Publisher’s version / Version de l'éditeur:
Transactions of the Engineering Institute of Canada, 6, A-1, pp. 1-11, 1963-07-01
READ THESE TERMS AND CONDITIONS CAREFULLY BEFORE USING THIS WEBSITE. https://nrc-publications.canada.ca/eng/copyright
Vous avez des questions? Nous pouvons vous aider. Pour communiquer directement avec un auteur, consultez la première page de la revue dans laquelle son article a été publié afin de trouver ses coordonnées. Si vous n’arrivez pas à les repérer, communiquez avec nous à PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca.
Questions? Contact the NRC Publications Archive team at
PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca. If you wish to email the authors directly, please see the first page of the publication for their contact information.
NRC Publications Archive
Archives des publications du CNRC
This publication could be one of several versions: author’s original, accepted manuscript or the publisher’s version. / La version de cette publication peut être l’une des suivantes : la version prépublication de l’auteur, la version acceptée du manuscrit ou la version de l’éditeur.
Access and use of this website and the material on it are subject to the Terms and Conditions set forth at
Variations of snow loads on roofs
Peter, B. G.; Dalgliesh, W. A.; Schriever, W. R.
https://publications-cnrc.canada.ca/fra/droits
L’accès à ce site Web et l’utilisation de son contenu sont assujettis aux conditions présentées dans le site LISEZ CES CONDITIONS ATTENTIVEMENT AVANT D’UTILISER CE SITE WEB.
NRC Publications Record / Notice d'Archives des publications de CNRC:
https://nrc-publications.canada.ca/eng/view/object/?id=c382da3b-5df2-44d1-b1d6-49baf75d5c44 https://publications-cnrc.canada.ca/fra/voir/objet/?id=c382da3b-5df2-44d1-b1d6-49baf75d5c44Ser
TH1
N2lr2
no.
189
c .2
BLDG
NATIOlVAL RESEARCH COUNCIL
CANADA
DIVISION O F BUILDING RESEARCH
VARIATIONS OF SNOW LOADS
ON ROOFS
B. G . W .
Peter,
W .
A .
Dalgliesh and
W . R.
Schriever
B U U I N G RESEARCH
-
LIBRP,RY
-
RESEARCH PAPER
NO. 189
OF THE
DIVISION OF BUILDING RESEARCH
Reprint From
TRANSACTIONS OF THE ENGINEERING INSTITUTE O F CANADA Vol, 6, No. A
-
1, April 1963Paper No. EIC 6 3
-
BR & STR . 5PRICE
25 CENTSOTTAWA
This publication i s being distributed by the Division of Building Research of the Na- tional Research Council. It should not be reproduced in whole or i n part, without permission of the original publisher. The Division would be glad to be of assistance in obtaining such
permission.
Publications of the Division of Building Research may be obtained b y mailing the ap- propriate remittance, (a Bank, Express, or Post Office Money Order or a cheque made pay- able at par i n Ottawa, to the Receiver General of Canada, credit National Research Council) to the National Research Council, Ottawa. Stamps are not acceptable.
A coupon system has been introduced to make payments for publications relatively simple. Coupons are available in denominations of 5, 25 and 50 cents, and may be obtained by making a remittance as indicated above. These coupons may be used for the purchase of all National Research Council publications including specifications of the Canadian Gov- ernment Specifications Board.
VARIATIONS OF SNOW LOADS
ON ROOFS
B.
G.
W .
Peter,W .
A.
Dalglieshnnd
W .
R.
Schriever all of the Building Structt~res Section, Division of Ruilditig Reseclrch, Notion01 Rest~~utsch Council, Ottcliuc~, Cunudu.Loads and Strengths the result of several factors rather arcnn i n 195g1 where an unbalanced Even a cursory examination of than a single factor. An excellent, b u t snow 1o;ld c o m b ~ n e d wit!] severdl structural failures in buildings will tragic, illustration of this is found in structural weaknesses related t o de- reveal that failures are nearly always t h e collapse of t h e Listowel hockey sign, materials, workmanship a n d in-
. Recent advances in calcu-
tory remarks about loads and the fied bv experience. Frequently, all safety of structures in general may
be appropriate.
