Review of en 1991-1-3 Part III - Multi-Pitched Roofs

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 27

To the Review of EN 1991-1-3

Snow Loads on Roofs


Part III: Snow load distribution
Multi-pitched roofs

Photo VDH: Melting down 65 cm snow per 8 h on a greenhouse roof

Autors:
Dr.-Ing. I. Pertermann, IB Puthli, Schüttorf
Prof. Dr.-Ing. R. Puthli, KIT, Karlsruhe Date: 24.02.2021
2
Table of content

Comments to prEN 1991-1-3:2020


Part III: Snow load distributions
Multi-pitched roofs

1. Introduction …………………………………………………………………………….…….… 4
1.1 General format – Changes ………..…………………………………..…………….… 4
1.2 Multi-span roofs ………………………………………………………………………. 4
2. Measurements on multi-span roofs ………………………………………………………..….. 5
3. Snow load distributions - models in different standards ………………………………......… 8
3.1 Multi-span roofs in the Eurocode - present and future …………………………..…. 8
3.2 Complete redistribution model in ISO 4355: 2013 …………………………...….… 12
3.3 Incomplete sliding model in ASCE 7-10 ……………………………………...…..… 15
4. General redistribution model for multi-span roofs – Proposal ………………………..…… 18

Literature to Part III …………………………………………………………………………..… 24

Note: This is a corrected version. In chapter 4 any drift losses above the trough are considered
within the model itself, not any more as an additional correction.

Symbols
Abbreviations:
Luv windward side of the structure
Lee leeward side of the structure
Symbols:
NOTE The following symbols are based on EN 1990 and EN 1991.

Latin upper-case letters:


Ce exposure coefficient
Ce,T special exposure coefficient for troughed roofs (roof valleys of multi-pitched roofs)
Cesl exceptional snow load coefficient
Cm surface material coefficient
Ct thermal coefficient
Sr,1 total roof snow load in kN for load case 1
Sr,2 total roof snow load in kN for load case 2
W width of a roof trough of a multi-pitched roof in m
3

Latin lower-case letters:


d depth of the snow in m
g acceleration due to gravity with: g = 9,81 m/s2
h; hr roof height in m
l roof width or span in m
l1 snow covered roof width in m
s roof span in m
s snow load in kN/m2
sb basic roof snow load in kN/m2 ISO-format
sd drift snow load in kN/m2 ISO-format
ss Slide load in kN/m2 ISO-Format; slide load limit in kN/m2
sgm roof snow weight at the slide limit in kg/m3
s0 characteristic ground snow load in kN/m2 ISO-format
sk,n characteristic ground snow load in kN/m2 for the reference period n
si,n,t characteristic roof snow load in kN/m2 for the location i, the shape coefficient µi, the reference
period n, the exposure coefficient Ce and the thermal coefficient Ct (Note: for Ct < 1; Ce = 1)
vm mean wind speed in m/s
vm,10 mean wind speed in m/s, measured in 10 m height
w width of the snow in the trough of a multi-pitched roof after sliding in m

Greek upper-case letters:


S Sum
Greek lower-case letters:
a angle of roof pitch, measured from the horizontal
r equivalent density by mass of a snow layer in kg/m3
g equivalent density by weight of a snow layer in kN/m3
µb basic shape coefficient ISO-format
µd drift load coefficient ISO-format
µs slide load coefficient ISO-format; slide load limit value
µs,mean mean value of the friction coefficient in a standardized sliding test
µi shape coefficient for the location i EN-format
µ1 shape coefficient EN-format for flat roofs
µ2 shape coefficient EN-format for mono- and duo-pitched roofs
µ2,b basic shape coefficient EN-format for duo-pitched roofs
µ0 shape coefficient for the location i = 0 above the ridge of a multi-pitched roof
µ3 shape coefficient for the location i = 3 above the gutter in a trough of a multi-pitched roof
n drift loss part from a trough (out of a roof valley)
4

1. Introduction
1.1 General format - Changes

For the calculation of roof snow loads, EN 1991-1-3:2003 is based on a multiplicative format, as
opposed to other standards, such as the ISO 4355, which have an additive format, because drifting
and sliding snow are added to the basic snow, uniformly distributed.

EN 1991-1-3: 2003: Roof snow load: si,n,t = µi · Ce · Ct · sk,n


where µi is the shape coefficient for the location i, with µ1 = 0,8 as
a reference value for flat roofs with a roof angle a = 0°
Ce is the exposure coefficient
Ct is the thermal coefficient
sk,n is the characteristic ground snow load* for a reference period n
Note: * Exceptional snow loads sAd = Cesl · sk,n can be treated in the same way.

The draft versions of the future Eurocode, e.g. the final draft prEN 1991-1-3:2020, indicate a shift of
the exposure coefficient from the equation for the roof snow load into the shape coefficient µi, where
it can be adapted to the shape of the roof. Within the shape coefficient the exposure coefficient Ce
can accommodate drift losses as well as drift surcharges better than before.

prEN 1991-1-3: 2020: Roof snow load: si,n,t = µi · Ct · sk,n


where µi is the shape coefficient for the location i, with µ1 = 0,8 · Ce as
a reference value for flat roofs with a roof angle a = 0°

1.2 Multi-pitched roofs

For the recent review of the second generation of the Eurocode EN 1991-1-3 according to the final
draft prEN 1991-1-3:2020, it has been claimed that international codes such as ISO 4355:2013 and
ASCE 7-15 have been compared. However, this cannot be true for multi-pitched roofs, as it will be
explained in detail in chapter 3.1, 3.2 and 3.3. For multi-pitched roofs, the redistribution of snow
within the limited space of a trough (valley) and the related drift losses from the trough are to be
specified, see proposal in chapter 4. This has not been done for prEN 1991-1-3:2020.
Also, the background needs to be understood, before any coefficients or values, such as a thermal
coefficient Ct or shape coefficients µi are taken into another standard. Especially the American codes
are calibrated quite differently.

Because multi-pitched roofs are standard for commercial production greenhouses, the research for
greenhouse related problems in the same and in other relevant documents lead to quite different
conclusions, such as the differentiation of roof snow loads and their distributions according to the
thermal conditions (Thermal coefficient Ct) and the surface roughness (Surface material coefficient
Cm). However, this seems to be beyond the scope of the Eurocode prEN 1991-1-3:2020. Therefore,
for glass and similar roofs, alternative regulations would be required. Only for commercial production
greenhouses this can be realised, see EN 13031-1:2019. Greenhouses open to the public and other
buildings with a glass roof are supposed to be covered by the Eurocode. It is shown in chapter 3.1
for which cases the model according to prEN 1991-1-3:2020, chapter 7.5.4 fails completely.

The background to general influences on roof snow loads is described in detail in the background
documents Snow EN 13031 Part I: Melting and Part II-0: Sliding and Drift. The following
chapters are partly based on the special background described in Snow EN 13031 Part II-2: Snow
Load Distribution – Multi-pitched roof.
5

2. Measurements on multi-pitched roofs


The most severe drifting on a multi-pitch roof was measured in a Survey of the National Research
Council of Canada on an unheated airport building in Ottawa, reported by Taylor (1980) and many
other authors after him. From the 3rd to the 5th of March 1971, the Eastern Canadian Blizzard was one
of the four most severe “storms of the century”. The blizzard produced up to 73 cm of new snow on
March the 2nd and continued with 40,6 cm at wind speeds up to 61,4 m/s on March the 4th. After 3
days, a ground snow load of 3,35 kN/m2 was attained in Ottawa. At the end of the season 444 cm of
snow had piled up, “the most in the city’s record”.

The multi-span roof had 3 spans of 11 m each and a roof height of 1,83 m between eaves and ridges.
The eaves height was small with 3 m above ground. The roof angle of this building was also very low
with 18,4°, giving a ridge height of 4,83 m. With these dimensions the building remained within the
ground snow drift zone of up to 5 m according to Tabler (2003), see also Background Snow EN
13031 Part II-0: Sliding and Drift. The rather exposed airport location may also have provided fetch
distances of 150 m to 300 m to develop a saturated drift flux (Takeuchi (1980)). Therefore, the drift
conditions were unusual for roofs of normal buildings but otherwise ideal.

Because of the severe snowstorm the snow load distribution due to drifting snow has nearly reached
the equilibrium state, where the drifted snow surface is streamlined enough for the wind flow to get
reattached to the surface. After that, drift growth on the roof stops and the snow load can only be
enlarged by further non-drifting snowfall. On this multi-span roof, the snow would continue to
accumulate in the same way as on a flat roof.

The near-equilibrium distribution in Figure 1 is shown together with the ground snow load level, as
it would appear on the roof without drift losses by wind. On the roof, a mean snow density of 300
kg/m3 was measured after the blizzard. This is a smaller drift snow density than predicted by Tabler
(2003), because it was the result of a single snowfall event of no more than 3 days. With the mean
snow density of 300 kg/m3 (2,943 kN/m3), the measured ground snow load of 3,35 kN/m2 would have
resulted in a snow height of 1,108 m above the roof surface (1,1383 m above the ground), if there
were no drift losses. This amount of snow was about 21% larger than the volume of the trough.
Therefore, some snow above the ridges was to be expected.

Figure 1: Incomplete drifting on a multi-span roof Ottawa 1971 according to Taylor (1980)

The measured snow load distribution in Figure 1 shows that not much snow settled in regions with
high wind speeds, the first windward roof pitch and the upper half of the second and third windward
roof pitches. Most snow has accumulated behind the first ridge, where the air flow was separated
from the roof surface and had to slow down. A drift formed in the zone with reverse air flow on the
leeward side. From there, the drift did grow to the windward side of the first valley filling it nearly
up to the ridge. The air flow may have reattached itself to the new surface; only slight discontinuities
below the second and third ridges are left.
6
The same drift growth process happened in the second valley, but with a smaller wind speed and a
smaller drift rate. However, the wind was still strong enough to create a nearly ideal horizontal roof
snow surface on the third and outer leeward roof face. It has to be noted, that the roof snow did not
end vertically above the outer edges, it was overhanging and started to creep down. Also, the large
wind- and leeward drifts at the ground are not shown in Figure 1.

