Snow Load Failures: Causes & Case Studies

Winter is quickly approaching and in some areas of the country, snow is already accumulating.  With more snow on the way and the inevitable roof damage and collapses that seem to follow at some point, Beau Menard, P.E., District Engineering Manager and Consulting Engineer with Engineering Design and Testing Corp. / EDT Engineers, P.C. in Charlotte, NC presented a visiting lecture to Penn State AE students on the topic of case study analysis of snow load failures.

Mr. Menard presented an introduction to the basic code items and procedures that need reviewed when investigating snow collapses.  He also put this all into perspective with a series of mini snow damage and collapse case studies that covered both common and unique contributing factors to failures from snow events.

Structural collapse and roof damage from snow can occur from a variety of sources and often results from a unique combination of events including snow drifting, roof geometry, storm overloads from ice and snow combinations, design deficiencies  and construction defects just to name a few.   A good background article on this topic can be found in the November 2008 issue of Structure titled “Structural Collapse from Snow Loads” by Michael O’Rourke.  Another good summary of snow collapses can be found from the Building Failures Forum guest post by Alyssa Stangl titled Snow-Induced Roof  Failures and Prevention Methods which also leads to a detailed discussion on the topic on the Failures Wiki site.

On a related topic, the Whole Building Design Guide contains a section titled Considerations for Building Design in Cold Climates: Avoiding Falling, Sliding or Windblown Ice and Snow from Buildings and Structures.  Another excellent supplemental resource is FEMA P-957 Snow Load Safety Guide.

Tags: , , , , ,

36 Responses to “Snow Load Failures: Causes & Case Studies”

  1. mkev
    November 20, 2014 at 10:15 pm #

    See this partial quote from the story above relative to our discussions on record snows and actual snow load vs design load in the Buffalo area:

    Even for the Buffalo area one of the snowiest and hardiest places in America this was one for the history books. The three-day total is close to the nearly 8 feet that the region typically gets over an entire year.

    “No matter how you cut it, this event will end up in the top five for the Lake Erie area,” said National Weather Service Director Louis Uccellini.

    Because the Buffalo area is so snowy, the building codes require homes and businesses to be able to handle up to 50 pounds per square foot on their roofs, which would be about as heavy as a slab of concrete 4 inches thick, according to Mark Bajorek, a structural engineer.

    As anyone who has ever shoveled snow knows, its weight depends in part on how wet or fluffy it is, not just on how deep it is. But Bajorek said some buildings may be close to that limit now, with more precipitation on the way.

  2. Christopher B.
    November 4, 2014 at 12:11 pm #

    In sloped-roof construction, what ASCE 7-10 thermal factor is appropriate for structures with ventilated, cold-eave detailing? The Fema P957: Snow Safety Guide p. 3-5, figure 4 shows that the temperature below the roof should be kept close to freezing to prevent ice dams. Could we then assume that residential roofs with this type of detailing should be treated as unheated structures?

  3. Jim P
    November 3, 2014 at 6:58 pm #

    A very interesting and informative treatise on snow loads produced recently by FEMA:
    Interestingly, the report states (Section 3.3.2), “[in] a report conducted by the Structural Engineers Association of Washington for the 2008–2009 snow season, heavy timber accounted for the second largest number of failures only
    to wood trusses.” The NDS allows a 1.15 allowable stress increase when considering load combinations including snow (as well as for other 2-month duration loads). If the 2008-2009 snow season failure rates are indicative of the general trend of failure rates, with wood trusses and heavy timber accounting for the top two building types which fail under snow load, should the NDS consider reducing the snow stress increase?

    • Nick D
      November 3, 2014 at 11:35 pm #


      As the treatise on snow loads produced by FEMA stated, the type of snow varies tremendously across different parts of the country. The code is only the minimum design values that designers must meet and could very well be increased. Section 2.2.3 states that “local authorities and structural engineers are most familiar with regional snow characteristics.”

