Snow Load Code Development and Building Failures and Investigations

Snow Load Code Development and Building Failures and Investigations

“Winter is Coming”   OK, just making sure all you GOT fans are paying attention…

State College hasn’t seen much in the way of snow flakes so far this season but there is no question that winter weather is quickly approaching in many areas of the country in addition to Pennsylvania.  Some locations with higher elevations have already experienced several inches of snow on the ground.  With more snow on the way and the inevitable roof damage and collapses that seem to follow at some point, Paul Rouis, PE,  Principal of Ryan Biggs Clark Davis  in Albany, NY  participated in the Visiting Practitioner Failures Lecture Series to Penn State AE students on the topic of Snow Load Codes and their evolution over the years.  Mr. Rouis also discussed several snow collapse case studies including pointing out lessons learned so that hopefully you won’t make the same mistakes as others.

Structural collapse and roof damage from snow can occur from a variety of sources and often results from a 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.  In addition, no snow collapse discussion would be complete without first looking at FEMA P-957 Snow Load Safety Guide.

While snow collapses are nothing to kid about, MKev does occasionally find humor or at least irony in the topic as can be seen in the photo below.

I guess this means I shouldn’t be standing here but how was I supposed to read the sign?

And to start out the commentary, remember the comment by Mr. Rouis about how before the early 1970’s many roofs didn’t have much in the  way of insulation so the escaping heat pretty much took care of long-term snow build up?   See my comments  below from last year about snow melting on an industrial scale.   And, ask me in class about the snow melt idea for one of the campus buildings that fortunately never came to pass.   MKev

Tags: , , , ,

29 Responses to “Snow Load Code Development and Building Failures and Investigations”

  1. Jordan O
    October 30, 2018 at 2:04 am #

    Mr. Rouis provided some important additional insight into snow load code development and building failures that helped to build upon the foundation we have so far. It makes sense that the code slowly developed into what it is today, but it is wild to think about how loose the restrictions were when considering snow load. In the realm of building failures, you have to wonder how many failures were mis-identified as poor design or construction when in reality it was a faulty code that was followed that lead to a collapse. As design evolved, more unforeseen issues were encountered. One such example is added insulation in the enclosure causing more snow to build up on a roof than before when heat could escape and melt snow off, leading to increased snow loads which then needed to be accounted for in design. A slow evolution like this is expected and it does often take multiple examples of mistakes to create a wave of change, but it hopefully has lead the industry to a point where we can begin to anticipate and design for future problems we may not have thought about yet that will allow for increased longevity and resiliency of buildings we design. Codes, design ideals, and even materials we use will all evolve along with the process to help us build the best buildings possible.

  2. Ryan L
    October 29, 2018 at 8:59 am #

    As Mr Rouis discussed the development of snow loading code and methodology for determining unbalanced loading capacity, I wondered how you can accommodate all of the variables under one specific methodology. He specifically mentioned the development of the thermal factor over the years, but I wonder how you balance the line between public safety and over engineering.

    From the first snow loading assignment, I believe if I was ever truly in charge of a structural snow loading design, I would over design every time.

  3. Sierra S
    October 25, 2018 at 10:42 am #

    Mr. Rouis was helpful in understanding the various components that need to be checked for snow loading. One of these being unbalanced loading. Due to wind, snow can load one bay but not the other. This is important to check because it causes pattern loading which can increase the maximum moment on a member. This is also why drift has to be looked at closely, both conceptually and mathematically. All the different ways snow can drift due to parapets, different building heights, etc. need to be considered and then the max build up potential.

    Another component to Mr. Rouis talk was the modification of existing buildings. What caught my attention was the Pre-engineered Metal Building. It shocks me that they moved the pre-engineered assembly without consideration for the different environmental factors. I am unsure how the building would be transferred or how many components they would have to take apart. However, when reassembly occurred the bracing at approximately mid height of the columns was forgotten. This bracing was critical for the structural capacity of the columns. Without knowing where the building was originally designed for I am shock it was able to withstand the snow storm without a complete collapse. This raises my question of how does transportation of preexisting buildings happen and is there any regulation?