The aim of structural design is to provide a structure that will perform a given function for its owners, first with safety against collapse, and sec- ond, with adequate protection against deformation which would impair its service. The crucial question is what
, L
that one can say is that present figures have apparently worked in the past. A particularly important field ill dctermining livc loads is the load imposed by climate, such as snow, wind, fain and ice loads. Although thcse loads are amenable, at least in the form of the basic climatic infor- mation, to statistical anaylsis, it will
w o
W l
R l
R o
MAGNITUDE OF L O A D S AND STRENGTHS
Fig. 1. Frequency distribution curves of actual loads and strength';.
A
is "safety against collapse," because, although present design concepts in gencral do not envisage the possibility of failure, every structure has some (although
-
- very small) probability of > u z w w u 3 z 0 W W (L [L (L LL 2 failure.The probability of failure depends on factors which can be grouped into two main statistical variables: strengths and loads. Fig. 1 is a n idealized frequency distribution plot of the actual strengths (load-carrying capacities) and the actual maximum live loads experienced by a large number of identical structures during their lifetime. The zone near inter- section F where the tail of the load curve overlaps the tail of the strength curve indicates the probability of the load-carrying capacity of a given structure being exceeded by the load. If the structure is made stronger, the right-hand curve is merely moved further to the right and the probabil- ity of failure is reduced, but never to zero. In other words, there is never any absolute safety against collapse2.
STRENGTHS
R
(LOAD CARRYING I C A P A C I T I E S )
I
LOADSW
be realized that the actual loads pro-
2
' 0-
2~
J W (LF
-
--
-
-
-
I bduced on a structure are greatly nffectcd by factors independent of the basic climatic data, such as by thc shape and type of structures, for in- stance, shape factors for wind loads, and by surrounding structures and topography. The main purpose of this paper is to discuss some of thesc factors with reference to snow loads for the assistance of designers.
Importance of Snow Loads
The largest load to which roofs are subjected in Canada is usually the snow load. Consequently, the choicc of the magnitude and distribution of the design snow loads takes on an extremely important role, both in rela- tion to the probability of roof failures and the cost of roof construction considered on a national scale. The effect of a change of, say, 1 0 p.s.f. in design snow loads across Canada, if applied to all the roofs built would amount to many millions of dollars per year.
specified by mdst bui
n uniformly distributed based mainly on the observed maximum sno ground. Modifications merely for the slope of t
;I few other conditions. Even a ca
observer will have noticed, however, that snow covers on roofs differ sig- nificantly from the uniform snow cover on the ground. Wind, the shapc of roof, the shelter conditions due to adjacent structures, heat loss, ancl solar radiation, introduce important complications leading to certain pat- terns of snow accumulation. Thus it is quite possible that design snow loads in the past have, for certain roofs or parts of roofs, been in error by a factor of two or three.
The inaccuracy of design snow loads for certain roofs in the past has been recognized for some time by the Associate Committee on thc National Building Code of Canada. In the 1960 revision of the National Building Code3 some steps have been taken to arrive at a more realistic assessment of snow loads for various shapes of roofs. Further progress in this direction is anticipated as will be explained later.
Typical Snow Accumulations on Roofs
Some fairly definite patterns of snow accum~lations can- b e recog- nized on certain types of roofs. What the actual patterns will be depends on the architectural design as well as the location of the roof.
