Ponding Load

[L_Bey]

I am designing a building with minimum snow load (South Carolina) and am looking at the rain load requirements. I typically work in the north east where roof snow loads can easily be in the 50 psf+ range so rain loads typically aren't an issue.

Looking at ASCE 7-16, rain load needs to be evaluated based on 15 minute duration/100 year return period, with the static head of the difference between the low point of the roof and the rim of the secondary drainage system (assuming that the primary drains are blocked), plus the hydraulic head associated with the secondary drain system. My secondary drains are scuppers at the exterior, with each drainage area having a secondary drain. Using the tables in the commentary it's pretty straightforward to calculate the hydraulic head for the given scuppers based on the 5.2(d_s+d_h) equation. However it does then say "account for ponding instability and ponding load", which would cause additional load on the roof. This additional load would be based on the deflection of the roof under the regular rain load, say 1" of deflection adding 1" to the water depth we're designing for. Then check that load, make sure the deflection isn't greater than the 1" allowed for, etc.

ASCE 7-22 has an additional term in that equation, d_p, to account for the ponding load. Is it reasonable to set the ponding depth to say 2" maximum, calculate the required rain load based on that number, and then design the roof framing to not exceed 1" deflection each for joists and beams (2" deflection maximum)? Would that satisfy the requirement to consider ponding? Or is there another consideration that I'm missing?

I've applied my total load (static head, hydraulic head, and ponding depth) to the whole roof, which results in a total load of 55 psf (6.5 in static head due to roof slope between internal drain and scupper, 2 in hydraulic head, 2 ponding). This is a bit conservative, since the actual depth of water between the internal drain and the scupper will vary with the roof slope. The ponding load will also vary between a maximum in the center of the bay and nothing at the columns. But trying to do a variable load doesn't seem like it would be worth the engineering time, or going to special joists.

Thoughts? Anyone else have a standard practice for bar joists and ponding? I'm beginning to see why all our flat roof designs have secondary internal drains at +2-3" above the primary instead of scuppers.

 

[lexpatrie]

I think your approach deviates from the design standard approach, but 2" of rain applied across the entire roof would be a lot more that what a "normal" ponding situation would create. There are points where the structure can't move downward (i.e. walls), so it is likely conservative.

As this is technically a stability analysis, the conservative approach needs to be looked at carefully, but it can be done this way safely. You may put "extra" material in to the roof, but that's not all that harmful. We just want to avoid the situation where the roof is heavier than it would normally be AND it still has a ponding defect.

I'd caution you that while a ponding check may create a structurally safe roof, it may (if adequate slope is not provided) violate the roofing manufacturer's installation instructions, AND the code.

The more recent approach I've seen tends to be to slope the bar joists 1/4" per foot, or the steel, AISC, as you know, has a ponding check, and SJI has a different ponding check. Keep in mind that sloping multiple bays TOWARD each other is now defined as a susceptible bay (in ASCE), but perhaps not in the actual building code, I recommend using the more stringent definition of the two....

Conversationally, there are two kinds of articles on ponding, the really technical ones that get wildly abstruse (cough Molina was it? Marino, et. seq.), and the basic "water bad, slope roof!" articles by, say, Patterson and his compatriots. Lawson has a few good articles (at least two) that strike the middle ground, which help understand the more abstruse technical and design articles, but aren't enough on their own to design with.

I thought I had links to some of the ponding papers in the FAQ.

1966 - Ponding of Two-Way Roof Systems - Marino

Not exactly comprehensive, sorry.

 

[DaveAtkins]

Designing for blocked drains and designing for ponding are two different things. What you have described is designing for blocked drains, but not a typical ponding check. A ponding check verifies that the members you choose are stiff enough so they will not progressively collapse due to water accumulating on their deflected shape. I use AISC Appendix 2 to check ponding.

 

[lexpatrie]

A blocked drain is inherent to a ponding check. So, No.

 

[DaveAtkins]

lexpatrie,

The excerpt you posted does not talk about ponding. Theoretically, a roof could be OK for "8.2 DESIGN RAIN LOADS" but still not be OK for ponding.

 

[ChorasDen]

8.2 states that "each portion of roof shall be designed to sustain the load of all rainwater that will accumulate on it, assuming all drainage systems that meet any of the following criteria are blocked", I read that as accounting for ponding effects, the curious part is it doesn't explicitly discuss secondary effects additional weight of water (due to deflected shape of supporting members), but merely indicates you need to account for the static weight of a volume of water wherever blocked drains are to be considered. However, any prudent engineer should be aware of feedback effects of rainwater and the developed deflected shape of the structural member as the water accumulates on the structure.

