Very Non-Standard Connection, Would like an Opinion.

[IngDod]

Greetings, I am currently designing a moment connection between an HSS column and an HSS beam, I am forced into this configurations and although I would very much prefer to use W-shapes it is out of the question. Initially I tried to simply weld the beam to the column, but this lead to a myriad of problems such as plastification of the column wall, punching of the column wall and such. Currently I am trying to use stiffening plates to keep the wall of the column from deforming. However I know of no hand calculation to design or even estimate the design of this connection, this led me to rely on a FEA program (SAP2000), which I very much dread since I have to rely completely on the computer. What I did is model the connection without the siffeners and check that its behavior is as expected (miserable failure of the the connection to behave as fixed) and then I proceeded with my stiffener plates scheme. The general behavior of the connection is as expected, the HSS walls are no longer deforming as the plates give them considerable resistance. I am seeing very high stresses near the corners of the plates.. But this is a linear FEM Analysis so no stress redistribution after yielding is being accounted for. I will attach Images and printouts of the connection and would very much like to hear your input, If anyone can suggest a method for hand calculation I would greatly appreciate it.

 

[AELLC]

Looks like a good idea.

What about a simple calculation for direct shear in the plates and the stress in the welds, due to load running in the same direction of weld along the side of the HSS column?

The definition of a structural engineer: overdesign by a factor of 1.999, instead of the usual 2.

 

[civeng]

How about using an end plate larger than the width of the column, welded to the HSS beam (all around)in shop, and then field welded to the HSS column (all around). This will eliminate the effects on the HSS column walls and give you a symmetrical weld which will not cause any eccentricities and will be enough weld to resist loads.

 

[IngDod]

@AELLC: As usual you have been most helpful, it occurs to me know that for the purpose of sizing the plates i could treat the connection as if the plates were gusset plates. Having this in mind I could check the plate for: Block shear due to the welds, Gross Section Shear Yielding, Gross Section yielding (using witmore width starting from where the beam first meets the plate, this would also give me a rational way of determining how long the plate should extend over the beam, based on the 30 degrees used for witmore). In the compression plate I would say that buckling is restricted due to the webs of the beam which are in contact and welded to the plate. I do wonder how to account for the beam webs welded to the plates when calculating yielding of the plates, as they must add resistance to the plate. I will definitely use your recommendation for the weld stress as its the only reliable way of designing the welds I can think of.

@civeng: Thanks, this is a very good idea. Another connection I must design has two beams meeting the column from opposite sides, this will most likely require a solution similar to the one you propose; a sort of "ring" plate around the column.

 

[AELLC]

Are you familiar with this? - it was very useful back when did steel design.

http://www.amazon.com/Design-Welded-Structures-Omer-Blodgett/dp/9998474922

The definition of a structural engineer: overdesign by a factor of 1.999, instead of the usual 2.

 

[AELLC]

Actually, this is the best place to buy -- worth every penny.

http://www.jflfoundation.com/ProductDetails.asp?ProductCode=DWS

The definition of a structural engineer: overdesign by a factor of 1.999, instead of the usual 2.

 

[IngDod]

Thanks I just purchased it

 

[JAE]

Would using 50 ksi steel help? The tubes are most likely 46 ksi, right?

 

[IngDod]

Indeed.. the tubes are 46ksi, however I dont think that plates of the same steel or superior are available commercially here. Best thing I could do is use cuttings from the beam's web as plates. But they go very low, only up to 0.9cm while the A-36 plate I'm considering is 2.54cm (1"). Now that I have a hand calculation method for the plate Im getting values much smaller (1cm plate) than that of what would seem appropriate with the FEA, as in the FEA I can see stresses going above the yield value in much of the A-36 plate when using the 1cm plate (enough lead me to believe that redistribution would not be enough). One thing that does holds true in the FEA is the witmore width, I get a very similar V shape on the stress distribution. Honestly I would trust the provisions in the AISC specs over a FEA any day of the week; but if the FEA suggest a more conservative approach I would prefer to err on the side of safety.

