Professional Documents
Culture Documents
Autors:
Dr.-Ing. I. Pertermann, IB Puthli, Schüttorf
Prof. Dr.-Ing. R. Puthli, KIT, Karlsruhe Date: 19.06.2020
2
Table of content
Symbols ………………………………………………………………….………….. 3
2. Parallel rows of PV or solar panels according to 7.5.2 (3) and (4) …………..... 5
Symbols:
Ce exposure coefficient
Ct thermal coefficient
The draft versions of the future Eurocode, e.g. prEN 1991-1-3:2020, indicate a shift of the exposure
coefficient Ce 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°
In this Part IV of the series Comments prEN 1991-1-3:2020, local effects due to drifting are
discussed as well as closely related drifts considered in other parts of the standard. Large parts of
these new regulations are just based on the old EN regulations. By including the exposure coefficient
Ce into the local shape coefficients and their limitations, systematic errors have been made.
Instead in the limits for the drift potential, where Ce could belong into the denominator, a factor of
0,8/Ce has been placed into geometrical limitations, such as the obstruction height, which cannot be
changed by drifting. This error appears everywhere, but not in the regulation for PV and solar panels
on flat roofs according 7.5.2 (3) and (4). There it is correct.
The systematic error by 0,8/Ce could also be an attempt, to calibrate (reduce) the drift surcharges
requiring windy, exposed conditions for Ce = 1,2 and Ce = 1. However, this should be done to “soft”
drift related limits, but not to “hard” geometrical limits. This version is assumed in the proposals.
Also, the thermal coefficient Ct is missing everywhere.
Comparative numerical evaluations over the whole range of relevant parameters show the
inconsistencies. However, many of these faulty regulations are not normative, because the verbs
“should”, “may” and “can” have been used. According to prEN 1991-1-3:2020, paragraph 0.3 this
means:
“The verb “should” expresses a highly recommended choice or course of action. Subject to national
regulation and / or any relevant contractual provisions, alternative approaches could be used /
adopted where technically justified.”
This allows the technically justified repair of faulty regulations via NCI or other means. Faulty limits
are anyway subject to a national choice (NDP). This comment Part IV gives background information
for a qualified national choice to 7.5.2 (3) and (4), 7.5.3 (4), 8.2 (1) and (2) and 8.3 (1) and (2) of the
future EN 1991-1-3. The corrected formulae and limits with the required course of action are
summarised in light blue field (boxes).
5
Paragraph (3) defines a global shape coefficient µp for the panel covered area of the flat roof
including large drift lengths of ls = 4 h / Ce (Formula (7.6)) on all four sides:
Paragraph (4) refers to clauses 8.3 (w < 2h) and 8.4 (w ³ 2h) for local effects. Meant are probably
8.2 (for obstructions) and 8.3 (for parapets).
In Figure 1 the roof snow load level sp/sk for the panel area is shown for different ground snow heights
hs,0 of 0,5 times to 2 times the panel height h. Depending on the roof exposure (Ce) the roof snow
loads decrease from the ground snow load level sp/sk = 1 to the flat roof snow load level sp/sk = µ1.
Figure 1: Roof snow load level for different ratios of ground snow height to panel height and
different roof exposures (with Ct = 1) according to prEN 1991-1-3:2020, 7.5.2 (3)
This regulation is not based on research (Brooks et.al. (2014), Grammou (2015)), but on engineering
judgement, see Fingerloos & Schwind (2019). However, the regulation in the German NA applies
to Ce = 1 and Ct = 1 only. It is limited to panel heights of 0,5 m. For higher panels the loads increase
by 10% to cover drifting and do not remain on the ground snow load level as a maximum.