 

[DaveAtkins]

I will agree with that, ChorasDen. If you account for the additional water due to the deflected shape of the members, then I suppose you don't have to follow AISC Appendix 2.

But it is an iterative process. The deflected shape allows a little more water to accumulate, which causes a little more deflection, which allows a little more water to accumulate, etc. That is what AISC Appendix 2 checks. It makes sure the deflection converges to basically nothing.

 

[ChorasDen]

I agree about the process acting iteratively, luckily, in my experience the solution converges quite quickly, so there is not much difficulty in determining. AISC applies to steel buildings, but the section that Lex posted looks to be from ASCE7, which applies to pretty much all buildings we design, regardless of material.

 

[lexpatrie]

Since the OP hasn't responded, I think they got what they wanted and I don't need to argue with you, Dave.

The excerpt I posted did address your comment because you claimed there's a distinction between a blocked drain analysis and a ponding analysis, and intrinsic in the rain load case definition is the requirement that the primary drain be considered blocked.


Source: ASCE 7-22

Further, the OP describes JOISTS (i.e. either wood joists, or open web steel joists), Open web steel joists are NOT under AISC and AISC doesn't apply. SJI publishes a guide for how to do ponding checks on joists. Wood joists are not under AISC either.

More to the point, though, is the OP mentions normally designing for snow loads and demonstrates a lack of familiarity with the design standard for ponding.

The concern I have with that is ponding is required for snow loads, too. So they should already be doing this for their northern state projects.


Source: ASCE 7-22

And for the compulsive, Chapter 7 is "snow loads"

 

[ChorasDen]

I think they got what they wanted and I don't need to argue with you, Dave.

*proceeds to argue*

 

[lexpatrie]

Math checks out.

 

[DaveAtkins]

If AISC Appendix 2 is used to check ponding, it includes provisions for open web steel joists. I have never heard of anyone being concerned about ponding on a roof constructed of wood rafters. I suppose it is possible, but I think the spans are generally shorter so deflections are smaller.

 

[ChorasDen]

Hate to pile on, but I can order and ship 60ft+ long wood trusses or engineered wood products very easily. With lead time, we can get those made to 100ft or more (shipping & manufacturing a single piece being the biggest challenge). Ponding is a universal concern, regardless of material.

 

[lexpatrie]

The what, 1996 LRFD for wood construction has a procedure for ponding on wood. ASCE 16 wasn't it?

 

[ChorasDen]

I'm not familiar with that one Lex, but what I think is more interesting, is that wood stiffness is based on the mean value of the representative population. If we consider ponding a collapse and life safety concern, should we be using the mean stiffness value to calculate deflection, or should we follow Appendix F of the NDS and use a 5th percentile stiffness value for the member? For reference, that takes a #1 Southern Pine from an modulus of 1.6*10^6 psi to a value of 0.94*10^6 psi. That'll really affect your analysis...

 

[DaveAtkins]

Good question. I think I would use the higher E value, because ponding is associated more with deflection than stress.

 

[lexpatrie]

Just when I think I'm out, they pull me back in.

@chorus,

You mean Emin? That's an interesting question, but my familiarity with Emin is it's more of a computational/tabulated artifice to unify LRFD/ASD design in wood to a single design equation (for bending and compression) rather than something that really "occurs." It's not typically used for actual deflection checks.

I could be misguided but that's my off the cuff answer.

Speaking further off the cuff, Emin affects the Fb* and Fc* values that you'd be using in a ponding check if I had to derive it from first principles, if you catch my drift. I'm not sure what the design standard actually said, I haven't looked at it lately.

At one point (say 2004? Condo/Apartment) I did specify a wood truss (say 16' spans, 24" spacing) be checked for ponding load, they ran it for 60 psf instead of the 38.5 or 42 snow, which seemed alright as an alternative, which sounds a lot like what our missing in action OP was planning to do, I think I took that 60 psf minus the 38.5 or 42, divided it by 5.2 to get an inches of equivalent water depth, and then looked at the deflection of the truss, which was less, therefore seemed OK.

(Duration of Load footnote.... Not sure there's an established rain load duration for wood.... I guess you might argue 1 hour, probably better to use two weeks if you aren't in a snow region, if you are in a snow region, use snow duration, two months, since there's a provision for ponding in ASCE 7 Chapter on Snow loads...). A lot of these ponding collapses in the real world seem to stem from an accumulation of water over time, say a few rainy weeks, coinciding with a blocked drain, but most of these articles on "actual failure" are VERY evasive about where it was, what it was, or when it happened, and NONE of them spend much time on rainfall in the period leading up to the collapse, or not the normal ones, I think Patterson had one that was an exceptional storm and framing perpendicular to slope, that accumulated load differently so it was a single storm on that one.

