Semi-Rigid Connection Modelling 4

[serdardundar]

Maybe this question will be longest ever in this site. Sorry for that.
I am working as the only civil engineer in a company specialized in cold formed residential steel structures. We are designing structures with 2-3 stories. I have graduated in 2000. I did not worked with an engineer since my graduate, and I have some problems with the type of connections being used in my company for years.
The connection is some kind of a semi-rigid connection: Flush end plate (not extended beyond the flanges.)There are two rows of bolts (one tension and the other in compression) each with one bolt of diameter of 20mm. The beam is C shaped with 25mm height-8cm flange-3cm lip- 0.3mm thick. The distance between the bolts is 17cm(There are two bolts in the connection). The end plate thickness is 8mm.The column is 20cm height and 0.3mm thick with same length of flange and lips And there are stiffeners in the column extending along the flanges of the beam , a doubler plate at the flange of the column where the beam is connected both have a thickness of 8mm. I am sure that this type of connection should be treated as a semi rigid or hinge type of connection. But if it is treated as a hinge connection the structure will be unstable under lateral loads only(No place left for bracing along one axis of the building) . The practice were modelling the connection as a fixed type of connection in previous designs (which is of course not true.) but i did not see how the connection is designed previously.
My approach is designing the beam as a simple supported beam under vertical loads, and designing the whole structure as the connections are fully rigid (modelling it as rigid frames in the 3d analysis). So that the beams are designed safer (not including the semi rigidity of the connection), and the columns are designed safer (because the connections are rigid the moment caused by the vertical loads are carried by the columns also).Resulting an uneconomical design.
But the moments calculated in the 3d analysis are so high that, if that moment is supposed to be carried by the couple formed by the tension and compression bolts; the top bolt will reach nearly its tension capacity. But i know that before the bolt reaches its tension capacity the end plate or the column flange will begin to yield. I think that the moment capacity of the connection is enough for lateral loads only. But this causes an inconsistency in the design.
I want to change the whole connection and make it fully rigid. I calculated the required thickness for the end plate to resist a moment of 213 tcm (moment capacity of the beam) according to the AISC approach and i found that the thickness of the plate should be 16mm if i extend the plate 8 cm above the beam and use another bolt at the top which is 4cm above the flange(Three bolts in the connection; two in the tension zone, one in the comp. zone) This is not aplicable in my case because the thickness of my column is 3mm and i can not change the thicknesses and there is no space in the flange to use two other bolts in the tension zone. I don't think welding will be solution for my case because my members are 3mm thick, so the welding should be done very very carefully in the site and according to my experience there is no attention paid in the welding at site.
Do anyone think of designing a fully rigid connection in my case? Or a method to calculate to what degree the connection is semi-rigid? Any help will be appreciated by this young, unexperienced and curious engineer. If any drawings or copy of my calculations are needed i can provide them via mail. Sorry for this long mail again. And thanks for everyones help, including the founders of this site...

 

[ishvaaag]

If you want the structure being the sustaining party, and not sustained by the masonry, all the loads need be taken by the structure. However there are plenty of steel structures in Spain where the continuous steel beams pass simply supported over bridge seats between 2 channels or double tees...this a practice very successful to the day, (the more conscious cases fortunately do include vertical bracing of the tiered system) as long the earthquakes have not shown the nose; since NO lateral restraint there is at the joint (for the worse cases), one needs the conclude that this (presently very acknowledgeable as bad) practice must rely in the ability of the -in such cases attached- masonries to force a "vertical forces only" behaviour for the beams and columns, the building taking the brunt of the lateral loads more or less in some monolithic way through concrete and masonries binding. Your firm may have been relying somewhat on this.

If the structure has to take all the loads, lateral included, rigid or at least semirigid must be at joints if no bracing is present. The single book I have that focuses on semirigid structures

Stability Design of Semi-Rigid Frames
Chen, Goto, Liew
Wiley Interscience

does not include semirigid behaviour data for a C like yours (not any cold formed), data which more or less seem be coming from experimentation.

hence your approach of making the connection rigid maybe sole choice if not wanting to outsource the investigation of the semirrigid behaviour, even if disgusting in using a too thick plate arrangement no doubt a source of stress concentration, and that very carefully must be welded for a only reasonable behaviour. I feel frankly uncozy with something at a joint where a 3 mm part needs be welded to a 16 mm one.

