation they should be the same unless the load in the member is purely axial.
Your difference is pretty small, in reality. I would not recommend designing within a .1 safety margin based on an analysis this crude.
There are likely other assumptions and simplifications in your model which are introducing larger errors than the 10% you are examining here. Some of them are listed in my post here https://forum.solidworks.com/message/408925#408925.
I suspect that Saggam is more worried about the significant difference in displacement, not the difference in stress. You are, of course, quite right about the possibility of errors due to other assumptions and simplifications.
Wow. I completely missed that.
The most important plot in weldments simulation is safery factor. Not deflection and not even stresses. I have conducted surveys in this direction, and concluded that the results between beam simulation and solid simulation can differ three times.
It's been our experience that the best way to be sure is to go ahead and model the welds in the 3D solid.
I finally got around to writing up the basics of it in a short book,
In the particular example here, the 'beams' are open section C-channels. This shape does not behave like a beam if there is any torsion.
The cross sections of an open section (C-channel vs tube) do not stay planar if the bar is twisted.