Roof Projections - Snow Loads - page 73

Chapter 8
Roof
Projections
Snow drifts frequently form at parapet walls and adjacent to rooftop units
(RTUs). At a parapet wall, the drift is clearly a windward drift because the
snow source is the roof, as opposed to snow originally on the top of the wall
itself. For an RTU, the situation is more complex. The drift on the upwind side
of the RTU is a windward drift for the same reasons as for parapet walls. The
drift on the downwind side is a somewhat reduced leeward drift composed
of snow originally on the roof upwind of the RTU plus a small contribution
from snow originally on top of the RTU itself. This leeward drift is somewhat
reduced because some of the roof snow is captured at the windward drift on
the upwind side of the RTU. Hence for wind out of the north, a windward
drift forms on the north side of the unit and a somewhat reduced leeward
drift on the south side, both made from snow originally to the north of the
RTU. For wind out of the south, a windward drift and a leeward drift again
form, but now they are made from snow originally to the south of the RTU.
For simplicity, ASCE 7 requires a windward drift on each side of the
unit, based on the larger of the two fetch distances. Hence, these roof projec-
tion drifts follow the same provisions as windward roof step drifts discussed
in Chapter 7 of this guide (Section 7.7 of ASCE 7-10). For parapet walls, the
drift height is taken as three-quarters of the value given by Equation G7-3 ,
where C u is the roof fetch distance upwind of the wall. The drift height at
two opposite sides of an RTU is also three-quarters of the value from Equa-
tion G7-3 , where C u is now the larger of the two roof fetch distances for the
direction of interest.
In a review of snow drift case histories, O'Rourke and DeAngelis (2002)
demonstrated that the three-quarters factor applied to windward drifts is rea-
sonable. The observed surcharge drift heights for six windward drifts were
compared with values predicted by the appropriate ASCE 7 provisions. The
resulting graph is presented in Figure G8-1 . In one case, the observed height
of 2 ft fi lled the space available for drift formation. If the parapet wall had
been taller, then a larger drift may have formed. This full-drift situation is
shown in Figure G8-1 by a horizontal line with question marks located to the