Stiffness of Partial Moment Connections

[FishTheStructure]

I typically try to avoid specifying full moment connections and instead list the factored moment that I require from a connection in an effort to decrease the steel costs.

However, I have a situation where my factored moment is only 10% of the moment capacity of the weaker member being connected. If the connection is designed for this factored moment, can I really expect the stiffness of the connection to behave as if it were connected for 100% of the moment capacity? If not, is a full moment connection required to assume that deflections can be based off of a fixed connection type?


Great spirits have always encountered violent opposition from mediocre minds - Albert Einstein

 

[KootK]

You could specify a connection stiffness value to the fabricator and ask for them to provide justification for the stiffness estimate that they provide. However, that's just passing the buck on a difficult problem to some extent.

Much depends on the type of moment connection that you end up with. If it's a classic welded or bolted moment connection the less weld or fewer bolts probably isn't a big deal. If you end up with an extended end plate connection on the other hand, the stiffness impact could be substantial. AISC's design guide on that issue may have something to say about connection stiffness.

I do think that your concern is valid. I rehabbed an excessively flexible facility a while back that had moment connections at every beam and column joint. The designer spec'd 10 kip-ft at each moment connection with no mention of stiffness or expected connection details. He got 1/2" wide flange column cap plates attached to beam flanges in a four bolt pattern, sans stiffeners. I checked the capacity and the 10 kip-ft is there and it jives with the lateral loads. However, there are signs of lateral building movement all over the place.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[asixth]

If the moment at the connection is low (10% of the member capacity) and I have wanted to utilize a fixed connection for the analysis I have always designed the connection for a minimum 50% of the section capacity.

I don't have any reference material to back this up.

It has come up on this forum a few times and usually results in a discussion of sorts about semi-rigid connections and how they should be analysed and designed.

thread507-321048

 

[FishTheStructure]

Thanks for the responses.

For my current situation I have multiple small moment frames supporting glass spandrel glass therefore I am more concerned about deflection than moment capacity. For my situation I have decided to use full moment connections. When I have more time I will look into this a little deeper in the hopes of coming to a more agreeable solution.

Great spirits have always encountered violent opposition from mediocre minds - Albert Einstein

 

[bookowski]

I think Kootk is spot on. I think it has less to do with the capacity than the type of moment connection and details. Anything that transmits forces through flexure, i.e. end plate connection or connections to flanges without stiffeners gets soft. Otherwise a reduction in bolts/weld should not be a linear relationship with flexibility, i.e. going from 6" of weld to 3" shouldn't halve the flexibility. Maybe you show the connection type that you want along with your demand.

 

[Lion06]

I agree that going from 6" of weld to 3" of weld won't halve the flexibility of the connection for a given flange plate, but going from a 8" x 1" flange plate to a 8" x 1/4" flange plate should have a big impact in reducing the stiffness of the connection. It's not realistic to think that the reduced moment demand will only affect the bolts and welds. It will absolutely affect the size of the flange plates.

We also need to consider the rotational stiffness over the range of expected moment demand. Going from a 8" x 1" plate with 6 bolts to an 8" x 1" plate with three bolts would have little impact on the rotational stiffness up to the capacity governed by the three bolts in shear. Going from a 8" x 1" plate with 3 bolts to a 8" x 1/4" plate with 3 bolts will result in a very large decrease in rotational stiffness throughout the entire range of moment capacity.

 

[bookowski]

Agreed that going from 1" flange plates to 1/4" would greatly affect the stiffness of the connection, idealized it should be about linear. However the impact on the overall stiffness of the frame would be much much less. A reduced connection stiffness is analogous to a reduced beam section over a short length of the beam, generally this has little impact on the global behavior but of course depends on the situation. I would guess that if you ran a simple single bay with full fixity vs 1/2 fixity connection you'd see 10% or less change in global stiffness - just guessing though.

 

[RFreund]

Maybe I'm missing something but I can't see how if the moment demand at the connection is 10% of your member size. Using an 8"x1" plate vs a 8"x1/4" plate is going increase the frame stiffness much. Assuming the frame has much larger dimensions that 8" tall. I mean essentially your talking PL/AE PL is so short your talking 100's of an inch right? I think it has more to do with how the connection is detailed i.e. end plate (more flexible) vs top / bot flange welded to the column (less flexible).

EIT

http://www.HowToEngineer.com

 

[KootK]

Joint flexibility due to the use of an extended end plate and joint flexibility due to the use of smaller flange plates are both extra joint flexibility. The only difference is magnitude. The short length of the connection is equally applicable, or equally inapplicable, to both scenarios. It's a good point Lion06.

