You are on page 1of 31

Background Report

to EN 13031-1: 2019, Annex C


Snow Loads on Greenhouse Roofs
Part I: Melting

Photo VDH: Rests of snow on plastic film near gutter end after heavy snowfall in the Alps

Autors:
Dr.-Ing. I. Pertermann, IB Puthli
Prof. Dr.-Ing. R. Puthli, KIT Date: 15.05.2020
2

Table of content
Symbols ……………………………………………………………………………... 3

1. Basic principles of EN 1991-1-3 …………………………………………………. 6

Part I: Melting

2. Greenhouse related requirements for controlled heating …………………….. 8


2.1 Transparency ………………………………………………………………………….. 8
2.2 Heat permeable cladding ……………………………………………………………… 8
2.3 Heating (internal air temperature under the roof) ………………………………….. 9
2.4 Unrestricted melt water outflow ……………………………………………………… 9
2.5 Additional safety measures ……………………………………………………………. 9

3. Thermal coefficient (influence of melting) ……………………………………... 9


3.1 Thermal coefficient according to EN 1991-1-3 ………………………………………. 9
3.2 Thermal coefficients Ct according to ISO 4355 ……………………………………... 10
3.3 Background of ISO 4355, Annex D ………………………………………………….. 10
3.3.1 Melt model …………………………………………………………………… 10
3.3.2 Heat transfer inside …………………………………………………………... 12
3.3.3 Heat and mass transfer outside ………………………………………………..15
3.3.4 Evaluation and Results ……………………………………………………….18

3.4 Use of Ct-formula for greenhouses ………………………………………………….. 20


3.4.1 Influence of the Roof angle ………………………………………………….. 20
3.4.2 Temperature range and heating ………………………………………………. 22
3.4.3 Limits for the heat transfer coefficient ……………………………………….. 23
3.4.4 Climatic limitations ……………………………………………………....….. 24
3.4.4.1 Air temperature outside (lower limit) ……………………………… 24
3.4.4.2 Snow load (upper limit) ……………………………………….…… 24
3.4.4.3 Snow load (lower limit) ……………………………………...….…. 25
3.4.5 Overview: Snow load limits and Ct-formula ……………………….….…..… 26

Literature to Part I ……………………………………………………………..… 28


3

Symbols
Abbreviations:

Luv windward side of the structure

Lee leeward side of the structure

SWE snow water equivalent

Symbols:

NOTE The following symbols are based on EN 1990 and EN 1991.

Latin upper-case letters:

Cesl exceptional snow load coefficient

Ce exposure coefficient

Cm surface material coefficient

Ct thermal coefficient

Lf is the latent heat of fusion of snow in kJ/kg

Mr(Dti) melt rate in kg/(m2·h) for the time interval Dti in h

Pr(Dti) precipitation rate in kg/(m2·h) for the time interval Dti in h

Qmelt energy flux density (rate) in W/m2 available for melting of snow at the roof surface (melt layer)

Qo energy flux density (rate) in W/m2 from the warm roof below to the melt layer

Qs,e energy flux density (rate) in W/m2 from the air outside via the snow to the melt layer

RT total thermal resistance of a component in m2K/W

Rsi internal surface resistance (surface to internal air) in m2K/W

Rsi,sw internal surface resistance for heat flow sideways in m2K/W

Rsi,up internal surface resistance for heat flow upwards in m2K/W

Rse external surface resistance (surface to external air) in m2K/W

Rl,j thermal resistance of the material layer j in m2K/W

Rf thermal resistance of the frame, e.g. cladding bars and gutter in m2K/W

Rg,k thermal resistance of the gas space k in m2K/W

Rts,m thermal resistance of the (thermal) screen m in m2K/W

U overall heat transmittance in W/(m2K)

Ug overall heat transmittance for glass in W/(m2K)


4
Uo special heat transmittance for snowmelt conditions in W/(m2K), heat transfer from the inner air
to the bottom of the snow layer

Use special heat transmittance for snowmelt conditions in W/(m2K), heat transfer from the bottom
of the snow layer to the air outside

Latin lower-case letters:

f(a) roof angle function for the thermal coefficient Ct

f(qi) influence of heating (internal air temperature) on the thermal coefficient Ct

f(Uo) influence of the thermal transmittance of the gladding on the thermal coefficient Ct

f(sk,n) influence of the snowfall rate (snow load) on the thermal coefficient Ct

g acceleration of gravity: g = 9,81 m/s2

hsf fresh snow height in mm

hsw wet snow height in mm

hw water content of a snow layer (liquid water height) in mm

hi ice content of a snow layer (pure ice height) in mm

he external surface heat transfer coefficient in W/(m2K), total he = hr,e + hc,e

hr,e external surface heat transfer coefficient for radiation in W/(m2K)

hc,e external surface heat transfer coefficient for convection in W/(m2K)

hro external surface heat transfer coefficient for Black-Body-radiation in W/(m2K)

hi internal surface heat transfer coefficient in W/(m2K), total: hi = hr,i + hc,i + hcd,i

hr,i internal surface heat transfer coefficient for radiation in W/(m2K)

hc,i internal surface heat transfer coefficient for convection in W/(m2K)

hcd,i internal surface heat transfer coefficient for condensation in W/(m2K)

hs,k heat transfer coefficient of the gas space k in W/(m2K)

k1 adjustment factor for snow cover

k2 adjustment factor for snow surface heat loss

s(ti) roof snow load in N/m2 at the time ti

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)

min s1,n,t minimum roof snow load with reference to the ground area in kN/m2

vm mean wind speed in m/s


5
Greek upper-case letters:

Dti time interval in h

DTi temperature difference in K between the internal air and the melt layer on the roof surface

DTs,e temperature difference in K between melt layer and external air

Greek lower-case letters:

a Angle of roof pitch in degrees, measured from the horizontal

lj thermal conductivity for a material layer j; (lj,sup - upper value; lj,inf - lower value) in W/mK

ls,eff effective thermal conductivity of a snow layer in W/mK

µi shape coefficient at the location i for roof snow load distributions

qi internal air temperature in °C

qe external air temperature in °C

qs,m melting temperature of snow in °C: triple point qs,m = 0,01°C

qcrit critical temperature in °C, lower limit for the melting of snow

rs,eff equivalent snow density of a snow layer in kg/m3

rsf fresh snow density in kg/m3

rsw wet snow density in kg/m3

rw density of water at melting point in kg/m3


6
1. Basic principles of EN 1991-1-3

EN 13031-1 and EN 13031-2 follow the basic principles for roof snow loads according to the
Eurocode, Part 1-3 (EN 1991-1-3), specifying only non-contradictory, complementary information
for greenhouses with special properties, which distinguish them from other buildings.

However, EN 13031-1 was reviewed and finally published, before the review of EN 1991-1-3 was
finished. Therefore, it is shown, that and how EN 13031 can be used with both generations of the
Eurocode.

For the roof snow loads, EN 1991-1-3:2003 formulates as a basic principle (P):

EN 1991-1-3:2003, clause 5.1(1) P and (2):


“The design shall recognise that snow can be deposited on a roof in many different patterns.”
“Properties of a roof or other factors causing different patterns are:
a) the shape of the roof
b) its thermal properties
c) the roughness of its surface
d) the amount of heat generated under the roof
e) the proximity of nearby buildings
f) the surrounding terrain
g) the local meteorological climate, in particular its windiness, temperature variations, and
the likelihood of precipitation (either as rain or snow).”

This is a large and ambitious list. What does the EN deliver?

For the calculation of roof snow loads EN 1991-1-3:2003 is based on the multiplicative format in
equation (1), opposite 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 (1)


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. prEN 1991-1-3:2020, indicate a shift of the exposure
coefficient from the equation (1) 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. Therefore, the format
according to equation (2) is also used in EN 13031-1:2020.

prEN 1991-1-3: 2019: Roof snow load: si,n,t = µi · Ct · sk,n (2)


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°

Also, the contents of clause 5.1(1) and (2) remain but are moved to clause 7.1 (1) including a NOTE.

Therefore, in 2020 the question remains: What does the EN deliver?


7

At least three of the influences of the above-mentioned properties a) to g) have not been included in
these formulae or in the three coefficients Ce, Ct and µi provided for the roof snow load:
a) Shape coefficients are provided for some, but only for a few standard roof types: µi
b) Heat transmittance of the roof (influence on melting: Ct = f(Uo)): not covered
c) Surface material coefficient (influence on sliding) Cm: not covered
d) Temperature and heat flux (influence on melting: Ct = f(qi)): not covered
e) Drift surcharges are covered for a few, not all roofs types by shape coefficients µi
f) Exposure coefficient Ce: National choice, otherwise Ce = 1: not sufficiently defined
g) Influence on exposure coefficient Ce: not covered
Rain on snow surcharge or ground snow loads including rain: National choice

The missing properties b), c) and d) and the very limited variation of f) and g) show, that EN 1991-
1-3:2003 and also prEN 1991-1-3:2020 do not provide enough information, especially for heat
permeable roofs with large heat fluxes due to heating inside (melting of snow) and for smooth or
slippery roof surfaces (sliding of roof snow). Both physical phenomena, the melting and the
immediate sliding of snow, are characteristic for greenhouse roofs.

Many of the specifications provided in EN 1991-1-3, especially those for exposure, drift and sliding
have been questioned, see e.g. Sandvik (1996), Schwind (2009), Meløysund (2010).

