Type 2 (Wind Only) Moment Frame 8

[RFreund]

I have a couple of questions regarding wind only moment frames. Meaning when analyzing gravity loads the beams in the frame are pin-pin. However when analyzing lateral loads the connection is fixed.

1. Is my description correct in describing the design process.
2. Are these typical?
3. Can you provide me a reference that discuses the design of these frames.

I just have a hard "stomaching" this design philosophy but it seems like many frames in the past have been designed this way.

Any other insight is appreciated.

EIT

 

[Ron]

NO! NO! NO! Use your modeling as a best effort to predict how the structure will react under ALL applicable loading conditions. Don't pick and choose. The worst case will control.

If the structural need is lateral support for wind, then the gravity loads will probably be fine....but don't change the analysis to just accommodate the one condition.

With available software, you can model the entire structure (or at least a significant portion thereof), so changing the loading conditions is a simple process....leaving the structural model conditions the same...

 

[Lion06]

Here is my two cents. I've designed a number of "wind clip" buildings, and I have, in fact, designed them exactly as you say, but I'll add a few comments.

The idea behind designing for pinned connections under gravity load is that the connections are designed for specific, wind-only moments. It may be true that gravity end moments exceed your lateral end moments, but that doesn't matter here. If your gravity end moment exceeds your lateral end moments then the connection simply yields under gravity loads. It's important that the plate or angle be designed to yield BEFORE the weld or bolts fail - this ensures the rotational ductility of the connection and that it can actually undergo the deformations necessary to allow it to act as a pin for gravity.

It's common to design the beams for gravity only as pins, then check those beams with fixed ends for lateral only. Do not combine gravity and lateral for this type of system when checking the beams. When you check the beams for lateral moments, be sure to use the entire length as unbraced for the negative bending.

Also, please note that this lateral system is not a candidate for the DAM. Well, technically speaking, it is, but it would be so burdensome to do that, that it's not worth it. You'll need to use the effective length method, but there are a couple modifiers you'll need to apply when calculating k. The nature of the connection is such that both end connections of a beam yield under gravity load. When a lateral load is applied, the windward connection tries to unload (which means it has moment capacity), but the leeward connection tries to load further (which means it's acting as a pin). So, the frame is really only fixed at the leeward ends of the beam for wind. See the commentary with the alignment charts for modifications to L and G for the girders.

Also, I typically use the actual drift for a 50 year wind for the drift that will be seen during a 10-year event. The reason is that a 10 year event is allowed to be multiplied by 0.7 for the 50-year event, and the flexibility of the connections typically results in a 50% increase in drift. So 0.7*1.5 = 1.05 * 50-year drift (I just use 1.0 * 50-year drift).

Lou Geschwindner has a couple papers on the topic. I'll post them.

 

[Lion06]

Please note that there is a process called shakedown and that the order and direction of loading impacts the final results. The above post is a generally accepted design procedure for wind-moment frame structures.

 

[Lion06]

Here is another paper.

 

[Lion06]

One more paper. I have highlighted on this paper the reference to the increase in drift to account for connection flexibility. I was mistaken, it is 1.5 to 2, not capped at 1.5.

 

[Lion06]

The reason it's ok to design the beams as pinned for gravity (and this is conservative), is because the beam will never see more negative moment than the connection can handle. You will design the connection for very specific moment capacities, so that this negative moment is never exceeded in the beam. Once you check the beam for that negative moment with the full unbraced length, you're good, as far as the beam is concerned.

Now when it comes to columns - Because you're either designing a lateral system with full fixity or a gravity system with zero fixity, you need to play some tricks with the program to capture the actual column moments. I typically will specify 50 % of a fixed end moment to get cranked into the columns from the beam. This is often more than will actually occur when you do the analysis, but it makes it so you don't have to think to hard about it. It's not changing your beam or connection design, and I don't mind being conservative on columns.

 

[Lion06]

As a side note, the leeward connection being pinned increases k pretty significantly for the columns. This is really only a viable lateral system for 7 stories and under. You'll need to grab almost every column and beam framing into the column as lateral elements in your wind-moment frame. Depending on the height of the building, you'll probably need to fix the base connections, too, but you can design that for 50% of the fixed moment, since that's what you're limiting the moment getting cranked into the column to. I think you can justify using less, but that's sort of the approach I've adopted.

 

[msquared48]

Why would you ever model anything other than reality?

Mike McCann
MMC Engineering

 

[Lion06]

Because it gets you out of having to do testing for the moment-rotation curvature of what could be called semi-rigid moment connections. Also, if you model it as fixed for combined gravity and lateral, you will get very large moments on the leeward connections (gravity FEM + lateral moments, because they are additive at that location), and that goes against the very principle of the system.

The whole point of the system is that the connections yield under gravity load, but still provide moment resistance for lateral loads. If you design for gravity + lateral moments then you're losing the economy of the system.