In prEN 1991-1-3:2020 the exposure coefficient Ce has been introduced correctly, but the influence
of the thermal coefficient Ct is not correct. This can be shown for the case µ1 < µp < 1, which refers
to the sloping part in Figure 1 between hs,0/h = 1 and hs,0/h = 1 / (0,8 Ce). In this case the roof snow
height is limited by the height of the panels. Above the panels drifting is possible, below it is not. The
6
shape coefficient would be µp = g · h / sk with the roof snow load sp = µp · Ct · sk = Ct · g · h. The roof
snow height would be hs = sp / g = Ct · h. This is not correct; the roof snow height should be hs = h.
Roof snow loads are calculated using Formula (7.1) with si = µ1 · Ct · sk. The roof height is calculated
by hs = si / g using the equivalent weight density g of the roof snow layer. The roof snow height
depends on Ct and would increase or decrease in cases where the thermal coefficient is not Ct = 1.
Therefore, Ct must have an influence on the equations and the limits. The influence can be worked
out by considering the different cases for the ratio of roof snow height to panel height and the
transition between them.
The model assumes, that drift losses as on other flat roofs are possible, when the roof snow
height hs = si / g = µ1 · Ct · sk / g is larger than the panel height h by more than the drift loss
part. For example, for cold roofs with Ct = 1, Ce = 1 and µ1 = 0,8 Ce the drift loss is (1 - 0,8
Ce) · sk. This is the case for ratios of ground snow height hs,0 = sk / g to panel height h larger
than hs,0 / h > 1,25 = 1 / µ1, see Figure 1 on the right side. The shape coefficient would be µp
= µ1 as on a flat roof without panels and the roof snow load sp = µ1 · Ct · sk.
However, the limit of hs,0 / h > 1,25 does only apply as long as the thermal coefficient remains
Ct = 1. Otherwise the roof snow height hs = µ1 · Ct · sk / g = µ1 · Ct · hs,0 changes as well. The
panel height h does not change. The limit between case 1 and case 2 has to be adapted with Ct
· hs,0 / h > 1 / µ1 or hs,0 / h > 1 / (Ct · µ1).
Case 2: Roof snow height equal than panel height: hs = h: µ1 < µp < 1
In case 2 the drift losses are limited by the height of the panels. Roof snow above the panels
can drift; below the panels there are no drift losses. The total drift potential is (1 - 0,8 Ce) out
of the precipitations (or the ground snow load as a reference value).
Case 2 is situated between 1 and 3 within the limits 1 / Ct < hs,0 / h < 1 / (Ct · µ1). In the
example in Figure 1 with Ct = 1, Ce = 1 and µ1 = 0,8 Ce this is within the limits 1 < hs,0 / h <
1,25. Within these limits the roof snow load does not change, it remains at sp = g · h with the
roof snow including any thermal influences filling the space between the panels up to the top.
The roof snow height remains at hs = sp / g = h. From sp = g · h = mp · Ct · sk the shape
coefficient can be derived as µp = g · h / (Ct · sk).
In case 3 there is no drift loss anymore. The roof snow load is limited by the ground snow
load. The thermal influences at the bottom of the snow layer are considered independently
from drift losses. Therefore, in case 3 the shape coefficient is µp = 1 and the roof snow load
sp = Ct · sk. The roof snow height is hs = Ct · sk / g = Ct · hs,0. Case 3 is limited by Ct · hs,0 < h
or hs,0 / h < 1 / Ct.
Warning for the NDP: For Ct < 1 snow is melting at the bottom of the roof snow layer. For such
cases, the equivalent snow weight density of g = 2 kN/m2 could not be appropriate anymore, see
Model Snowmelt.pdf in www.greenhousecodes.com.
7
Including Ct correctly does not change the limitation for the shape coefficient itself with µ1 ≤ µP ≤ 1.
But the limits for the application change and the shape coefficient given in Formula (7.5) has to be
corrected as follows:
µP = g · h / (Ct · sk)
Everywhere else in the document, the thermal influence Ct is to be included in this way and the
limit g · h / sk is to be corrected by g · h / (Ct · sk).