The draft versions of the future Eurocode, e.g. prEN 1991-1-3:2020, show no sign of basic changes
concerning snow melting. However, the reference to ISO 4355 in the existing norm was replaced by
a paragraph to the choice of the thermal coefficient Ct, based on ISO 4355.

prEN 1991-1-3:2020, 7.4 (2) NOTE 1:


“Locations where the duration of the snow load is long enough can be selected on the basis of the
characteristic ground snow load greater than a threshold value sk,min. sk,min = 1,5 kN/m2 unless the
National Annex gives a different value for use in a country.”

A characteristic snow load value of 1,5 kN/m2 is no limit for the application of a thermal coefficient
Ct. It is just the smallest characteristic snow load in Norway.
However, there are many other conditions and limits, also for the snow loads, see Comments to ISO-
4355-2013 Part I to III.pdf.

Paragraph 7.4 is no sufficient description of the conditions, application rules and safety measures
required for the application of Ct < 1 for greenhouse roofs. These rules in connection with a
“controlled heating operation” are specified in the reviewed EN 13031-1:2019 and will also be
found in the second part EN 13031-2 with the background being given in Part I of this report.

Therefore, Part I of the Background report has had few editorial corrections, but no changed content
in comparison to the previous version to the draft prEN 13031 from 2017-06-22.
8

Part I: Melting
2. Greenhouse related requirements for controlled heating

The reduction of roof snow loads due to melting (using Ct < 1) for greenhouses is based on very strict
rules. It must be ensured, that sufficient heating is provided when needed, despite possible problems
that may occur during the design working life of the structure. If a greenhouse is left to be out of
service during winter: Ct = 1 is to be assumed for the design; no reduction.

2.1. Transparency

In contrast to other buildings, greenhouses require the fast removal of snow from their mainly
transparent roofs, to allow the full solar radiation necessary for most of the plants cultivated or stored
here. Greenhouses shelter plants, not people. That is why the first and central requirement for Ct <
1 is the transparency of the roof cladding. The property itself is not defined, because transparency
does not contribute to the physical processes of snow melting. The transparency has to be high enough
to allow plants to grow, using the natural light outside.
A second advantage or side effect is, that snow and ice remain visible through the roof cladding, to
allow removal of any excess snow, or if this is not possible, to evacuate the area. The risk can be
located and assessed visually under a transparent roof. This is not possible for other roofs.

Figure 1: (Photo VDH) Light impairment in a greenhouse during heavy snowfall

2.2. Heat permeable cladding

The second condition for the use of Ct < 1 is given by the lower limit for the heat transmission of the
cladding. Here, values of Uo ≥ 1 W/(m2K) are generally accepted as sufficient, see ISO 4355, Annex
D, equation (D.2) and also EN 1991-1-3, clause 5.2(8), where U > 1 W/(m2K) is given.
9
In case of doubt, e.g. if the cladding is advertised as a heat protective cladding, the heat transmission
requirement of Uo ≥ 1 W/(m2K) must be checked by a calculation or test, see ISO 4355, EN 673, EN
674, EN 675, EN ISO 6946, EN ISO 10077-1 and EN ISO 10077-2. That can be required for multi-
layer glass or plastic with reflective layers, additives, surface structure or coating. TNO Delft and
other research facilities can provide certificates.

2.3. Heating (internal air temperature under the roof)

The third requirement concerns the provision of the heat. The heating must be able to maintain an
internal air temperature qi under the roof during and after snowfall, with a defined minimum internal
air temperature min qi well above 5°C and preferably up to 18°C, until the snow is removed. This is
called a controlled heating operation.
During this time, ventilators should be closed and the screens, also the thermal screens, should be
opened or retracted. When a gutter heating (snow melting tubes near the gutter) is used, the screens
under the gutter heating can remain closed. That increases the air temperature directly under the roof.
The specification of min qi = 12°C or min qi = 18°C, is a national choice depending on the snow
climate, operational experience of the users and their technical equipment. For characteristic snow
loads sk,50 above 1,5 kN/m2 or more than 50 mm SWE (Snow-Water-Equivalent) of new snow per
day, experience with the melting of such snowfall amounts and a heating operation with min qi =
18°C is recommended, otherwise min qi = 12°C can be used.

2.4. Unrestricted melt water outflow

The fourth requirement for a roof snow load reduction is, that the melt water can be drained quickly
without risk of icing, of bridging internal gutters or clogging internal drains. Here the experience of
the greenhouse producers and the gutter design play a role. In cold winter climates, a gutter heating
can be supportive. If a gutter is designed sufficiently for maximum precipitation rates (mostly in the
summer half year) and clogging is prevented, it is also sufficient for the melt water outflow in winter.

2.5 Additional safety measures

Further safety measures can be chosen nationally. Greenhouses may only be considered as heated
(use Ct < 1), if an automatic back-up system is present, capable of taking over the heating, before
the temperature drops below 5°C. That should include an emergency electrical power supply.
Qualified climate and weather surveillance with alarm system should be available, e.g. provided by
a so-called climate computer.

3. Thermal coefficient (influence of melting)


3.1 Thermal coefficient according to EN 1991-1-3

EN 1991-1-3:2003, clause 5.2(8) recommends the use of a thermal coefficient Ct as follows:


“The thermal coefficient Ct should be used to account for the reduction of snow loads on roofs
with high thermal transmittance (> 1 W/(m2K)), in particular for some glass covered roofs, because
of melting caused by heat loss.
For all other cases: Ct = 1
NOTE 1: Based on the thermal insulating properties of the material and the shape of the
construction work, the use of a reduced Ct value may be permitted through the National Annex.
NOTE 2: Further guidance may be obtained from ISO 4355.”

EN 1991-1-3:2003 introduces the thermal coefficient Ct but does not specify any values or describe
any influences. However, because of the challenges to specify the properties of a large range of
10
different roof claddings and types of roof structures, this is no task for the National Annex of the
countries. Research needs to be done. With the exception of some Scandinavian National Annexes,
in the other European countries no further information is given. In a few countries even Ct = 1 is fixed
as the only option. This means that greenhouses cannot be designed with EN 1991-1-3 + NAD alone.

Here the responsibility of the greenhouse Standardisation begins.


The direction for further investigations is also shown, using the ISO 4355 based on international
research beyond Europe, e.g. from Japan, Canada and USA. The ISO 4355:1998 has been reviewed
recently. With the ISO 4355:2013 a modern snow load standard is available, which offers solutions
for the greenhouse specification.

3.2 Thermal coefficients Ct according to ISO 4355

The recent and the previous version of ISO 4355 show two different methods to specify the thermal
coefficient Ct in Annex D, based on research and experience in Norway, Sweden, Finland and
Denmark and in Annex F, based on research and design experience in Japan.

The more direct method according to Annex D is to use a Ct-formula covering the different influences,
which are the heating, represented by the internal air temperature, the heat transfer coefficient for the
cladding and the snowfall rate, itself not known, but correlated to the characteristic snow load. The
formula is based on a model by Sandvik (1988), see clause 3.3.

The second method according to Annex F requires the statistical evaluation of meteorological data
using a detailed physical model for the accumulation and ablation of the roof snow cover with melting
at the bottom. For the calibration of the model and for the validation of the results experimental data
have to be gathered and evaluated too. The basic principles for detailed models where formulated by
Aoki et.al. (1988).

3.3 Background of ISO 4355, Annex D

The Ct-formula is based on a model for the accumulation and ablation of roof snow with melting at
the bottom by the meteorologist Rune Sandvik (1988), using statistical data of snowfall events in
Norway. Not snow heights, but precipitation heights, available for 12-hour measurement intervals
together with other data, corrected for drift losses by wind, evaporation losses and surface wetting,
have been used. This statistical approach has the advantage that other methods may not have, e.g.
CSA S367-9 or Rees et.al. (2002) and Liu et.al. (2006a, 2006b) which are all based on snowfall rates.
However, design values for snowfall rates are hardly available.

3.3.1 Melt model

As in any other thermodynamic calculation of snow accumulation and ablation the model is based on
an energy and mass balance, however with reference to the bottom of the snow layer, where the
melting occurs.

Mass balance:
For the chosen time interval Dti (in the model 12 hours) between ti and ti+1 the roof snow load s in
N/m2 increases by the precipitation added to the already existing snow load s(ti) in N/m2 reduced by
the molten amount of snow, which has been removed from the roof (unrestricted outflow), as follows:

s(ti+1) = s(ti) + g · S( Pr(Dti) - Mr(Dti) ) · Dti (3)


where Pr(Dti) is the precipitation rate in kg/(m2·h) for the time interval Dti in h
11
Mr(Dti) is the melt rate in kg/(m2·h) for the time interval Dti in h
g is the acceleration of gravity: g = 9,81 m/s2
Dti time interval in h, chosen as small as possible, according to the availability of
meteorological data
si(ti) roof snow load in N/m2 at the time ti

Energy balance:
The melt rate can be estimated via the energy balance of equation (4) with respect to the potential
melt surface at the bottom of the roof snow layer.

