The goal of the research is to determine the causes for more than 400 snow related roof failures officially reported in 2010-2011 winter in the Northeast United States such as in Massachusetts, Connecticut and New York. The failure data collected by building officials will be analyzed to determine if the failures are either due to excessive snow loads or due to structural capacity being less than that envisioned by building code provisions. These key questions will be answered by analysis of the roof load case histories provided by the structural engineering officials and community in combination with detailed weather information on snowfall and wind for selected sites from the Northeast Regional Climate Center. The research results will provide the basic design information to engineers and code officials to improve snow load demand provisions and material resistance standards prescribed in the current structural design provisions. The project outcomes will lead to determination of appropriate response needed to ensure better building performance in the future.

The data and analysis results will be made available to the Structural Engineering Institute of the American Society of Civil Engineers (SEI/ASCE) for its publication and wider dissemination to the structural engineering community for follow-up studies. A session on Roof Failures in the Northeast will be organized for the April 2012 ASCE/SEI Structures Congress. The report and session will result in practicing structural engineers and building officials gaining a fundamental understanding of how real buildings respond to real snow loads to design safer and more economical buildings in the future. The undergraduate students will be involved in the project to provide unique experience and insight into the behavior of various roofing systems.

Project Report

The purpose of this project was to investigate the very large number of roof collapses that occurred in the Northeast during the 2010-2011 Winter. The key issue is whether the collapses were due on the one hand to snow loads well in excess of that prescribed in building codes, or due on the other hand to structural capacity significantly less than envisioned by building codes. Weather data from multiple sources was used to estimate both the ground snow load and roof top drift loads for the 2010-2011 Winter. These in turn were compared to requirements in the current ASCE 7 load standard. Ground snow load simulated from weather data showed that 2010-2011 winter values were less those from the ASCE 7-10 map. The ratio of simulated 2010-11 ground snow load to ASCE mapped ground snow load ranged from about 85% for Connecticut to about 59% for Rhode Island. Hence the structural performance problems in 2010-11 were not due to ground snow loads well in excess of those in ASCE 7-10. Similarly, weather information – specifically snowfall, wind speed, wind direction and duration of wind storms – was used to simulated 2010-11 drift snow loads for various roof geometries, at selected locations in southern New England. Although the simulated drifts were in some cases substantial, they were not significantly larger than those prescribed by the ASCE 7-10 provisions. Building performance databases from state officials in Connecticut and Massachusetts were gathered as well as case histories from structural engineering practitioners. These databases proved useful in three ways. First, they allowed determination of the typical "problem" building. The majority of the problem structures were single story, wood framed buildings with pitched roofs. Roof structural components were most commonly wood beams and roof surface materials were most commonly reported as non-slippery. Also somewhat surprisingly, about 40% of the problem structures were unheated. Secondly, the practitioner case histories allowed determination of the apparent failure mechanism. For case histories where the apparent failure mechanism was known, approximately half were attributed to large snow loads, either uniform or drifts, which apparently exceeded the current structural capacity. For the other half of the case studies, the failure was apparently due to structural capacity problems such as an initial construction deficiency, a man-made reduction in structural capacity, or natural deterioration of the structure over time. Finally some of the practitioner case histories included roof snow load measurements. These in turn were compared to the flat roof snow load prescribed in ASCE 7-10. Separate comparison were made for heated and unheated structures since the ASCE 7 flat roof snow load is a function of the roof thermal condition. For both groups, the majority of the roof load measurements were less than or reasonably close to those proscribed in ASCE 7-10. In a limited number of cases, the measured roof load was significantly larger than the ASCE 7 flat roof load. However, in each of the "outlier" cases, the measured load was also in excess of the estimated ground snow load. As such they were not consistent with a flat roof load and conceivably include a drift surcharge or impounded roof snow melt due to clogged or inadequate drainage. In summary, although the 2010-11 winter in southern New England was quite snowy, the estimated ground snow loads were less than design values in the ASCE 7 load standard. Similarly, simulated roof snow drift loads for the 2010-11 winter were substantial in selected locations and for selected wind direction. However, again, they did not significantly exceed the prescribed design loads in ASCE 7-10. Finally, measured roof loads were generally less than those prescribed in the ASCE 7-10 In general the 2010-11 winter roof problems were due to initial construction defects, poor maintenance, improper modification of the structure, and natural deterioration of structural capacity over time. It seems likely that "problem" roofs for which localized drift loads were the apparent failure mechanism, were designed prior to the adoption of modern (post 1988) snow drift provisions. As such, although there were over 350 roof collapses, there is no evidence that the root cause was inadequate design snow loads in building codes.

Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Rensselaer Polytechnic Institute
United States
Zip Code