Snow Loads on Roofs and Decks

For a large portion of the country the proper design of the roof and supporting elements of a building or other structure requires consideration of the load (weight) of snow on the roof. Many building departments have assisted designers by providing the `Design Roof Snow Load’ for their jurisdictions (or loads where there are local variations). Providing a `blanket’ number for design may make the design process easier, but it also opens up the design process to leaving some things out. Things left out include effects that cause the amount of snow on different surfaces, or different locations on a single roof, to vary, sometimes greatly. And another thing left out, sometimes, is common sense. Related to this, or not, it is the current trend of building authorities (at least around here) to provide, instead of a `roof’ snow load, a `ground’ snow load, and then leave it up to the designer to design the `roof’. And with good reason.

2. Ground Snow Load

Ground snow load is basically the weight of the snow on the ground. But not just on any given day; it is the amount of snow (weight) that might be expected to occur on the ground, say, once in fifty years. Or, more precisely, the amount of snow that has a one-in-fifty chance of occurring in any given year (presumably during winter). This amount of snow (or `event’) is a matter of weather, climate, hydrology, etc. Hopefully the building department having authority over the area in which you hope to build can give you this value. If not, you may be able to pull the value out of a `model’ code, like the International Building Code (IBC), Figure 1608.2. This table gives ground snow loads for much of the country. And, in the case of the IBC, you will be directed to another document as to how to go from `ground’ to `roof’. This other document is ASCE-7, Minimum Design Loads for Buildings and Other Structures, by the American Society of Civil Engineers. And when we go from ground to roof later in this discussion it will be by basically using ASCE-7. But, for a lot of the west, it (IBC Figure 1608.2 and ASCE-7) shows `CS’, meaning the ground snow load must be obtained by case study. In Idaho the `case-study’ information can be found in Ground and Roof Snow Loads for Idaho, University of Idaho Department of Civil Engineering. In Washington State we use Snow Load Analysis for Washington, by the Structural Engineers Association of Washington. If the local building authority can’t point you to a `case-study’, contact the Structural Engineers Association of your state, or start Google-ing.

For both Idaho and Washington maps are provided that give `normalized’ ground snow load information. By `normalized’ I (we) mean `Ground Snow Load per foot of elevation above Mean Sea Level (MSL)’. So, let’s say your proposed site is at elevation 2850 ft MSL, and it is in an area of 0.030 (lb. per sq ft/ft or psf/ft). Then the `50-year’ ground snow load would be 0.030 psf / ft x 2750 ft = 82.5 psf (lb. per sq ft). That is a `chunk’ of snow. That is the weight of over a foot of standing water. (Water weighs 62.4 lb per cubic foot, or exerts 62.4 psf per foot of depth.) The site is probably up in the mountains of northern Idaho somewhere. (Presumably you already knew that.)

3. Roof Snow Load

To get the roof snow load for our structure, we turn (as directed) to ASCE-7. ASCE-7 provides the following equation to go from the ground snow load (p_g) to a `flat roof’ snow load (p_f).

… p_f = 0.7 C_e C_t I p_g

where,

C_e is a coefficient (factor) that takes into account the exposure of the roof structure (to wind, etc.),

C_t is a coefficient that takes into account the thermal properties of the roof/structure, and

I is an `Importance’ factor, kind of an extra factor of safety that `inflates’ the load for important structures like hospitals.

I will not repeat (or replace) the code here by repeating all the factors and how to get and use them, except to point out a few interesting things.

If the structure is sheltered and is not heated, and it is not some kind of important structure, then the factors are:

C_e = 1.2, C_t = 1.2, and I = 1.

If we dump these values into our basic flat roof equation we get,

… p_f = 0.7 (1.2) (1.2) (1) p_g … = 1.0 p_g …

the flat roof snow load and the ground snow load are the same … which makes sense (doesn’t it?).

And if the roof is flat and extremely well insulated, and sheltered, we should expect the same. (Although I would argue against a flat roof in any `snowy’ area.)

If you have a greenhouse and keep it heated, then C_t = 0.85; see, the `code’ makes sense; it is taking into account that you’re going to be melting some of the snow.

For a lot of structures … partially exposed, roof R-values less than 25, and not of great importance, we get, (C_e = 1.0, C_t = 1.0, and I = 1)

… p_f = 0.7 (1) (1) (1) p_g … = 0.7 p_g … or 70% of p_g.

For the `locale’ of the example above we would calculate,

… pf = 0.7 (82.5 psf) = 58 psf.

But, in country with that much snow, I would recommend against a flat roof. I would design a sloped roof.

4. Sloped Roof Snow Loads

The `code’ (actually, ASCE-7) let’s us take into account that a sloped roof will collect less snow, generally, and the more sloped, the less snow. This is accomplished using Figure 7-2 (of ASCE-7) which takes into account roof slope, the thermal properties of the roof (via C_t), and roof surface (slippery and unobstructed, or not). And it does this by allowing us to multiply p_f by the factor C_s to get p_s, the sloped roof snow load. In equation form,

… p_s = C_s pf

Let us take the example of a well-insulated (cold) metal roof with C_t = 1.1, sloped at 8 on 12, and in which the snow is not obstructed from slipping off. We will get C_s = about 0.65; so, in our example,

… p_s = 0.65 (58 psf) = 38 psf.