Qmelt = Qo - Qs,e = Uo · DTi - Us,e · DTs,e (4)


2
where Qmelt is the energy flux density (or rate) in W/m available for melting
Qo is the energy flux density (or rate) in W/m2 from the warm roof below
Qs,e is the energy flux density (or rate) in W/m2 from the air outside via the snow layer
Uo is the heat transmittance in W/(m2K) from the internal air to the bottom of the snow
layer, where melting occurs, with: Uo = 1/RT,o and RT,o = Rsi + S dj/lj where dj is
the depth of the roof cladding in m and lj its heat conductivity in W/mK
Us,e is the heat transmittance in W/(m2K) from the melting snow layer through the snow
and into the air outside, with: Us,e = 1/RTs,e and RTs,e = Rse + ds/ls,eff where ds is the
depth of the snow in m and ls,eff the effective heat conductivity of snow in W/mK
Rsi surface heat resistance inside in m2K/W from the air inside to the inner roof surface
Rse surface heat resistance outside in m2K/W from the snow surface into the air outside
DTi is the temperature difference in K between the internal air and the melt layer on the
roof surface: DTi = qi - qs,m
DTs,e is the is the temperature difference in K between melt layer at the bottom of the snow
and the external air: DTs,e = qs,m - qe

The melt rate Mr(Dti) in equation (4) is related to the energy flux rate for melting Qmelt by the latent
heat of fusion Lf of snow: Qmelt = Mr(Dti) · Lf , assuming Lf = 333,5 kJ/kg for melt temperatures
around 0°C.
Note: kJ/(m2·h) is convertible into W/m2 = J/(m2·s) by a factor 1000/(60·60) = 1/3,6.

Mr(Dti) = 3,6 / Lf · ( k1 · Uo · (qi - qs,m) - k2 · Us,e · (qs,m - qe) ) (5)


where Mr(Dti) is the melt rate in kg/(m2·h) for the time interval Dti in h
Lf is the latent heat of fusion of snow in kJ/kg at melting point: Lf = 333,5 kJ/kg
qs,m is the melting temperature of snow, assuming qs,m = 0°C
qi is the internal air temperature in °C
qe is the external air temperature in °C

With the factors k1 and k2 the different melt conditions on the roof can be adjusted. They depend on
the air temperature outside qe and require the calculation of the critical temperature limit for the
possibility of snow melting on the roof: qcrit = -Uo/Us,e · qi (6)
If qe ≤ qcrit: k1 = k2 = 0: no melting (too cold) → increase ds → decrease Us,e and qcrit
If qe > qcrit: if qe < 0°C: k1 = k2 = 1: melting, heat loss from the snow surface (normal)
if qe ≥ 0°C: k1 = 1 and k2 = 0: no heat loss from the snow surface (warm)

In the following chapters the model and the assumptions will be reviewed in the light of international
research, see list of Literature and References.
12
3.3.2 Heat transfer inside

The Sandvik model includes the heat transfer from the air inside to the inner roof surface, using
standard internal surface resistances. The standard value Rsi = 0,13 m2K/W is on the safe side, but it
can be improved for roof surfaces with heat flow upwards. It does also not cover condensation, which
is normally present during the melting and would increase the internal surface heat transfer, see
equation (7) as given by Tantau (2013).

Rsi = 1/hi = 1/(hr,i + hc,i + hcd,i) where hcd,i the heat transfer coefficient for condensation. (7)

The model and also ISO 4355:2013, Annex D, require a special heat transfer coefficient Uo. The
coefficient Uo is supposed to cover the heat transfer inside the greenhouse to the inner surface of the
cladding and the heat transport across the cladding by conduction to the outer roof surface. Because
it does not include the transport to the outside air, Uo is larger than U or Ug (for glass). However, in
contrast to heat protection purposes, not the highest, but the lowest values are needed.

Estimation of Uo: by relevant measurements and certification according to relevant codes


by one of the accepted calculation methods as follows:
Method 1: from lj or Rj: sum of all material and gas layers j and k: S (dj / lj) and S (1/hs,k)
with Rsi = 0,1 m2·K/W
→ Uo = 1 / (Rsi + S Rl,j + S Rg,k) (8)
Method 2: from U or Ug: with Rse = 0,04 m2K/W
→ Uo = 1 / ((1/U) - Rse) = U / (1 - U · Rse) (only if U also applies to roofs) (9)
Otherwise: Correction for the orientation!
Replacement of Rsi,sw = 0,13 m2·K/W (wall) by Rsi,up = 0,1 m2·K/W (roof)
If lower / upper value of the thermal conductivity and thickness are known:
Replacement of upper value li,sup by lower value li,inf
→ Uo = 1/ ((1/U) - Rse + (Rsi,up - Rsi,sw) + di · (1/li,inf - 1/li,sup)) (10)
Example: Monolithic single layer glass roof (4 mm) of a commercial production greenhouse.
The influence of the frame (material Aluminium with l = 160 W/(m K) is neglected,
being on the safe side. Aluminium frames increase the U-value of the glazing.
Method 1: 4 mm glass: depth of 1 layer: d1 = 0,004 m
l1 = 0,76 W/(m·K) (code takes: 1,0 W/(m·K) = upper value)
Uo = 1/(0,1 + (0,004/0,76)) = 9,5 W/(m2·K)
Method 2: 4 mm glass: U = 5,8 W/(m2·K) for vertical windows
Applying the 2 corrections given above step by step:
Uo = 1/((1/5,8) - 0,04) = 7,55 W/(m2·K)
Uo = 1/((1/5,8) - 0,04 + (0,1 - 0,13)) = 9,76 W/(m2·K)
Uo = 1/((1/5,8) - 0,04 - 0,03 + 0,004·(1/0,76 - 1/1,0)) = 9,64 W/(m2·K)
Note: The 1,5% difference to the calculation above is the result of the
“rounded” reference values. Differences < 3% are O.K

For multi-layer glass units or plastic sheets with larger cavities the heat flow and circulation in the
cavities (and Rs,k) may change with the orientation. In comparison to a vertical orientation the
convection diversifies, the U-value becomes larger. That is why it is important to ask for already
13
adapted U or Ug-values for roofs, not for windows, in a certification. For calculations see EN 673:
2011, chapter 5.4.
For the purpose of protection against heat loss, the upper value of li,sup (defined as 90/90: 90% of
data reaching a 90% single-sided confidence interval; and lD for corrections for temperature, moisture
and ageing influences, see also EN ISO 10456: 2007 and EN ISO 10077-1: 2006/AC: 2009, is used
in tables and certificates (worst case), however the melting of snow (phase transition) must be based
on the lower value of li,inf.
If no better values for the special climatic conditions inside / outside greenhouses (during snowfall,
convection / conduction through air, long- and shortwave radiation, condensation inside, evaporation
/ sublimation outside) are provided, standard values have to be used, which are:

According to EN ISO 6946: 2008 and EN 673: 2011


Overall heat transfer coefficient U [W/(m2·K)]: U = Ug = 1/RT (11)
Overall thermal resistances R [m2·K/W]: RT = 1/U = Rsi + S (Rl,j + Rs,k) + Rse (12)
or with the individual heat transfer coefficients of the surfaces hi and he [W/(m2·K)]:
RT = 1/hi + S (dj/lj + 1/hs,k) + 1/he (13)
Conduction trough a single material layer j: Rl,j = dj/lj (14)
with: dj = depth of the layer j [m] and lj = thermal conductivity [W/(m·K)]
Conduction, convection and radiation in a single gas space k between two layers:
Rg,k = 1/hs,k (15)

Regular standard values of surface resistances → according to EN ISO 6946: 2008, Table 1:
Internal surface resistance: Rsi = 0,1 m2·K/W for heat flow upwards (roof)
Rsi = 0,13 m2·K/W for heat flow sideways (side wall, windows)
External surface resistance: Rse = 0,04 m2·K/W

These Reference values for the surface heat transfer coefficients hi and he of glass or other panels
cover the following, so called Standard conditions:

According to EN ISO 6946: 2008, Annex A and EN 673: 2011, chapter 7 and 8:
with: Black-body-radiation hro [W/(m2K)]: hro = 4 · s · Tm3 (16)
Stefan-Boltzmann-constant s [W/(m2K)]: s = 5,67 · 10-8 W/(m2·K4)
Mean absolute temperature of surface and environment Tm: Tm = 273 + (qi or qe) [K]
Emissivity of parallel surfaces in radiation exchange: e12 = 1 / (1/e1 +1/e2 -1) (17)
If surface 2 (sky) >> surface 1: e12 = e1 (18)

Internal: heat transfer due to radiation and free convection (circulation):


Radiation: Emissivity surfaces: e = 0,9; air temperature inside: qi = 20 °C:
hro = 4 · s · Tm3 = 4 · 5,67 · 10-8 W/(m2·K4) · ((273 + 20) K)3 = 5,7 W/(m2K)
hr,i = e · hro = 0,9 · 5,7 = 5,13 W/(m2K) (19)
Convection: for heat flow sideways (e.g. window): hc,i = 2,5 W/(m2K)
→ hi = hr,i + hc,i = 5,13 + 2,5 = 7,63 W/(m2·K) ≈ 7,7 W/(m2·K) (20)
→ Rsi = 1/7,63 = 0,131 m2·K/W ≈ 0,13 m2·K/W

According to EN 673: 2011, chapter 7.2: Contradictory regulations, same background:


Radiation: hr,i = 4,1 e* / 0,837 with e* = 0,837 for clear, uncoated, dry glass (21)
Convection hc,i = 3,6 W/(m2K) for free convection, vertical glass panes
→ hi = hr,i + hc,i = 4,1 + 3,6 = 7,7 W/(m2·K)
→ Rsi = 1/7,7 = 0,12987 m2·K/W ≈ 0,13 m2·K/W
14

Convection: hc,i = 5 W/(m2K) increased for horizontal glass (heat flow upwards) EN ISO 6946:
→ hi = hr,i + hc,i = 5,13 + 5 = 10,13 W/(m2·K) ≈ 10 W/(m2·K)
→ Rsi =1/10,13 = 0,0987 m2·K/W ≈ 0,1 m2·K/W

External: heat transfer due to radiation and turbulent convection (due to wind)
Radiation: Emissivity surfaces: e = 0,9; air temperature outside: qe = 0 °C:
hro = 4 · s · Tm3 = 4 · 5,67 · 10-8 W/(m2·K4) · ((273 + 0) K)3 = 4,6 W/(m2K)
hr,e = e · hro = 0,9 · 4,6 = 4,14 W/(m2K)
Turbulent convection: mean wind speed near the surface: vm = 4 m/s
hc,e = 4 + 4 · vm = 20 W/(m2K) (22)
→ he = hr,e + hc,e = 4,14 + 20 = 24,14 W/(m2·K) ≈ 25 W/m2·K
→ Rse = 1/24,14 = 0,0414 m2·K/W ≈ 0,04 m2·K/W

To remove roof snow as fast as possible, screens under the roof, which reduce the heat transfer to the
cladding, should be pulled back. However, technical devices could get accidentally stuck. Others may
see no need to remove any screens during snowfall. Table 1 shows some typical contributions of
screens Rl to the total thermal resistance RT.