It is important to understand, that the snow was not “redistributed” by wind after it accumulated, as
often claimed, the drifting snow settled there at the first place due to the prevailing wind field
saturated with drifting snow. More snow settled down in areas with lower wind speed, changing the
wind field in such a way, that the wind can flow with less resistance over the obstacle. Drifting is a
self-regulatory process.

In total the roof snow after drifting was not larger than the ground snow level, as it is shown above
the roof surface in Figure 1. This is also shown by calculation. The measured and calculated data and
ratios are summarized in Table 1.

Span 1 Span 3 Span 3


Windward side Valley 1 Valley 2 Leeward side
Ridges: 1 2 3

windward leeward windward leeward windward leeward

Measured roof snow heights in cm:


61 & 41 36 122 & 155 185 114 & 41 23 97 160 94 20 89 & 155
Measured mean density roof snow: 300 kg/m3
Measured ground snow load at the site: 3,35 kN/m2 Ground snow height: 112 cm
Global snow load values calculated from values above:
Roof snow load in kN/m (per m roof length):
7,634 22,034 11,86 16,078 10,313 15,054
Total roof snow load: 82,97 kN/m roof length
Ground snow load in kN/m (per m roof length):
18,425 18,425 18,425 18,425 18,425 18,425
Total ground snow load: 110,55 kN/m roof length
Ratio of roof snow load to ground snow load:
0,414 1,196 0,644 0,873 0,5597 0,817
0,414 0,9198 0,716 0,817
0,805 0,758 0,688
Total ratio of roof snow load to ground snow load: 0,75
Local values at ridges and valleys:
Roof snow load in kN/m2:
1,056 5,428 0,675 4,964 0,587
2
Ratio of roof snow load to ground snow load sG = 3,35 kN/m :
0,315 1,62 0,201 1,401 0,175
Table 1: Measured roof snow on a cold multi-span roof with a low roof angle (18,4°) in
Ottawa after the “Blizzard of the century” March 1971

Because of the low roof angle and the fact, that more snow arrived than the valleys would
accommodate, there was still 36 cm, 23 cm and 20 cm snow above the 3 ridges. On the windward
side of the first span the snow height is small, but the roof not cleared. For higher roof angels than
7
18,5° and a larger roof valley or slightly less snow, the windward side of the first span and the ridges
would have been cleared of snow. From CFD investigations it is known that higher wind speeds above
4 m/s and roof angles above 17° to 20° are main influences (Tominaga et.al. (2016)). The roof surface
roughness and roof temperature might also be highly influential. These limits are still controversial.
Therefore, it is a suitable approach for prEN 1991-1-3:2020, to assume complete windward sliding
for all pitched roofs a > 5°. Other codes differentiate and might have other limits such as 15°
according to NBCC. However, such limits should be consistent for all roofs within a standard, which
is not the case in prEN 1991-1-3:2020.

With the given snow heights, measured at 2 to 4 points along the roof pitches, a total roof snow load
of 82,97 kN per meter roof length can be calculated. The measured ground snow load of 3,35 kN/m2
after 3 days only resulted in a snow height of 1,85 m in the first valley being slightly larger than the
roof height of 1,83 m. In the second valley the snow height 160 cm is 86,5% of the maximum height.

The total roof snow load is only 75 % of the total ground snow load, not more, as often implied in
publications. Only the first leeward roof pitch had 20% more snow than the ground snow value of the
roof pitch might have. This was an intermediate drift state and the snow would have been distributed
more evenly in the equilibrium state with a little more wind. The first span had in total 80% and the
first valley 92% of the ground snow level. The local snow load value in the first valley (1,82563 m
roof snow) was not larger than 1,6 times of the ground snow value.
For a comparison, a flat roof in the same exposed airport location in Canada would have to be
designed for a roof-ground snow load ratio of 60% using a basic roof snow coefficient of Cb = 0,8
and a drift coefficient of Cw = 0,75 according to NBCC. Although this short-time snowfall event did
not represent the characteristic snow load level, it showed that on multi-span roofs a little less snow
is removed by wind than on flat roofs.

Conclusion: It can be seen that in the equilibrium state for the unbalanced load case with unlimited
wind and snow supply for multi-span roofs, the snow surface would be parallel to the ridge heights
with a snow-free windward roof pitch, a triangular distribution on the outer leeward roof pitch and
zero snow on all the ridges (s0 = 0). The maximum value above the gutter would be s3,max = g · h.

In this specific limit state only a certain amount of snow can be contained within the limited space of
the valley. Assuming there are no drift losses but losses by melting snow, the snow load potential is
sk,lim = (g · h/2) / Ct. Potential drift losses are a national choice. If drift losses from a valley are to be
allowed, they should be smaller than on a flat roof as a reference and limited by the overall roof size
as well as the number of valleys in a row.

The equilibrium state is a limit for usual design snow load distributions for non-equilibrium stages
when there is not enough snow (sk < sk,lim) or too much snow (sk > sk,lim) available due to snowfall.
As in prEN 1991-1-3:2020, the limit cannot be described by a maximum value for the shape
coefficient µ3 alone, the total snow load in the valley is limited. Also, it does not depend on the actual
snow load sk, which can be smaller or larger. For the format according to prEN 1991-1-3:2020 with
si = µi · Ct · sk the snow load potential and shape coefficients for the limit can be given as follows:

Equilibrium state limit for the snow load distribution on multi-pitched roofs:
Snow load potential to fill the valley completely: sk,lim = (g · h/2) / Ct
Above the ridge (i = 0): s0,min,lim = 0 µ0,lim = 0
Above the gutter (i = 3): s3,max,lim = g · h µ3,lim = 2
Where Ct is the thermal coefficient;
h is the depth of the valley in m;
g is equivalent weight density of the accumulation in the valley * in kN/m3
8
Note: * The equivalent density is a national choice. In prEN 1991-1-3:2020 g = 2 kN/m3 is
recommended for cold roofs with Ct = 1. For warm roofs with controlled heating according to EN
13031-1:2019 g = 3,5 kN/m3 should be applied.

For snow loads below the limit (sk < sk,lim) any distribution can be chosen as long as it gives a smooth
transition to the equilibrium state as in nature, compare Figure 1. For warm roofs with slippery roof
surfaces, complete sliding should be considered with µ0 = 0. For cold roofs and for small roofs, where
sliding is restricted, incomplete stages of redistribution can be chosen. This is a local problem.
However, for snow loads above the limit (sk > sk,lim), the excess snow above the ridge level is
distributed as on a flat roof. The drift losses from the excess snow should be consistent with the flat
roof model.
None of these principles have been considered for the model in prEN 1991-1-3:2020, 7.5.4.

Grammou (2015, 2019) investigated snowdrifts on photovoltaic (PV) panels mounted on flat roofs.
The water channel that was used, with fluid characteristics and sand particles matched natural drift
conditions better than artificial snow in a wind channel with small models. The water flume model
was calibrated in such a way as to produce similar distributions as snow in the moderate climate of
the middle of Europe (80% of German territory). Small fluid speeds were sufficient to “blow” the
ridges of the PV panels free of sand. The largest measured shape coefficient in the valleys between
the PV panels was µ2 = 1,57 < 1,6 for a distribution with plateau and µ2 = 2 for a triangular
distribution. The accumulations in the middle were larger than at the edges, where the sand could be
removed completely with increasing duration (equilibrium).

An important finding not only applicable to roof mounted solar panels was, that in all of the tests
there was less sand on the roof model than on the bottom of the water channel. It was shown,
that snow is preferably blown down and does not settle in drift mounds above ridge level. The height
of the solar panels forms a hard barrier as solid snow fences would do (Tabler (2003)). Extreme
unbalanced snow accumulations on a roof are produced together with large overall drift losses.
Snow drifts are shaped like the wind profiles that formed it.

3. Snow load distributions - Models in different standards


3.1 Multi-span roofs in the Eurocode – present and future

International research shows, that drifting cannot produce more snow on the roof than on the ground,
as the drift models in EN 1991-1-3:2003, chapter 5.3.4 or in prEN 1991-1-3:2020, chapter 7.5.4
imply, with up to 144% and now even up to 154% of the ground snow load on the roof (for a = 30°
and Ce = 1,2; Ct = 1,2).

The recent model in the Eurocode EN 1991-1-3:2003 seemed to aim just on using the roof angle
influence as for other pitched roofs. In the trough the snow loads increased locally up to the “magic
limit” of µ = 1,6 = 2 · 0,8 in the unbalanced load case (ii). That was all. Between 5° and 30° despite
the increasing snow load in trough, the snow load at the ridges remains unchanged. For 30° the total
roof snow load above the trough is 144% of the ground snow load (Ce = 1,2). For the balanced load
case (i) the snow above the trough used to get smaller with roof angels above 30° and disappear with
roof angles above 60°, see Table 2.

The new regulation in prEN 1991-3:2020 shown in Table 2 is also no redistribution model. For the
ridges the same roof angle function is used as for other pitched roofs. The complete failure of the
model for larger snow loads in comparison to the trough height and large roof angles is only just
avoided by the following three regulations or limits:
9
• Limit 1: With µ2,b(30°; 1) = 0,8 in load case (i), the snow load remains at 80% of the ground
snow level despite of the roof angle. However, for larger roofs this might be too small.
• Limit 2: The maximum shape coefficient µ3(a) in the trough is now limited when reaching
the trough height h by µ3(a) ≤ g · h / (Ct · sk). This becomes problematic for larger roof snow
loads.
• Limit 3: The maximum µ3,max can be chosen nationally (NDP), otherwise µ3,max = 1,6 applies
as before.

However, despite of these 3 limits, probably taken from the new German National Annex DIN EN
1991-1-3/NA:2019 and partly because of it, the model remains inconsistent. In Germany the exposure
coefficient Ce and the thermal coefficient Ct are not taken into account (Ct = Ce = 1).