    • Clayton T
      October 30, 2018 at 8:23 am #

      I believe most pre-engineered building manufacturers are given almost nothing for building information other than basic things like the location and use of the building, resulting in minimal design and minimal load capacity but also a very cheap option for the owner. However, you raise a good point in this case with moving an existing structure because something like a higher snow load which was the driving factor in this structure could make a vast difference on whether or not the building would experience a failure because of those minimum design procedures. I am also wondering if, through the building code, the same code parameters apply to those pre-engineered building as they do to all other or if there are some exemptions due to the nature of the building.

  4. Jackson H
    October 25, 2018 at 8:51 am #

    Mr. Rouis’s presentation provided good examples of some snow loading conditions that we usually neglect, but can actually cause significant damage or collapse, namely the overhangs that cover building entrances. As students, we usually only consider the snow load in the largest roof areas and in the controlling load cases, but from the pictures we were shown in class, even the smaller roof areas that we usually wouldn’t consider drifting on can experience significant drift loads and can collapse if they are not designed properly. I know this is important for my thesis building because the roof of the building is very convoluted and has many smaller areas that are all adjacent to other roofs on a variety of levels, so there are many different areas that could all experience different levels of drifting, and it is important to make sure that each roof area, no matter how small, gets designed properly for any potential loading situation.

    Going through historical codes and how they evolved in to the modern codes that we use today was interesting because it doesn’t seem like the actual design values that we use for ground snow load have changed that much, but how we arrive at these design values and the factors that go into determining roof snow load have changed significantly. It seems like every year we get less and less snow, I wonder if the regional snow load required by code will ever decrease? I don’t think we have had snow accumulation over 3-4 feet in the area that I’m from since I’ve been alive.

    • mkev
      October 28, 2018 at 10:24 pm #

      Some might blame global warming but snow patterns are quite cyclical in nature. Just when you think the loads are going down you get the right combination that surprises you. State College easily goes 10 to 20 years between big snows at times.

  5. SamZ
    October 25, 2018 at 8:41 am #

    Mr. Rouis’s presentation about snow loading was enlightening going over the development of code loading for snow and some of the more complicated snow loading cases. What I found particularly interesting however were the case studies he presented. In the case of the salt dome I appreciated learning a bit about a typical failure mode of a dome as well as the effects of little to no quality control in the construction process can do to a structure. The reoccurring snow load problem with pre-engineered metal buildings was another item of note. This summer I analyzed a metal building to determine if an additional load of 2 psf for a new layer of roofing was too much load. As it turned out the post construction sprinkler system had already was over stressing some of the purlins. Hearing about the pre-engineered metal buildings did not surprise me, they may be cheap and material efficient but they are engineered so close to the prescribed design loads. Pre-engineered buildings don’t allow for flexibility for the owner down the road or in the case of an unforeseen condition such as a drift sliding down a roof as a mass or sno-jax added to the roof.

    • mkev
      October 28, 2018 at 10:19 pm #

      Good point SamZ. We often see localized overloads in engineered metal buildings. The frames have capacity but the purlins do not. Frame bracing is often attached to the purlins which can then cause a progressive collapse type condition.