Nr. Type of screen Thermal resistance Rl in m2K/W 1 + 2 1 + 2 + 3


1 Day screen 0,08
2 Thermal screen 0,14 0,21
3 Black-out-system 0,36 0,58
Table 1: Heat resistance of screens inside a greenhouse according to Tantau (2013)

Total thermal resistances R [m2·K/W]: RT = 1/U = Rsi + S Rl + S (Rj + Rs,k) + Rse (23)
If only a U-value is known, U and Uo can be corrected as follows:

Uo = 1 / ((1/U) + S Rl) and Uo = 1 / ((1/Uo) + S Rl) (24)

Greenhouse roof cladding, examples U in W/(m2K) Uo in


for heat protection a W/(m2K)
Direction of the heat flow sideways b upwards c for snow melt
according to
for walls for roof
ISO 4355
Single-layer: 4 mm Soda lime glass d 5,7 6,9 9,50
Single-layer: plastic film (0,2 mm PE or ETFE) e 5,8 7,1 9,92
Laminated (safety) glass d 4/4/0,76 5,5 6,6 8,75
Double layer glass d unit filled with Argon 4/16/4 2,6 3,0 3,40
Double layer plastic film (2 x 0,2 mm PE or 4,0 4,5 5,46
ETFE) e, filled with air (50 mm)
a
These values are based on reference standard conditions according to EN 673 and EN ISO 6964:
- external surface resistance Rse = 0,04 m2K/W for e = 0,9; qi = 0°C; wind vm = 4 m/s;
- internal surface resistance dependent on the heat flow as in b and c
b
heat flow sideways: Rsi,sw = 0,13 m2K/W for e = 0,9; qi = 20°C; free convection hc,i = 2,5 W/(m2K)
c
heat flow upwards: Rsi,up = 0,1 m2K/W conditions as in b, but increased convection hc,i = 5 W/(m2K)
d
Thermal conductivity of glass: linf = 0,76 W/(mK) and lsup = 1 W/(mK)
e
Thermal conductivity of PE or ETFE plastic film: linf = 0,24 W/(mK) and lsup = 0,52 W/(mK)
Table 2: Uo-values for typical greenhouse claddings under standard conditions
15

For the heat transfer coefficients of the roof cladding, Sandvik (1988) considers general Uo-values of
3 W/(m2K) and 5 W/(m2K). Typical greenhouse claddings have considerably larger Uo-values even
for standard conditions, see Table 2.

3.3.3 Heat and mass transfer outside

In the Sandvik model at the outer roof surface a melt layer is simulated, from which the melt water is
removed (outflow). To melt any snow, the external roof surface temperature has to be qse > 0°C.
Surfaces with the temperatures under 0°C will also reach melting condition, after a snow layer has
been built up. Snow is an excellent insulation material. The melting can only use the energy available
at the surface after the heat loss into the ambient air is taken into account. The model uses the standard
external surface resistance Rse = 0,04 m2K/W for radiation (e = 0,9; qe = 0°C) and turbulent convection
(vm = 4 m/s), see chapter 3.3.2.

On the outer surface a layer of roof snow can build up, taking into account the remaining snow after
melting plus any new snowfall. The snow layer separates the melt layer from the ambient air with the
surface heat loss from the snow cover, the higher the layer becomes, the better the melting.
The conductive heat transfer through the increasing snow layer is calculated using an equivalent mean
density of 250 kg/m3, leading to an equivalent heat conductivity of 0,18 W/(mK).

Note: Before cold snow can be molten it needs to be warmed up to 0°C (ripe snow). That does not
require much energy in comparison to the melting, so that it is often neglected, also in the Sandvik
model. The additional energy for warming can be integrated into melt rates as in equation (4) very
simply by using a correction factor f(cpi) for the latent heat Lf:

Lf is to be replaced by: Lf · f(cpi) = Lf · (1 - cpi/Lf · DTs,e) (25)


where Lf is the latent heat of fusion of snow in kJ/kg at melting point: Lf = 333,5 kJ/kg
cpi is the specific heat of snow and ice: cpi = 2,1 kJ/kg/K
DTs,e is the temperature gradient between new snow and ripe snow: DTs,e = qs,m - qe

Note: In a more detailed approach this energy is added to the snow as energy flux density for
precipitation advection with the aspects: snow / rain, warming / cooling, melting / refreezing.

The effective heat conductivity of a snow layer depends on the snow density. The density can change
and increases with the depth of the layer, so an equivalent value is required. On top of the
accumulating roof snow layer cold freshly fallen snow can be found, on the bottom is ripe snow at
melting point together with melt water, a mixture called slush. A snow layer retains a certain small
quantity of melt water, before the outflow begins. Dependent on the application in the literature
retention heights vary between 10% and 40% of the snow height or between 2,5 and 5 cm water total.
The irreducible (or residual) water content however is as low as 3-6% volumetric water content. That
is why some applications allow a limited outflow of 10% of the full outflow or an increasing outflow
from zero to 100% before reaching the retention height. However, if the outflow is not restricted, e.g.
on a sloped roof, the retention limit does not need to be larger than the irreducible water content.

The mass and volume change in a melting snow layer can be simulated using the hydrological Snow-
Compaction-Method according to Knauf (1980), based on a densification formula by Bertle (1966).
For a ratio of melt water height to ice content (hw/hi) the density at the bottom of the snow layer is
described by equation (26).

Wet snow density: rsw = rsf / (1 - 0,474 · (hw/hi)) with: (hw + hi) / hsw = rsw / rw (26)
16
where rsw is the wet snow density in kg/m3 taking into account the compaction
due to the rise of the liquid water content
rsf is the fresh snow density at the begin of the process in kg/m3
rw is the density of water: rw = 1000 kg/m3
hw is the water content (pure water height in the snow layer) in mm
hi is the ice content (height of pure ice in the snow layer) in mm
hsw is the total wet snow height in mm

With the Snow-Compaction-Method, as still in use, see Gerlinger & Haag (2011), densities increase
to 400 to 600 kg/m3 at the bottom of melting snow. For a simplified one-layer melt model the mean
value between the fresh snow density on top and the wet snow density at the bottom can be assumed.

Fresh snow densities have naturally a large scatter because the conditions under which snow is
created on the long way down to earth are volatile and not measured routinely. Reliable measurements
are available for the air temperature (2 m above ground) and the wind speed (10 m above ground).
Based on such data a couple of formulae have been developed to approximate the mean value of the
fresh snow density for further applications, see Figure 2.

Figure 2: Comparison of different formulae for fresh snow densities

These densities are not used for design purposes, due to the large scatter, but to estimate parameters
for snow accumulation and ablation models, e.g. in hydrology, glaciology or engineering.
The different approaches have been analysed and the following formula can be chosen as best suited
for this purpose. The fresh snow density according to Hedstrom & Pomeroy (1998) in equation (27),
see pink line in Figure 2, is the best known and most used mean density.

Hedstrom & Pomeroy (1998): rsf = 67,9 + 51,25 · exp(qe/2,59) with exp(x) = ex (27)
where rsf is the fresh snow density in kg/m3
qe is the external air temperature in °C (measured data)
17

The heat conductivity of a snow layer is always an equivalent value, because snow is not an ideal
solid material, but porous, containing mostly air with water vapour, some ice and liquid water. Solid
materials transfer heat only through conduction. In liquids and gases, which can both be understood
as streaming fluids, heat is also transferred by convection (sensible heat) and by radiation. Water
being present in all three phases and changing between them drives latent heat fluxes.
However, radiation (both short-wave and long-wave) gets extinct very rapidly near the surface.
Turbulent convection and latent heat fluxes in the pore space of snow have been investigated, but
without applicable results. The variability and complexity of snow grain shapes and their
metamorphosis make it difficult. The changing, open pore space is not comparable to any other pore
space, e.g. in sand. Theoretical formulations could not describe the processes in such a way, that
laboratory results and measurement could be verified. But the task is still in progress.

Instead it was found, that the effective heat conductivity as a function of the snow density could cover
all internal heat fluxes as well. Snow can be treated as a quasi-solid material. Only the surface heat
transfer has to be treated differently.

Figure 3: Comparison of different formulae for the equivalent heat conductivity of snow

From the different approaches analysed, the formula by Calonne et.al. (2011) in equation (28), see
green line in Figure 3, gives mean values for the snow types present on a greenhouse roof
(precipitation particles, rounded grains, melt forms). The formula is based on a new evaluation of all
data available. The dark blue line by Riche & Schneebeli (2013) is based on the previous work and
gives maximum values for anisotropic, faceted crystals and depth hoar, particles that are not likely to
dominate during roof snow melt.