From research on elevated open back PV-panels on a flat roof in the water channel it is known, that
large local accumulations (plateau up to µ = 1,57) occur only in connection with small deposition
areas, where the edges and parts of the panel covered area are free of snow (see Grammou et.al.
2019). The panels had large lateral distances of b >> 3 h and large free edges. Only for very small
snow heights in comparison to the panel height, µ = 2 is possible (with a triangular snow distribution
between the panels with a mean value µ = 1).
Brooks et.al. (2014) showed water flume model snow drift patterns for all wind directions. Closed
back panels were used, the panels covered the roof fully, no free edges. Both studies found that PV
surfaces facing downwind (or with the back sides toward wind from the North) created larger vortices
than wind from other directions. The presence of the panels does not change the formation of the
separation bubble at the windward edge of the building, same as with a flat roof. The panels interact
locally with the separation bubble and affect size and position of the snow accumulation. The drifts
form between the panels somewhat closer to the roof edge as on a flat roof.
With a thermodynamic FEA-program and hourly + daily meteorological data Brooks et.al. (2014)
calculated characteristic flat roof snow loads with and without panels. Depending on the winter
climate, flat roof snow loads in the range of 0,38 to 0,68 sk were increased by snowdrifts around the
panels by 15% to 45%, remaining well below the ground snow level. Local shape coefficients were
between 0,7 and 1,12. Suggestions for design were not given, because many more investigations
should be done, and questions answered before any design regulations are created.
However, local shape coefficients of µ = 4 have not been encountered in both studies.
Both studies agreed with the following conclusion:
For the roof as a whole, the ground snow load limit is never exceeded.
This is correct in prEN 1991-1-3:2020, but not correct in DIN EN 1991-21-3/NA:2019.
8
Paragraph (1) gives the shape coefficient µ6 (Formula (8.1)), paragraph (2) the limits for the shape
coefficient and paragraph (3) the drift length ls as follows:
µ6 = 0,8 / Ce · g · h / sk with 0,8 ≤ µ6 ≤ 2 and ls = 2 h / Ce ≤ 15 m
No background for the drifting around such obstacles shall be given here (research results are
available worldwide), in the following only the formula and their limits are checked for consistency.
Figure 2: Roof snow load level at max µ6 for different ratios of ground snow height to obstruction
height and different roof exposures (with Ct = 1) according to prEN 1991-1-3:2020, 8.2
As can be seen in Figure 2, the shape coefficient µ6 does not meet the shape coefficients for the roof
snow loads. The regulation is not consistent, as explained in detail in the following:
The consistency of the proposed corrections of formula and limits is shown in Figure 3. The diagram
has the same lower limit for large snow loads as in Figure 1 for PV-panels on flat roofs, the roof
snow load. Only the upper limit with the drift potential is different here.
Figure 3: Consistency of the proposed correction: Roof snow load level at max µ6 for different
ratios of ground snow height to obstruction height and different roof exposures (with Ct = 1)
10
In comparison to the local obstructions according to 8.2, a redistributed potential is given depending
on the ratio b/hp. Also, the limiting drift potential is higher with µ7,max = 4 instead of 2.
Paragraph (1) gives the shape coefficient µ7 (Formula (8.3)) and the limits for the shape coefficient
and paragraph (2) the drift length ls as follows:
µ7 = 0,8 / Ce · 2 b / ls with µ7 ≤ min (0,8 / Ce · g · hp / sk; µ7,max) with µ7,max = 4 (NDP)
and ls = 4 hp / Ce with 5 m ≤ ls ≤ 15 m
However, in the attempt to include exposure coefficient Ce and thermal coefficient Ct into the
calculation, systematic errors have been made, as it is shown in Figure 4.
Figure 4: Roof snow load level at max µ7 for different ratios of ground snow height to parapet
height and different roof exposures (with Ct = 1) according to prEN 1991-1-3:2020, 8.3
For this drift model despite many limits, the lower limit for the shape coefficient µ7 is missing, see
Figure 4. A lower limit is required for the transition to the roof snow height, if the snow height is
larger than the parapet height, the same as in clause 7.5.2 (3) and Figure 1 for PV and solar panels
on flat roofs and in Figure 2 for local obstructions.
