Welded Connection Strength

[slickdeals]

Happy 2013 to all.

See attached sketch, which shows an axial load P transferred through two plates to a WF member (web slotted). My question is in the sketch.

 

[Bagman2524]

Seems kind of risky. As long the the pull is direct tension the load would distribute evenly. You might want to design for 10% side/vertical pull as well.

 

[slickdeals]

Personally I would never do equally sized welds, but use bigger weld at the side of the flanges. I was curious though if the connection would fail if equally sized welds would be used?

 

[connectegr]

How long are the welds? My first thought is shear lag. If the welds are of adequate length, then the weld size will probably be fine as proportioned in the sketch.

http://www.FerrellEngineering.com

Providing fabrication and erection efficient structural design of connections. Consulting services for structural welding and bolting.

 

[desertfox]

Hi slickdeals

Happy New Year to you too!

Have a look at this link:-

http://www.roymech.co.uk/Useful_Tables/Form/Weld_strength.html#Design_St

It seems to me that increasing the length of weld would increase the weld strengh capacity and what safety factor are you using?

 

[TLHS]

I totally didn't understand what you were asking and wrote an entirely different post before I worked it out.

Assuming shear lag worked, I would likely design to 0.125P everywhere in this condition.

Both cases symmetrically load the W section. if you take the welds about the centre of the section, they have zero moment. You can argue about trying to distribute based on proportional stiffness, but there's no significant difference in stiffness between the spot at the top of the plate and the spot at the bottom. The distance is pretty insignificant.

However, sizing the welds at 0.2P and 0.05P would unsymmetrical load the plates creating a moment of 2*(0.2P-0.5P)*0.5t.

Just as a note, this is a really confusing way to label welds. Using the far side symbol when you've got more than two obvious sides can only lead to tears, especially if you point at the two spots that are diagonally opposite from each other. I'd normally either run an arrow to each joint, or make it so one pointed to the top left and one to the bottom left so it was clear that the opposite for each was the corresponding right side. Also, I think you forgot to put a 'Weld 4' in your sketch.

 

[TLHS]

I was assuming the plates were slotted, because that's probably what I'd do. If the lower image is a section part way through the connection then I'm wrong and that likely changes my answer

 

[slickdeals]

TLHS, if the plates were slotted it would be easy I was reviewing a crane tie back detail supplied by another engineer, which involved slotting the primary member. This question is an extension of something similar.

 

[hokie66]

Slick and TLHS,

Perhaps you can explain to me why you think it matters which is slotted.

 

[SteelPE]

hokie

I can tell you why I would slot the plate and not the beam... and it has nothing to do with this problem. From time to time I am asked to cope the end of a beam to form a seat at the end. In this instance I like to slot the plate and not the beam because as the shear builds up in the web of the beam you have a continuous path for the shear once you are beyond the cope. From this problem, I would probably lean towards coping the plate and not the beam (it probably doesn't matter in this instance).

 

[slickdeals]

My line of thinking is based on how I am visualizing it.
If the web of the beam is slotted, then the force has to be delivered by the plates to 3 "pieces" and then travel through beyond the end of this slot to the full beam. Each piece will carry the axial load based on its area. As a result, I think the welds should be sized based on which piece the force is being delivered to.

Based on what I just typed, I think it may not matter whether the plate or beam is slotted.

The bigger question is still the sizing of the welds, all equal sized welds (0.125P) or based on load distribution?

 

[wannabeSE]

Hokie,

I can think of one potential reason for slotting the plates rather than the beam. The slot reduces the net area and rupture capacity. In the US, the beam is probably A992 steel with F_y = 50 ksi and F_u = 65 ksi while the most commom plate is A36 with F_y=36 ksi and F_u = 58 ksi. If A36 plates and an A992 beam have the same yeilding capacity without slots, then the beam may have significantly less rupture capacity. Depending on the width of the slot and length of the weld (shear lag), the rupture capacity may control the connection capacity.

 

[BAretired]

I'm not sure that it matters much, but I would prefer to slot the plates because it is easier to do. Sizing each weld for 0.125P would be my choice but for the small cost involved, it might be a good idea to increase lap length, hence weld length by a nominal amount.

If the portion of web between the two plates is insufficient to carry 0.5P at yield, I would extend the lap sufficiently to account for shear lag.