    • Ryan L
      October 29, 2018 at 9:08 am #

      Sam, I have seen the same regarding pre-engineered metal buildings. Specifically I see a lot of owners of these building reinforce certain areas for specific purposes however it makes a very confusing footprint for future owners/occupants. For example, I saw a split use unit where an auto-body shop had put in a timber support system for a ceiling hoist, however the system was covered by interior finishes for insulation. A company sharing the building put a heavy load on the bottom of some small trusses and after a heavy snow winter, the roof line showed significant damage

  6. Smithr
    October 25, 2018 at 8:41 am #

    One of the most important takeaways I had from Paul’s lecture was some of the differences in the development of design codes. Early codes treated snow like a uniform live load and did not consider the movement of snow. It does seem a little crazy to me because we see snow drift occurring on the ground, but sometimes we don’t tie two and two together until it creates a financial burden. Second generation codes accounted for local topography and roof slope. This to me is the biggest step because the local weather patterns are going to make the most drastic changes in the majority of buildings. This was actually recognizing that every area is a little bit different and we need to account for this. The third generation refines our search even more to include snow exposure, thermal condition (due to inclusion of insulation), building occupancy/importance and wind effects. To me this is covering the last 10-15% of buildings that are not covered by the second generation. This might seem nominal, but how many buildings actually do collapse from snow load failures? Probably less than that 10-15%. Additionally, in most of these case studies, there are a few features that make the building unique and ultimately a factor in the failure method.

    • mkev
      October 28, 2018 at 10:15 pm #

      Keep in mind that while some of the early codes were not that sophisticated, the engineering community treated many cases conservatively. I can remember many engineers using 30 PSF on the roof in PA in locations that today likely map out as 25 or 30 on the ground. So even though they were uniform, there was some extra capacity built in.

      In reality, very few roofs collapse from snow. You hear about it in the news when they do but it is a very small percentage. I don’t have the statistics but my guess is under 1% for sure. I don’t know of any major building on campus that failed due to snow. Some localized damage here and there but no collapses.

  7. Josiah M
    October 25, 2018 at 8:26 am #

    Mr. Rouis gave an interesting lecture of snow loads and their development over the years. It’s really cool to see how snow loads got to be what they are today, especially when you consider that they were basically a single number suggestion roughly 100 years ago. Mr. Rouis spoke about how the codes came to be what they are today. He mentioned many cases that lead to code provisions such as failures due to snow drift and failures due to unbalanced snow loading. The most interesting of these cases to me, is the development of a thermal factor in the code. Before a certain period of time, it was uncommon to insulate a roof and, as a result, the snow would just melt right off because of the heat loss in the building. This became a problem during the energy crisis when people started to become concerned with their usage of energy. Their concern led them to insulate structures more completely, meaning less heat loss from the building, especially from the roof. Because of this, snow began accumulating on structures which eventually caused failures. The code community responded to this by updating codes to adjust for this factor. It was very cool to hear about how these events link to code development.

    This also leads me to a couple if questions. How thoroughly do we investigate the impact of these new methods for building construction? Is it easier to keep things as they are, or is it easy to append the code to take these new methods into account?

    • Sierra S
      October 28, 2018 at 5:37 pm #

      The evolution of the code is shocking to me. The IBC was developed by combining the minimums of the codes from around the country. Different areas focused on the issues in their region. For example, California focused on seismic design. I am shocked that they decided to only pull only the minimums from each one. This allowed larger room for failures to occur and the code to be modified accordingly. So while the code changes, as Josiah mentioned, due to changes in construction it also changes due to be more conservative so similar failures do not occur again.

      A question brought up in my wood class the other day is if it is the engineers job to go beyond the code to predict likely events that could occur in the building they are designing or if the code is what needs to change. For example, twice this year there has been a failure of flooring system. These failures occur due to overloading the capacity of the space for a party. The code does not anticipate this high of a load largely due to fire safety and exit limitations. However, the students at college parties do not follow this expectation. This led to the discussion if it was the job of the code to require a higher load consideration or the job of the engineer. The issue with the code is that these failures typically only occur due to college parties and very rarely so requiring an higher design load for everywhere is not realistic. I believe this issue will eventually produce a change in the code but many different considerations have to be taken into account prior to that.