 

[serdardundar]

To Ishvaaag;
Thanks for your reply. I should have mentioned some other properties of my situation. First we don't use masory walls or concrete walls. The usual practice is using OSB (oriented strand board) for siding of the structure. Wall studs with 60cm spacing is used to benefit from the lateral load carrying capacity of the OSB. But my problem arises when the lateral force is applied in the direction parallel to the frames (in the perpendicular direction to the frames; there are only joists and purlins which are covered by OSB).In paralell to the frames direction there are nearly no OSB to benefit from its lateral load carrying capacity.
I would be happy to hear what would be your approach to the problem. Would you design a fixed connection and change the whole connection geometry or designing the column base as fixed type (very very difficult in my case with a thickness of 3mm column).I could not find a way to design a fixed connection without using weld (which i do not trust because it is not done appropriate in site with 3mm thickness of steel) or using a too thick plates(16mm plate welded to a 3mm thick beam and bolted to a column with a 3mm thick flange!!). How would you design a fixed connection in my case? I am looking forward to your reply...
I have tens of recently published papers about semi-rigid connections (SRC). (Papers of Simoes da Silva,W.F. Chen,J.P. Jaspart) I examined all of them but i could not find a way to apply them in my designs. Did you ever use the component method or any of the other methods to calculate how rigid the connection is? Maybe i should post this as another forum.

 

[IJR]

serdar

I must admit I havent fully read your question(It is too long) but here is my input

1) Modelling is a matter of judgement and that takes some time. You are the one who knows how your structure behaves. The more precise you model, the more the chances that you might exaggerate non critical and take for granted the most critical.

2) Bases and beam-column connections in steel structures can quite easily be detailed as hinged or fixed, though behavior might fall within these limits, usually there is not much lost to worry about. It is a matter of careful detailing and recognition of how much load is likely to develop there. So dont get too theoretical.

3) I agree with you that beam column connections in most steel framed buildings especially with floors (heavy) do attract a lot of moments. Bolted connections using heavy endplates can sustain such moments but usually it is best to let moments be resisted by full penetration welds at column beam interface and use bolts some distance away where moments are light. Another technique is to cut down the effects of moments there by using a haunch. We use haunches a lot to cut down bolt forces.

4) Never distrust weldings. Shop welds are checked by well established techniques and so are some field welds. Most of the serious work is usually handled by careful field weldings.

5) I think you should give up serious modelling(whatever moment rotation curves you are going into) because those analysis are used in developement of standard details or special details. The kind of moment connections you are likely to use is probably a well established one and someone else probably fully analyzed it for us to use. Fixity of such connections can usually be checked by simple empirical relationships given by codes such as Eurocodes 3.

Bottom line: Try to understand standard details as given in AISC manuals or Eurocodes 3 and apply a lot of judgement. And have faith in welding.

regs
IJR

 

[ishvaaag]

Well, really I am not an expert in semirigid structures, not at all, what I posted was just fruit of general knowledge. So I have never "modeled" a semirigid connection from scratch, less those not even acknowledged in the book(s).

So if i has to meet semirigidity for the unknown connection I surely would try to get empirical. This would have another benefit, since you have thin walled or near so members (even singly symmetrical Cs), and you would know for real to what extent distorional modes of failure of the section affect the behaviour and the model, which is one of my main concerns respect the behaviour even if you make a joint wholly in itself rigidized.

Lacking one safe semi-rigid approach if only on the lack of data, of course we need to resource to the lateral rigidity estimated in any conventional way. Of course if there are not even frames for one direction, nor planking can in reliable way provide it, you need gain it by end fixities.