      I wonder if this is more of an issue of the local areas needing a greater ground snow load value?

  4. Christopher B.
    November 2, 2014 at 11:19 pm #

    Is there a standard method for calculating snow loads on a green roof? For instance, should the soil be considered saturated with a snow load superimposed; or is dry soil with a standard snow load adequate? Also, does a green roof affect the thermal factor used to calculate roof snow loads because of the insulated thickness of the roof assembly?

  5. Yewande A
    November 2, 2014 at 6:26 pm #

    Following further discussion in class, I had some comments and a question about snow drift. Snow drift could cause considerable damage to building structures. I came across an image in (page 4) where the peak snow load of the drift was 130 psf and the ground snow load was 20 psf while the upper roof snow load was 15psf. The large difference in this figures put in perspective the magnitude of peak drift snow loads compared to ground snow load and the additional load it adds to the building. The article by Michael O’Rouke on Structural Collapse from Snow Loads highlighted that about 70% of snow related structural performance problems are from snow drift. In addition to structural performance problems, snow drifts could block access doors to a building, rooftop penthouse doors, mechanical equipment, air intake louvres, plumbing vents and cause damage to adjoining structures. Since not all possible scenarios are covered by ASCE, an experienced engineer could probably make design decisions from experience and infer from the scenarios that are covered in the standards. How else can engineers get guidance on designing for scenarios related to snow drift that are not yet covered by the standards?

  6. ShuvrajitG
    November 2, 2014 at 1:45 pm #

    Mr. Beau’s talk on snow load failure was quite interesting. It was surprising to know that agricultural buildings in North Carolina don’t have to meet the codes. My question is if the agricultural buildings don’t need to follow codes, it is quite probable that the structures are not designed by an engineer. In that case the structure could be under-designed and might fail under any load be it snow etc. In such circumstances how do insurance companies justify paying the owner for the damaged structure, when the structure couldn’t have taken that load?

  7. Sam d.
    October 30, 2014 at 9:39 am #

    I really enjoyed the lecture on snow loading. I am curious about agricultural buildings in places where they aren’t included in the code. When contractors build those buildings without an engineer looking at the design, do they make mistakes leading to failure because they are trying to cut costs or is it more likely because people are unaware of how snow-loading conditions affect the building? Are there any easily accessible resources for contractors building agricultural buildings to become more knowledgeable of reasonable construction practices?

    • Shuvrajit G
      November 3, 2014 at 10:55 am #


      I found some design guidelines for agricultural buildings. Although these guidelines are for the UK but they have very detailed design standards for agricultural building construction.

      The link will give details on the various agricultural building codes.

      I also found an interesting link. This document is a planning guide for agricultural buildings.


    • Xiaodong J
      November 4, 2014 at 1:40 am #

      Hi, Sam,
      For first question, on my own opinion, that kind of constractor is usually not aware of the loads, which includes snow load.

      For the second question. I think it really depends on regions. Typically, in the U.S., the agricultural building is exempted from “building” definition; and it doesn’t even require permit to start construction. Therefore, it is hard to tell. But I found that Canada has Ontario Building Code 2012 that states requirements and guides for agricultural building (of course, it is in Canada). There is a ppt. referrence of introduction of Ontario Building Code 2012.
      Also, in the U.S. some city building deparments might also have its own requirements and guide of agricultural building construction.

      Wish it might help.

  8. Todd H.
    October 29, 2014 at 8:51 pm #

    Thanks to Mr. Menard for the valuable lecture on snow loading and subsequent failures.

    In cases, like the chicken coops, where shoddy construction occurs due to the lack of requirements for agricultural buildings, how does an engineer go about convincing an owner to pay for improvements in construction and to make the structure withstand loading conditions that may or may not ever happen? Are there any legal or ethical issues that prevent you from presenting an owner with case studies in an attempt to convince them?