Calonne et.al. (2011): ls,eff = 0,024 - 1,23 · 10-4 * rs,eff + 2,5 · 10-6 * rs,eff2 (28)
where ls,eff is the effective thermal conductivity of a snow layer in W/mK
rs,eff is the equivalent value of the snow layer density in kg/m3
18

The mean density value of 250 kg/m3 in the Sandvik model represents cold fresh snow on top of the
layer with a density of 100 kg/m3 and ripe snow and slush with a density up to 400 kg/m3 at the
bottom. This is a good assumption for cold and moderate climate in Europe. For wet snowfalls with
higher fresh snow densities up to 150 kg/m3 and ripe snow and slush density up to 600 kg/m3 a higher
mean density up to 350 kg/m3 can be assumed. Calculations for the snow melting on heated
pavements and cold room experiments show the gradual rise of the thermal conductivity during the
melt process (Nuijten et.al., 2016).
The equivalent thermal conductivities for those two climate cases are:
• Cold / moderate: snow density rs,eff = 250 kg/m3: thermal conductivity ls,eff = 0,15 W/mK
• Warm: snow density rs,eff = 350 kg/m3: thermal conductivity ls,eff = 0,29 W/mK

The standard value for the external surface resistance Rse = 0,04 m2K/W is not well suited for snow
surfaces and does not apply to wet surfaces, because of additional latent heat losses due to evaporation
or sublimation. Also the heat transfer by radiation and by convection over a fresh snow surface is
slightly different (e = 0,99; qe < 0°C). But despite that, the standard value can be an approximation,
especially for heavy snowfall conditions with dry snow, not too cold and not very windy. It is good
as a mean value over one or half a day. Smaller time intervals require a more specific approach.
In the night after the snowfall under a clear sky the heat loss from the snow surface can be large. The
snow surface cools down. Therefore, it is recommended to remove the snow as quickly as possible,
preferably during the snowfall and on the day after using the short-wave radiation gain, partly
compensating for any heat losses at night.

3.3.4 Evaluation and results

As an example, for the evaluation using his model, Sandvik (1988) showed the accumulation and
ablation process for two different roofs (Uo = 3 W/(m2K) and 5 W/(m2K)) under an extreme heavy
snowfall event 1970 in Kjevik, see Figure 4. The snowfall lasted nearly 7 days, brought up to 54 mm
SWE per day (with drift correction: 70 mm SWE), and led to 1,5 m ground snow after 6 days.
However, the roof snow loads on roof 1 and 2 remained much smaller.

The characteristic snow load on the ground for the area Kristiansand is: sk,50 = 4 kN/m2, which leads
to the following Ct-values for Kjevik`s roof snow:
• Roof 1 (Uo = 3 W/(m2K): Ct = s1,n,t / sk,n = 0,921 / 4 = 0,23
• Roof 2 (Uo = 5 W/(m2K): Ct = 0,537 / 4 = 0,13 *
Note: * due to a higher snowfall rate in another year: Ct = 0,67 / 4 = 0,17
As can be seen in this example, the Ct-evaluation was carried out, setting other roof snow load
coefficients to unity. Because of the flat roof situation, the shape coefficient was µ1 = 1. Because the
snow did melt, no drift loss was considered, and the exposure coefficient was Ce = 1.

The model has been verified by statistical evaluations in other Scandinavian countries, e.g. in Sweden
by Dahlberg et.al. (1988).

In 1990, a calculation formula for Ct based on these evaluations was first introduced in the Norwegian
standard NS 3479 “Design load for structures”. There, a general conversion factor for the ratio of roof
to ground snow loads of 0,8 was given, probably via the shape coefficient, as usual in European
standards. For this ratio of 0,8 the Ct-value has been increased by more than 1/0,8, as can be shown
using the Ct-formula on the two roof examples.
• Roof 1 (Uo = 3 W/(m2K): Ct = 0,353 > 0,23 · 1/0,8 = 0,29 → reserve 21%
• Roof 2 (Uo = 5 W/(m2K): Ct = 0,279 > 0,17 · 1/0,8 = 0,21 → reserve 33%
19

a) Ground snow load (red line) and roof snow loads in kN/m2

b) External air temperature in °C (measurement 2 m above ground)

c) Precipitation in mm (left, blue columns) and mean wind speed in m/s at 10 m height (right, green line)

Figure 4: Ground snow load (red line) and roof snow load for two roofs (1: Uo = 3 W/(m2K), 2:
Uo = 5 W/(m2K)) for meteorological station Kjevik, Norway, February 1970

The Ct-formula is conservative for this extreme snowfall event. The snowfall event from Kjevik,
Kristiansand, Norway is not the most severe event in Northern Europe. Wern (2015) reported a new
record from Kilgarden, Sweden with 75 cm new snow on the first day of December 2012, if the
legendary 109 cm from the first of November 1921 in Njunjes, Lapland is neglected. According to
Taylor (1980) the absolute record was measured in Canada with 195 cm new snow on April the 15th
1921, with a more modern record of 145 cm on February the 19th 1999.
20
When in 1998 the Ct-formula was introduced into ISO 4355:1998, Annex D, no general conversion
factor of 0,8 did exist, but a standard exposure coefficient of Ce = 0,8 was recommended.
In the reviewed ISO 4355:2013 there is a general conversion factor of 0,8 in front of the roof snow
load formula. The standard exposure coefficient has increased to Ce = 1, the basic shape coefficient
for flat roofs is also: µb = 1. In comparison to the original NS 3479 for both ISO Norms there was no
further correction necessary.
This would also not be necessary in connection with the EN 1991-1-3, which has a general shape
coefficient for flat roof of µ1 = 0,8, but no general conversion factor. The usual choice for the exposure
coefficient is Ce = 1. Any increase, also for large roofs, see ISO 4355:2013, clause 6.1 is not
necessary.

Conclusion:
For the use of Ct < 1 according to ISO 4355; Annex D, the exposure coefficient is Ce = 1. Larger or
smaller values would not be appropriate, because Ct has been estimated using the complete, wind
corrected precipitation data. The fast melting roof snow cover is too moist and has no time to drift,
as will be explained in Part II of this report.

The Sandvik model and the Ct-formula according to ISO 4355, Annex D is the most reliable for
building design in comparison to other models (CSA S367 for Membrane Structures, ASHRAE for
Surface Snow Removal).
Because of the possible deviations from the standard parameters and mean values used in the model,
additional limits and restrictions are formulated for the use of the Ct-formula, concerning the heat
transfer coefficient for the cladding Uo and the heating (air temperature qi). The experiences in
Scandinavia and in the greenhouse industry in Europe show, that the Sandvik-formula is conservative.

3.4 Use of Ct-formula for greenhouses

For heated greenhouses equipped for the controlled melting of roof snow, the thermal coefficient Ct
in accordance with ISO 4355, Annex D can be used, completed by a roof angle function, see equation
(29). It allows the calculation of Ct for certain categories of greenhouse claddings and heating
regimes. Ct can also be estimated for each single project. The latter will be useful for countries with
large variations of the ground snow load. In the future it can be used to approximate the costs for
energy saving decreases of the U-value, e.g. for new claddings or as target values in standards.

Ct = { 1 - f (sk,n) · f (qi) · f (Uo) } · f (a) (29)


Ct = {1 - 0,054 · (sk,n / 3,5)1/4 · (qi – 5°C) · [sin((0,4 · Uo - 0,1)·180°/p)]3/4} · cos(a)

With the following influences:


• Snow load (correlated to snowfall rate): f (sk,n) = 0,054 · (sk,n / 3,5)1/4
• Inside air temperature (heating): f (qi) = qi - 5°C
• Heat transfer coefficient roof: f (Uo) = [sin((0,4 · Uo - 0,1)·180°/p)]3/4
• Ratio of ground area to roof area: f (a) = cos(a)

3.4.1 Influence of the Roof angle

In addition to the Ct-formula for flat roofs given in ISO 4355, Annex D (1998 as well as 2013), the
function for the roof angle influence can be added:
f(a) = cos(a) with a: roof angle (30)
21
In the National Annexes to EN 1991-1-3 of some Scandinavian countries a roof angle function of
f(a) = cos (2a) has been introduced. Besides the roof angle influence on melting, it was probably
meant to compensate for the missing influence of snow sliding from warm slippery surfaces in the
EN 1991-1-3. The ISO 4355 specifies such an influence traditionally in the shape coefficient µb by
the surface material roughness coefficient Cm. However, this would not lead to any reduction of the
snow load in the inner bays of multi-span greenhouses.

EN 1991-1-3 with the bilinear shape coefficient µ1 for all roof materials, smooth or rough, warm or
cold, does not consider any sliding effect on the uniform roof snow load. ISO 4355:1998 however,
had a reduced shape coefficient µb especially for slippery, warm roofs with Cm = 1,333 for Ct < 0,9.
The difference between (Ce,ISO-98 · µb) and (Ce,EN-03 · µ1) within practical limits can be compensated
for by working out the mathematically correct difference, resulting in the following roof angle
functions in equation (31) to be used with the shape coefficient of EN 1991-1-3: 2003:

Roof angle: 0° ≤ a ≤ 30°: f(a) = (cos(2a))1/2 (31)


30° < a ≤ 45°: f(a) = ½ (cos(2a))1/2 / (1 - a/60°)

These roof angle functions apply only to external slopes and cover a snow load reduction due to
sliding on warm, slippery surfaces, not due to the melting itself.

Conclusion: The roof angle function for snow melting due to the larger surface area should be
separated from that for the sliding of snow from external roof pitches. Instead of applying the roof
angle functions in equation (29), the sliding of roof snow can be covered separately by the
appropriate shape factors according to ISO 4355: 1998, 5.4 using Cm > 1. They can be adapted to the
EN-format, see Part II. For melting alone, a roof angle function in equation (28) for the thermal
coefficient Ct is used.