EN 1991-1-3:2003 Shape coefficients µ2 (a) and µ3 (a)


Shape coefficients 0° ≤ a ≤ 30° 30° < a < 60° a ³ 60°
Load case (i) µ2 = 0,8 µ2 = 0,8 (60° - a) / 30° µ2 = 0
Load case (ii) µ3 = 0,8 + 0,8 a / 30° µ3 = 1,6 -
Format Roof snow load: s = sk · Ce · Ct · µi
Where a is the roof angle (inner trough area µ3 is calculated with a = 0,5 · (a1 + a2));
µ3 is the maximum shape coefficient above the gutter;
µ2 is the shape coefficient above the ridge depending on the roof angle.

prEN 1991-1-3:2020 Balanced Load Arrangement


Shape coefficients Uniform distribution for roof angle a in °
Windward Above the troughs Leeward
Load case (i)
µ2,b (a1, Ce) µ2,b (a = 30°; Ce = 1) = 0,8 µ2,b (a2, Ce)
Basic shape coefficient for outer roof surfaces ai for i = 1 und 2:
5° < ai ≤ 30° µ2,b (ai; Ce) = 0,8 Ce
as
Saddle 30° < ai < 70° µ2,b (ai, Ce) = 0,8 Ce · (70° - ai)/40°
roof
ai ³ 70° µ2,b (ai, Ce) = 0
Unbalanced Load Arrangement (outer roof surfaces as in case (i))
Load case (ii) Above the ridges: Shape coefficients as for outer roof surfaces load case (i)
Above the gutter: µ3 (a) = 0,9 + 0,7 · a/30° ≤ g · h / (Ct · sk) ≤ µ3,max
Format Roof snow load: s = sk · Ct · µi
Where a1,a2 is the roof angle of the roof pitches; inner troughed area: a = 0,5 · (a1 + a2);
h is the height of the trough in m;
g is the density of the snow with g = 2 kN/m3, if not otherwise specified (NDP);
µ3,max is the maximum above the gutter with µ3,max = 1,6, if not otherwise specified (NDP)

Table 2: Shape coefficients and snow load distributions for multi-span pitched roofs
according EN 1991-1-3:2003+A1:2015 and FprEN 1991-1-3:2020

How problematic this regulation can get, especially for large roof areas with many small roof spans
in comparison to the snow load, can be demonstrated on the shape coefficients and the ratio of the
total snow load in the trough to the ground snow load.
For a large multi-span roof with equal roof heights and roof angles the total load above the trough
can be calculated as follows:
10
S STrough = sk · Ct · ½ { µ2,b(a; Ce) + MIN (µ3(a); g · h/(Ct · sk); µ3,max)) }

Where Ct is the thermal coefficient and Ce is the exposure coefficient;


a is the mean value of the roof angles within the trough: a = ½ (a1 + a2);
µ2,b(a;Ce) is the basic shape coefficient above the ridge (see Table 4);
µ3(a) is the shape coefficient above the gutter: µ3(a) = 0,9 + 0,7 a/30° ≤ g h / (Ct sk) ≤ µ3,max
g is the density of the snow g = 2 kN/m3 (National choice by NDP);
h is the height of the roof (depth of the trough) in m;
µ3,max is the maximum µ3,max = 1,6 (National choice by NDP)

In Table 3 the results for normal cases with Ct = 1 und Ce = 1 are shown.

Roof Shape Relevant coefficients including limits for shape coefficient µ3


pitch coefficients µ3 ≤ lim µ3 = g · h / (Ct · sk) ≤ 1,6 with g = 2 kN/m3
angle (for Ce = 1) Characteristic snow loads on the ground sk in kN/m2
a in ° µ2,b µ3 0,65 kN/m2 1 kN/m2 1,5 kN/m2 2 kN/m2 4 kN/m2 6 kN/m2
g · h / (Ct · sk) = 3,077 2 1,333 1 0,5 0,333
10° 0,8 1,133 1,133 1,133 1,133 1 0,5 0,033
20° 0,8 1,367 1,367 1,367 1,333 1 0,5 0,033
30° 0,8 1,6 1,6 1,6 1,333 1 0,5 0,033
40° 0,6 1,833 1,6 1,6 1,333 1 0,5 0,033
50° 0,4 2,067 1,6 1,6 1,333 1 0,5 0,033
60° 0,2 2,3 1,6 1,6 1,333 1 0,5 0,033
70° 0 2,533 1,6 1,6 1,333 1 0,5 0,033
If the trough is full: lim µ3 ≤ 1,6 is used instead of µ3
Case (i) Shape coefficient µ2,b(30°;1) = Ratio roof to ground snow load
a in ° µ2,b(a; Ce) s2/sk = Ct · Ce · µ2,b with: µ2,b(30°; 1) = 0,8 above the trough; for Ct = 1
10° - 30° 0,8 0,8
40° 0,6 0,8
50° 0,4 0,8
60° 0,2 0,8
70° 0 0,8

Case (ii) Ratio of total snow load in the trough to ground snow load
a in ° µ2,b sTrough/sk = Ct · (µ2,b + MIN(µ3; lim µ3; 1,6))/2 with Ce = Ct = 1
10° 0,8 0,956 0,956 0,956 0,9 0,65 0,417
20° 0,8 1,083 1,083 1,067 0,9 0,65 0,417
30° 0,8 1,2 1,2 1,067 0,9 0,65 0,417
40° 0,6 1,1 1,1 0,967 0,8 0,55 0,317
50° 0,4 1 1 0,867 0,7 0,45 0,217
60° 0,2 0,9 0,9 0,767 0,6 0,35 0,117
70° 0 0,8 0,8 0,667 0,5 0,25 0,017
Table 3: Shape coefficients, limits for µ3 and ratio of roof to ground snow load for an example
for roof height h = 1 m (Ce = 1; Ct = 1) according prEN 1991-1-3:2020
11
Without load case (i) the model for load case (ii) would fail completely and become unsafe for large
ratios of snow load to roof height and for large roof angles (red ratios < 0,8). The reason is simply,
that the snow is not redistributed within the limited space of the triangular trough; and if the trough
is full, above the ridges, despite the fact, that this is very easy, just trigonometry.

Because of the not realistic roof angle influence in the model and due to the ignored principle, that
snow has to be redistributed within a limited space, for Ct = 1 und Ce = 1 the total snow loads vary
between 80% and 120% of the ground snow load level. It is not plausible, that the roof angle is the
cause of such a large range. Also, the trends are not plausible. For higher snow loads in comparison
to the roof height and for larger roof angles, the total snow load decreases.

With the help of the example of a multi-span roof with symmetrical roof troughs with 30° roof angle
with µ2,b = 0,8 Ce and µ3 according to Table 3, further trends can be shown, when the exposure
coefficient Ce and the thermal coefficient Ct are varied as shown in Table 4.

Ratio of Typ of Thermal Ratio of roof to ground snow load


snow height building coefficient
to trough (thermal Exposure coefficient Ce
height Influence) 1,2 1 0,8
Freezer hall,
Ct = 1,2 1,536 1,44 1,344
Small: Ice sport arena (µ2,b = 0,96; µ3 = 1,6) (µ2,b = 0,8; µ3 = 1,6) (µ2,b = 0,64; µ3 = 1,6)
µ3 relevant Normal case Ct = 1
1,28 1,2 1,12
(µ2,b = 0,96; µ3 = 1,6) (µ2,b = 0,8; µ3 = 1,6) (µ2,b = 0,64; µ3 = 1,6)
Large: Freezer hall, Case (i): µ2,b = 0,8: S S/W/sk = 0,96
Ice sport arena
Ct = 1,2
lim µ3 (µ2,b = 0,8 · Ce; µ3 = g · h/(Ct · sk) < 1,6: Case (i) relevant)
< 1,6 Case (i): µ2,b = 0,8: S S/W/sk = 0,8
relevant
Normal case Ct = 1
(µ2,b = 0,8 · Ce; µ3 = g · h/(Ct · sk) < 1,6: Case (i) relevant)
Table 4: Ratio of roof to ground snow load (S S/W) / sk and shape coefficients for the inner
troughs of multi-span roofs according to prEN 1991-1-3:2020

For small snow heights, the model does not allow a realistic design, especially in case of Ct = 1,2. For
large snow heights, the model is less safe than for small snow heights. This is the opposite trend in
comparison to a flat roof. Furthermore, in case of unheated buildings with Ct = 1 (no ice sport or
freezer hall), for very large roof areas with many small roof spans, in comparison to flat roofs, the
question arises, where 20% of the ground snow load might have gone.

General conclusion (Summary):

In comparison to other models according to prEN 1991-1-3:2020, the model for multi-pitched roofs
according to 7.5.4 does not apply consistent rules about the redistribution of snow by drifting and
sliding. The model fails to adapt the exposure coefficient Ce according to its local influence in the
trough or at the ridges as well as its overall global influence. Of course, this is not easy, as ISO
4355:2013 shows, see chapter 3.2. No consideration was given to the influence of the surface
roughness or any thermal influences, which may cause sliding.

For the recent review of the second generation of the Eurocode EN 1991-1-3 according to the final
draft prEN 1991-1-3:2020, neither international research has been considered, nor as claimed other
international codes such as ISO 4355:2013 and ASCE 7-15 have been compared.

The model is also not calibrated for all cases of multi-span roofs, it is supposed to cover. No
comparison to flat roofs, small or large, is made.
12
As a result, the model does not allow a realistic design and can even be unsafe. Especially important
is such a model failure for lightweight roof structures, where the snow load is more important than
the dead weight, e.g. for glass or membrane roofs.

Conclusion for greenhouses:

The model for multi-span roofs in prEN 1991-1-3:2020, Figure 7.6 shows just two joint gable roofs
with one valley between them, a duo-span situation. The snow load distribution seems to aim at large
roof height to snow height ratios. The sliding of roof snow on slippery warm surfaces and the
redistribution of snow in a small limited space are not considered. There are safety issues for large
snow loads on large roof areas.