First, the plank may be providing the rigidity, and then the question is how to calculate the contribution of plank. Since we in Spain do not practice much with wood, I am not familiar with the precise specifications for shearwalls in plank, but some there must be. This is most surely the solution to your problem, which would need then to always ensure the controls for the plank walls be effective in shear be in place.

Other than this and for a direction in which there's no frame action, hinges would give a mechanism, so you need mainly make fixed ends, welding everything, even resourcing to the torsion capacity of the beams...then to pass to the joints, your problem. How? In no other alternative "build a stiff box" at joints. A brute test you can make, build a 1 span frame like this, and pull with a cable from atop till ruin. Study at what forces the failure happen and see if distortional modes concurr. If so, and within the reach of the lateral forces, in more than the stiff joints you will need to include stiffeners or battens in the columns.

As a last note, even in the systematical analysis of the behaviour of the end plate, 2 bolted connections I wouldn't feel very confident in the behaviour always be similar for this in some aspects uncomely detail.

 

[serdardundar]

To IJR;
Thanks for your reply. It was very helpful. But i think i could not made myself clear at some points. I do trust welding at site but in my case both the beam and column are C shaped with 3mm thicknesses. Assume the weld is done perfectly at the side still i can not weld both sides of the beam (in and out of the C) because of the thickness so moment capacity of the connection will not be even half of the moment capacity of the beam. What may be the max. thickness of weld that i can utilize in my connection? Is there any other way to solve the problem with bolts only?
If i design the beam as a simply supported beam (capable of resisting ql^2/8) and design the columns as if the connections are fully rigid (so that they carry some of the span moment of the beam due to vertical loads) and design the connection only against the moment caused by the lateral loads would i be on the safe side? Is this approach acceptable from engineering point of view? (I know it is not a usual approach and i am not sure if it yields a safe design always.)
I could not obtain Eurocode 3, could you please post those empirical relationships in the forum? I am looking forward to your reply. Regards,

 

[IJR]

Serdar

Let me thank ishvaag for the great discussion first

I think I can see your point now. You have a small structure and you are using thin walled profiles(3mm thick C profile is "thin&quot.

Now the following is based on my personal preferences.

1) It is very difficult to get a decent moment connection with such a profile.

2) Why dont you brace your frame?. If you brace, then you can connect all beams and columns with web bolts (shear only) and assume this to be a pin connection. Note a pin in steel structure can easily be realized(A rotation of 3 rad causes a pin and most webs can deform this much)

3)If you insist on moment connection, then I suggest a haunch and bolts.

4) You can still make beam connections rigid by using knee braces.

* * * * * * * * beam
* *
* *knee brace
*
*
*
*
*column

In this configuration, all ends can then be simply connected(shear connections)

The fact that you are using 3mm thick profiles makes me think your structure is lightweight. Such a structure is best braced and connections made very simple. Moment frames are for relatively heavier structures where heavy profiles can be used.

With respect to connections I suggest you first get a copy of AISC manual on connections for a feel(check out libraries in Technical Universities). Eurocodes 3 should be available in libraries specializing in Standards.

respects
IJR

 

[serdardundar]

Thanks for the last reply. Now we are talking about same things. Yes i have a light gauge small structure with 2-3 stories. And yes my structure is thin walled.
1. I agree that it is very very difficult to make rigid connections in my case. But some reasons forced me to think about it.
2. I thought about bracing frames. But as i said in the previous mails there are only floor joists and purlins in one direction of the building. In this direction i can provide braces in the exterior walls and this will do the job. But in the direction of the frames i can never use diagonal or eccentric bracing because the braces will cut the rooms and kill the interior useful places. Just like an industrial storage building; one can not use braces in the direction parallel to frames except in the endwalls. But in a residential building i can not use braces in the endwalls either because it is the front side of the house where there are too many architectural openings.
3.So i began thinking about making the connection rigid. Thanks for the knee brace advice. I did not think of using it. In my case the ceiling height below the beam is not enough for a knee brace (The height of the section carrying the ceiling plaster board is 3cm)Can i use a knee bracing at the top of the beam? I mean if i use an angle at the top of the beam with two stiffeners (one above the beam web and one above the beam lip) and bolts both in the beam flange and in the column flange (Still i will face some tension at the top bolts in the column flange but this time somewhat less).
4. Can you please describe what you meant by "haunch and bolts"? I don't think i understand it correctly. How can i use a haunch when the column is continuous and the beam is connected to it at nearly midheight.
Thank you very much for your concern about my questions.
Regards...

