Essential structural considerations in roof design

Calculation of the weight of snow is also a challenge. When snow first falls, its density is exceptionally light, 6.4 pcf (1 kN/m³) and within 24 hours, the density doubles and can increase to 25.5 pcf (4 kN/m³) over time. If only the depth of the snow has been measured, the change in density with time makes estimating the snow load difficult.

The snow load provisions in ASCE-7 are dealt with in Chapter 7 and are calculated from the following equation:

p_f = C_b C_e C_t p_g

p_s = C_s p_f

= C_s C_b C_e C_t I_s p_g

P_g = Ground snow load based on a reliability analysis

C_e = Exposure factor

C_b = Ground to roof factor set at 0.70

C_t = Thermal factor

C_s = Sliding factor 1.0 ≥ C_s ≥ 0.0

In the 2022 edition of ASCE-7, the thermal factor was increased by 20 percent in value to reflect the effects of increased roof insulation or vented roof, and increased 30 percent where a complete building was used as a refrigerator.

In 2020, a major project was launched to generate ground snow loads across the country to reflect a more consistent level of safety, matching what is required by ASCE-7. Achieving a more consistent level of safety was done following a “reliability approach.” This method could only be done at this time because of the amount of computing capacity required to do this.

The work was done in the following manner:

The snow load was compared to the resistance of a steel roof:

• The steel resistance probability distribution was used based on research by the American Institute of Steel Construction (AISC).
• The snow load probability distribution was calculated based on a ground to roof conversion probability distribution.

o The ground snow load probability distribution based on annual maximum snow load records across the country.

o Snow depth to load conversion equations for different parts of the country at locations where only snow depth was measured.

Using all these distributions, Monte Carlo simulations were done. Snow loads that would cause failures were found at each site for low-occupancy buildings, normal occupancy buildings, and high importance only for an acceptable number of times.

The new snow maps have improved the prediction of ground snow loads across the U.S.; and the project results improved the prediction of snow loads across the country. The new ground snow load maps have less areas requiring special studies to establish ground snow loads, and the new maps have dealt with areas where local codes used different loads than in ASCE-7.

The differences between NBC and ASCE-7 are the ground snow load p_g and the basic roof factor C_b. In Canada, snow loads are a combination of the basic roof factor and the thermal factor. The 1953 edition of the NBC first dealt with design snow loads. The roof loads were equal to the ground snow load, with reductions allowed for sloped roofs only. The load values were approximate and resulted in over-design for some roofs and under-design for others, particularly in areas subject to high-drift loads.

Between 1957 and 1968, The National Research Council of Canada (NRC) undertook a country-wide survey of snow loads on roofs. This survey provided evidence on the relationship between ground and roof loads and enabled the committees responsible for the 1960 edition of the NBC to adjust the code requirements.

In 1960, roof loads were set at 80 percent of the ground load, and they were adjusted to allow for the increase in the load caused by rainwater absorbed by the snow. At the time, NRC researchers recognized if a roof was cold and in a sheltered area, the snow load remained at 100 percent of the ground snow.

However, this possible variation (now C_b) was not documented in the structural commentaries or the NBC. By 1965, all roof loads were directly related to the snow load on the ground. The basic design load remained at 80 percent of the ground load, but a snow load of 60 percent of the ground load (C_w) was allowed for roofs exposed to the wind.