BA

 

[dhengr]

Slick:

It would be nice to know the proportions (dimensions) of the members, materials used, and the magnitude of the loads, to know which element is pushing/pulling the others along for the ride, in this complex process and detail. Then you have the additional complexity where you can’t really be sure that one of the plates isn’t actually carrying .6P and the other .4P. Is the WF really very highly loaded with the load P? What are the plate and web thicknesses and sizes? Then I have an additional question, I don’t know if you are designing this, or it exists and you are trying to rationalize the design. Is this a brace from your previous tower crane thread? Otherwise, I assumed that problem was put to rest some time ago.

You might be better off to leave the WF web alone, so you wouldn’t have to handle the WF through a slotting process. Slot your two plates at the same time you match drill the pin holes. Exactly how you would fab. these plates depends on your shop’s equip., but because they are pin holes they should probably be drilled or machined, not punched or burned. These slots would have a nice radius of one half the web thickness (at their inner ends) where they meet the end of the WF web, a stress reliever if you wish, but this would not be welded. Can you physically reach in there to make the welds properly? In some details and situations, I agree with SteelPE’s thinking, and would not want to notch the WF web for stress flow reasons, in other instances it might not matter. In any case, the welds and detail shouldn’t over stress the WF web in the length required to make the load transfer to the entire WF section, or at least you have to rationalize how this happens. I don’t know that I would use different sized welds, but I might stop the inner welds shorter than the welds out nearer the flanges. Although, this is probably more complex detail and confusion than it is worth, as these inner welds and web will just go along for the ride.

At first, the load transfer will be influenced by the load in/from the plates and all eight welds will try to transfer load equally or you would have some real funny compatibility issues (strains at a given point/weld length location) btwn. the plates/WF web and welds, per inch of length, as you move into the WF. As you move further into the WF, the web btwn. the plates will start to yield (be over loaded/over stressed), and its welds will just start sloughing load off to the welds out nearer the flanges. That this inner web area yields a little really doesn’t hurt much in this confined region, it just allows the needed load transfer out to the flanges. Slick’s changing the weld sizes, .05P and .2P is some effort to match the weld size with this effect. But, his smaller welds might likely be over stressed at the start point, and I don’t like that.

At the same time, the shear lag in the WF web is starting to take some of the load out into the flange areas. In some posts it sounds like there is a little confusion about the terms ‘shear flow’ and ‘shear lag.’ In this case, shear flow is a means of representing the force/inch of length or stress in the welds. While shear lag has to do with the fact that when you apply the load P, in a fairly concentrated way, it does not instantly transfer into the entire WF shape. But, through shear lag, or shear deformation of the webs, and over some distance, it will transfer into the entire section.

As Connectegr suggests this is a shear flow and shear lag problem; in two ways, I think. First these types of welds tend to have a diminishing cap’y. per inch of length as they get longer; although given the proportions you show I don’t think that will be a problem, your welds don’t look long enough. Then, shear lag is also what finally transfers the loads out into the flanges, or the entire member. And, you can’t overstress the web in tension or shear in this transfer region.

My 4 cents worth, or at least some food for thought.

 

[Ron]

Agree with BAretired...easier to slot plates than a beam....otherwise, doesn't make any difference.

Slickdeals...as long as your loading is purely linear, your premise is fine.

 

[TLHS]

The reason the location of the slot affects my answer is that a non-slotted beam maintains continuity and can pass load from the inside welds through the web back up to the flange. If the beam is slotted, this load transfer can't happen.

Take a look at the sketch. I've simplified this down to having a spring for the middle web section and a spring for each of the two flange+web sections. If you treat the member pieces as independent springs, and t as small, you end up with the non-slotted case having both welds with the same force because the continuity of the member is maintained and it can pass load between the components of the beam. In the slotted case you end up with the weld loads on each side of the plate being proportional to the stiffnesses of the portion of the beam it's connecting too. The only continuity between the sections is the welds, and if you tried to use them to redistribute you'd just end up with both welds being the same size as you would if you designed them based on proportional stiffness (plus you'd be designing to a load path where a load goes down and then back up the same weld which is silly). Obviously I'm assuming we're sticking in the elastic zone. You also get a moment in the flange pieces because you're loading them off centre, but you could deal with that pretty easily by upping your weld a little and balancing the moments from the top flange piece and bottom flange piece.

 

[hokie66]

TLHS,

I disagree with your reasoning. Welds are metal fusion, not springs. It may be simpler to slot the plates, but for tension only connections, I fail to see the difference in the two approaches.