    • mkev
      October 30, 2014 at 8:08 pm #

      You would always design to what you feel are acceptable and reasonable standards even if the code in that location doesn’t cover it. And, you and the owner should expect installation in a workmanlike manner. In the case of agriculture buildings, you would likely have designed a better connection and expected the contractor to execute. What you need to be careful of is convincing the owner to pay you (the engineer) or a qualified professional to perform site inspection services so that you can catch the problems. Most owners don’t want a problem that will put them out of business for a period of time and as such they are usually reasonable about these things.

  9. Adam J.
    October 29, 2014 at 8:50 pm #

    Thank you Mr. Menard for an insightful lecture on collapses due to snow loads in North Carolina. Growing up in north western PA and attending Penn State Behrend for a year, I figured that the southern states would not experience significant snow loads to control over roof live load. However, after your lecture I realized that every now and again the southern states do get large snow storms.

    My question relates to ASCE’s snow load calculation. While a large number of conditions are considered in respect to thermal and exposure values, even when the worst case situation of a sheltered structure, intentionally kept below freezing, the flat roof snow load is equal to the ground snow load. Since the ground snow load assumes some melting throughout the year, is there ever a case where a structure may experience higher than the given ground snow load, besides drift and sliding? If so, are there any suggestions for choosing the design snow load other than an engineer’s educated guess?

    • Caroline K
      October 30, 2014 at 12:07 am #

      I don’t have a response to Adam, but am curious about similar factors. Would the design engineer ever be expected to predict a situation not mandatory to check for in the code? For example, if the roof was tilted in a way to cause snow to slide and pile up, the building was oriented so excessive drift formed, and there were other fairly common sense factors contributing to the snow load in one area of roof, would the engineer be expected to foresee that the load could be higher than that calculated in the code?

    • Todd H.
      November 3, 2014 at 10:39 pm #

      I did not find any information that answered the question. During blizzards and extremely harsh winters, the snow loads may possibly exceed the loads out of ASCE 7, but factors of safety in ASCE 7 and the structural design likely reduce the amount exceeded. In addition, I would believe that this is a case where the idea of “reasonable judgement” would come into play. If any issues arise due to excessive snow or lack of melting and the worst case possible from ASCE 7 and/or local requirements was utilized, the structural engineer would likely be free of liability.

      If this is a real concern for a certain project, a possible course of action would be to contact officials and industry professionals who work/have worked in the region to find out if there has been a history of snow accumulation that is not reflected in ASCE 7.

    • Beau Menard, P.E.
      November 7, 2014 at 9:02 am #


      I also went to Behrend for two years. One time we had over 2 feet of snow fall overnight. By that standard, the snow we experience in the south is not big. Remember the storm in two of those case studies only produced 8 inches of accumulation. The actual load from that snow was far less than the ground snow load.

      Remember the design load is based on your importance factor. In situations where the importance factor is greater than 1, you design load can be higher than the ground snow load. Remember too, that the standards set in the codes are (and should be) the minimum. Meaning that it is the design engineer’s discretion to increase the loads as they see fit. Trust me, as a design engineer, you will sleep better at night by being conservative in your load applications.

  10. Julia H
    October 29, 2014 at 8:29 pm #

    I really enjoyed Mr. Menard’s presentation last week; it was definitely eye opening to see how some buildings south of PA can react to smaller snow loads. I do have a lingering question about ice and ponding water on roofs.

    During my internship this summer I was looking at some structural drawings that had an allowance for ponding water on the flat roof, since the building was located in an area that got a lot of ice during the winter. I was wondering whether this allowance (~5 psf) was just a typical value or if there is a special ponding water map in ASCE 7-10. The concept of ponding water isn’t something I am too familiar with, but I know that in areas that experience drastically quick weather changes it could be a problem.