This was back when I wasn't probing architectural drawings for detailing of roof slopes (well, not on the first one). For reference at one point (different project, a school) I got into a tussle with my boss over specifying weld washers on a metal deck, (required in the fire rating the Arch selected for a school, not my fault but naturally the Arch wasn't showing the weld washers, imagine the change order if anybody caught on), and another tussle (probably the same project) over a roof slope of 0.24" per foot versus the required o.25" per foot. Hypothetically this was deficient because the steel joists were set on coursing (probably not something I lectured that architect about, but it's possible as I was fond of lecturing people back in the day, particularly architects) and the Arch decided it was close enough and didn't flag it for me. Some ponding analysis does NOT change the steel joist design. It was a long argument with my boss at the time, and the check was already performed and worked "as-is" (it's ALSO possible that's the project the boss said to design the joists at two feet longer than measured and for a ballasted roof in case the architect wanted it and changed their mind, and then got annoyed I didn't rerun everything last minute to reduce the weight of the joists based on the actual span,.... those two things may play together in the "ponding didn't change the design" aspect on that one).

I did find this, (Woodworks article below), but I don't recall the intricacies of the LRFD wood ponding stuff, I think I remember it being disappointingly vague and short. I surely have it around somewhere in my e-horde.

Considerations for Roof Ponding in Low-Slope Wood-Frame Roofs, Woodworks website​

[ These "expert tip" articles always annoy me because there's no author and no date. Woodworks doesn't seem to be interested in that little qualm from little old me, though.]

Sorry, yet another TLDR answer.

ASCE 16-95, LRFD for wood construction

On an unrelated note the 60' span is the threshold for special inspection of wood and cold-formed trusses... as of the 2009 International Building Code, as I recall, that provision is collapse during construction related, probably that Florida Dollar Tree or Dollar General.... Nope, Never Mind, that was 2021.

Investigation of the November 8, 2021, Partial Collapse of Wood Roof Trusses during Construction of Dollar General Store, Orange City, Florida, OSHA, May 2022, Alan Lu.

Wait, it's not by Mohammed Ayub?

I did find an outline of one in Oregon....https://osha.oregon.gov/OSHAPubs/reports/fatal-descriptions/acc-fatal-cal-yr08.pdf

Not on topic, I'll stop now.

Ok I found it anyway.

Investigation of the July 13, 2007 Collapse of Roof Trusses in Township of Franklin, NJ, Ayub, OSHA, October 2007.

 

[ChorasDen]

You mean Emin? That's an interesting question, but my familiarity with Emin is it's more of a computational/tabulated artifice to unify LRFD/ASD design in wood to a single design equation (for bending and compression) rather than something that really "occurs." It's not typically used for actual deflection checks.

What? I don't follow what you are trying to say here. The concept of bending stiffness in wood is based on the sample population average, without adjustment. What this means is that for any given graded piece of wood, you have a 50% chance the stiffness is lower than assumed, and 50% chance that it exceeds the value assumed. This is due to deflection typically being considered a serviceability concern, and not a life safety concern, so there are no duration of load or percentile adjustments taken from the base stiffness value calculated when member design values are determined, unlike for bending or shear.

Speaking further off the cuff, Emin affects the Fb* and Fc* values that you'd be using in a ponding check if I had to derive it from first principles, if you catch my drift. I'm not sure what the design standard actually said, I haven't looked at it lately.

No it doesn't, or at least, not really. If you are taking a bending member that is unbraced, sure, but typically on roofs, you'd expect a roof diaphragm on the top compression edge, , or something stiff enough to support the ponding load, and gyp or something similar on the bottom face, if needed for bottom edge compression checks.

(Duration of Load footnote.... Not sure there's an established rain load duration for wood.... I guess you might argue 1 hour, probably better to use two weeks if you aren't in a snow region, if you are in a snow region, use snow duration, two months, since there's a provision for ponding in ASCE 7 Chapter on Snow loads...). A lot of these ponding collapses in the real world seem to stem from an accumulation of water over time, say a few rainy weeks, coinciding with a blocked drain, but most of these articles on "actual failure" are VERY evasive about where it was, what it was, or when it happened, and NONE of them spend much time on rainfall in the period leading up to the collapse, or not the normal ones, I think Patterson had one that was an exceptional storm and framing perpendicular to slope, that accumulated load differently so it was a single storm on that one.

Interesting point here, curious if ponding leads to further stiffness loss due to an incomplete waterproofing membrane, which further increases ponding load.