However, in this case not only the flat roof snow load with µ1, but the pitched roof snow load with
µ2, possibly on a leeward roof side are the limits.
11
The shape coefficient µ7 = 0,8 / Ce · 2 b / ls is the result of a triangular redistribution of the roof snow
potential of 0,8/Ce · Ct · sk from the roof length b to the drift length ls: 0,8/Ce · Ct · sk · b = µ7 · Ct · sk
· ls /2. The drift length ls has to be smaller than b. Otherwise ls = b applies. At the end of the drift
length ls the roof is snow-free after redistribution. The resulting shape coefficient µ7 = 0,8 / Ce · 2 b /
ls = 1,6 b /(Ce · ls) = 0,4 b / hp increases with the ratio b / hp until the limit value µ7,max = 4 is reached.
This is the case for ratios b / hp = 10. This ratio is known from the research on snow fences by Tabler
(2003), see also Part II of the Background Report EN 13031 - Sliding and Drift.pdf. The ratio b
/ hp = 10 refers to the maximum length of the windward drift in the equilibrium state for solid fences
with zero porosity and unlimited fetch distance for snow drifting on the ground. These values have
been established for North America. They should at least be questioned for Europe.
The upper limit “4” is also relevant for small snow loads with snow heights much smaller than the
parapet height. It represents the drift potential, which could be a bit larger for sliding snow on the
sloped roof.
The upper limit should have the form “4 / Ce” and is a National choice (NDP), same as in 8.2. Here
the calibration with 0,8/Ce would be appropriate, e.g. by µ7 ≤ 4 · 0,8/Ce = 3,2 / Ce.
After the adaptation of the limits (range) it becomes clear by numerical comparison, that for higher
snow loads Formula (8.3) is not consistent with the roof snow loads, see also Figure 4.
If the shape coefficient µ7 is limited to the fixed obstruction height h, the factor 0,8 / Ce is not correct.
Furthermore, the thermal coefficient Ct needs to be introduced into this limitation. As in Formula
(8.1) it must be: µ7 ≤ g · hp / (Ct · sk). The consistency of the proposed correction shows Figure 5.
Figure 5: Consistency of the proposed correction: Roof snow load level at max µ7 for different
ratios of ground snow height to parapet height and different roof exposures (with Ct = 1)
12
For very small ratios of ground snow height to parapet height, the drift can get as large as the potential
with 3,2/Ce as limit (NDP). With tis limit the ratio b/hp = 10 also changes for Ce = 1 to b/hp = 8 and
for Ce = 1,2 to b/hp = 6,67. For larger snow heights the parapet is the limit. As soon as the snow height
becomes larger than the parapet, the roof snow loads are to be used.
The old numerical limit for the drift length of 5 m is not justified, when the shape coefficients are
calculated with respect to the aspect ratio, snow heights, parapet height and drift potential. The upper
limit can remain for practical reasons until it is verified using available literature.
Remove the old limit “5 m” in 8.3 (2), Formula (8.4) and correct: “The drift length ls in Figure 8.2
should be determined from: ls = 4 hp / Ce ≤ 15 m.
Figure 6: Roof snow load level at max µ2,p for different ratios of ground snow height to retention
device height and different roof exposures (with Ct = 1) according to prEN 1991-1-3:2020, 7.5.3 (4)
The new regulation is meant for roof snow heights hs larger than the height of the retention device hp
to be reduced to the height hp to avoid full sliding. However, the shape coefficient µ2,p is not applicable
for roof snow heights smaller than the device. In this case µ2 applies. A very important limitation is
missing here.
13
Assuming that the local effects on the retention device are covered by chapter 8, the roof snow load
itself can be used. The limiting condition for the use of µ2,p is hs = µ2 · Ct · sk / g > hp or µ2 > µ2,p.