  8. Katie W.
    October 25, 2018 at 8:15 am #

    Mr. Rouis’s presentation showed us how the snow load code was developed. It was interesting to see how it was barely considered to all the factors put into it nowadays. I really thought the way the Vermont code from 1990 was laid out was a good idea, because it left out any confusion from trying to follow the contours. However it would not be feasible on a national level unless there was a great many more maps in the appendix. Looking at the case studies he showed into class, it reminded me a lot of my internship from two summers ago. A lot of what I did was keeping track of the buildings on salt storage sites. There are so many add-ons and adaptations to buildings that weren’t there originally that I wonder if they have considered snow loading. I remember two pole buildings that were close together, where they put a roof in between the two to store more materials. Both the pole buildings will have sliding snow onto that makeshift structure, which is concerning. There was also a lot of lean-to structures attached to sides of buildings which would also be susceptible to increased snow loading. Changes to a building can really effect the snow load and how its distributed as shown in the case study about the added PV panels which ended up catching more snow than expected. A looked over the FEMA guide, which seemed to be aimed at stakeholders in the building itself and not the engineer. I think this is a good way to keep them informed when thinking about adding on to or changing their structure.

    • mkev
      October 28, 2018 at 10:09 pm #

      Infill structures like you discussed between two buildings are really susceptible to snow overload. Sliding and drift can build up quickly.

  9. Clayton T
    October 24, 2018 at 11:54 pm #

    Mr. Rouis gave an insightful discussion on snow load parameters in design and failures as well as the progression of snow load code throughout the years. The development of snow codes takes place through historic data and better understand environmental conditions, while structures may provide their necessary function for many years with varying conditions they could fail due to design discrepancies, lacking load calculations or incorrect construction means and methods. For many years, the concept of snow drift was not considered but according to Mr. Rouis’ estimate it accounts for about 75% of failures attributed to snow loading. An example he presented is a simple awning over doorways with drift from the higher adjacent roof causing a collapse. With variance in snow loading based on location in the country, construction issues can cause major failures whether they are designed correctly or not. This was the case in the collapse of salt storage dome, wood construction with multiple construction discrepancies allowing for improper load distribution causing a collapse of a mid-dome assembly. As a take away, the structural integrity of a building subjected to snow load is highly dependent on design code, design process and construction quality utilized.

    • Jackson H
      October 29, 2018 at 10:35 pm #

      I remember Mr. Rouis telling us that drift load cases account for 75% of snow load failures. Even with drift provisions that are currently in place, that percentage still seems high. I wonder if back before there were snow load provisions, there were more snow failures? Or if building design has become less robust, and as a result, snow drifting conditions now need to be taken more into consideration since building structures are leaner, where as before the tendency to overbuild structures made up for the lack of consideration for drift. Were drift failures always the most common snow failure? Or did they become more common once the industry moved towards lighter framing? If such a significant percentage of snow failures are still attributed to drifting, are the codes that are currently in place adequate? Or as we often see, are these drifting failures attributed to causes beyond the control of the designer?

  10. Steven B
    October 24, 2018 at 11:27 pm #

    Mr. Rouis was able to demonstrate how the codes related to snow loads have evolved from a simple graph or a single paragraph to the multipage document that we have in the current codes. This evolution will continue with research into various variables associated with snow load being conducted.
    He emphasized the ability to visually walk through an issue before jumping into the drawing and calculations. Look at what realistically could happen when a variable is changed. Is it realistic to have the accumulation of snow from drift and sliding? Can they both occur at the same time on the roof structure you are analyzing. How will one affect the other? What does you experience tell you? There is no substitute for experience. When designing structure in areas that you have little experience, it is best to have your designs peer-reviewed by someone more familiar with the area that the structure will be constructed in.

    • Abby S
      October 30, 2018 at 7:49 am #


      You bring up a few good points about the development of the snow code and different ways to think about the loading conditions. I think it is very important to visualize the issue and consider if it is realistic or not. In my opinion, thinking practically about these situations is something that requires time and effort. I think this can be done without a significant amount of experience, but experience in the field with these situations would certainly be a benefit. I also agree that peer-review is critical when you are not completely sure about the conditions you are designing for. It’s amazing that the snow load provisions in the code have developed from a single paragraph to the complex document it is today, so thinking critically and realistically are huge factors in interpreting the code.