 

[hokie66]

Thanks for your eloquent discussion, Lion. Sometimes we have to be reminded of how "rigid" analysis differs from tried and true design methods.

 

[RFreund]

Wow - Lion that was more than I could have hoped for - Your the man.

I'll have a look through and see if I have any questions. One question is why do you say it would be very burdensome to apply damn?

EIT

 

[RFreund]

Lion any chance you can re-post the last paper, for some reason it does not open for me.

EIT

 

[Lion06]

The reason it would be almost impossible to apply the DAM, as far as I'm concerned, is this. Let's look at notional loads in one direction (say in the positive X direction (which would be to the right on the page). Say you have 10 frames with 3 columns and two beams in each frame. The left end of each beam would need to be modeled as fixed (since this is the only connection that has LATERAL moment capacity, because the direction of lateral moment is opposite the direction of gravity moment so it's trying to unload) and the right end would need to be pinned (because this yields under gravity load and is trying to load further under lateral load, but has already yielded). Now, for notional loads in the negative X direction you'll need to flip those pinned and fixed ends because the direction of the lateral load has changed. You'll need to do the same thing for lateral loads in both Y directions. Don't even get me started on the torsional wind cases with lateral loads in both directions simultaneously. I think it's too much book keeping as to which connections need to be fixed and which need to be pinned - too many models to check.

 

[WillisV]

The FMC (Flexible Moment Connection)/Type 2 connection as put forth by Disque and later carried on by Geschwindner is a simplification of true PR analysis that has its place for simple buildings meeting certain limits (more on that below), but in my view for more complicated structures its use would be as a preliminary design tool and would be followed with a true analysis of the structure taking into account the actual moment rotation curves and loading history of the connections.

Eurocodes actually impose strict limits on the use of the FMC method to low rise simplistic type buildings (regular geometry, 8 floors max, 4 spans max, 16ft floor-floor max, 40ft span max with spans not more than twice or less than half of adjacent spans). Though these limits are not present in U.S. codes (the method is not as widely used in the U.S. anyway), it is my opinion that it would be prudent to use them as a guide.

The FMC method has several shortcomings as one would expect from an approximate method. First of all, connection deformations are ignored for the determination of lateral forces, thus the drift is unknown (you are relying on far reaching "fudge" factors to amplify your drift based on recommendations derived from a small class of 2-d frames). Also, since the force-deformation response characteristics of the connection are ignored, the distribution of forces is only correct if the connection can travel through its actual moment rotation curve in the shake down process without a non-ductile failure such as bolt shear.

White and Goverdhan lay out a method showing how to use the Direct Analysis method with the actual moment rotation curves to perform a true PR design, using the same example problem as Geschwindner and Dique that I would highly recommend reviewing as a more rigorous alternative to the FMC. (you can get this paper in the proceedings for Steel Connections VI conference for free here

http://www.aisc.org/WorkArea/showcontent.aspx?id=26572)

Note that the FMC shortcomings are (sort of) recognized in the 14th Ed. steel manual which states on p11-2 that "When used, the analytical model of the PR connection MUST include the force-deformation characteristics of the specific connection. As an alternative, flexible moment connections (FMC) may be used as a simplified approach to PR moment connection design, particularly for preliminary design."

I am not against the FMC method, just wanted to make sure everyone who uses it considers its possible shortcomings and considers a true PR analysis if warranted.

 

[Lion06]

Willis-
If you use the DAM with moment rotation curves, you would need to design for gravity + lateral moments, correct? This is not quite the same as the flexible wind moment connection. I agree there are a lot of limitations with the flexible wind moment connections, and that it's important to understand them before using this lateral system, but it has been used successfully on a pretty large number of projects.

AISC 360-05, section 11 gives a nice blurb on the uses of the FMC.

 

[WillisV]

Lion - correct - no it's not the FMC, it's better (and what FMC is approximating). The AISC manual blurb from the 13th ed. manual has been updated to what I quoted above in the 14th ed. manual to be more in line with current research and to more clearly denote that the FMC is an approximate method best suited to preliminary analysis.

 

[WillisV]

Note that the Section title in the 14th Ed. Manual has even been changed from Design of Flexible Moment Connections to Design of Partially Restrained Moment Connections to move the emphasis away from solely the use of FMC.

 

[Lion06]

PR connections, from what I've heard are not usually a cost-effective option, because they require testing to determine the moment rotation curve. Is that a true statement?

 

[WillisV]

Yes and No. If you are using one never used before, yes, just like in seismic if you were not using a pre-qualified connection you would need to get a P-R curve. However, for the generally used top and bottom angle connections (same ones you probably used in your FMC analysis) there already have readily available published moment rotation curves that can be used, with references provided to them in the 14 Ed. manual, so there is no additional cost other than design time. Speaking of which, as you pointed out it is often actually harder to try to figure out how to use direct analysis with FMC to meet all the simplifying assumptions than it is to just use it with real PR analysis.