    • Zach B.
      October 30, 2014 at 9:27 am #

      Section 8.4 of ASCE 7-10 states that if a bay is susceptible to ponding, then they must be designed to it using the larger of the rain or snow load equal to the design condition for a blocked drain system.
      A susceptible bay is one where the roof slope is less than 1/4 in. per foot.
      There doesn’t seem to be a lot of information on how to calculate the rain loads which leads me to believe that it is not a condition that is checked for often. Would I be correct in assuming this? Is this because roof snow loads usually control?

    • Nick D
      November 4, 2014 at 7:44 am #

      I did not find any additional insight into ponding loading but did find this discussion of ponding published by AISC:

      I found it interesting that even the allowable mill tolerance in the members (1/8″ over 10 ft. for steel shapes) could make a large impact on flat roofs. Which makes me wonder, are roof members ever cambered to avoid ponding?

  11. Tyler P.
    October 29, 2014 at 1:40 pm #

    These last few lectures on snow loading has been interesting. I am still surprised that Agricultural buildings do not need to be built to any code for North Carolina. Is it common for states throughout the US to not have codes for agricultural buildings? I feel like it is only a matter of time until someone is injured or killed in one of these collapses until code will be adopted in all the states.

    Another snow related question I have has to do with snow load reductions. On my internship this summer, we did not reduce snow loads by the 0.7 factor for warm roofs. Therefore, we just used the snow load present in ASCE 7 without using reduction factors. Is is common for engineers to do this for a larger factor of safety? Would designing for this non-reduced snow load cause member sizes to increase enough to really make that big of a difference in price in the grand scheme of things?

    • mkev
      October 30, 2014 at 7:59 pm #

      Keep in mind that even in a location where the design is not bound by code, you are bound by ethics and standard practice. I suspect most structural engineers use accepted practices and loading requirements (ASCE 7 perhaps???) to protect human life and provide a quality structure for the owner. If it fails, and even if there is only economic loss, it is a situation you don’t want to be involved in nor would most engineers.

      The .7 factor is more than just for warm roofs, but in effect it is a reduction in the load. You can find older codes / references / standards without the 0.7 but they likely have different maps. Some of the state codes in use in the past had maps of the ROOF LOAD and as such an additional factor was not needed since it was built into the map. It would be interesting for you to ask your work colleagues their reasoning. Another possibility could be that you are working very near a CS area and just want to be a little more conservative in case the map contour was not sufficiently accurate. In any event, the code values are minimum and each engineer is responsible to make their own decisions.

      In terms of cost, it adds a little so you would have to review each case individually. That said, what is controlling the design? If your roof is controlled by deflection, by the time you increase the member sizes for stiffness, you may have already increased your strength capacity anyway.

      • Adam J.
        November 3, 2014 at 9:16 pm #

        Prof. Parfitt,

        Along the lines of meeting the code requirements for buildings, in the case of owners who have had other buildings collapse and wish to have their current buildings evaluated, is it proper practice to value them against the currently adopted code? Or would an engineer determine the required loading on the original design loads? I am not sure since it is hard to determine whether or not the %25 renovation mark for complete code compliance would be met during the preliminary analysis.

      • Tyler P
        November 4, 2014 at 9:07 am #

        When I had talked to my colleagues about there reasoning for using the 30 psf snow load instead on the reduced snow load, they said that it was an added safety factor. In many cases on projects I was working on, there were roof top units being added or moved to an existing structure. I believe that since the old plans used 30 psf snow load for the roof, that is was used again and a reduced snow load would be used only if the members was slightly over-stressed. Which brings me to another question I have. When adding new roof top units, to a building that was built under a different code, is it acceptable to use the reduced snow load or do you have to use which ever code is worst case?

  12. Xiaodong J
    October 28, 2014 at 9:59 pm #

    Thanks Mr. Menard for the great lecture of snow load failure. As one of the conclusion of this lecture, miatakes of construction are the usually the source of failure. However, a building with correct structural design and construction might also fails. For example, when a roof is experiencing a full design roof snow load and wind is blowing. The dynamic wind load can causes roof to have the up-lift force, which can reduce the load on the roof structure. But, if the up-lift force is applied with a frequency that could cause roof structure to vibrate. The result of that is the roof might experience a downward force beyond the designed load value at certain moment, and cause building to fail.

