Simulating uniform roof snow loads

Abstract Snow tends to slide from unobstructed slippery sloped roofs of sufficient steepness. This behavior can reduce the required design load and deposit additional loading on lower adjacent structures. Computer-based data acquisition systems have made it possible to obtain the necessary data to understand sliding snow. Microcomputers can collect data, transmit the information while unattended, and control laboratory equipment. Data from field sites, combined with data generated using artificial snow, comprised the database; thus, the conditional probability of sliding was expressed using temperature as the random variable. The probability of sliding on any given day was expressed using the equation of joint probability; with these probabilities and actual meteorological data, roof loads for each winter day over a period of 25 years were simulated. A log Pearson type III extreme value distribution yielded 50-year mean recurrence interval roof loads that correlated with measured values.