Figure 2: Sliding of snow on a greenhouse Type Venlo with plastic film after an unexpected
snow fall event (during construction: emergency heating up to +7°C)

For typical multi-span pitched greenhouse roofs with many small roof troughs with small ratios of
roof height to snow height, the models of EN 1991-1-3:2003 and prEN 1991-1-3:2020 are not
suitable. Therefore, the greenhouse standard EN 13031-1:2019 has specialized regulations for the
type of roofs, it covers. They are rather based on research and modern snow load standards, such as
ISO 4355:2013 and ASCE 7-10, but avoiding their shortcomings, which will be discussed in chapter
3.2 and 3.3. They are not presented here, but in the Background Report Snow EN 13031 Part II-2.

3.2 The redistribution model in ISO 4355:2013


The drift model for multi-pitched roofs in ISO 4355:2013, Annex B, chapter B.3 is based on the
redistribution of the basic roof snow load, adapted by the general conversion factor of fc = 0,8, the
exposure coefficient Ce and the thermal coefficient Ct, but for a basic shape coefficient of µb = 1. In
this limit state, the snow is allowed to slide down completely (sliding load case 2 in ISO 4355:2013),
but only after age related losses due to drift, melting, sublimation / evaporation have occurred. If the
snow quantity is too large for the roof trough (valley), the excess snow is distributed evenly above
the ridge level (sliding load case 1 in ISO 4355:2013). For the snow after sliding a mean density of r
= 300 kg/m3 is recommended.
13
ISO 4355:2013 formulates two principles:
• “However, for multi-pitched roofs, the snow slides and results in a redistribution of load on
the same roof.”
• “The sliding load case accounts for the potential sliding snow and possible drifted snow.”

ISO 4355:2013 seems to consider a final state distribution occurring after some duration. Therefore,
drift losses as large as for a flat roof are allowed. Also, the influence of the roof size is not mentioned.
For large multi-span roofs, an increased exposure coefficient Ce ≤ 1,25 as for large flat roofs could
be used. With that the ground snow load level would be reached on the roof. However, such a
regulation would also be not suitable if the thermal coefficient Ct < 1 according to ISO 4355, Annex
D, which is based on precipitation data and is to be used with µb = 0,8 and Ce = 1.

For smaller roofs it remains questionable, whether the general conversion factor and exposure
coefficient Ce in the formulae can have the same influence as on a flat roof. Drifting can reduce snow
on the ridges, but most of the drifted snow will fall back into the troughs and not be removed from
there. Also, sliding from the ridges down into the trough would reduce the possibility for further drift
losses. As for multi-level roofs in ISO 4355:2013, Annex B, chapter B.5, a roof trough could be
treated as a sheltered zone below the ridge: Ce0 = 1,2. For snow on the ridges and above ridge level
Ce0 < 1,2 can be used. Again, this could be problematic for warm roofs with Ct < 1 according to ISO
4355, Annex D.

Given are the following formulae:


ISO 4355:2013, Annex B, chapter B.3 Multi-pitched roofs

Basic load case:


sb = 0,8 s0 · Ce · Ct · µb (B.5)
Sliding load case 1:
h < (0,8 s0 · Ce · Ct · µb · W / (r · (tan (90°- b1) + tan (90°- b2))))1/2 (B.6)
d = (0,4 s0 · Ce · Ct · µb / r) + h/2 (B.7)
ss1min = 2 r ((0,4 s0 · Ce · Ct · µb / r) - h/2) (B.8)
ss1max = 2 · r · d (B.9)
Sliding load case 2: How is “w” to be calculated?
h ≥ (0,8 s0 · Ce · Ct · µb · W / (r · (tan (90°- b1) + tan (90°- b2))))1/2 (B.10)
d = (0,8 s0 · Ce · Ct · µb · W / (r · (tan (90°- b1) + tan (90°- b2))))1/2 (B.11)
ss2min = 0 (B.12)
ss2max = 2 · r · d (B.13)
where: d and h are not defined
sb is the basic roof snow load in kN/m2
s0 is the characteristic ground snow load in kN/m2
Ce is the exposure coefficient
Ct is the thermal coefficient
µb is the basic shape coefficient
b1 and b2 are the roof angles of the two pitches in degrees
r is the snow density in kg/m3 in the trough after sliding (r = 300 kg/m3 is suggested)
W is the width of the roof trough in m
w is the width of the snow in the trough after sliding in m

In any case, the drift model for multi-pitched roofs in ISO 4355:2013, Annex B, chapter B.3 needs to
be calibrated. Furthermore, the formulae given in ISO 4355:2013, chapter B.3 cannot be used for
14
calculation without major changes because of the following printing mistakes and contradictions
between them.

The acceleration due to gravity g = 9,81 m/s2 is missing. Instead of the density by mass r in kg/m3
the “weight” density r · g in kN/m3 is to be used.

Figure B.4 is incomplete and contradictory. It does define the width “W” of the roof trough and the
width “w” of the snow in the trough after sliding and the roof angles b1 and b2, but it does not define
the heights h and d. However, as in other parts of the standard, h seems to be the height of the roof
as the obstacle and d seems to be a snow depth.

However, which snow depth? The snow depth “d” could be the equivalent roof snow depth after
sliding distributed uniformly over the width w. Some of the formulae seem to support this. However,
more likely “d” is the maximum roof snow depth in the trough after sliding, as it is used in other parts
of the document. In this case, there are some mayor corrections required.

ISO 4355:2013, Figure B.4 - “Snow load distribution – multi-pitched roof” shows a trough half filled
with snow after complete sliding. This would be the slide load case 2. This load case is shown in the
figure as a steep triangle over the reduced width w of the snow in the trough. This part of the figure
is in contradiction to the snow load distribution shown above it, which belongs to load case 1.

Furthermore, the given equations are wrong. For load case 2 equation (B.11) gives the snow depth
“d” of the basic load case (B.5) assuming the snow density r after sliding is corrected by multiplying
it with g = 9,81 m/s2. Two times this value above the gutter (B.13) and zero above the ridge (B.12)
describe a triangular snow load distribution over the entire width of the trough W, no complete sliding
as long as the snow does not fill the trough completely. Only in this case the equations would be
correct. However, in this case the figure would be misleading.

The selection criterion (B.6) and (B.10) compares the height of the trough “h” with the snow depth
of the basic load case (B.5) using r · g. However, if “h” and depth of the uniformly distributed basic
snow load would be equal, the snow would not fit in the triangular trough. Half of it would not fit in
and be left on top, as in load case 1. Therefore, this cannot be the criterion to separate both load
cases from each other.

Conclusion: This model cannot be used without basic corrections.

In the following corrections it is assumed that ”d” is the maximum depth of the snow in the trough
after sliding. Slide load case 2 (trough is not full) is corrected first, because slide load case 1 (trough
is full) is based on it. Also, “slide load case” is used instead of “sliding load case”.

Correction:
Slide load case 2: Snow height smaller than roof height
The triangular load after sliding Ss in kN/m roof length can be written as: Ss = r · g · d/2 · w.
The roof snow load before sliding can be written as: SR = 0,8 · s0 · Ce · Ct · W.
The snow load is to be redistributed and applying SR = Ss results in the following:
0,8 · s0 · Ce · Ct · W = r · g · d/2 · w
d = 1,6 · s0 · Ce · Ct · W / (r · g · w)
For the geometry of the roof trough and the slide load in the trough the following geometric ratios
apply: tan (90°- b1) = cot b1 = W1/h = w1/d
tan (90°- b2) = cot b2 = W2/h = w2/d
15
Roof width: W = W1 + W2 = h · (tan (90°- b1) + tan (90°- b2)) = h · (cot b1 + cot b2)
Snow width: w = w1 + w2 = d · (tan (90°- b1) + tan (90°- b2)) = d · (cot b1 + cot b2)
The depth of the snow after sliding can be written as:
d = 1,6 · s0 · Ce · Ct · h / (r · g · d)
d2 = 1,6 · s0 · Ce · Ct · h / (r · g)
d = (1,6 · s0 · Ce · Ct · h / (r · g))1/2
The roof height h can be replaced by the roof width W by introducing the roof angles, but this
complicates the calculation as follows:
d = (1,6 · s0 · Ce · Ct · W / (r · g · (cot b1 + cot b2)))1/2
d = (1,6 · s0 · Ce · Ct · W / (r · g · (tan (90°- b1) + tan (90°- b2))))1/2
Because the snow depth in the trough has the height d, not 2 d, the condition for the differentiation of
slide load case 1 and 2 is also to be corrected:
h < / ≥ d = (1,6 · s0 · Ce · Ct · h / (r · g))1/2
h < / ≥ d = (1,6 · s0 · Ce · Ct · W / (r · g · (cot b1 + cot b2)))1/2
h < / ≥ d = (1,6 · s0 · Ce · Ct · W / (r · g · (tan (90°- b1) + tan (90°- b2))))1/2
For the design with load case 2 the width w of the snow in the trough has to be estimated using the
trigonometrically relations to the snow depth and the roof angles:
w = d · (tan (90°- b1) + tan (90°- b2)) = d · (cot b1 + cot b2)
The snow load values are: ss2,min = 0 and ss2,max = r · g · d

Slide load case 1: Snow height larger than roof height.


The snow load consists of a triangular part filling the trough and a rectangular rest above the trough.
The roof snow load before sliding can be written as: SR = 0,8 · s0 · Ce · Ct · W.
The roof snow load after sliding Ss in kN/m roof length can be written as:
Triangular part in the trough: Ss,T = r · g · h/2 · W.
Rectangular part above the trough: Ss,R = r · g · Dd · W.
The snow load is to be redistributed and applying SR = Ss,T + Ss,R results in the following:
0,8 · s0 · Ce · Ct · W = r · g · (h/2 + Dd) · W
h/2 + Dd = 0,8 · s0 · Ce · Ct / (r · g)
Dd = 0,8 · s0 · Ce · Ct / (r · g) - h/2
The maximum height d of the snow in the middle of the trough can be corrected as follows:
d = h + Dd = h/2 + 0,8 · s0 · Ce · Ct / (r · g)
As a result, the snow load values for sliding and redistribution can be corrected as follows:
Above the ridge: ss1,min = r · g · Dd = 0,8 · s0 · Ce · Ct - h/2 · r · g
Above the gutter: ss1,max = r · g · d = 0,8 · s0 · Ce · Ct + h/2 · r · g
Despite the open questions with reference to the drift losses in comparison to flat roofs and the
required major corrections, the snow load redistribution model for multi-pitch roofs in ISO 4355:2013
is more realistic than any other model so far, compare Figure 1 and Figure 2.

