Prediction of local snow loads on roofs

Abstract

Large snow loads on roofs during the winter season of 2005 - 2006 led to the collapse of several buildings in Norway. In Central Europe, during the same winter season, there were several serious accidents related to heavy snow loads where many people were killed or injured. Hence, there is a need for evaluating the background for snow loads used in the design of buildings and an assessment of the reliability of buildings which are subjected to roof snow loads.

The snow load represents one of the most important structural loads in Norway. The magnitude of the snow load varies throughout the country depending on local climate. In the current Norwegian standards, characteristic snow loads on roofs are found by means of simple expressions for the relationship between snow depth on the ground and snow loads on the roof. In addition, rules are given in order to account for the effects of wind, roof geometry and heat transfer on roof snow loads.

An investigation is performed in order to obtain a reliable indicator as to whether existing buildings in Norway meet current regulatory requirements concerning safety against collapse caused by snow loads and wind actions. The analysis comprises studies of 20 existing buildings in five high-snowfall and five high-wind municipalities in Norway. The investigation demonstrates that most of the buildings considered have higher calculated probability for collapse owing to snow loads than the regulations now require. It also indicates too low calculated reliability for a considerable number of buildings in Norway, when evaluating the possible implications of the findings. An unexpected result in this study is the discovery that many buildings have even lower calculated reliability than the historical increase in design loads should imply.

Weather data from meteorological stations in Norway for a reference period of 30-years, 1961-1990, are used to quantify the effects of wind exposure on roof snow loads according to the definitions in the Norwegian standard NS 3491 and the international standard ISO 4355. It is shown that the procedure in an informative annex of the standard does not reflect the actual effects of wind exposure on roof snow loads in Norway, the main reasons being oversimplifications in the definition of the exposure coefficient and the extreme variations of the climate in Norway. As a result of the present work, the Norwegian snow load standard NS-EN 1991-1-3, which recently has been published, includes an improved definition of the exposure coefficient.

Whether it is beneficial to differentiate roof snow loads in view of material costs, is studied by means of a selected house concept. This study lead to unexpected conclusions which challenge the prevailing view that increased calculated capacity results in unacceptable increased costs for the individual house owner. In this investigation, a timber detached house is designed for different roof snow load levels. Some differences are found when evaluating the degree of building material consumption, but the economic effect is small. When comparing the costs of increased reliability of all houses to the total damage insurance payments, conclusions may be drawn that it is more reasonable not to increase the reliability.

From an environmental perspective reduced material consumption is however highly appreciated. Since the costs of increasing the calculated reliability and the negative effects on the environment probably are limited, an individual property owner would most likely prefer to invest in increased reliability.

Models for predicting roof snow loads and snow density on the ground are developed using meteorological data as input which extends the application of the models. Snow load and density measurements performed at 105 sites by Professor Høibø at the Agricultural University of Norway (now the University of Life Sciences, UMB) in the period 1966 to 1986 are analysed. New knowledge of the influence of local climate on resulting maximum snow loads is achieved. A clear correlation is found between observed climate and the measurements. The study also reveals that the relation between local climate and snow load is complex. The results are a step forward in the process of understanding these relations. The work has revealed that wind velocity as a single parameter probably is of less importance in relation to maximum snow loads than previous research has indicated.

Additional work is necessary and should focus on further developing the method for predicting snow loads on roofs, which in turn can be used to improve standards and regulations. The snow measurements performed by Høibø were done within a small area in Norway. Additional work should also aim at investigating the results in view of data from other parts of the country.

The methods developed in the current study for predicting snow loads on roofs are important in order to understand the climate’s significance on the accumulation of roof snow loads. In a longer perspective it can be used to improve the European standards recommendations with respect to design roof snow loads. It is also demonstrated in which way the methods can be used to estimate roof snow loads in areas subjected to large snow falls with short duration. In this way, snow clearance of roofs can be carried out in time. Roof snow loads for buildings located in the city of Kristiansand in Norway are calculated. Methods for collection and preparation of the necessary meteorological input are presented, including development of parameters for building sites where limited meteorological data exist.