I believe that by extending the plate all the way to the lenght of the column sidewalls I will be effectively avoiding any of the failure modes of the HSS walls (except shear due to the weldings). Any thoughts on this?

Thanks, this forum and its users are a tremendous help.

 

[SAIL3]

I usually use the min overlap of pl on beam to be the width of the beam. Using the witmore approach is ok and also checking for shear lag as per AISC. I second the recommendation of civ in making the pl a ring on the col. The weld of the pl to the bm will be a partial penetration flare-bevel weld whose effectiveness is 5/8's of depth of weld. You are on the right track with your concept and also in questioning and confirming the model output. Not knowing the magnitude of loads or member sizes, I am curious why you had to add an extra seat @ the bottom of the bm, unless to take the verical shear, if so, I would add a closure pl @ end of the bm. In HSS design, I often find that the connections often drive the size of the member that I choose to facilitate a reasonable and practical conn.

 

[IngDod]

Thanks for your reply, I followed the advice of "wrapping" the plate around the column and it is actually the only way I can develop enough resistance to the tension force in the top plate. I am using 3/4" fillet welds between the plates and the HSS, on the top and bottom surface.. I chose 3/4" because my understanding is that this is the maximum allowed for A-36 steel using 70ksi welding. I have a question, I envisioned the weld as a simple fillet between the plate and the column wall; why would I need a partial penetration weld?. I don't know if you are referring to the bottom plate or the second plate, the bottom plate I added because the compression force coming from the haunch is enough to punch in the column wall; the middle plate I added only for redundancy and I am considering removing it. I usually do the same for connection of HSS trusses.. but in my situation I have to use a relatively large HSS column (20x20cm) and all the beam sections that I could choose from where either 26x9 or 30x10, in both cases the flanges are slim enough to cause punching and plastification of the column.. so i find myself in need of this stiffeners.
Im trying to have the connection take a moment of 16000 kgf.m or 116 kip.feet, this beam runs 40 feet between columns but at 20 feet it rests over another beam that runs perpendicular to this one.

Here no one construct using this stiffeners (no one actually calculates the connections), so my client may think I am crazy when I give him the drawings, But theres no way this connection would take any moment without such stiffeners.. And I know for a fact that others assume this things to be fixed since they put haunches at the ends of the beams.. why else would you use haunches if not to increase the moment capacity at the the fixed end of the beam? My guess is that in reality this connections act as pinned and all that moment goes to the middle of the beam.. the only thing saving this guys is the 1.2DL and 1.6LL.

 

[IngDod]

I am attaching a .pdf that contains my current thinking on what failures to check the connection for. I would appreciate it if someone could give me their opinion about it.

 

[AELLC]

In block shear, I would not include tension allowance.



The definition of a structural engineer: overdesign by a factor of 1.999, instead of the usual 2.

 

[IngDod]

Do you mean disregarding the tension plane and using only the shear planes? Even by neglecting this I still have enough capacity, the controlling limit state is shear yielding at the whitmore width.

My hand calcs give a 45000 kgf capacity for a 1cm thick plate. Which I would be quite happy with (I need only about 35000). I try to test this on the FEM and Im getting stresses as shown on the attached picture. The deep blue color are stresses above the yield stress of the plate. I would like to somehow reconcile this results with my hand calcs... It seemed to me that when I use whitmore width for the plate yielding limit states I am assuming that all the yielding occurs along the whithmore width.. In the FEM This is not happening.. I have the "triangle" resulting from the whitmore width and the lines starting at 30 degree.. however the inside of the triangle is not yielding, while its perimeter is.. When we do whithmore width.. are we assuming that all the high stress is being redistributed from the point of force aplication "down" trought the plate?. The stresses in the picture occur when I apply a tension force of 40000 kgf. I would attempt a non-linear material analysis but this takes days to run. Basically I am just trying to validate my hand calcs so I can get rid of the FEM, which honestly I quite dislike using; but I dont want to ignore some important limit state by mistake.

 

[AELLC]

When you get this whole procedure down, you should automate it with Excel, then you won't need to run it on FEA.

The definition of a structural engineer: overdesign by a factor of 1.999, instead of the usual 2.