3.3 The incomplete sliding model in ASCE 7-10

An incomplete redistribution model for roof snow on multi-span roofs can be found in ASCE 7-10
(see Table 5) with local snow loads of 0,5 pf above the ridges and 2 pf / Ce in the trough (valley).

The local factor 2 pf / Ce in the valleys seems to be large, but in the USA the roof snow is calculated
with a general conversion factor of 0,7 only. This can be compensated by thermal factors Ct between
16
1,1 and 1,3 for cold roofs; as opposed to Ct = 1, which requires a warm roof on a closed building with
qi >> 0°C, the reference case in the USA.

The influence of melting on the roof snow loads is slightly more dominant than the influence of
drifting. Therefore, single values of Ce or Ct cannot be compared directly to the appropriate values in
the Eurocode, because the ranges are different. The total roof snow load should be compared, as can
be seen in Table 5a, based on the differentiation in Table 5b.

Multi-span pitched / arched roof ASCE 7-10


chapter 7.4.4, 7.6.3 and Figure 7-4

Balanced Load Case 1:


Flat roof snow load: pf = 0,7 · Ce · Ct · pg (for Importance Factor Is = 1 and Roof Slope Factor Cs = 1)
with: Ce Exposure Factor, with 0,7 ≤ Ce ≤ 1,2, see classification in Table 8
Ct Thermal Factor, with 0,85 ≤ Ct ≤ 1,3, see classification in Table 8
pg Characteristic ground snow load, with 1,2 kN/m2 ≤ pg ≤ 14,4 kN/m2

Total snow load per trough: Ratio of conversion pf / pg


Thermal Exposure Factors Ce
Factor Ct 1,2 1,1 1 0,9 0,8 0,7
Load
1,3 1,092 1,001 0,91 0,819 0,728 0,637
Case
1 1,2 1,008 0,924 0,84 0,756 0,672 0,564
1,1 0,924 0,847 0,77 0,693 0,616 0,517
1 0,84 0,77 0,7 0,83 0,56 0,4704
0,85 0,714 0,6545 0,595 0,5355 0,476 0,3998

Unbalanced Load Case 2:


Locally at the ridges: Min: ps1 = 0,5 pf Locally in the valley: Max: ps2 = 2 pf / Ce
Total snow load: Ps = (ps1 + ps2) / 2 · W (with: W – total width of the trough)

Total snow load per trough: Ratio of conversion (Ps / W) / pg


Thermal Exposure Factors Ce
Factor Ct 1,2 1,1 1 0,9 0,8 0,7
Load
1,3 1,183 1,16025 1,1375 1,11475 1,092 1,069
Case
2 1,2 1,092 1,071 1,05 1,092 1,008 0,987
1,1 1,001 0,98175 0,9625 0,94325 0,924 0,90475
1 0,91 0,8925 0,875 0,8575 0,84 0,8225
0,85 0,7735 0,7586 0,74375 0,7289 0,714 0,699
Table 5a: Snow redistribution model for multi-span roofs according to ASCE 7-10

The incomplete redistribution model has two load cases, balanced loads (Load Case 1) and
unbalanced loads (Load Case 2). The balanced load case 1 replicates the flat roof snow load. This
load case would be used to design structural parts at the ridges of the roof. The unbalanced load case
2 gives a local maximum for the gutter and also the maximum total load for the structure.

The important findings for load case 2 for multi-span roofs are, how a part of the precipitations is
redistributed from the ridges back down into the trough, another part is removed (overall drift loss).
17
ASCE 7-10, chapter 7.3
Generalized information taken from ASCE 7-10, Table 7.2 and 7.3
Classification of thermal conditions and terrain category / climate / roof exposure
Thermal 1,3 Structures intentionally kept below freezing
Factors Unheated open and open-air structures
1,2
Ct Structures kept just above freezing; ventilated roofs only if U < 0,23 W/m2K between heated
1,1
and ventilated space
1 Other structures
0,85 Continuously heated greenhouses (qi > 10°C; roof with U > 2,5 W/m2K; safety precautions)
Exposure 1,2 Warm / calm climate, sheltered roof 1)
Factors Moderate climate, sheltered roof 1)
1,1
Ce 1 Cold climate, sheltered roof 1); warm / calm and moderate climate, partially exposed roof
0,9 Warm / calm + moderate climate, fully exposed roof 2), cold climate, partially exposed roof
0,8 Cold climate, fully exposed roof 2); very cold climate / terrain 3), partially exposed roof
0,7 Very cold climate / terrain 3), fully exposed roof 2)
1)
Sheltered roof: by terrain, higher structures or trees present during the expected working life of the structure
2)
Fully exposed roof: no shelter on all sides, no parapets higher than the expected roof snow, no roof obstructions
3)
Very cold climate / terrain: Alaska, if site with no trees within 3 km radius; roof above the treeline in windswept
mountainous areas
Table 5b: Differentiation of roof snow loads according to ASCE 7-10

The unbalanced load case 2 is realistic for certain ratios of roof snow height to roof height and for
cold rough cladding. For high roofs with slippery cladding in combination with small snow loads or
a low thermal coefficient (warm roof) it is not realistic. A complete sliding is very likely for such
roofs. In this case the load in the valley would be too small.
The load at the ridges can also be smaller / larger than 0,5 pf.
Notes and Comments to Table 5:
1. The general roof-ground conversion factor ks = 0,7, defined as reference standard in the USA
cannot be compensated by the exposure factor maximum of Ce = 1,2 for sheltered roofs in warm
climate alone (0,7 · 1,2 = 0,84). To reach the ground snow level on a roof, thermal factors Ct > 1
are required too (on unheated buildings, open air structures and freezer buildings, ice-skating
halls). That means in reverse that large roof snow reductions are rather the result of melting than
of drift losses. The Eurocode (ks = 0,8 included in the shape coefficient, with 0,8 ≤ Ce ≤ 1,2 for
the exposure of the roof and with Ct = 1) has not covered this point. ISO 4355:2013 is also not
clear about it.
2. The largest ratio of roof to ground snow can be calculated using Ce = 1,2 and Ct = 1,3 giving 0,7
· 1,2 · 1,3 = 1,092 in load case 1. This is only 9% above the ground snow level and this gain can
be explained due to condensation + refreezing / resublimation. However, 18% gain in load case 2
raises questions. Total roof snow loads in excess of the ground snow load as for Ct = 1,3 are only
possible locally and at the roof edges (first valley, leeward side). For the global loads of large
roofs, the ground snow load is the limit.
3. It is doubtful, whether the exposure factor Ce can have a large influence on the reduced snow
loads on greenhouses, because the presence of roof snow during rapid melting is very short.
However, if Ct is kept constant at Ct = 0,85 and if lower Ce correlate to larger ground snow loads,
the sk-dependence of Ct is covered by this regulation. In contrast, ISO 4355, Annex D gives a
thermal coefficient with a dependence on sk. That Ct is calibrated with Ce = 1 and µ = 0,8.
4. For unbalanced loads, the limited variation of roof snow loads for different Ce is a result of the
drift limitation for snow in the valley. This should also be applied to balanced loads.
18
5. Because the model works with constant ratios of roof to ground snow load and does not check
whether the trough can accommodate the snow, it fails for very large ground snow loads in
comparison to the trough size in the same way as the Eurocode-models.

4. General redistribution model for multi-span roofs - Proposal


A proposal for a general redistribution model for multi-span roofs is given in Table 7. Load case (i)
is based on the bi-linear roof angle function for the basic shape coefficient µ2,b. In load case (ii) the
shape coefficient µ2 is used for the ridge and µ3 is used for the gutter. ISO 4355:2013 shows this snow
load distribution in Figure B.4. Because of the shortcomings of Figure B.4 as discussed in chapter
3.2, the geometry and snow load distribution for the roof trough is given in Figure 6.

Figure 6: Geometry and unbalanced snow load distribution load case (ii) in roof troughs

For the internal troughed roof area a special exposure coefficient Ce,T is defined. As for flat roofs the
special exposure coefficient for troughs depends on the roof size. In this case this can be defined by
the number of troughs per roof. As in other cases the reference value for a = 0° (flat roof) is 0,8.
Ce,T = 1,25 - n · (1,25 - Ce)
Where Ce is the exposure coefficient for the roof location
n is the drift loss part from the trough (Flat roof reference value: 1 - 0,8 Ce)
19
From small roofs with one or two troughs drift losses are possible, as the measurement in Ottawa
1971 has shown with Cb · Ce,T = 0,75 (75% of the ground snow load), see Figure 1. For a flat roof at
the same location on the airport in Ottawa, Canada, according to NBCC, the flat roof snow load level
would have been Cb · Cw = 0,8 · 0,75 = 0,6 (60% of the ground snow load).

In ASCE 7-10 it is recommended to allow half of the flat roof drift losses from the ridges. The other
half is distributed back down into the trough. The drift loss part would be: n = ½. Using the special
exposure coefficient Ce,T, the ratio of roof to ground snow load would be 0,8 Ce,T = 1 - ½ (1 - 0,8 Ce).
For normal exposure Ce = 1 the special exposure for the troughed area would be 0,8 Ce,T = 0,9. This
leads to 10% drift loss from a small roof trough in comparison to 20% from a small flat roof. Such a
percentage is already indicated in prEN 1991-1-3:2020.

For large roofs (flat roofs as reference: max Ce,F = 1,25) at the inner troughed roof area applies: 0,8
Ce,T = 0,8 · 1,25 = 1 (roof snow load = ground snow load). Special exposure coefficients for different
troughs in comparison to the exposure coefficient on flat roofs are shown in Table 6.

