First independent try - stress fields around T-fillet welds

I can hardly start to list why you should not take these simulations seriously.

• I haven't checked "convergence"
• I haven't checked plausibility (though they do broadly fit with engineering knowledge of stress concentrations near changes in cross-section)
• would you really consider the meaning of work from someone who has yet to know how to implement a plane-of-symmetry (the way I negated twisting of the T-fillet was to define a cross-joint "X" fillet(!))
• etc
• etc
• !!!

Just look at the pictures for their "beauty" and think of what it meant to me getting the program to respond in my first ventures away from step-by-step tutorials.
There's quite a number of steps before the "solver" will accept your model and produce a non-null output.
The "small steps" are happening...!

For what it's worth

• the model is 2-D
• however, I have declared plane-strain - so these objects whose cross-section should behave as if infinitely long (??????)
• I've given a Poisson's Ratio of 0.3 - which represents steel (particularly important as I've constrained (?) the model to plane-strain???)
• The F/A "mean" stress is 213MPa (60% of 355MPa - matching my fatigue tests of 2011)
• I've declared a Young's Modulus of 2e11 - though that should be having no effect on the stresses being seen in this linear-elastic model (??????)

"T-fillet weld"

If there is an interesting thought from looking at this;
it's that the load is producing an bending away from the terminating / intersecting plate, which is placing the unwelded side of the joint in much higher tension while reducing the stresses near the weld. That should be advantageous for fatigue-resistance - as fatigue failures are mainly around welds. However - this observation does not match what is seen in real life - welds fail by fatigue and one does not get the impression that anything is "protecting" the weld from those stresses.

"Cross-joint fillet / X-fillet weld"

I've "cranked-up" the discretisation, making the finite-element mesh finer, for a more resolved, accurate, result.

I've "mirrored" the original weld, producing this entire "X"-joint which is modelled in full. This form is balanced with zero bending moment in relation to applied forces along the "main plates" (the structural members reaching out left-and-right (hand sides)).
In order to just look at stress concentrations around the "fillet-weld toe".

That's a very improvised way of getting around that I didn't yet know how to create a plane-of-symmetry in a FEA model
(so it's as if the model had the "mirrored" part, for all its physical effects - but you don't expend computer resources simulating it - so you can concentrate your computing resources on the half which is actually modelled)

Moving onwards...

Leave these for now and forget about them.
I'm trying to walk before I can crawl!

(R. Smith, 17May2016)