I'll try a more nuanced version:

1) If you wind up with a flexible, non-conventional moment connection detail, expect joint stiffness to take a significant hit.

2) If you wind up with a conventional moment connection detail, expect joint stiffness to take a modest hit. I suspect this would be linear-ish as Bookowski suggested. Your moment/rotational stiffness ratio would remain roughly constant so you could expect about the same ballpark level of joint stiffness for whatever level of connection moment is specified.

Asixth's method is sounding better all the time.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[canwesteng]

I'm with RFreund. Different flange plate sizes will have a marginal effect on stiffness imo, to the extent you can still assume full moment connection. The issue is if the moment can be resisted at the connection without stiffeners if it is quite small.

 

[Lion06]

rfreund - the elongation of 1/4" flange plates is small for an 8" length, but it would be 4 times larger than for 1" flange plates. That directly relates to the moment-rotation curve for the connection. The stiffness of the connection will not have a linear relationship with the drift of the frame, but it will have an impact.

I made a very quick model of a single story frame with two columns and one beam, pinned bases and fixed beam/column joints. It's a very simple frame with lateral load only. One has the full stiffness for the entire length of all three members. The other has a 6" section at both ends of the beam with a I/Ig = 0.1 to approximate a connection stiffness of 10% of the beam stiffness. This is to assume the moment demand is 10% of the beam capacity and the flange plates are optimized. That much of a rotational stiffness reduction in the 6" at each end resulted in a frame drift of twice that for no stiffness reduction at the connection. This is a crude model with no vertical loading, so second order effects are non-existent.

 

[bookowski]

See attached unscientific study. Single bay frame 20ft long with the connections modeled as a 6" segment at ends, stiffness varied from 5% to full. Once you get above around 20% it doesn't seem particularly sensitive. I'm sure this would vary with configuration but it provides some info in general.

I agree that a connection with 10% stiffness should be considered in your analysis - but that's probably not a likely connection. Again maybe you want to specify some lower bound, i.e. flange plates have X% of beam flange area or X% of capacity etc. Personally I would rather have real forces on the drawings and impose a lower bound some other way just so that anyone looking at the drawings in the future (including myself) can make sense of the forces shown.

 

[RFreund]

Lion06 - the old StructuralEIT handle if I remember correctly? I should've known not to 'challenge' your statements. Your Model sounds sufficient for this discussion and kudos for doing so. You're statements (and the results) start to make more sense when I imagine these very large/thick beam and column flanges being connected by sheets of paper (your initial argument). In your model, (hopefully you saved it???) what happens if you make the entire beam the same smaller moment of inertia? What is the deflection difference?

EIT

http://www.HowToEngineer.com

 

[Lion06]

Good memory, rfreund. I just enjoy the discussion.

If I make the entire beam have a I/Ig ratio of 0.1, then the frame drift is increased by a factor of 6, instead of a factor of 2.

 

[bookowski]

Interesting article on the topic for various configurations, seems to reinforce the idea that it doesn't make a huge difference at the lower end in variations of connection stiffness (for drifts):

http://www.ejse.org/Archives/Fulltext/2010/2010v1/20103.pdf

 

[canwesteng]

Sorry for the bump - just did some connection design myself and I came across in AISC design guide 16 that moment connections should not be designed for less than 60% of member capacity iirc. There are also some design factors added in if the connection is going to be assumed to be rigid (for the flush end plate connection design I did, a gamma.r factor of 1.25). In the future I think calling out rigid moment connections on the drawing and adding the design moment to the drawing might be best

 

[JeanLucPicard]

It's very hard to achive a 100% fixed connection... It depends on the web, flange thickness, distribution of bolts, stiffeners thickness etc...
There are few methods for checking, try reading EN1993-1-8, there are calculations to perform to check the predicted behaviour... If not, you could menage to model a simple beam with equvivalent springs that will mimic your connection and another model with fixed edges. If moments and forces do not differ too much, you can state that it is OK to assume fixed connections.

Live long and prosper!

 

[bookowski]

I believe that 60% limit is for end plate connections only and is one of the assumptions built into the design rationale, don't think it applies to other connection types.

See link that I posted earlier to paper on the topic, doesn't seem very sensitive until you get to very reduced numbers. Also ran a little test model and found similar results, also posted above.

 

[canwesteng]

Could well be it only applies to end plate connections, that's what I was working on.