    My question is that, is that possible to happen and how often it might happen; or the wind danamic frequency is very different from the building responce frequency and the vibration would never be an issue.

    Thanks, I know it is not a static snow load issue. but I’m just curious about if that would be an issue when full designed snow load and wind dynamic effects both occur.

    • mkev
      October 30, 2014 at 7:46 pm #

      Since you state the building is assumed to be properly designed, it should be capable of resisting all the ASCE 7 load combinations which includes wind and snow etc. Dynamic excitation of a roof as you describe is very rare and in fact I don’t recall a single case of a failure or serviceability (in a building). Low rise buildings in particular don’t usually have dynamic issues. Keep in mind that the mass of the snow is also going to work to your advantage. If you can find an example where such an event has occurred, it would certainly be interesting to discuss and review.

    • Beau Menard, P.E.
      November 7, 2014 at 8:45 am #


      I agree with Prof. Parfitt on this. I have never heard of wind induced vibration that resonates with the building frequency to cause dynamic excitation. The more concerning issue with full snow on a roof combined with wind is drifting, and/or unbalanced snow loads. Those scenarios are also described in the ASCE 7 and should be taken into account during the design phase.

  13. Yewande A
    October 28, 2014 at 9:45 am #

    Thanks for a great presentation on snow load failures. Sometimes engineers overdesign (specify larger structural members) structures to exceed the safety requirements and probably to reduce the chances of failure or collapse. Overdesigning could lead to higher costs, more time and resources will be used also the likelihood that the building would be subjected to the excessive load it has been designed for is slim. Design guides are available for engineers and designers to determine the appropriate safety factors to be used which should be sufficient to provide the required support. I feel overdesigning could be sometimes beneficial, for instance, in an area where unexpected snow loads are experienced and the current standards did not take that into consideration. How does an engineer strike the balance between trying to avoid failure of the structure by overdesigning and just following the specifications outlined in the design standards?

    • mkev
      October 28, 2014 at 2:55 pm #

      A complicated question when you get down to the details. When all factors are well defined, and you have confidence in the people constructing and operating the building, over design for strength can increase costs. Some of what is often called over design is for performance issues. For example, a floor can carry the load but requires additional stiffness (larger members) to prevent floor vibration. Codes require strength but don’t say much about floor vibration. Your client will complain about floor vibration so the “over design” to cover performance issues is certainly justified.

      What I suggest to new grads or young engineers is to first design the structure to the code and standards. Then look at it and see if you have confidence in all the conditions. If you are not comfortable with those conditions, one way to address it is to make a more robust or redundant structure. Some will call this over design, some will call it robustness, some will call it flexibility or ability to carry unanticipated loads. Good question and one that we should discuss again in class.

    • Beau Menard, P.E.
      October 28, 2014 at 3:45 pm #


      You bring up a good question, and Prof. Parfitt’s response is accurate. I would further add that most often structural design is based on serviceability limits (i.e. deflection) rather than stress. Due to the stringent requirements for serviceability, often times stress, or overload failure, is not a concern.

      As you could see from the presentation, none of the instances of failure were the result of an overloaded condition. Furthermore, the damage to the main structural members occurred as a secondary effect. The primary failures occurred on a smaller component of the whole system.

      With regard to you question concerning over design and cost, for the most part the structural components of a building are minimal compared to other features, such as architectural or even mechanical/electrical. One of the biggest costs associated with a structures are the connections. Unique and complex connections require specialized tradesman to construct. Cost savings, as far as the structural components are concerned, can be earned when typical/redundant connections are utilized.

      I hope that answered you question(s). I very much enjoyed my time back at State. Thanks for being a great audience.

    • Yewande A
      November 2, 2014 at 12:59 pm #

      Thank you for your response and for taking time to explain this Prof Parfitt and Mr Menard. My question was well answered.