 

[ishvaaag]

Just 2 references

Residential Steel Framing Handbook
Robert Scharff and the Editors of Walls & Ceilings Magazine
Mc Graw Hill

This I read but almost forgot, these things we almost don't practice here. Spain is a RC, (for industrial PC) and maybe laminated steel realm for buildings' structures.

This book is not focused on any kind of calculations, but describes thoroughly what the industry does. So I guess from it you can find if your concerns are in some way typically being dealt with.


If you want to use plank or plywood shearwalls I found in my library

Wood Engineering and Construction Handbook 2nd ed
Keith F. Faherty and Thomas G. WilliamsonMc Graw Hill
in p. 8.58 and around you have shearwall capacities

of course UBC and alike must give direct specifications on how to

In any case you can use shearwalls put anywhere, as long the conditions of equilibrium are met. So you put if you want some of the shearwalls in the inner partitions. Of course the roof or floors need be stiff enough or be reinforced in such way as to be able to take their lateral forces frpm the outside to the shearwalls.

Where I say shearwalls could also be braced panels not affected by doors or windows.

I no doubt would go for shear wall or brace action, which by the way seems to be much of the practice for this kind of (to some extent) light structure buildings.

 

[IJR]

Here is a haunched beam

**********************************************
* U section beam
**********************************************
* *
* *
* *
* * A piece from U section welded to the beam
*
------ Lh--------

Lh is the length of haunch, typically twice the total expanded heigth

Of course you must model this using tapered elements(Nonprismatic section in SAP2000) because haunched beams tend to attract larger moments at ends. But the height is good enough to accomodate a large number of bolts or a reasonable size welds. To me this seems to be the best solution.

If a haunch presents a problem(architectural), then cut the column where it intersects the beam(at mid-height as you say), add a thicker U section at the joint and use this thick portion to make your connection.

*column
*
*
/\!!!!!!!!!!!!!! beam
*
*

/\ is a thicker piece of column and * is your normal size column. It is like a pipe joint in water supply systems. You have a stiff joint. And you can weld quite easily there.

But this is going to add some fabrication costs, so check out to see if there are too many of them. A small number should not be a problem

Finally, why dont you increase the size of your columns or choose a section with thicker flanges?.

As ishvaag and I have mentioned before, light structures are best braced and braces can be hidden in walls. So talk the owners into it. A typical brace will be a length of non deformed reinforcement bars used in RC construction. Cheap and very effective.

respects
IJR

 

[StructuralDzine]

If I understand your description properly this sounds similar to what is being used in Australia. When tested these types of connections generally perform poorly. I suggest you have a look at these references;

http://www.topica.com/lists/Structuraldzine/read

"A new knee joint"

JE Mills & J Miller, "A new knee joint for cold formed channel portal frames" ASEC Conference Proceedings, The Institution of Engineers, Australia, 2001

 

[RiBeneke]

The discussion so far has hinted at the structural strength of the panel or plank components of the buildings you are working with.

I suspect that the flat panel components contribute a very large proportion of the strength to the building.

Try building yourself a cardboard model of your building, fastening the edges of the cardboard panels only with adhesive tape. You will find that the model has quite considerable stiffness, provided the panels do not have very large or continuous openings cut into them.

Back to the real thing ... I think that a realistic analysis of the structure must take the panel stiffness, in the plane of each panel, into account. (This includes the floors). Then you will probably find that the bending stiffness of your beam joints is not all that important. However, what is important is the shear strength of the fasteners between the steel and timber panels, and the layout of door and window openings. The main function of the steel members will then be to stiffen the wood panels so that they do not buckle out of plane, and of course to carry dead and live floor loads.