Problem with LRFD Combination 3 (ASCE 7-10)

[pyseng]

I am not trying to start another LRFD vs ASD discussion here. But I was just wondering if anyone can either validate my problem or offer any insight that might help me make better sense of this issue. I have reached out to the ASCE committee on this issue but they didn't seem to quite understand what I was talking about.

As given in the title the crux of my issue revolves around LRFD Combination 3... specifically, the 0.5 W term.

If we take a quick step back and look at the ASD combinations, we find that we can categorize them as follows (ignoring Lr and R for simplicity):

GRAVITY EFFECTS:

1. D​

2. D + L​

3. D + S* (*snow plus drift)​

4. D + 0.75 L + 0.75 S​

LATERAL EFFECTS:

5. D + (0.6 W or 0.7 E)​

7. 0.6 D + 0.6 W​

8. 0.6 D + 0.7 E​

COMBINED EFFECTS:

6a. D + 0.75 L + 0.75 (0.6 W) + 0.75 S​

6b. D + 0.75 L + 0.75 (0.7 E) + 0.75 S​

I find the ASD to be rather elegant and straightforward. And as a designer of low rise structures in Northeast US, I like how snow drift only appears in COMBO 3 (when snow is the primary load effect). This allows me to deal with snow drift in a more localized, conservative manner. And it makes my modelling effort much more simple and I have more confidence in the results. If you ever tried to explicitly define snow drift loads in RAM on a building with multiple low/high roof transitions, parapets, and roof projections, you will know what I mean. In my opinion, the way ASD is setup sort of leads you down this path of thinking of snow drift like a sort of "components and cladding" load effect, one that needs to be applied to some elements, but not taken through the structure. I may or may not be wrong about that, but that is my own sort of engineering judgement way of thinking that I think applies in 98% of cases. At the very least, it certainly allows you to neglect snow drift in a separate lateral system model, which in and of itself is nice.

Now... let's try to do the same categorization for LRFD.

GRAVITY EFFECTS:

1. 1.4 D​

2. 1.2 D + 1.6 L + 0.5 S​

3. 1.2 D + 1.6 S* + (L or 0.5 W) (*snow plus drift)​

LATERAL EFFECTS:

4. 1.2 D + 1.0 W + L + 0.5 S​

5. 1.2 D + 1.0 E + L + 0.2 S​

6. 0.9 D + 1.0 W​

7. 0.9 D + 1.0 E​

This is actually pretty nice, looks like the consolidated a little actually... they got rid of the dead + lateral cases. Also, notice how snow drift is only required in COMBO 3, again, because snow is the primary load effect. But wait... wh... (L or 0.5 W)? What is that lateral load case doing there? How am I supposed to CHOOSE between L or W? They are two different load effects. I would understand better if this was split into two cases... 3a. 1.2 D + 1.6 S* + L AND 3b. 1.2 D + 1.6 S* + 0.5 W. Or maybe if they stipulated that that W term is only components and cladding wind... that way it sort of acts like a gravity load. But what I hate most is that this ONE term means that I can no longer create a separate lateral system model and keep snow drift out of it.

Am I right or missing something? I know I can get a bit winded in my rants, but if there was a better way to explain this too, that would be helpful to send over to the committee again.

Thank you!

 

[Aesur]

Why are you splitting your load combinations into gravity and lateral?

As for wind and seismic effects, there are vertical components of each that apply to the "Gravity cases"; in which case I believe you should be including snow drift in cases 6a and 6b as well.

As for choosing between L and W, you don't choose, you check both.
For instance, combination 3 LRFD has the following combinations:
3a. 1.2 D + 1.6 Lr + L
3b. 1.2 D + 1.6 Lr + 0.5 W
3c. 1.2 D + 1.6 S + L
3d. 1.2 D + 1.6 S + 0.5 W
3e. 1.2 D + 1.6 R + L
3f. 1.2 D + 1.6 R + 0.5 W

 

[phamENG]

I agree with Aesur. Those combinations are not meant to be grouped as gravity, lateral, and combined. They're all combined. If the load combination happens to only have gravity loads, then you pick your favorite stability poison and, more often than not, you add some fraction of the gravity load as a lateral load. Where you have wind or seismic, you apply both the vertical and horizontal components to the structure at the same time for the MWFRS. And as Aesur says, you have to check both versions of load combination 3. ASCE 7 doesn't provide every permutation of the load combinations - have you ever looked at the list from a PEMB model output? It's pages long of 8pt font.

I'll admit I'm a bit further south so I don't have to deal with a lot of snow (most of the drifts I deal with are barely larger than the roof live load), but based on the experience I do have and my knowledge of the subject, I disagree with your comparison of snow drifts to C&C wind loads. C&C is variable based on the member's tributary area due to the variability of wind pressures. Localized wind pressures can get VERY high, but they're VERY localized. Not sure if you've ever seen an output from a bank of pressure sensors on a wind test structure - they're all over the place. So it's quite plausible that a single bolt holding a wood plate to the top flange of a beam will see 35 or 40psf in it's tiny tributary area, but the probability of every single bolt and, therefore, the entire beam, seeing that same load at the same time to produce that load effect is essentially nill. So, the beam can be designed for a lower pressure because that is plausible (and probable within the limits of the code's reliability targets). So the C&C wind load is being taken through the whole structure, but in a probabilistic way not unlike live load reduction. Snow drifts, on the other hand, are a bit more predictable. The snow drift occurs in a "known" location on known members. It is quite possible and probably that it would occur across the entire face of a parapet. So that load should be considered and its effects through the entire structure should be considered.