    • Josiah M
      October 30, 2018 at 8:43 am #


      You mentioned that “there is no substitute for experience”, I think that’s a great point. It’s very important to be able to visualize the sorts of loading types that may be seen in the building that you are designing. If you can’t do this, you might overlook things like drift or unbalanced loadings.

      But, on the other hand, I believe that the code could be slightly more helpful in helping engineers visualize these scenarios. The writers of code seem to rely more heavily on language, which can be confusing at times. In addition, the equation used to calculate things like roof snow load and drift are somewhat left to individal judgement. Like Professor Parfitt mentioned before, even design professionals produce differnt answers.

  11. rgstanza
    October 24, 2018 at 11:18 pm #

    I had never even considered the process of removing snow in a specific manner to avoid unbalanced loading conditions until reading the Snow-Induced Roof Failures and Prevention Methods article. As designers, I would think that we should be taking this unbalanced loading condition into account even if there is a certain protocol the owner should follow to avoid this. If we do not, and a roof fails due to this condition, are we responsible for the collapse, is blame placed on the owner, or is it another one of those gray areas?

    • mkev
      October 28, 2018 at 10:01 pm #

      I have never heard of an engineer getting blamed for this. I think when a roof is so loaded up that problems are appearing most people seem to think it is time to call a code official or structural engineer. Those who clear large snow falls do so at their own risk for collapse or damaging the roof.

      There are documented collapses from snow removal. Twice I have seen buildings where owners asked the fire department to “blow” the snow off the roof using fire hoses and water. Talk about Rain on Snow! Needless to say it not only did not work but probably made things worse.

    • Katie W.
      October 30, 2018 at 1:11 am #

      I think you bring up an interesting point in trying to remove snow and possibly causing even more issues. In our load combos, you usually choose between roof live load and snow load. You never use both at the same time and the roof was never designed that way. Having people go onto the roof, or other methods, like Professor Parfitt mentioned could mess with the loading even more. Either way, the removal process is not usually designed for, and should be done very carefully. Unless specified, I don’t see how that could be the fault of the engineer.

  12. Abby S
    October 24, 2018 at 11:10 pm #

    Mr. Rouis’ lecture detailed the importance of evaluating all aspects of a building and how they can affect performance under snow loading. I found the two most important takeaways to be the considerations of combined loading conditions and modifications to existing structures. Mr. Rouis stressed that it is critical to consider a variety of snow load patterns and conditions, such as sliding and drift. He also brought up the case of rain-on-snow, which was also stated to be a factor in 10% of the case studies that Mr. O’Rourke analyzes in his Structure article. I think this is important because we typically consider snow loads but have not focused heavily on ponding effects, and even less or not at all on the combined effects of rain and snow. This proves there is much more to snow load analysis than simple flat roof snow load effects. With weather than changes so quickly here, I would be interested to learn more about this type of analysis and how common it is.

    Another takeaway from the lecture is the major impact that seemingly minor changes to a building can have on its structural integrity. Mr. Rouis discussed a case where PV had been added to an existing roof, which had also been altered such that the roofing was lifted up from the roof truss to create roof monitors. Although the height difference between the PV and roof was not incredibly large, the panels allowed snow to build up along one side and increase the load. The roof monitor modification also removed the compression bracing from the top chord of the truss and allowed it to buckle easily. An engineer analyzing the possibility of PV panels missed this structural issue, so it is critical to inspect every aspect of the structure as it was meant to be built and is actually built.