Considering the different energy fluxes which can be present during snowfall and subsequent melting,
the roof angle influence f(a) = cos(a) can be explained as follows:

The precipitation itself is independent from the roof angle, so is the advective heat transfer by
precipitation. However, on the other side the snow melting is dominated by the conductive heat
transfer through the roof surface. In comparison to a horizontal roof, with a roof angle a, the roof
size is increased by a ratio of f(a) = 1/cos a. That is why more snow can melt (independent from any
sliding) with a larger roof angle a. It is not the angle, but the size of the roof surface.

The short-wave radiation gain can also increase, dependent on the orientation of the roof with
respect to the sun. The position of the sun is defined by its height above the horizon (0° at dawn) and
the azimuth along the horizon: ds = 0° for North, 90° for East, 180° for South, 270° for West, 360°
for North again. The function for the influence of the roof angle a is given as follows:

f(a) = sin a · (cos(dsun - droof) / tan hs) + cos a (32)


with: a is the roof angle measured from the horizontal
hs is the “sun height” or “zenith” = angle of the sun direction above the horizon
dsun - droof is the azimuth angle difference between the direction of the sun and
the roof normal

Calculating the short-wave radiation has been worked out by the solar industry in detail, see
appropriate literature and software products. For the melting of snow, the short-wave radiation gain
is often neglected. It does not play a large role, because heavy snowfall events and the major melting
22
operation happen at night. If a heavy snowfall persists during the next day, there will not be much
sun. To neglect the short-wave radiation gain is on the safe side.

The long-wave radiation exchange (often heat loss at night) of the roof depends on the area of the
sky, which can be "seen" by the roof. E.g. a horizontal roof "sees" 180° of the half sphere in all
directions, a vertical wall only 90° in one of the directions. The roof angle influence, called size factor,
in relation to the horizontal roof is shown in Table 3.

Roof orientation Size factor for long-wave radiation


Horizontal roof (a = 0°) 1
Vertical wall (a = 90°) 0,5
Plane roof pitch with roof angle a < 90° (1+cos a) / 2 = (cos (a/2))2
Multi-span, duo-pitch roof with roof angles a (1+cos (2a)) / 2 = (cos a)2
Table 3: Roof angle influence on long-wave radiation

Of course, heat radiates in every direction. The parts of the roof, which cannot “see” the sky, stand in
radiation exchange with the surrounding, which can be colder than the roof. That is why the long-
wave radiation is often not reduced by the roof angle size factor in Table 3.
On a multi-span roof, the "surrounding" is the same roof with the same emissivity and temperature
(no exchange, the size factor applies).

The turbulent heat exchange (sensible and latent) is driven by the temperature and water vapour
pressure gradients in the air. The sun creates them. The gravity also plays a part. Turbulences are
created by the wind. Wind pressure coefficients change across wall and roof. Due to the influence of
the sun the roof angle influence is weaker, but similar to that one for long-wave radiation.

Note: During heavy snowfall events, the heat losses by long-wave radiation and by turbulent sensible
and latent heat exchange are limited, because the driving gradients in temperature, pressure and
moisture are small. During snowfall the air moisture saturation is close to 100%. The air temperatures
are moderately low. Very heavy snowfalls in Europe (wet snow events) are often around 0°C (melting
point).

Conclusion: Using the roof angle function f(a) = cos a for the thermal coefficient Ct is on the safe
side. It covers the conductive heat transfer between roof and snow correctly. It increases with larger
a. The heat losses at the upper snow surface due to long-wave radiation and turbulent exchange
support the trend, they get smaller, as well as the gain by the short-wave radiation, which increases.
Because these surface influences are small in comparison to the conduction at the roof surface and
influenced by more parameters, which change during the day, their exact calculation would not be
appropriate for global models with snow melting at the bottom. They are covered by limiting the roof
angle function to: f(a) = cos a.

3.4.2 Temperature range and heating

For the application of the Ct-formula the air temperature under the roof, which can be taken into
account, is limited.
Air temperature inside under the roof: 5°C ≤ qI ≤ 18°C

The lowest value of 5°C is a reserve, to ensure, that the snow melting can start. Melt temperatures
measured in experiments are never exactly at the theoretical melting point. Snow contains not only
ice crystals, but also traces of liquid water and a lot of air. These elements can be warming, while the
23
ice melts. That is why in melting layers temperatures above 0°C can be measured. It has to be ensured
the required energy can be provided.

Above 18°C internal temperature, there is no experience (melt rate, tolerance for plants,
documentations, data). However, the heat transfer from inside is the most important influence for the
melting of snow during snowfall and for melting of accumulated snow on a roof. That is why in
reality the temperature can be as large as possible. The limit of 18°C is only used for design purposes
and it is a robustness measure, making the Ct-formula conservative.

3.4.3 Limits for the heat transfer coefficient

In connection with the use of Ct the heat transfer coefficient of the cladding is limited:
1 W/(m2K) ≤ Uo ≤ 4,177 W/(m2K)

The lower limit is given in ISO 4355: 2013, Annex D, equation (D.2) with min Uo = 1 W/(m2K).
This seems to be an agreement based on experience, more than a physical limit to the melting of roof
snow.
The sine-function for the simulation of the influence of the heat transfer coefficient f(Uo) has
mathematical limits. The influence function of the heat transfer coefficient f(Uo) reaches its minimum
for the minimum value of the sine-function: sin (0,4 · Uo – 0.1) = 0, giving the purely theoretical
lower limit value of min Uo = 0,1 / 0,4 = 0,25 W/(m2K).

Although snowmelt affects most roofs above buildings inhabited by people or livestock, Ct < 1
according to ISO 4355, Annex D should not be used for roofs with a thermal transmittance lower than
Uo = 1 W/(m2K).

Note: A certain melt effect additionally to the drift loss is already included in the general
conversion roof snow / ground snow of most standards. In EN 1991-1-3:2003 the shape
coefficient of µ1 = 0,8 represents this ratio. However, only the exposure coefficient Ce > 1
contributes to an increase of this ratio, not the thermal coefficient Ct > 1, e.g. for open
buildings, canopies, freezer buildings or ice-skating halls. For these roofs the conversion
ratio of 0,8 could be too small. This inconsistency in EN 1991-1-3:2003 results from the
insufficiently defined exposure coefficient, related to wind only, not to temperature.

In ISO 4355:2013, Annex D, equation (D.1) to (D.3) the thermal coefficient Ct is given together with
an upper limit value of 4,5 W/(m2K). It can be expected that for the limit value the sine-function will
reach the maximum. This is not the case. The numerical value of 4,5 W/(m2K) is only an
approximation of the mathematically exact limit. It is sufficiently exact for Ct-values with up to one
digit. For values with more than one digit, the following exact limit can be derived:

Influence function of the heat transfer coefficient: f(Uo) = ( sin (0,4 · Uo - 0.1) )3/4 (33)
The sine-function reaches its maximum for: (0,4 · Uo - 0,1) · 180°/p = 90° (34)
0,4 Uo = p/2 + 0,1
→ giving max Uo = (5 p + 1) / 4 = 4,1769908 W/(m2K) = 4,177 W/(m2K).

Because modern design calculations are done by computer, the approximation is to be replaced by
the exact upper limit max Uo = 4,177 W/(m2K).
24
3.4.4 Climatic limitations

There are also conditions for the climate when using the Ct- formula. These limits seem to have
physical reasons. It should not be too cold or too warm and wet. The Sandvik model does not cover
those special conditions. Nonetheless the snow melts there as well.

3.4.4.1 Air temperature outside (lower limit):

The mean temperature at snowfall has to be larger than min qe = -8°C. For colder temperatures the
Ct-value has to be increased by a factor 1,2 (with max Ct ≤ 1).

Under colder ambient temperatures, ice lenses on top or in the melting snowpack are possible. Ice
layers (refreezing, also within the snowpack) have not been included in the energy-mass-balance
model, which was used for the development of the Ct-formula in ISO 4355, Annex D. Pure ice has a
larger heat conductivity than snow: 2,3 W/(mK) at -10°C in comparison to the mean value of 0,18
W/(mK) for snow taken into account.

3.4.4.2 Snow load (upper limit):

The snow load function f(sk,n) = 0,054 (sk,n/3.5)1/4 correlates the climate (and its ground snow load
sk,n) to the snowfall rate or intensity and duration, which really govern the melting process. The roof
snow melts during and immediately after the snowfall and does not accumulate on the roof as it does
on the ground. Higher ground snow loads are the result of colder temperatures with longer
accumulation periods and very often more than one snowfall event, not necessarily of much larger
snowfall intensities. That is why, for larger sk,n the Ct-values tend to get smaller.

Figure 5: Reduced roof snow loads using Ct according to ISO 4355: 2013 without upper limit
25
The Ct-formula is based on a simplified energy and mass balance model using statistical data of
snowfall events in Norway, according to Meløysund (2010) for ground snow loads sk,50 between 1,5
kN/m2 and 9 kN/m2. A lower limit for the snow loads is given but an upper limit is missing. For
very large sk,n the value of Ct can be reduced faster than the ground snow load increases, leading to
decreasing roof snow loads, see Figure 5.

Such a trend is may be possible in dry, cold climate, if the snowfall rate decreases. The Norwegian
data may have proved this tendency for up to 9 kN/m2. However, in the Alpine climates of central
and southern Europe with even larger snow loads, there is no known statistical basis for such a trend.
For very high snow loads negative values for Ct result, which are obviously wrong.