Because the shape coefficient µ2,p is limited to the fixed height hp of the retention device, the factor
0,8 / Ce is not correct. Because the roof snow height is hs = sp / g = 0,8 / Ce · hp, only for Ce = 0,8 the
roof snow height hs would be equal to the height of the retention device hp. For Ce = 1 the roof snow
load would be only 0,8 of the height of the retention device, for Ce = 1,2 only 0,667 of hp. These
reductions cannot be explained.
Furthermore, the way the paragraph is formulated, invites to misinterpretation. The sliding off has
nothing to do with the values in Table 7.2 and 7.3. It concerns the unbalance load arrangement
according to 7.5.3 (2) and Figure 7.3 with a value of zero at the windward side of the roof. This
introduces a large asymmetry. Snow retention devices can limit the asymmetry if designed properly.
For a correction in Paragraph 7.5.3 (4), text and NOTE should be adapted as follows:
“The unbalanced snow load arrangement in Figure 7.3 applies when snow is not prevented from
sliding off the roof. For the roof pitch with a retention device such as snow fences or a parapet at the
lower edge, the snow load is not reduced in cases with µ2 · Ct · sk / g ≤ hp (or µ2 ≤ µ2,p).
NOTE: For cases with µ2 > µ2,p the shape coefficient µ2,p = g · hp / (Ct · sk) can be used, where g and
hp are defined in 8.3 (1), unless the National Annex gives a different value for use in a country.”
The consistency of the proposed correction is shown in Figure 7. For very small snow loads with
roof snow heights below the retention device, the shape coefficient µ2 is used. As soon as the roof
snow loads get larger, the height of the retention device becomes the limit. The shape coefficient for
the limit µ2,p does not depend on the roof exposure. The thermal coefficient would have an influence,
but in Figure 7 only Ct = 1 is shown.
Figure 7: Consistency of the proposed correction: Roof snow load level at max µ2,p for different
ratios of ground snow height to height of the retention device and different roof exposures Ce
14
Literature
General Standards and References:
- EN 1991-1-3: 2003 + A1: 2015: Actions on structures – Part 1-3: General actions – Snow loads
- prEN 1991-1-3: 2018: 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)
- DIN EN 1991-1-3/NA: 2019: Eurocode 1 Einwirkungen auf Tragwerke – Teil 1-3: Allgemeine
Einwirkungen – Schneelasten; Nationaler Anhang (in German).
- 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.
- Formichi, P. (2019): Report from CEN/TC250/SC1, Project Team SC1.T2 – Second Intermediate
Draft, 43th Meeting February 2019 in Paris.
- ISO 4355: 1981: Bases for design of structures – Determination of snow loads on roofs
- 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
- AIJ Japan Building Code, Commentary on Recommendations for Loads on Buildings. Chapter 5
Snow Loads, pp. C5-1-C5-28.
- JIS Japanese Industrial Standard C 8955: 2011: Design guide on Structures for Photovoltaic Array
- NGMA 2004, National Greenhouse Manufacturers Association, Structural Design Manual:
Structural Loads: Chapter 6 and Annex C.6: Snow loads.
- ASCE 7-10, 2010: Chapter 7: Snow Loads.
- Brooks, A.J., Gamble, S., Dale, J., Gibbons, M., Williams, R., Davies & Irwing 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.
- Fingerloos, F., Schwind, W. (2019): Zur Neuausgabe des Nationalen Anhangs DIN EN 1991-1-
3/NA „Schneelasten“ in 2019-04. In: Bautechnik 96 (2019), Heft 4, pp. 352-359.
- 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 mounted solar
arrays – research results on wind-induced shape coefficients. In: Steel Construction – Design &
Research 12 (2019), No. 4, pp. 364-371.
- 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.
Snow Load Background information of general interest will be accessible for Download under
www.greenhousecodes.com. The following documents are currently available:
15