Fillet weld in compression 1

[2021member]

Hi,

I want to collect some opinions on welds in compression and FEA.

When analyzing say a WF column to a base plate, with just a vertical load the weld would be in compression. However, when using FEA with a small gap between the column and the base plate, I get very high VM stress in a very small area of the weld. With fit up and roughness, I think a small gap is very realistic.

1) Have you ever checked the weld in compression?
2) Do you agree with the small gap?
3) What should you do if the VM stress is beyond tensile? Do you worry about fatigue?
4) Would you design the weld to lower the VM stress to below tensile?

Thank you

 

[dauwerda]

It is definitely something that should be considered. That said, say it does yield in compression - what happens? More bearing area is engaged and stresses go down. Remember yielding (especially in small, localized areas) does not constitute failure.

1) Yes
2) Yes, unless you are specifying machining for bearing
3) See initial response, local yielding is a normal part of design. Maybe, are there load reversals or lots of loading and unloading?
4) Depends

 

[Tomfh]

1. Yes (even though in reality it’s often a nonsense in these type of situations where the fabricator welds a post to a baseplate)

2. Yes, in theory

3. You need to take the FEA results with a grain of salt. This is exactly the sort of situation where FEA will give you a hotspot.

4. Not solely because FEA shows a red spot.

 

[human909]

Use of FEA in weld design is a specialist art. You really need to know what you are doing. Ideally you should avoid FEA for weld design unless you absolutely have to. If you do use it for structural steel connections the a program such as IdeaStatic is ideal. But it isn't cheap.

If you do decide to use your generic FEA program for weld design you really need to use non-linear plastic analysis. You should consider using a failure criterion of strain not stress. And you need to be very careful with your assumptions, meshing and all sorts of aspects.

Fatigue design when it comes to welds is even more complicated and indepth though should only be used if there is genuine cyclical loads which isn't the case in most structures. I did do a day long conference on this very topic. We barely scratched the surface during the presentations as there are MANY MANY aspects to consider and suitably model the welds.

 

[2021member]

Thank you all for your feedback.

 

[JLNJ]

We used to ask that the bottom of columns be "milled" to ensure full bearing. I believe that AISC many moons ago said that cutting the column "square" is adequate for bearing and small deformations from imperfections do not affect performance. You case may be similar.

 

[icebloom]

For more lightly loaded connections we typically check that the welds can take the compression which usually is not a problem - if there is a small gap left during fabrication the weld will just yield and you will get contact with the baseplate. The key is to avoid fracture of the weld in other load cases, as even if your joint can take the compression in pure bearing the weld is providing lateral restraint to the base of your column preventing buckling, which is obviously lost if it fractures. This is not usually a concern under predominantly static loads. When we have serious compressive loads we either use a full penetration butt weld or specify that the joint interfaces need to be prepared for full contact bearing (in conjunction with fillet welds all around). In the Australian steel code here, there are some prescriptive requirements regarding the flatness of the surfaces to be joined that the fabricator needs to meet during fabrication such that you can claim the full contact bearing. Eg they must machine any undulations or ridges to meet those requirements. When specifying full contact bearing, the welds also need to be designed for some minimum percentage of the compressive load mandated by the code.

Even without specifying full contact bearing to the code, you will usually get pretty good contact between column and baseplate unless you have a bad fabricator, in which case any small gaps will result in weld yielding followed by contact. You'd need a pretty large gap to result in weld fracture before contact following weld yielding is achieved, at which point the connection would either be rejected due to incorrect fabrication or require some kind of rectification. Again, for serious loads the way to solve this is to spec full contact bearing (if that exists where you practice which I imagine it would) or a full penetration butt weld.

Further to the FEA point, welds are very difficult to model due to the non linear behaviour, especially if using a plate model FEA where your plates are 2D elements.

If you do use it for structural steel connections the a program such as IdeaStatic is ideal

As human909 mentioned programs like this are the best way to capture the behaviour. In idea statica now you can actually model contact in conjunction with welds and so based on stiffness the load is distributed between the two. Licenses cost an arm and a leg but the software is amazing.