That’s a nice reduction from 82.5 psf.

But, it comes at a cost. The roof must be of a slippery material, and there must be noobstructions to sliding action. No snow stops, no `valleys’, no anything that will prevent sliding. The photograph accompanying this article shows a nicely sloped metal roof, but, aha!, at the front door a covered entry-way roof that inhibits sliding. Sorry, at least for that portion of the roof, don’t take the `unobstructed slippery surface’ reduction. You could, however, legitimately use the C_s that gives a (lesser) benefit for slope only. (C_s would be a higher value, giving less reduction for the same slope.)

5. Drifting Snow

Drifting snow needs to be considered, obviously. I’ll discuss two common situations. First, a gable roof where snow from one side of the roof blows off one side or over the top and more snow settles on the other side (unbalanced snow load). We get snow where I live and I see this all the time. You’ll have to buy or borrow the code (ASCE-7) to get the numbers, but it makes sense. The second common case is that of `valleys’. Obviously. Snow will drift into valleys with accumulations up to as much as twice the flat roof or sloped roof value. Add changes in the roof plane or roof profiles that could `drift’ the snow … guess what … the snow will drift. In general the snow will drift off the high spots and fill in the low ones. And while the calculations that we are required to use to account for such may not be `easy’ – they do, in the end, in my experience, make sense and reflect what actually can happen, and in many cases, has. So, I leave this section with a couple rules of thumb: 1) snow may blow off one side of a roof and accumulate on the other side to magnitudes 20 or 30 or so percent greater than the `equation’ values above, and 2) in valleys the snow may accumulate to as much as twice that of a single gable or hip roof.

6. Lower Roofs and Decks

Lower roofs and decks are `magnets’ for snow. Think about it. Snow blows off the roof, or an upper roof, and where does it go? Or, snow slides off a roof, and where does it go? If it ends up on a lower roof, or your deck, you better account for it. And, yes, the code provides a rational approach to do so. As I run the numbers I often get accumulations of up to THREE TIMES the roof snow load ending up on lower surfaces. If you don’t believe that, then take some time this winter and travel around snow country and look at what snow does. And while you might argue that you’ll shovel it off as fast as it accumulates, the building official will argue that you might be gone on vacation, or you might sell the house to someone who spends the winter in a warmer climate. So, be safe, follow the code, and design against potentially huge amounts of snow. And while steep, slippery surface roofs are fabulous for accumulating less snow, it does go somewhere. Some photographs of snow sliding, and not sliding, are here.

7. Things ON the Roof

The weight of things on the roof, such as air handling or other mechanical equipment must certainly be taken into account in design, and the effect they will have at causing drifting snow. I include here parapets. Snow will accumulate aside / behind parapets, and drift up in front of and behind obstructions, including steps in the roof surface. And since we cannot generally control the direction of the wind, this motivates a rather robust roof design all around such obstructions.

8. Ice Damming

Snow tends to melt off warmer parts of the roof, travel down slope, and then freeze over unheated eaves. As a result, significant ice-damming may occur and the eaves must be designed accordingly. For `warm’ roofs the eaves must be designed for 2 p_f … twice the flat roof snow load. Sometimes this ice is on the eave surface itself, sometimes over the edge, as icicles, or both. (This past winter we had icicles over 10 ft long … fascinating, but also betraying the lack of insulation in our roof.) Along with this it should be noted that unheated roof rakes will tend to carry more snow than adjacent (heated) main roof surfaces.

9. Closing Remarks

First, in closing, let me mention maintenance. Well-meaning maintenance personnel wanting to keep mechanical equipment on the roof clear of snow may well pile it somewhere else (and dangerously so). I am not sure how to explicitly deal with this in the `design’ of a structure, except to return to the idea of providing a rather robust roof system (all) around such equipment. Second, in areas of high wind or winds from a predominant direction, unexpected things can happen. I have observed huge drifts at the ends of gable structures, where I would have otherwise expected the snow to end up (instead) along the sides (from the steeply sloped slippery surfaces). Wind vortex action (say, along a parapet) may take snow from one location (next to the parapet) and put is somewhere else (where accumulations were not expected). Also, while an `obstruction’ may preclude you from counting on the snow sliding off, it shouldn’t be counted on preventing it (from ending up on the lower surfaces). So, in conclusion, `follow the code’, for, though complicated, it will take care of much of what we have discussed here, `wrought to us’ through the labor of others (and structural failures). And remember that the code provides minimum loads. Then apply (your growing) common sense and understanding of what snow does on structures. You may very well find yourself adding to these minimums.

And if you want a simple way to measure snow that is already on the roof, or ground, one is provided here.

10. References

International Building Code, International Code Council, 4051 West Flossmoor Road, Country Club Hills, IL 60478.

Minimum Design Loads for Buildings and Other Structures, ASCE Standard ASCE/SEI 7, American Society of Civil Engineers, www.asce.org.

Ground and Roof Snow Loads for Idaho, R.L. Sack, A. Sheikh-Taheri, University of Idaho, Department of Civil Engineering, Moscow, Idaho 83844.

Snow Load Analysis for Washington, Structural Engineers Association of Washington, P.O. Box 4250, Seattle, Washington, 98104.