To avoid this effect, an upper limit for the ground snow load lim sk,n can be derived by finding the
maximum value of the roof snow load function by differentiation: δ (Ct · sk,n) / δ (sk,n) = 0

→ Maximum for: 1 – f(qi) · f(Uk,n) · 0,054 · (1/3.5)1/4 · 5/4 · sk,n1/4 = 0


1 = f(qi) · f(Uo) · 0,04935 · sk,n1/4
→ giving: lim sk,n = (1 / (0,04935 · f(qi) · f(Uo)))4 = (20,263425 / (f(qi) · f(Uo)))4 (35)

Example:
For max qi = 18°C (f(qi) = 13 K) and max Uo = 4,177 W/(m2K) (f(Uo) = 1) the following limit as
the lowest possible value can be derived: lim sk,n = (20,2634 / (13 · 1))4 = 5,903 kN/m2. Any other
condition leads to a higher limit load. Thus, the limit applies only to very large snow loads.

The upper snow load lim sk,n is associated with the lower limit for the thermal coefficient min Ct
= 0.2. Instead of comparing the snow load sk,n to the upper limit lim sk,n, a lower limit for the thermal
coefficient min Ct has been introduced in EN 13031 as a limiting condition. Including the roof angle
function, the condition reads: → Check: Ct ≥ min Ct = 0,2 · f(a)

If Ct is smaller, min Ct has to be used. However, using min Ct for the calculation together with larger
snow loads than lim sk,n would also increase the roof snow loads further, proportionally. That would
only be realistic for drastically increasing snowfall rates.
It might be more realistic to limit the reduced roof snow loads to an upper value based on min Ct ·
lim sk,n, as in Germany with s1,t = 0,75 kN/m2 for heated sales greenhouses according to DINV
11535:1998 (superseded). The insurance data in Germany (for characteristic ground snow loads
between 0,65 kN/m2 and 13,86 kN/m2 at A = 1500 m, SLZ 3) did not record any problems for sales
greenhouses with a roof snow load of 0,75 kN/m2, see green line in Figure 5.

For the example here with max qi = 18°C and max Uo = 4,177 W/(m2K) the upper roof snow load
limit is: s1,n,t = Ce · min Ct · lim sk,n · µ1 = 0,94 kN/m2 for a typical multi-span greenhouse roof (Ce
= 1; µ1 = 0,8). If a roof angle function f(a) is used, for a = 22° an upper roof snow load limit of s1,n,t
= 0,876 kN/m2 results, which is slightly larger than the old value of 0,75 k/m2.

Summary: For large ground snow loads applys:


If Ct < 0,2 · f(a), take min Ct = 0,2 · f(a) and sk,n = lim sk,n = [20,2634 / (f(qi) · f(Uo))]4

3.4.4.3 Snow load (lower limit):

Very small snow loads should not be reduced any further. They could result from one single snowfall
event, for which the standard shape factor of 0,8 could already be too small. If the snowfall were too
wet, the assumed snow density and equivalent heat conductivity in the Sandvik model would not be
on the safe side anymore. Also, the Standard surface resistance taken into account may not cover heat
26
losses from the wet snow surface or the roof, if the snow cover gets too thin. A minimum roof snow
load value should remain. It can also cover heavy rain precipitations in the summer half year.

However, the lower limit for the ground snow load of min sk,n = 1,5 kN/m2 in ISO 4355 refers to the
minimum snow load in Norway, used during the statistical evaluation. With the evaluation in Sweden,
see Dahlberg (1988), the limit was reduced to 1 kN/m2. This is still large. Snow melting has been
demonstrated for smaller snowfalls. That is why for greenhouses, the lower limit according to ISO
4355 is replaced by a robustness limit for the roof snow load after melting.

For commercial production greenhouses EN 13031-1 limits the remaining roof snow load after any
reduction to:
min s1,n,t = 0,25 kN/m2

The value is chosen because of positive experiences (proven by insurance data) with the superseded
greenhouse standards EN 13031-1:2001 and DINV 11535-1:1998, which both allowed roof snow
loads of 0,25 kN/m2 in the entire country, as long as a strict heating regime was followed.
Additionally, 0,25 kN/m2 turned out to be a slide load limit for heated membrane roofs with a slope
angle of about 25°, see Otsuka et.al. (1992). For greenhouses in Germany a 24-hour emergency
service is guaranteed, to repair any damage to the technical equipment required for plant survival,
especially the heating. As an example a roof snow load of s1,n,t = 0,25 kN/m2 (µ1 = 0,8; Ce = 1; f(a) =
1) would result from a ground snow load of sk,n = 0,57 kN/m2 for any heat permeable cladding with
Uo ≥ 4,177 W/(m2K) with heating operation for qi = 18°C. With qi = 12°C the ground snow load
could be sk,n = 0,4 kN/m2.

For greenhouses open to the public a larger reserve is required. Multi-layer glazing or other heat
protective claddings are used more often. The Scandinavian standards for normal buildings could
serve as examples, e.g. SFS-EN 1991-1-3/NA in Finland. The greenhouse standard EN 13031-2 limits
the roof snow load to:
min s1,n,t = 0,50 kN/m2.

As an example a roof snow load of s1,n,t = 0,50 kN/m2 (µ1 = 0,8; Ce = 1; f(a) = 1) would result from
a ground snow load of sk,n = 0,78 kN/m2 for any heat permeable cladding with Uo ≥ 1 W/(m2K) with
heating operation at qi = 18°C. With qi = 12°C the ground snow load could be up to sk,n = 0,7 kN/m2.
Larger ground snow loads do not reach the lower limit.

3.4.5 Overview: Snow load limits and Ct-formula

Figure 6 and 7 show the thermal coefficient Ct according to ISO 4355: 2013, Annex D for typical
service temperatures of greenhouses of qi = 18°C and qi = 12°C. The roof angle influence is neglected:
f(a) = 1, because it could vary. Here the values for a horizontal level are shown.

The necessary limitations for large and for small snow loads are marked in red. For large snow loads
the thermal coefficient should be limited to min Ct = 0,2, see horizontal red line. For small snow
loads the required robustness level for a possible interruption of the heating need to be considered.
The effects of the known and used levels for the lower roof snow limit, see dotted curved red lines
are also shown: min s1,n,t = 0,5 kN/m2 for greenhouses open to the public, as well min s1,n,t = 0,25
kN/m2 for commercial production greenhouses and s1,n,t = 0,75 kN/m2 for sales greenhouses as the
still used very old values according to the superseded DINV 11535-1.
27

Figure 6: Thermal coefficient Ct and limits for qi = 18°C and different heat transmittances Uo

Figure 7: Thermal coefficient Ct and limits for qi = 12°C and different heat transmittances Uo
28

Literature

Standards and References:


- CSA S367-09 Air-, cable-, and frame-supported membrane structures
- 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
- DINV 11353-1: Gewächshäuser Teil 1: Ausführung und Berechnung
- EN ISO 6946: 2007: Building components and building elements – Thermal resistance and
thermal transmittance – calculation methods
- EN 673: 2011: Glass in building – Determination of thermal transmittance (U-value) –
Calculation method
- EN ISO 10456: 2009: Building materials and products – Hydrothermal properties – Tabulated
values and procedures for determining declared and design thermal values
- EN ISO 10077-1: 2010: Thermal performance of windows, doors and shutters – Calculation of
thermal transmittance – Part 1: General
- Final Report of the Commission of the European Communities DG III-D3 (1998-03), 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
- 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
- JIS Japanese Industrial Standard C 8955: 2011 (E): Design guide on Structures for Photovoltaic
Array
- NS 3491-3: Prosjektering av konstruksjoner, Dimensjonerende laster, Del 3: Snolaster, Norway
- SFS-EN 1991-1-3/NA: Actions on structures – Part 1-3: General actions – Snow loads; National
Annex Finland
- World Meteorological Organisation: Solid precipitation measurement intercomparison. Final
Report. WMO No. 67 (1998)

Background information of general interest will be accessible for Download under


www.greenhousecodes.com. The following documents are currently available:
- 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
- Roof Angle Function for Thermal Coefficient Ct.pdf
- Comparison of Formulae: Heat Conductivity and Density of Snow.pdf
- Model Snowmelt.pdf
Citations from these documents are not marked as such, because of the same authorship.

