The EC3's theory establishes the use of a "spreaded tension" all over the section of the weld.
This tension will be reworked through a serie of coefficients to take back the calculated stress to "the real one".
The coefficients that permit all of this are obtained by tests over welded samples, with a certain shape and material (especially construction material like S235). One must always try to trace it back to welded joints like these to maintain valid the calculations.
FEM permits to obtain tension punctuals evaluations over whatever welded links geometrie.; trace we back to the model of a normalized tension over the entire section means loosing much of FEM's potentialities.
However, I believe that, if one want really following the EC norm, the method of the normalized/"speaded" tension evaluation is applicable (through those measures which I've asked in the firstes posts of this thread).
By going back to FEM's potentialities, for welds it is proposed the FKM use, which they need:
- the punctual evaluation of max tension (sigma, tau) of the section;
question: I start by saying that I've not had access to the FKM-Guidelines yet, but in the example (by Paul Kloninger) one considers only the tau_yz and not the tau_xy. This last one should be 0 in the example (no torsion stress) but I would say it should be considered generally. In this case, should it be considered by this formula radq(sigma_z^2 + (tau_yz+tau_xy)^2) [as in the Paul's example at page 6] or this one radq(sigma_z^2 + 3(tau_yz+tau_xy)^2) ?
- the definition of a rounded weld seam (so, the weld seams must be modeled and insert within the model)
- a dense mesh on the weld seams
From what I've understood, it doesn't need an elasto-plastic material neither for weld seams nor for the groun material (so the simulation remains in linear field) since one utilizes the plastic shape coefficient n_pl, if the structure's strain remains between 1-5%.
In the Paul's document it hasn't been written, but I would say that if the rule of 2*a>=t is respected, then you can consider the weld as a "full deep" (I don't know the right term in english...). So there is material continuity among plates and so you can avoind the use of a contact (non-linear analysis).
By utilizing other coefficients obtained by tabs (which I've not seen yet) like alpha_w, j_s, j_p, j_z, one arrives to the calculation of a global coefficient j_ges where:
j_ges > 1 OK
j_ges < 1 BAD
Do I understand correctly?
A personal curiosity
At the end of page 6 there is the term "stainless steel" as "rostfrei Edelstal". If I translate (from ENG to GER) the term "stainless steel" I have "rostfrei Stahl", while if I translate back to front (from GER to ENG) the term "Edelstahl" I have "stainless steel". So it seems like to say "stainless stainless-steel". Which is the right translation?
------------------------------------------------------------------------------------------------------------
Regarding welds, I take this opportunity for invite you to vote two ideas:
- Fillet weld with edges preparation
- Edge preparation length for welding