Special Exposure Coefficient Trough Multi-pitched Roof


Ce,T = 1,25 - n · (1,25 - Ce)
Location Exposure Coefficient Ce
Part n
Drift Loss Potential 1,25 1,2 1 0,8
Internal roof area, large no drift loss
roof, many troughs (0%) n=0 1,25 1,25 1,25 1,25
drift loss from
Small roof, single trough
the ridges (50%) n = 0,5 1,25 1,225 1,125 1,025
100% drift
Flat roof (Comparison)
losses n=1 1,25 1,2 1 0,8
Table 6: Special exposure coefficient for troughed roof area of multi-pitched roofs

Note: Are drift losses as large as for a flat roof allowed as in ISO 4355:2013, the drift loss part is n
= 1 and the exposure coefficient becomes Ce,T = Ce. The model is consistent with other models in ISO
4355. However, this is not recommended for the Eurocode. It does also not agree well with the
measurements 1971 in Ottawa.

On the outer roof pitches on the windward and the leeward side of a multi-span roof, the same rules
as for other sloped roof surfaces apply. Sliding is possible. However, there are no saddle-roof drift
surcharges across multi-span roofs. The drifting is supposed to be included in Ce,T > Ce.

For a consistent model as shown in Table 7 for the unbalanced Load Case (ii) for large and small
troughs or roof areas and large and low snow loads, two snow load distribution cases need to be
separated, Case 1 and Case 2. Is the volume of the roof snow smaller than the trough volume, the
distribution of the snow within the trough is triangular (Case 1). Is the volume of the roof snow larger
than the trough volume, in addition to the triangular distribution in the full trough, there is a
rectangular distribution of the access snow on top of the trough (Case 2). Which case applies can
only be found out by checking.

For the check, in the first step it is assumed that the snow volume is smaller than the trough volume.
The snow height of the assumed triangular distribution d1 = (sk · Ct · Ce,T · 1,6 h / g)1/2 is calculated
and compared with the height of the trough h.

In Case 1 the assumed snow height d1 is smaller than the trough height h (h ≥ d1) or depth of the
trough. In this case the assumption about the triangular snow load distribution is correct and the
maximum snow height above the gutter is d = d1. Using the geometric relations between triangular
20
trough and triangular snow load distribution within the trough, in a second step the reduced width w
of the snow cover in the trough can be calculated with w = w1 + w2 = d (cot a1 + cot a2).

Load Case Balanced Load Arrangement


(i)
Shape Coefficient µ2 (a) for uniform distribution
Roof angle in ° Windward Inner troughed area Leeward
5° < a ≤ 30° µ2,b = 0,8 Ce µ2,b = 0,8 Ce
30° < a < 70° µ2,b = 0,8 Ce (70° - a1) / 40° µ2 = 0,8 Ce,T µ2,b = 0,8 Ce (70° - a2) / 40°
a ³ 70° µ2,b = 0 µ2,b = 0
Format Roof Snow Load: s = sk · Ct · µ2
Load Case Unbalanced Load Arrangement
(ii)
Wind- For the inner troughed area
Sliding and
leeward 1) Check: h ≥ / < d1 = (sk · Ct · Ce,T · 1,6 h / g)1/2
+
Drift µ2 = 0 h ≥ d1: Case 1 h < d1: Case 2
or
d = d1 d = h + Dd with:
µ2 = µ2,b Dd = 0,8 Ce · sk · Ct /g - h/2 · Ce/Ce,T
w = w1 + w2 w = w1 + w2 = W1 + W2
w = d (cot a1 + cot a2) w=W
smin = g · Dd
Above smin = 0
smin = 0,8 Ce · sk · Ct - h/2 · g · Ce/Ce,T
the ridge
µ2 = 0 µ2 = 0,8 Ce (1 - g · h/(1,6 Ce,T · sk · Ct))
smax = g · d smax = g · (h + Dd)
Above
the smax = (1,6 h · g · sk · Ct · Ce,T)1/2 smax = 0,8 Ce · sk · Ct + g · h (1- ½ Ce/Ce,T)
gutter
µ3 = ((1,6 h · g · Ce,T) / (sk · Ct))1/2 µ3 = 0,8 Ce + g · h/(sk · Ct) · (1- ½ Ce/Ce,T)
The exposure coefficient in the troughed area is Ce,T = 1,25 - n · (1,25 - Ce)
Recommendation: Internal troughed area of large roofs: no losses: n = 0: Ce,T = 1,25
Small roof, one trough: Losses from the ridges only: n = 0,5; Ce,T = 1,25 - ½ (1,25 - Ce)
Format Roof Snow Load: s = smin … smax or: s = sk · Ct · µi with i = 2 and 3
Ce exposure coefficient, with 0,8 ≤ Ce ≤ 1,2;
Ce,T special exposure coefficient for troughed roofs, with Ce,T ³ Ce;
n drift loss part from the trough (NDP; suggested is n = 0 (no drift loss) for large roofs with many troughs; otherwise n
= 0,5 (drift loss from the ridges, not from the lower part of the trough (50% of the area);
Ct thermal coefficient Ct > 1 for very cold roofs;
a1, a2 roof angles a in ° for the two roof surfaces of the trough i = 1 und i = 2;
h trough height (mean value, usually equal to the roof height) in m;
W total width of the trough in m (W = W1 + W2);
d maximum snow height in the trough in m;
w width of the snow cover in the trough in m (w = w1 + w2);
g equivalent snow weight density in the trough (NDP; suggested is g = 2 kN/m3 for cold roofs, for warm roofs rather g
= 3 kN/m3 to 3,5 kN/m3.)
1)
For outer roof surfaces in load case (ii) the snow load distribution should consider sliding.
Table 7: Proposal for a general snow load distribution for multi-span roofs – in the format
according to prEN 1991-1-3:2020
21
For very small snow loads in large troughs, the shape coefficient maximum in case 1 can become
larger than µ3 = 2, however together with a very small snow-covered width w in the trough. The
designer can decide how to distribute the concentrated snow across the structural element to be
calculated, e.g. a single glass plate. This is the designer’s responsibility and needs no further guidance.

In Case 2 the assumed snow height d1 is larger than h (h < d1). Then the assumption of a triangular
distribution is not correct. The distribution consists of a triangle with the height h and the width W of
the trough. Above ridge level there is the rectangle across the full trough width W with the unknown
height Dd above the ridge. In Case 2 in the second step the correct snow height d = h + Dd needs to
be calculated.

Remark to the Format:


Table 7 is large and seems complicated, because it contains three alternative calculation methods
for load case (ii). Shape coefficients µ2 and µ3, snow load formulae and most simple relations for the
local snow loads smin above the ridge and smax above the gutter can be calculated. It becomes clear,
that the calculation with simple relations for smin and smax is the easiest. The risk for printing mistakes
is smaller too.

Calibration and comparison:


The calibration of the proposed model can be seen in Table 8. The results are calculated for a thermal
coefficient of Ct = 1. For Ct = 1,2 the results would be larger by a factor of 1,2. This would not be
correct. However, this failure cannot be repaired within this model. The value is already much too
large for a flat roof. Thermal coefficients Ct > 1 have to be calibrated, see ASCE 7-10.

Furthermore, if the proposed model is used for glass roofs together with a thermal coefficient Ct <<
1, e.g. for greenhouses open to the public, the roof snow loads are short-term loads due to
precipitations of few hours to three days. Drift losses should not be taken into account (Ce = 1,2).
However, if the thermal coefficient Ct according to ISO 4355, Annex D is used, despite of that, Ce,T
= Ce = 1 can be applied. This thermal Coefficient Ct is developed on the basis of wind-corrected
precipitation data and calibrated to match a roof to ground snow conversion of ks = 0,8 (by factor,
shape or exposure coefficient), see Background report Snow EN 13031 Part I.

Table 8 shows, that the total snow loads above valleys or troughs, from where it is difficult to remove
snow, do not depend on the roof angle, except for very small roof angles in transition to the flat roof.
In this proposal the load cases (i) and (ii) are consistent, as it should be.

Not shown are results for the snow load distribution Case 2, because the total snow load depends
additionally on the height of the snow above the ridge level Dd in comparison to the trough height h.
For the part Dd instead of the exposure coefficient of the trough Ce,T, the exposure coefficient of the
flat roof surface Ce can be used. The larger this height Dd becomes, the more the total roof snow load
will tend towards the flat roof snow load.

For small snow loads and large roof troughs Case 1 is relevant, for large snow loads and small troughs
Case 2. Generally, the limit between Case 1 und 2 can be estimated for a ratio of though height to
snow load (h/sk)lim with the dimension of m3/kN:
(h/sk)lim = Ct · (1 - n · (1 - 0,8 Ce)) · 2/g
Where h is the maximum trough height (depth) in m
sk is the characteristic ground snow load in kN/m2
Ce is the exposure coefficient for the roof location with 0,8 ≤ Ce ≤ 1,25
Ct is the thermal coefficient
22
g is the equivalent (force) density of the snow in kN/m3
n is the drift loss part from the trough (flat roof reference value: n = 1; single
trough, small roof: n = 0,5; internal roof area, large roof, many troughs: n = 0)

For equivalent snow densities of g = 2 kN/m3 and standard conditions with Ct = Ce = 1 the limit
ratio is (h/sk)lim = 1 - 0,2 · n

Load Case (i): Balanced Load arrangement


Flat roof: s1,F = µ1,F · Ct · sk (Shape coeff. µ1,F = 0,8 · Ce) Exposure coefficient: Ce see below
Trough: s2 = µ2 · Ct · sk (Shape coefficient µ2 = 0,8 · Ce,T) Exposure coefficient trough: Ce,T = 1,25 - n · (1,25 - Ce)
Total Snow Load in the Trough: Ratio s2 / (sk · Ct) for Load Case 1 1)
Drift Loss Exposure Coefficient Ce
Part n 1,25 1,2 1 0,8
Internal roof area, large roof n=0 1 1 1 1
Single trough n = 0,5 1 0,98 0,9 0,82
Flat roof (Comparison) n=1 1 0,96 0,8 0,64