    • Eric I
      October 28, 2018 at 9:45 pm #


      I think both of your points prove that one of the most difficult parts of structural engineering is all of the conditions that we must anticipate as the designer. Even the smallest change to a building can have massive effects on the structure. With your example about the PV panels, we all can easily see that they will add weight to the roof but recognizing that they create snow drift problems is not quite as obvious. Even if one does recognize this, quantifying the effects presents another challenge due to the complexity of the codes and that the exact condition may not be referenced. Continuing with your rain-on-snow example, we all know to analyze the rain and snow loads separately, but again, considering the combined effects takes a deeper understanding. This was the first time I had heard of this condition but it does happen quite frequently, especially in our State College climate.

      The current codes are obviously very extensive but there are always cases that require sound engineering judgement. Having the ability to extrapolate information to new conditions that aren’t explicitly stated in the codes is such an important skill. We as engineers almost need to be able to see into the future so that we can account for the “less obvious” events such as snow drift against PV panels. Also, we need to be very informed about what is being removed or added to a building. For example, we have studied numerous buckling failures in this class and the root cause has often been the accidental removal of compression bracing. Checking for sufficient strength is again, very obvious, but recognizing that the removal of roof deck for a roof monitor opening creates an unbraced length in the compression flange is almost a secondary thought. These things are easily missed due to the high level of detail required to identify the problem but we must always be aware of all the potential issues being created.

  13. Eric I
    October 23, 2018 at 9:15 pm #

    Mr. Rouis provided us with very valuable insight regarding the confusing topic of snow loading. One of the most interesting parts of his discussion was about the evolution of the snow load codes. He showed several historical methods that were no more complicated than simple graphs which were only dependent on a few geographical variables. These methods also treated snow like all other uniform live loads. As time went on, codes started to consider more things such as local topographical factors, roof slope/geometry, wind effects, unbalanced loading and drift. Reading through the current code shows that we have come a long way in determining snow loads and this is a great example of how the engineering community has come to better understand the forces we are trying to resist. However, there does need to be a balance between accuracy and simplicity. One of the common complaints with the snow provisions, which we have discussed in class, is that they are starting to become too complex and hard to understand. Unfortunately, this can lead to calculation errors simply because the procedures are too difficult to follow. As our careers progress, it will be interesting to see if the codes take a step back and develop less complicated procedures to follow.

    Another interesting topic was the case study about the gymnasium. The maintenance staff simply added sno-jax to the roof in an effort to stop the snow from sliding off the roof. While this may have seemed like an easy fix for the staff, it created a loading condition that the structure had never seen. Although the failure was ultimately due to the absence of proper bracing, there are two lessons to be learned here. First, we as an industry must get better at verifying that our design was constructed correctly. Contractors and builders do miss things sometimes but they also may not understand the importance of something like bracing. Second, we need to be very careful when things are added to buildings that weren’t originally intended. Something as simple as sno-jax can create unforeseen conditions that may have extreme consequences.

    • mkev
      October 24, 2018 at 2:14 pm #

      Good points Eric. Mr. Rouis also mentioned buildings where lots of insulation was added which would sound like a good thing but it did result in a number of older buildings retaining more snow over the winter season, some of which led to damage or collapse, especially in drift situations.

    • Jordan O
      October 25, 2018 at 2:48 am #


      You make a great point discussing the industry wide need for better verification of proper construction. I’ve spoken in several of these discussion posts about the need for more wide spread commissioning to ensure a design is built the way it is meant to be. Simple errors that might not be glaring during the construction process can compound on one another and lead to a much more catastrophic failure. It is difficult to guarantee that builders have a good grasp of why things are specified the way they are. Having a knowledgeable and experienced professional on site to not only ensure proper construction, but to also help educate contractors and builders, can play a large role in the performance of that project and projects the building team works on in the future.

    • Smithr
      October 25, 2018 at 8:50 am #


      Good point about the complexity of the snow loading calculations. When I worked for a GC, often times when we weren’t getting high quality work out of crews, we would try to simplify the tasks for them. In some cases, digging in too deep and fully understanding a system is not worth the time. It takes a good manager to realize this.

Leave a Reply