  14. Garrett S
    October 28, 2014 at 9:20 am #

    Thank you very much for the insight into snow loads and related building failures. It was interesting to see that some of the failures may have been avoided if better construction practices had been employed.

    Does your company ever experience cases with building failure during construction? The indoor environment is not always maintained during construction, leading to less melting of snow that accumulates on the roof/structure. Even if not much snow falls, it could have a load larger than anticipated because the snow that would be assumed to have melted due to heat transfer did not melt. Would this situation fall under the building owner’s insurance or would it be the liability of the contractor?

    • mkev
      October 28, 2014 at 2:47 pm #

      I will let Mr. Menard respond to the question related to his company’s experience. In general, a large percentage of structural collapses occur during construction due to construction overloads etc. I have heard of a few structures that collapsed during construction due to snow primarily because they were in an incomplete stage. For example, not all the bracing or sheathing etc. had been installed and so there were some weaknesses or instabilities present that would not be there after the construction was complete.

      If you go way back to the start of the year, I had a slide of a snow collapse in State College at the old Lowes store greenhouses. It was one of those big storms we had in the 90’s. The greenhouse heat was not yet turned on and so the snow did not melt or reduce and was combined with sliding into the valley between buildings. I did not investigate that one but the possibility exists that not all the bracing was in place at that point either…I just don’t know the answer to that part. A similar problem exists due to wind. What is stable after construction is finished may not be during the construction process which results in a number of wind failures at relatively low wind loads.

      Generally, the contractor is responsible for the building until it is turned over to the owner so in the case of a failure during construction, the contractor’s insurance would cover something like a snow event as long as the contractor was following standard guidelines and their was no secondary factor involved (such as poor workmanship).

    • Beau Menard, P.E.
      October 28, 2014 at 3:58 pm #


      To answer your first question, yes. My company evaluates all sort of building failures, starting at the initial construction phase. With regard to failure in construction during a snow event, we do not often see overloading situations in the south. I can reach out to my colleagues in the northern states to see if they have come across that situation. One thing that should be considered in the question you pose, is, does the additional weight of the un-melted snow exceed the expected/design construction loads (not just the expected snow load).

      With regard to you question pertaining to the insurance coverage, my best advice would be not to worry about it. As an engineer you are not an insurance adjuster, and nor should you be. There are various legal ramifications that can occur when you get into coverage discussions. My advice would be stay away. As a professional engineer, you are licensed and held to the ethical obligations of the state you are licensed. Your duty, first and foremost, should be to the safety of human life. When you are tasked with determining causation, your investigation should be unbiased, regardless of who your client is. Trust me, insurance companies and lawyers would rather have bad news upfront. So you shouldn’t worry about coverage issues. Your concern should be to uncover the facts of the investigation.

  15. Jim P
    October 28, 2014 at 8:37 am #

    Thank you Beau for the informative discussion.
    It is my understanding that AISC does not include metal building design in the steel manual anymore (I think the last edition to include metal building design was the 6th or so). Does this lack of regulatory oversight contribute to why metal buildings fail? How do metal building designers/erectors prove due diligence?

    • mkev
      October 28, 2014 at 1:40 pm #

      AISC does not exclude metal buildings as a category. Generally anything that that is comprised of cold formed steel (purlins, girts, etc.) comes under the American Iron and Steel Institute (AISI) (standards can be seen at The parts of a “metal building” that are rolled shapes, plate girders, connections, bolts etc. should still be reviewed under the AISC standards and guidelines. Dr. Louis Geschwindner (Penn State Professor Emeritus and former VP of AISC) confirmed this and notes that “..MBMA (Metal Building Manufacturers Association) has an active member on the Specification Committee (of AISC) and much of what AISC does impacts metal buildings…”

      MBMA also publishes recommendations, guidelines and guide books but see AISC and AISI for the official standards involved.

Leave a Reply