Load Case (ii): Unbalanced Load Arrangement


Case 1 (h ≥ d1) Case 2 (h < d1)
Local, above Ridge: s2 in kN/m2 s2 = 0 s2 = smin = g · Dd
Local, above Gutter: s3 in kN/m2 s3 = smax = g · d s3 = smax = g · (h + Dd)
Snow Height d in m d = d1 Dd = 0,8 Ce · sk · Ct / g - h/2 · Ce/Ce,T
d = (sk · Ct · Ce,T · 1,6 h / g)1/2 d = h + Dd
Width of the Snow Cover w in m w = d · 2 cot a = d · W/h w = W (Width of the Trough in m)
Total Snow Load in the Trough in S ST = smax/2 · w = d · g · w / 2 S ST = (smin+smax)/2 · W = g (Dd + h/2) · W
kN/m S ST = 0,8 Ce,T · sk · Ct · W S ST = (0,8 Ce · sk · Ct + g · h/2 · (1- Ce/Ce,T))· W

Total Snow Load in the Trough: Ratio (S ST / W) / (sk · Ct) for Load Case 1 1)
Drift Loss Exposure Coefficient Ce
Part n 1,25 1,2 1 0,8
n = 0 (Ce,T = 1,25) 1 1 1 1
n = 0,5 (Ce,T = 0,625 + Ce/2) 1 0,98 0,9 0,82

Classification of the thermal conditions and the Exposure of the Site / Winter Climate:
Thermal 1,2 Buildings, intentionally kept below zero (freezer buildings, ice skating arenas)
Coefficient
1 Other structures
Ct
<< 1 Greenhouses with transparent cladding and controlled heating (qi >> 5°C; cladding: Uo > 1 W/m2K;
further safety requirements see EN 13031-1:2019
Exposure 1,25 Ce,F: Flat roof – Limit for large roof size Lc ³ 400 m
Coefficient
1,2 sheltered site, e.g. Terrain category 4 in DIN EN 1991-1-4
Ce
1 normal, e.g. Terrain category 3 in DIN EN 1991-1-4
0,8 windswept (4,5 m/s), exposed site, e.g. Terrain category 0,1 oder 2 in DIN EN 1991-1-4
1)
In load case 2 the total snow load depends also on the ratio of snow height above the trough Dd and trough height h.
Table 8: Results for the total Snow Load – General Snow Load Distribution Model

The snow load distribution on the airport roof in Ottawa, Canada, during the Blizzard of the Century
1971 according to Figure 1 and Table 1 can be approximated with the proposed model. The Canadian
standard NBCC is taken into account with the basic shape coefficient Cb = 0,8 and the exposure
coefficient for a very windy and exposed airport site with Cw = 0,75. For a flat roof as a reference a
23
roof-to-ground snow load conversion of Cb · Cw = 0,8 · 0,75 = 0,6 would result, comparable to a
value of 0,64 according to the Eurocode with µ1 = 0,8 · Ce with Ce = 0,8. As ground snow load the
measured value of s = 3,35 kN/m2 (based on g = 2,943 kN/m3) is taken into account, because this
value is higher than the recently known characteristic ground snow load value for the Ottawa
Macdonald-Cartier Airport of sk,50 = 2,8 kN/m2.
Note: The lower value is based on a set of more recent measurement data from 1983 - 2011,
see Brooks, et.al. (2014). From this publication a characteristic flat roof snow load of s1,50 =
1,904 kN/m2 is known, giving a roof-to-ground snow load conversion of 0,68 > 0,6 according
to NBCC. The lower values with less drift loss might be due to a milder and wetter climate.
For the two exposed roof troughs with a small height of h = 1,83 m (a = 18,5°), a medium value for
the drift losses of n = 0,5 is assumed, giving an exposure coefficient for the trough of Ce,T = 1,25 - n
· (1,25 - Ce) = 1,25 - 0,5 · (1,25 - 0,75) = 1 with Ce = Cw = 0,75 for the small flat roof according to
NBCC.

In a first step a triangular snow load distribution is assumed. The maximum snow height of this
distribution would be d1 = (sk · Ct · Ce,T · 1,6 h / g)1/2 = (3,35 · 1 · 1 · 1,6 · 1,83 / 2,943)1/2 = 1,82563
m. This height is nearly as high as the trough h. The test gives: d1 = 1,8256 m ~ 1,83 m = h. The
distribution is correct and the snow load distribution Case 1 with d = d1 applies.
In a second step, the width w of the snow load distribution is calculated. In this case it is nearly equal
to the width of the trough w = d · W/h = 1,82563 · 11/1,83 = 10,9737 m ~ 11 m = W. With this
information the snow loads can be calculated as follows:

Balanced Load Case (i):


Windward / leeward: Ce = 0,75: s2,b = µ2,b · Ct · sk = 0,8 · 0,75 · 1 · 3,35 = 2,01 kN/m2
Inner trough surfaces: Ce,T = 1: s2,b = µ2,T · Ct · sk = 0,8 · 1 · 1 · 3,35 = 2,68 kN/m2

Unbalanced Load Case (ii):


Above the ridge: smin = 0 (µ2 = 0)
Above the gutter: smax = g · d = 2,943 · 1,83 = 5,3857 kN/m2 (µ3 = 1,60767)

The local value above the gutter smax = 5,39 kN/m2 meets the maximum measurement value of 5,428
kN/m2 in the first valley (< 1 % difference) with precision. The measurement value in the second
valley is lower. The measured snow loads on the ridges are not zero as in the drift model for the
unbalanced load case (ii). However, the model also has a balanced load case (i), which covers these
and higher local loads above the ridges in case of snowfall conditions with less wind.

The snow loads on the roof spans and on the roof in total are also met. These are 29,668 kN/m (80,5%
of the ground snow) on the first span, 27,938 kN/m (75,8% of the ground snow) on the second span
and 25,367 kN/m (68,8% of the ground snow) on the third span, resulting in 82,937 kN/m (75% of
the ground snow) in total. The model gives 25,866 kN/m2 (70% of the ground snow) on the first and
third spans due to Cw = 0,75 on the outer roof surfaces, but 29,62 kN/m2 (80% of the ground snow)
on the second span, resulting in 81,36 kN/m (73,6% of the ground snow) in total. In total, this is a
difference between measurements and model of less than 2%.

Only on the first leeward roof surface the measurement values show a larger drift than covered by the
proposed model: 22,034 kN/m (119,6% of the ground snow). To simulate this, a drift surcharge from
the windward roof surface would be required. However, no reliable proposal can be made for such a
surcharge, because the influence of the roof angle has not been investigated sufficiently. The drift
surcharge factor d with a maximum for roof angles of 30° or 45° derived from ill-suited measurements
on saddle roofs (Høibø, 1988) is not useful. In this case the roof angle is 18,4° only.
24
As a result of this drift surcharge, the measured snow load in the first trough is also larger: 33,894
kN/m (91,98% of the ground snow). However, it has to be kept in mind, that the more recent
evaluations imply less severe drift conditions for the Ottawa Airport (Cb · Cw = 0,8 · 0,85 = 0,68 >
0,6 = 0,8 · 0,75). Also, a more conservative choice of the drift loss part n < 0,5 can be made to increase
the snow load in the first valley to 92%. At the Blizzard of the Century 1971, despite of the strong
winds, the drift duration of three days for one single snowfall event was shorter than in usual winter
seasons. Using Cw = 0,85, a smaller drift loss part of n = 0,2 for a single snowfall event would give
Ce,T = 1,17 resulting in 34,49 kN/m (93,6% of the ground snow) as in the measurement.

For a comparison with the model according to ASCE 7-10 the reference value of 0,7 has to be taken
into account. Also, the total snow loads in the proposal according to the Eurocode cover all thermal
conditions with Ct = 1, whereas ASCE 7-10 differentiates them with values Ct = 1 to Ct = 1,2. The
Eurocode-Model can be placed in the middle of that range at about Ct = 1,1 according to ASCE 7-10.
Furthermore, according to ASCE 7-10, 60% of the drift losses from the ridges (50% of the area) are
redistributed back into the trough. Only 40% drift away from the roof. That refers to a drift loss part
of n = 0,2. This value seems arbitrarily chosen, not suited to be used on all roofs. Instead, the already
accepted approach according to prEN 1991-1-3:2020 is used with normal drift losses (1 - 0,8 Ce) from
the upper part (ridges) and with n = 0,5 for single troughs only. For the internal troughs of large roofs
n = 0 applies (no drift losses). However, the drift part n should be made a national choice (NDP) in
the same way as the drift factor d for saddle roofs.

Literature
Standards and References:
- EN 1991-1-3: 2010: Actions on structures – Part 1-3: General actions – Snow loads
- prEN 1991-1-3: 2020: Actions on structures – Part 1-3: General actions – Snow loads (Final draft)
- EN 13031-1:2019: Greenhouses – Design and Construction – Part 1: Commercial production
greenhouses
- Final Report of the Commission of the European Communities DG III-D3 (1998-3), Scientific
Support Activity in the Field of Structural Stability of Civil Engineering Works: Snow Loads.
- Final Report of the Commission of the European Communities DG III-D3 (1999-09), Scientific
Support Activity in the Field of Structural Stability of Civil Engineering Works: Snow Loads.
- Formichi, P. (2017): Report from CEN/TC250/SC1, Project Team SC1.T2 – EN 1991-1-3 Snow
Loads (M515), 40th Meeting June 2017 in Brussels.
- ISO 4355: 1981: Bases for design of structures – Determination of snow loads on roofs
- NS 3491-3: Prosjektering av konstruksjoner, Dimensjonerende laster, Del 3: Snolaster
- ISO 4355: 1998: Bases for design of structures – Determination of snow loads on roofs
- ISO 4355: 2013: Bases for design of structures – Determination of snow loads on roofs
- ASCE 7-10, 2010: Chapter 7: Snow Loads.
- NBCC: In: Ontario Regulation 88/19: Building Code filed May 2, 2019

