Fatigue Stress Checks With Compression and Shear

[jmggks]

I have a welded knee brace. The knee brace and a beam above it make up a bracket that supports the runway for an automated crane system in an industrial plating machine.

The axial force through the knee brace resolves into compression and shear at the face of the column. Loading on the weld where the knee brace attaches to the column face will be cyclic but not reversed. When I consider fatigue at this location, is my stress range (1) based on the shear component only or (2) based on the scalar sum of the shear and compression loads or (3) based on the vector sum of the shear and compression loads?

My gut instinct is that only the shear matters to fatigue because (A) compression does not cause fatigue and (B) compression is carried by bearing and does not pass through the weld.

I'm looking for reassurance on this. Fatigue seems like a complex subject. Thanks for your comments.

 

[BridgeSmith]

I'm not sure I understand the configuration you're describing, but I would be careful of making the assumption that bearing load is not carried by the weld unless you're sure the base metal surfaces will be in adequate contact when the weld is made. If there is a gap between the pieces and they are connected using fillet welds, the weld carries that force in shear.

 

[jmggks]

Here is a sketch.

 

[BridgeSmith]

Yeah, if it's fillet welded around the perimeter of the knee brace tube, I wouldn't be confident that the fillet welds aren't going to see shear due to the vertical force as well as the horizontal component. You might be ok not considering fatigue of the base metal (the tube) for the vertical force, but I would at least consider fatigue on a failure plane through the weld and size the weld accordingly.

 

[jmggks]

Thanks HotRod10 - that's exactly what I have been wrestling with. If I use the assumption that the weld carries both components of the load, then I need to reinforce the connection in the field. Two options I am considering are to build up the existing welds or add plates on the sides of knee brace like the attached sketch.

Any thoughts on those options, or other options you would recommend?

 

[BridgeSmith]

I'm not really the guy to ask on the finer points of welds, especially the feasibility of adding a layer to an existing weld. That's over my head. I do know that partial and full penetration groove welds have a significantly higher fatigue resistance than fillet welds, but there's probably a simpler solution than grinding off the existing welds and groove welding. Hopefully, someone more familiar with this type of fabrication can provide some insights and suggestions.

 

[dhengr]

Jmggks:
I’m not too impressed with that fix/modification. Why not take a piece bar stock which is 2-3” narrower (two times the corner radii, plus [not minus] weld size) than the tube is wide; 24” long or some such, you want your welds on the flat portion of the tube; cut it in half diagonally to match the col. vert. face, this makes 2 pcs., left and rt.; grind, nibble the back of the diag. cut to clear the existing vert. fillet weld to the col.; clip and grind the two sq. corners round so you can weld around them and across the outer end of the bar, water tight, but to drain at the bot. Stop the vert. weld to the col. .5” short of the edge of the bar stock and do a good job of filling the starts and stops, so as not to leave craters, etc. Stop the edge weld on the bar 1” short of the existing vert. fillet weld at the tube/col., never (well, almost never) weld into a weld and corner like that. That’s likely a serious stress raiser. Be very careful (almost never) weld around sharp corners or edges, stop short of them. You will almost always leave a notch/nick at the sharp corner and if this is stressed, it is a crack starter. Be very careful that your design does not overload (to highly stress) the termination of a weld (i.e. combined tension, bending and shear, etc.), these weld starts and stops are always potential stress raisers.

 

[jmggks]

Thanks dhengr.

Would it be possible for you to roughly sketch what you are suggesting? I am struggling to follow the description.

 

[dhengr]

Jmggks:
Read my post twice, and think and sketch while doing that. I can’t scan things right now. What are the dims. of the knee brace tube? What is the vert. flat face dim (not the vert. dim.) of the tube, its height dim. less the two corner radii? Your bar width is that flat width, minus the/your two side fillet sizes, +/-. The bar thickness should allow a back chamfer to clear the existing vert. weld btwn. the tube and col., and leave some material to weld to, a thick enough land. What is the angle btwn. the knee brace and the col., that is your diag. or slope cut angle across the bar. Center the slope cut on the bar length so you get two identical pcs., one rt. and one left. Round the four outer corners on the bar, so you are not welding around a sharp corner from the sides to the end weld. You add those details to your sketch with sufficient notes and I’ll comment.

 

[jmggks]

dhengr:

I really appreciate your time.

The tube beam is HSS12x6x1/4" and the knee brace is HSS6x6x1/4". The knee brace is at 45 degrees. The beam projects 22" from the column. The column flange width is 11.25". Tube to column weld is 5/16" fillet all around, knee to column weld is 3/16" fillet all around, knee to beam weld is 3/16" fillet on the faces and flare bevel flush on the sides.

So take a piece of bar stock 4.75" wide x 24" long and cut it diagonally, so that makes (2) 4.75" x 24" triangles. I'm not clear where you are suggesting to put these. I follow the edge bevel / land / bar thickness requirements for an added plate to clear the existing weld.

Thanks.

 

[dhengr]

Jmggks:
I would be real leery about using a flare-bevel groove weld if I thought I had a fatigue problem, that’s an awful weld to do well. Why isn’t that upper knee weld just as important as the lower weld to the col? Maybe these 4.75” reinforcing bars should extend from the col., following the knee brace, all the way up onto the 12x6” tube beam webs. I said…, “What is the angle btwn. the knee brace and the col., that is your diag. or slope cut angle across the bar. Center the slope cut on the bar length so you get two identical pcs., one rt. and one left.” We now know that the slope cut is 45°. On the longitudinal center line of the 4.75” bar, at the 12” length mark, make the 45° slope cut. Kinda like making one of the plumb cuts at the ridge or eave on a roof rafter. You now have two pcs. which will exactly fit the webs of the knee brace at the col., and they are 12” long at their centerline. Make the bar at least .5” thk., so you can cut .25”+ chamfers on the slope cut and still have about .25” mat’l. to weld to. What leads you to believe you have a fatigue problem now? We know nothing about the existing design, loads, stress levels, quality and conditions of the existing fab., details, welding, etc. Sometimes, the more garbage you add to something like this, the more chances you are adding for defects, poor conditions, etc., rather than making things better. That’s just something for you to think about.

 

[jmggks]

Thanks dhengr, I follow your suggestion now.

In all of this, keep in mind that I am a mechanical engineer, so if I say something that is ignorant in the structural world, have mercy.

The hoist manufacturer traditionally has used a design of tube columns, tube beams, and tube knee braces, all with matching width in/out of the page in my sketch. They fillet weld the top / bottom of the connections and flare bevel weld the sides to flush. This is the design that they used for many years before my employer bought them and I got involved. This design came from a structural engineer consultant that the hoist company used. The current system needs much stiffer columns, hence the wide flange columns.

The design is intended to have the stress range on the weld throats cause less that 8 ksi stress, the threshold for fatigue on fillet welds. The concern over fatigue is because we found out after the structure was built that the hoist weight is 65% higher than expected. This makes the expected stress on the welds > 8 ksi. The system goes into service in the next 2 months.

... and you are right, the concerns at the bottom of the knee brace will be repeated at the top, with the difference being the flare bevel welds on the sides at the top instead of fillet welds at the bottom.

... and just FYI these connections may or may not ever see the load that I want to design for, and our goal is unlimited fatigue life at design load. The connections will likely see 75% of the design load once every 5 minutes, and 50% of the design load once every minute.