 

[IngDod]

Oh I do everything in excel, I call hand calc's everything I do with a spreadsheet... I have a spreadsheet or maple (similar to mathcad) file for pretty much every situation I've had to face in the past. It saves so much time.. but lately I seem to be facing completely different stuff in every different project. My ongoing hobby is to get all the spreadsheets to draw data from one access file, I've manage to do this with section properties and rebars; for example in my current HSS capacity spreadsheet I can push a button and then choose from a list of sections pulled from the access file. If I want to add a new section I just open the access file and add a line with the data then the section is available in all the spreadsheets that get data from this database. My long term goal is to be able to have a database keeping information for each structure I work on.. say the nodes, the members and the moment and shears.. so I can get this values into each spreadsheet design and then save the design data back into the database.. But I never have time.

 

[IngDod]

I think I found my error in the model.. I was applying the loads at only a few points at the end of the plate... Now I am testing different configurations of loads on the plate.. The one that I believe is closer to reality is where I apply the Tension force coming from the beam's top flange all along the weld line that joins the top flange of the beam with the plate. I will attach a pdf showing the force application and resulting von mises stress distribution in case anyone is willing to give it a look and comment.

 

[dhengr]

IngDod:
The FEA looks about like you would expect it to look. And, your FEA printout/picture is telling you exactly what to do. Show us a couple dimensioned views of that connection, what size are the two members? What are the side pls. made of, that make up the hunch? Are they full height, top of beam tube to the bottom sloped surface, a bottom flg. in compression? You see that you can eliminate the middle stiff. pl., it doesn’t do much, except complicate the fab’ing. The bot. stiff. pl. might be placed at the same slope as the bot. of the hunch. And, you might try reshaping the top stiff. pl. so it takes the top tension out to the sides of the column tube, as a “Y” shaped (forked) tension field, not in bending across the stiff. pl. and the face of the column tube. Maybe that top stiff. pl. has an elliptically shaped cut-out, with the minor (major?) axis being the width of the col. tube, and running half way out the length of the hunch over the top of the beam tube to a half circle termination. This completely eliminates top stiff. pl. mat’l. in the yellow area of your latest picture. The outer edges of this top stiff. pl. should taper out on top of the beam tube about the length of the hunch, maybe a little further, as you’ve shown in your latest sketches. Try running something like this on the FEA and see what it looks like stress-wise. It should completely change the blue stress picture around the column corners. You will always see high stresses in your FEA, at corners, the multi-axial stresses kill you. And, we know this from our Theory of Elasticity study. We’ve known for a long time welding into and around corners can cause problems. But, a second angle of attack (a consideration), regarding the high stress areas in your FEA pictures is that, as long as a small region in a detail is well confined, by surrounding material or weld, a little yielding is usually not a big problem. There is still compatibility and just some redistribution of stresses/forces to the adjacent material. All of this needs a little more study, once we know the proportions/sizes of the connection and the members, and the loads and moments. I have to reread your last couple posts, and think on them a bit.

The book that AELLC suggested would be well worth your while, and Lincoln has several other good books on welding and weld design too, several of them by Omer W. Blodgett. “Design of Weldments” is one of them, with some different material in it. You have some really difficult welding conditions on this detail, and the FEA doesn’t show the potential problems in the welds. But some of them are at the same locations as the high stresses shown on the FEA.

 

[IngDod]

Thanks, I tried the FEA with the shape you described. Honestly I didn't quite understand the part about the shape close to the column face. I tapered the top plate as in one of my previous sketches.. It doesn't seem to have that much of an effect, other than saving material perhaps. I am adding a pdf wih the new results as well as the dimensions and materials of the connection. The moment im designing for is 16000 kgf.m or 116 kip.ft, the baseplate for this results have a 2" thickness. All other dimensions are on the pdf. I removed the middle plate, there's no appreciable effect on the connection. Thanks for your help.

 

[dik]

Try to find a copy of Packer and Henderson's text, "Hollow Structural Section Connections and Trusses" one of the best HSS connection design texts out. Also CIDECT has some excellent publications.

Dik