Background information of general interest will be accessible for Download under


www.greenhousecodes.com. The following documents are currently available:
- Background Snow EN 13031 Part I.pdf: Melting
- Background Snow EN 13031 Part II.pdf: Sliding and Drift
- Background Snow EN 13031 Part II-1.pdf: Snow Load Distribution – Duo-pitched Roof
- Background Snow EN 13031 Part II-2.pdf: Snow Load Distribution – Multi-pitched Roof
- Background Snow EN 13031 Part II-3.pdf: Snow Load Distribution – Arched Roof
25
- Comments prEN 1991-1-3:2020 Part I.pdf: General - Flat Roofs
- Comments prEN 1991-1-3:2020 Part II.pdf: Pitched Roofs
- Comments prEN 1991-1-3:2020 Part IV.pdf: Local Effects – Obstructions – Parapet – Retention
Devices – Solar Panels on Flat Roofs
- Comments prEN 1991-1-3:2020 Part V.pdf: Arched Roofs
- Comments to ISO-4355-2013 Part I.pdf: The use of the thermal coefficient for large roofs
- Comments to ISO-4355-2013 Part II.pdf: Critical review of the limits of the thermal coefficient
- Comments to ISO-4355-2013 Part III.pdf: Influence of the roof angle on the thermal coefficient
- Comments to ISO-4355-2013 Part IV.pdf: Drift / slide model multi-pitched roof - Corrigenda
- Comments to ISO-4355-2013 Part V.pdf: Drift / slide model for duo-pitched roofs - Comment
- Extreme snowdrift Multi-span Roof Ottawa 1971.pdf
- Comparison of Formulae Heat Conductivity and Density Snow.pdf
- Roof Angle Function for Thermal Coefficient.pdf
Citations from these documents are not marked as such, because of the same authorship.
Literature: References and sources

- Apeland, K. (2000): Global and European Standardization, ISO and CEN. In: Snow Engineering,
Recent Advances, Proceedings of the 4rd International Conference on Snow Engineering,
Trondheim, Norway, pp. 157-160.
- Brooks, A. J., Gamble, S., Dale, J., Gibbons, M., Williams, R., Davies & Irwin Inc. (RWDI),
2014: Determining Snow Loads on Buildings with Solar Arrays. In: CSCE 2014, 14th
International Structural Specialty Conference, Halifax, NS, Canada, pp. 1-10.
- Chiba,T., Tomabechi, T, Takahashi, T. (2012): Study on evaluation of snow load considering roof
snow-slide on gable roofs. In: 7th International Conference on Snow Engineering, Fukui, Japan,
pp. 231-241.
- DaGaetano, A.T., O`Rourke, M.J. (2003): A Climatological Measure of Extreme Snowdrift
Loading on Building Roofs. In: Journal of Applied Meteorology, Volume 43, pp. 134-144.
- Delpech, P., Guilhot, J. (2008): Quantitative assessment of snow load on complex and extended
roof shapes. In: 6th International Conference on Snow Engineering, Whistler, Canada, p.
- Dufresne de Virel, M., Delpech, P, Sacre, C (2000): Wind tunnel investigations of snow loads on
buildings. In: Snow Engineering, Recent Advances, Proceedings of the 4th International
Conference on Snow Engineering, Trodheim, Norway, pp. 171-178.
- Gandemer, J., Palier, P., Boisson-Kouznetzoff, S. (1997): Snow simulation within the closed
space of the Jules Verne Climatic Wind Tunnel. In: Snow Engineering, Recent Advances,
Proceedings of the 3rd International Conference on Snow Engineering, Sendai, Japan, pp. 347-
352.
- Grammou, N. (2015): Ermittlung von geometrischen Formbeiwerten für Flachdächer mit
aufgeständerten Photovoltaikanlagen anhand eines physikalischen Analogiemodells. Dissertation
TU Darmstadt, Germany.
- Grammou, N., Pertermann, I., Puthli, R. (2019): Snow loads on flat roofs with elevated solar panel
arrays – Research results on wind-induced shape coefficients. In: Steel Construction 12, No. 4,
pp. 364-371.
- Høibø, H. (1988): Snow Load on Gable roofs - Results from Snow Load Measurements on Farm
Building in Norway. 1st International Conference on Snow Engineering, Santa Barbara,
California, USA, CRREL Special Report No. 89-6, pp. 95-104.
26
- Irwin, P.A., Gamble, S.L., Taylor, A.D. (1995): Effects of roof size, heat transfer and climate on
snow loads. Studies for the 1995 NBC, Canadian Journal of Civil Engineering, 22, 9, pp. 770-
784.
- Isyumov, N., Mikitiuk, M. (2008): Sliding Snow and Ice from Sloped Building Surfaces – Its
Prediction, Potential Hazards and Mitigation. In: 4th International Conference on Snow
Engineering, Whistler, Canada 2008.
- Ito, T., Tomabechi, T., Mihashi, H. (1997): Characteristics of roofing materials as related to
adfreezing of snow. In: Snow Engineering, Recent Advances, Proceedings of the 3rd International
Conference on Snow Engineering, Sendai, Japan, pp. 209-214.
- Jelle, B.P. (2012): The Challenge of Removing Snow Downfall on Photovoltaic Solar Cell Roofs
in order to Maximize Solar Energy Efficiency – Research Opportunities for the Future.
Department of Materials and Structures, SINTEF Building and Infrastructures, Trondheim,
Norway.
- Li, L, Pomeroy, J.W. (1997): Estimates of Threshold Wind Speeds for Snow Transport Using
Meteorological Data. In: Journal of Applied Meteorology, Volume 36, March 1997, pp. 205-213.
- Lutes, D.A. (1970): Snow loads for the design of roofs in Canada. Proceedings of the Western
Snow Conference, Victoria, British Columbia, Canada, pp. 61-67.
- Meløysund, V., Lisø, K.R., Hygen, H.O., Høiseth, K.V, Leira, B. (2006): Effect of wind exposure
on roof snow loads. In: ScienceDirect, Building and Environment 42 (2007) pp. 3726-3736.
- Meløysund, V. (2010): Prediction of local snow loads on roofs. Doctoral Thesis, Norwegian
University of Science and Technology, Trondheim, Norway.
- Mihashi, H., Takahashi, T. (1996): Japanese recommendations 1993 for snow loads on buildings.
In: 3th International Conference on Snow Engineering, Sendhai, Japan 1996.
- O`Rourke, M., Koch, P., Redfield, R: (1983): Analysis of roof snow load case studies: Uniform
loads. Cold Regions Research & Engineering Laboratory, CRREL Report 83-1.
- O´Rourke, M, Auren, M. (1997): Unbalanced snow loads on gable roofs. In: Snow Engineering,
Recent Advances, Proceedings of the 3rd International Conference on Snow Engineering, Sendai,
Japan, pp.201-208.
- Otsuka, K., Homma, Y. (1992): Removal of snow from membrane structures. In: Proceedings of
the 2nd International Conference on Snow Engineering Santa Barbara, USA, pp. 263-274.
- Pomeroy, J.W., Brun, E. (2012): Physical Properties of Snow. Available online:
http://www.inscc.utah.edu. Downloaded 2013-03-24.
- Sakurai, S., Abe, O., Joh, O. (2012): Practical method for predicting snow accumulations on roofs
based on wind pressure characteristics. In: 7th International Conference on Snow Engineering,
Fukui, Japan, pp. 463-475.
- Sandvik, R. (1988): Calculation of Maximum Snow Load on Roofs with High Thermal
Transmittance. Proceedings of the first International Conference on Snow Engineering, Cold
Regions Research & Engineering Laboratory (CRREL), Report 89-6, pp. 317-324.
- Sandvik, R., Apeland, K. (1996): A comparison of the new ISO 4355 with CEN ENV 1991-2-3.
IABSE report 74, 1996, pp. 101-108.
- Tabler, R.D. (2003): Controlling Blowing and Drifting Snow with Snow Fences and Road Design.
Final Report, National Cooperative Highway Research Program Transportation Research Board
of the National Academies.
27
- Takeuchi, M. (1980): Vertical profile and horizontal increase of drift-snow transport. In: Journal
of Glaciology, Vol. 26, No. 94, pp. 481-492.
- Taylor, D.A. (1979): A survey of snow loads on roofs of arena-type buildings in Canada. In:
Canadian Journal of Civil Engineering, Vol. 6, No. 1, March 1979, pp. 85-96.
- Taylor, D.A. (1980): Roof snow loads in Canada. In: Canadian Journal of Civil Engineering, Vol.
7, No. 1, pp. 1-18.
- Taylor, D.A. (1983): Sliding snow on sloping roofs. Canadian Building Digest, National Research
Council of Canada.
- Terasaki, H, Fukuhara, T. (2016): Roof snow slide-off experiments using membrane deformation.
In: 8th International Conference on Snow Engineering, Nantes, France, pp. 330-333.
- Thiis, T.K., O`Rourke, M. (2012): A model for the distribution of snow load on gable roofs. In:
7th International Conference on Snow Engineering, Fukii, Japan, pp. 260-271.
- Tominaga, Y. (2017): Computational fluid dynamics simulation of snowdrift around buildings:
Past achievements and future perspectives. Cold Regions Science and Technology,
http://dx.doi.org/10.1016/j.coldregions.2017.05.004, pp. 1-13.
- Tomabechi, T., Ito, T., Takakura, M. (1997): Snow Sliding on a Membrane Roof. In: Snow
Engineering, Recent Advances, Proceedings of the 3rd International Conference on Snow
Engineering, Sendai, Japan, pp. 341-344.
- Tominaga, Y., Okaze, T., Mochida, A. (2016): CFD simulation of drift loads for an isolated gable-
roof building. In: In: 8th International Conference on Snow Engineering, Nantes, France, pp. 214-
220.
- Tory, J.R. (1992): The Codification of European Snow Loading. In: 2nd International Conference
on Snow Engineering, Santa Barbara, California, pp. 453-464.
- Yan, K., Zhou, X., Kasperski, M. (2010): Wind drifted snow load simulation on gable roof. 5th
International Symposium on Computational Wind Engineering (CWE 2010), Chapel Hill, USA.
- Zhou, X. (2013): Simulation method of sliding snow load on roofs and its application in some
representative regions of China. In: Natural Hazards 2013, 67, pp. 295-320.
- Zhou, X., Li, J., Huang, P, Gu, M., Sun, L. (2016): A new method of predicting slide snow load
for sloped roofs. In: 8th International Conference on Snow Engineering, Nantes, France, pp. 197-
183.

You might also like