Literature: References und sources used generally and for Part I - Melting:
- Bertle, F.A. (1966): Effect of snow compaction on runoff from rain on snow. United States
Department of the Interior Bureau of Reclamation, Water Resources Technical Publication,
Engineering Monograph No. 35.
- Calonne, N., Flin, F., Morin, S., Lesaffre, B., Rolland du Roscoat, S., Geindreau, C. (2011):
Numerical and experimental investigations of the effective thermal conductivity of snow. In:
Geophysical Research Letters, Vol. 38, L23501.
- Dahlberg, M., Hansson, L., Lindt, T. (1988): Snölast pa Glastak. Examensarbede, Lund Institute
of Technology, Department of Civil Engineering, Sweden.
29
- Ducloux, H., Nygaard, B.E. (2014): 50-year return period wet-snow load estimation based on
weather data for overhead design in France. In: Natural Hazards and Earth System Sciences, 14,
pp. 3031-3041.
- Fierz, C., Armstrong, R.L., Durand, Y., Etchevers, P., Greene, E., McClung, D.M., Nishimura,
K., Satyawali, P., Sokratov, S.A. (2009): The International Classification for Seasonal Snow on
the Ground. IHP-VII Technical Documents in Hydrology No. 83, IACS Contribution No. 1, Paris,
France.
- Hedstrom, N.R., Pomeroy, J.W. (1998): Measurements and modelling of snow interception in the
boreal forest. In: Hydrological Processes 12, 1998, pp. 1611-1625.
- Kolar, G. (2008): Anwendung des Schneedeckenmodells CROCUS am Standort der
automatischen Wetterstation Teufelsegg im Firngebiet des Hintereisferners. Master Report,
University of Innsbruck, Austria.
- Knauf, D. (1980): Die Berechnung des Abflusses au seiner Schneedecke. Schriftenreihe des
Deutschen Verbandes für Wasserwirtschaft und Kulturbau, Heft 46, Analyse und Berechnung
oberirdischer Abflüsse, Teil III.
- Gerlinger, K., Haag, I. (2011): LARSIM – Meeting 2011 (in German and French) – Modellierung
des Schnees im Wasserhaushaltmodell LARSIM: Berechnung der Setzung und der aktuellen
Schneeschmelze.
- Lehning, M., Bartelt, P., Brown, B., Fierz, C. (2002): A physical SNOWPACK model for the
Swiss avalanche warning, Part III: meteorological forcing, thin layer formation and evaluation.
In: Cold Regions Science and Technology 35, 2002, pp. 169-184.
- Leichtfried, A. (2005): Schneedeckenmodellierung Kühtai 2002/2003, Sensitivitätsstudien.
Magister Report University Innsbruck, Austria.
- Luciuk, M. (2010): Night Radiative Cooling. The effect of clouds and relative humidity.
- Meister, R. (1985): Density of New Snow and its Dependence of Air Temperature and Wind.
Workshop on the Correction of Precipitation Measurements, Zurich, Switzerland, pp. 73-79.
- Meløysund, V. (2010): Prediction of local snow loads on roofs. Doctoral Thesis, Norwegian
University of Science and Technology, Trondheim, Norway.
- Nielsen, A. (1988): Snow-Melting and Snow Loads on Glass Roofs. In: Proceedings of the 1.
International Conference on Snow Engineering Santa Barbara, USA, pp. 168-177.
- Nikolov, D., Wichura, B. (2009): Analysis of spatial and temporal distribution of wet snow events
in Germany. IWAIS XIII, Andermatt, Switzerland.
- Otsuka, K., Homma, Y. (1992): Removal of snow from membrane structures. In: Proceedings of
the 2. International Conference on Snow Engineering Santa Barbara, USA, pp. 263-274.
- Pertermann, I., Puthli, R., Ummenhofer, T. (2014): Temperatureinfluss auf Dachschneelasten von
Gewächshäusern – Abtauen von Schnee über wärmedurchlässigen Dacheindeckungen. In:
Stahlbau 83 (2014), No. 12, pp. 860-872.
- Pomeroy, J.W., Brun, E. (2012): Physical Properties of Snow. Accessed online
(www.insecc.utah.edu/-campbell/snowdynamics/reading/Pomeroy.pdf).
- Riche, F., Schneebeli, M. (2013): Thermal conductivity of snow measured by three independent
methods and anisotropy considerations. In: The Cryosphere 2013, 7, pp. 217-227.
- Sandvik, Rune (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.
- Schwind, W. (2009): Die neue Schneelastnorm DIN 1055-5, kritisch hinterfragt. In: Bautechnik
86, Berlin, Germany, pp. 22-29.
- Sturm, M., Holmgren, J., Konig, M., Morris, K. (1997): The thermal conductivity of seasonal
snow. In: Glaciology, 143, 43, pp. 26-41.
30
- Tantau, H.-J. (2012): Low energy greenhouse - method to analyse heat flux and PAR
transmittance. In: Landtechnik, Building and Planning, 67, No. 3, pp. 196-204.
- Taylor, D.A. (1980): Roof snow loads in Canada. In: Canadian Journal of Civil Engineering, Vol.
7, No. 1, pp. 1-18.
- Teutsch, C. (2009): Neuschneedichteanalyse in den Ostalpen. Magister Report, University
Innsbruck, Austria.
- Wern, L. (2015): Snödjup i Sverige 1904/05 – 2013/14. SMHI Meteorologi Nr. 158, Sweden.
- Yen, Y.-C. (1981): Review of the thermal properties of snow, ice and sea ice. CRREL Technical
Report 81-10, Cold Regions Research and Engineering Laboratoty, Hanover, New Hampshire,
USA.

Additional Literature: For properties of snow, snow physics, melt models


- Anderson, E. (1976): A Point Energy and Mass Balance Model of a Snow Cover. NOAA
Technical Report NWS 19, U.S. Department of Commerce, National Oceanic and Atmospheric
Administration, National Weather Service.
- Anderson, E. (2006): Snow accumulation and Ablation Model – SNOW-17, Background.
- Akbari, H., Hosseini, M. (2010): Heating Energy Penalties of Coo Roofs: The Effect of Snow
Accumulation on Roof: Heat Island Group, Concordia University Montreal, Canada.
- Aoki, K., Hattori, M, Ujiie, T. (1988): Snow melting by heating from the bottom surface. In:
Proceedings of the International Symposium on Cold Regions heat Transfer. University of
Alberta, Edmonton, sponsored by American Society of Mechanical Engineering 1987, pp. 189-
194.
- Asztalos, J. (2004): Ein Schnee- und Eisschmelzmodell für vergletscherte Einzugsgebiete.
Diplomarbeit Technische Universität Wien, Austria.
- Bergdahl, P., Martin, M. (1984): Emissivity of clear sky. In: Solar Energy, 32 (5), pp. 663-664.
- Blöschl, G., Kirnbauer, R. (1991): Point snowmelt models with different degrees of complexity
– internal processes. In: Journal of Hydrology 129, 1991, pp.127-147.
- Endrizzi, S. (2007): Snow cover modelling at a local and distributed scale over complex terrain.
Doctoral Thesis, Universita degli Studi di Trento, Trento, Italy.
- Förster, K. (2013): Detaillierte Nachbildung von Schneeprozessen in der hydrologischen
Modellierung: Dissertation, Technical University Hamburg, Germany.
- Fukami, H, Kojima, K., Aburakawa, H. (1985): The Extinction and Absorption of Solar Radiation
within a Snow Cover. In. Annals of Glaciology 6, 1985, pp. 118-122.
- Hockersmith, S.L. (2002): Experimental and computational investigation of snow melting on
heated horizontal surfaces. M.S. Thesis, Oklahoma State University, Stillwater, USA.
- Jordan, R. (1995): Effects of Capillary Discontinuities on Water Flow and Water Retention in
Layered Snow covers. In: Defence Science Journal, Vol. 45, No. 2; April 1995, pp. 79-91.
- Kimbar, G. (2015): Dynamics of snow melting on tents during possibly threatening precipitations.
In: Technical Transactions Civil Engineering 2-B/2015, pp. 433-440.
- Kuchment, L.S., Gelfan, A.N. (1996): The Determination of the Snowmelt Rate and the Melt
water Outflow from a Snowpack for Modelling River Runoff Generation. In: Journal of
Hydrology, May 1996, pp. 23-36.
- Liu, X., Rees, S.J., Spitler, J.D. (2006 a): Modelling snow melting on heated pavement surfaces.
Part I: Model development. In: Science Direct Applied Thermal Engineering 27 (2007), pp. 1115-
1124.
- Liu, X., Rees, S.J., Spitler, J.D. (2006 b): Modelling snow melting on heated pavement surfaces.
Part II: Experimental validation. In: Journal of Applied Thermal Engineering.
- Mensah, K, Choi, J.M. (2015): Peak and Annual Snow Load Pattern for Effective Snow Melting
System Design in Republic of Korea. In: International Journal of Air-Conditioning and
Refrigerating, Vol. 23, No. 4, 2015, 12 pages.
31
- Mensah, K, Choi, J.M. (2015): Review of technologies for snow melting systems. In: Journal of
Mechanical Science and Technology, 29 (12) 2015, pp. 5507-5521.
- Murphy, D.M., Koop, T. (2005): Review of the vapour pressure of ice and super cooled water for
atmospheric applications. In: Q.J.R. Meteorological Society 131, pp. 1539-1565.
- Nielsen, A., Claesson, J. (2011): Melting of snow on a roof. Mathematical Report. Göteborg:
Chalmers University of Technology, Sweden.
- Nielsen, A., Claesson, J. (2010): Snow melting and freezing on older townhouses. Aalborg
University, Denmark.
- Nuijten, A. (2014): Modelling heated pavements. NTU Norwegian University of Science and
Technology – Trondheim, Norway.
- Nuijten, A., Holand, K. V., Kasbergen, C., Scarpas, T. (2016): Modelling the thermal conductivity
of melting snow layers on heated pavements. In: 8th International Conference on Snow
Engineering, Nantes, France, pp. 269-275.
- Rees, S.J., Spitler, J.D., Xiao, X. (2002): Transient Analysis of snow-melting system
performance. In: ASHRAE Transactions 108(2), pp. 406-423.
- Richter, T. (2015): In German: Simulation zur Abbildung des thermischen Verhaltens von
Oberflächen am Beispiel einer beheizten Freifläche zur Schnee- und Eisfreihaltung. In:
Bauphysik-Kalender 2015, Verlag Ernst & Sohn, pp. 451-519.
- Rahmatmand, A., Harrison, J., Oosthuizen, P.H. (2014?): An improved method for predicting
snow melting on photovoltaic panels. Department of Mechanical and Materials Engineering,
Queens University, Kingston, Ontario, Canada.
- Zeinivand, H., De Smedt, F. (2010): Prediction of snowmelt floods with a distributed hydrological
model using physical snow mass and energy balance approach. In: Natural Hazards, Volume 54,
No. 2, August 2010, pp. 451-468.

You might also like