 

[pyseng]

I was grouping the load combinations together based on their primary load effect, mainly for convenience to (hopefully) better articulate my point. The primary load effect in LRFD Load Combination #2 is snow. Because it is a primary load case, snow drift is required to be considered. By including wind in this combination, the framers of these combinations created a situation in which snow drift was required to be considered in combination with lateral loads. This situation is avoided in ASD combinations. In my opinion, the fact the ASD does not require snow drift in combination with lateral loads is a huge advantage.

I would respectfully argue that the ballooning of load cases is part of the problem here as well. The "or" cases imply that the controlling load effect can be selected and the others discarded. I don't believe the intent behind the (Lr or S or R) term is to require a subdivision down to Case a/b/c. Instead, figure which load controls for the element and ignore the others. Consider, for example, why do they not condense LRFD combinations 6 and 7 down to a single combination... 0.9 D + (1.0 W or 1.0 E)? It would make sense to do so with the understanding that the "or" term is really just there to represent these sub-combinations. Also, the pages long output IS the problem and I find LRFD makes it worse.

A couple side notes:
1) Aesur: Cases 6a and 6b are specifically NOT required to consider snow drift.
2) phamENG: Well reasoned argument regarding snow drift v. components and cladding, I change my mind.

 

[Harbringer]

...the ballooning of load cases is part of the problem here as well...the pages long output IS the problem...

Agreed. You don't have to do a 3D model with hundreds of load cases to efficiently and safely design a low rise building (although you can). Some of the garbage output I've seen in calc packages makes me wonder if some people even know what they are designing.

 

[phamENG]

I agree that the number of permutations can get a bit ridiculous, but the principle isn't. pyseng, if you think about it you are going through those permutations in your design. By going through and selecting the worst one, you've at least qualitatively considered the three permutations of that particular combination if not quantitatively. The benefit of a 3D modeling program running through all of them is that it catches the strange or unique cases. What if you have a girder supporting two sections of roof - one with parapets and one free draining. The joists with the parapet may be controlled by rain loading, while the joists that are free draining are controlled by roof live load. Your girder then will need to be checked for both, no? I'm sure there's a configuration where you can bring snow in, too, but it's late and I can't think of one. So if you're sure that S will always control, then you can throw the others out. But you've still considered them to make sure they don't control the design.

You make an interesting point with the wind and seismic load combinations...they could have easily wrapped those into one. The commentary is silent on the reasoning. My guess is because they both need to be considered quantitatively. You can't throw out seismic just because wind is the greater load because seismic can have certain prescriptive detailing requirements. Sounds like we may agree here on at least some level.

Regarding the inclusion of snow drift with wind....how do you get snow drifts? The wind has to blow the snow there, correct? So it seems perfectly reasonable to me to consider wind acting on the structure with that snow load, since a blizzard likely brings some decent wind with it.

Thank you for the kind words - nice to know reasoning counts for something every now and again!

 

[Aesur]

@pyseng, I agree with you that you use the worst case of each combination, I was simply pointing out that you need to review each of those combinations to determine which controls and go from there. For instance, it's common in my area that Lr is always used over R or S simply because of the climate of this region so we rarely look at R or S loads.

I must admit I am limited in snow loading experience having done only a hand full with drifts, can you point me to where in the code it says you can ignore drift in cases 6a and 6b?

 

[phamENG]

Aesur, it's Exception 1 listed directly below the load combinations in 2.4.1.

 

[Aesur]

@phamENG, thanks, I don't know how I missed that. I may have over designed a few things...

 

[Settingsun]

Regarding the question about why the wind and earthquake load combinations aren't done as a single 'W or E' combination, it might be because you can't decide which governs just by looking at the design loads. They have very different acceptance criteria.

 

[pyseng]

I definitely agree that there will always be edge cases, but I disagree that the benefit of the 3D model is to catch those cases, that is the responsibility of the designer in my opinion. One interesting note there is that RAM Structural doesn't even let you consider Roof Live Load and Snow, you have to manually choose which one to consider.

And again, we agree on the idea that Snow drift + Wind makes sense in combination from a physical standpoint. However, the fact still remains that ASD does not include a Snow drift + Wind. In my opinion, this simple fact has considerable implications and, personally, makes ASD more appealing. On that same note, I actually do take a bit of issue with the idea that there is this sort of physical inconsistency between the two approaches - I would say that either ASD should start including Snow drift + Wind, or LRFD should take it out, but it doesn't make sense for one to include it and the other not to.

 

[phamENG]

Sorry - I don't think I was clear. I agree that it's the designer's responsibility. I'm saying that an advantage to 3D modeling programs in this context is that they can use the immense computing power available to run through those permutations in about as much time as it takes me to take a sip of my coffee. In doing so, it assists the designer in selecting the controlling combination (or permutation thereof) based on worst case load effects. Garbage in garbage out, and even gold coming out can be misused/abused.

I'm not quite ready to agree that ASD or LRFD should be modified to match in this regard. As I understand it, they each take a different statistical approach to reliability in an effort to arrive at a similar point. LRFD is intended to be more 'accurate' - considering the variability of loading and variability of material and workmanship separately rather than lumping them together as ASD does. Perhaps this caused the developers of LRFD to hone the load combinations a little more? Something worth looking into when I have time.

To our Canadian friends who might or might not be ignoring this: how do you guys handle these? One, you guys are used to lots of snow calculations, and two, it seems everything in Canada is Limit State Design (more or less equivalent to our LRFD).

 

[jayrod12]

In my experience up here in the frozen north, we do much of what has been indicated above, we use our judgement where possible to eliminate the load combinations that will not govern the certain scenario we are reviewing. Our typical load combinations look like the attached screenshot(please note this is the 2010 edition, not the current 2015 but it's what I had available at the time). As a note, we always consider snow drifts.