Forces in Weld

[Anonymous]

Hi All,

I would like to ask you could you suggest please, how one may extract internal forces (or reactions) i three main directions X, Y and Z from fillet weld part modelled as brick 3D finite elements? These would be used further for external calculation of the weld itself according to some standard, outside Simulation Mechanical.

[Anonymous]

The internal reaction forces on the legs of the weld will be available if you choose "Ignore" for the reaction type of the base parts.  This is found in the Settings button in the Solution tab of the Analysis parameters.

[Anonymous]

@Anonymous I am almost there. Please see figure below. I did as per your recommendation.

Fig. A - Total load of 10 000 N applied in Z direction (axes not shown).

Fig. B - Front view at the model - no contact between welded plates to examine how load will be transferred through weld.

Fig. C - Weld is divided into two parts, magenta selection indicates part that has "ignored" "Reaction Method" for internal reaction force calculator.

Fig. D - Single surface from which sum of Z reactions is being read out.

I could expect that my weld sum of reaction as per Fig. D will be approx. 10 000 N / 2 = 5 000 N, while FEA gives me 4100 N.

Why is this happening? Additionally, global reactions of the model (not welds) shows good value of 10 000 N.

[Anonymous]

@John_Holtz may I ask you please for your input? I think this case may be a quick one for you while I am still scratching my head... thanks!

[Anonymous]

Did you test varying levels of mesh refinement for the welds?

[Anonymous]

Yes, the result is it does not help either. Is your experience that only very very small mesh can capture this?

[Anonymous]

Attached is model archive. Hopes this helps to investigate what is going on...

[John_Holtz]

Hi @Anonymous

I think the problem is that there are too many parts all coming together at the location of the weld, so ignoring one of the many parts may not be sufficient. The parts involved are:

• part 7 - one half of the weld
• part 8 - the other half of the weld
• part 5 - one section of the plate
• part 6 - the other section of the plate.
• The other weld is also split in half and shares a line of nodes with part 5 and part 6.

All 4 of those parts share the same nodes along one edge of the weld. By ignoring the reaction of part 8, you are missing the reaction from part 7 into parts 5 and 6.

If you ignore only parts 5 and 6 in the reaction force calculation, the reactions at the bottom of the weld are 4996 on one side of the column and 5004 on the other weld. This agrees with the 10000 load applied.

[Anonymous]

@John_Holtz, My main purpose is to use the easiest way to readout forces/reactions acting on throat plane between welds' halves (Part 7 and 8) being not focused if other parts should be ignored or not (too much possible configurations in real geometries).

Please see this idea: can I use e.g. Sliding/NoSeparation contact between weld's halves and set very large coefficient of friction (to simulate that in fact there is no sliding) and after calculation readout contact forces from this plane in three X, Y and Z components?

[John_Holtz · Accepted solution]

Hi @Anonymous

Friction is usually hard to converge, but you could try a test. I also do not know if sliding/no separation provides a contact force or not.

A safer alternative would be to trim the welds so that the face on the throat only touches the weld.

[Anonymous · Accepted solution]

Hello @Anonymous,

If I may jump into this thread, I would highly recommend you to watch this two webinars on calculating welds in FEA. 

https://www.youtube.com/watch?v=K6MPCb6V02M
https://www.youtube.com/watch?v=oSJNiXpqGTQ

Although the first one performs the calculations in Nastran InCad, it's possible to replicate the process in SimMech. The second video that I reffer to, shows a part of this process in SimMech.

I hope this helps.

Best regards