Some mention should be made here of the density of snow, since this will allow depths of snow to be visualized in terms of loads. Freshly fallen snow has a density of about 0.1 gm/cc (one tenth the density of water), but due to a change in the shape of the snow crystals the density increases with time so that the value most commonly found on roofs varies from about 0.18 to 0.3 gm/cc. In spring, or other periods of melting the density may leach about 0.4 grn/cc. In terms of a load in p.s.f. 1 ft. of snow will usu- ally represent a load of 10 to 15 p.s.f. and 1 ft. of wet snow will produce a load of about 20 p.s.f. Higher densi- ties d o occur occasionally, for in- stance, ice on roofs in areas of re- freezing melt water causes a load of about 50 p.s.f. for each foot of thick- ness.
Some of the more common patterns of snow accumulation will now bc discussed for various types of roofs.
Flat Roofs
Flat roofs, without any vertical projections from the roof such as parapets, penthouses or other higher
building parts, usually show a rather uniform snow cover. The amount of snow on these roofs, in relation to the ground load, varies collsiderably and d e p e i ~ l s mainly 011 thc degree that
the roof is sheltered from the wind. In heavily \vooded areas low flat roofs will accumulate loads equal to the snow load on the ground, but single-level flat roofs in exposed sites seldom accumulatc more than a few inches of snow.
Multi-Icvel flat roofs present a more complicated problem since the lower rqof's are particularly suscept- iblc to high snow loads, and on such roofs it is commoil to find drift loads that cxcced the ground load (Fig. 2). Drifts of 4 to 6 ft. in depth are not uncommon after a single snow storm. Not only the maximum depths but also the average depths are much greater on such roofs than on adjacent higher roofs.
Fig. 2. Example of heavy drift load on the lower of two adjacent flat roofs in Lethbridge (courtesy of R. E. Peacock, Lethbridge).
Ca~lopies, Lean-to's, etc.
Canopies, balconies, lean-to's and roofs located lower than and adjacent to a higher roof are often subjected to very high snow loads (Figs. 3 to 5). These high loads usually occur on the leeward sicle of the building, but since wind-producing drifting may blow from various directions, drifts can occur on any side of a building. The cross-section of these drifts is frequently triangular with the maxi- mum load adjacent to the wall. The depths of these drifts often exceed the depth of snow on the ground, but thc actual loads are difficult to pre- dict, sincc they depend ilot only 011
the difference in elevation of the two roofs but also on the c o ~ ~ r i b u t i ~ l g area of thc upper roof n~id thc wind tlirectio~~. When the projection of the higher roof above the lower roof is moclcrate thc drifts often accumulatc to the edge of thc upper roofs.
When the uppcr roof is s'opcd or curvcd to\\rartls thc lower roof a d drains onto thc lower roof, melt water may be aclded to the accumulatccl snow and, if rctainecl, increasc the load further. This mclt watcr oftci~ rcfrcczes and in a succession of
Fig. 3. Example of a snow drift over 10 ft. deep on a lean-to roof at Gander Airp01-t (courtesy H. M. Chafe, Gander).
storms, thaws and cold periods a build-up of a thick layer of ice occurs. In a case observed on a can- opy below an arch roof in Ottawa, an ice layer 2 6 in. deep (load of over 100 p.s.f.) extended out 1 2 ft. Fur- thermore, in the spring the snow from the upper roof may slide onto these lower roofs, causing impact as well as static loads. The impact effect is known to have contributed to fail- ure in a lower roof. I t is thus annarent
I L that depending on the size and con- dition of the upper roof, a low roof adjacent to a high roof should b e designed for a load considerably greater than the basic roof load.
Sloped Roofs
Two types of snow accumulatioi~s have been observed 011 sloped or
peaked roofs, again depending mainly on shelter conditions. Sloped roofs in sheltered areas collect fairly uniform loads (Fig. 6); sloped roofs in areas
exposed to the wind will either be relatively free of snow or have larger accumulations of snow on the lee- ward side of the roof. Sloped roofs with ridges parallel to the wind direc- tion are often swept bare during a storm, whereas those with t h e ridge at right angles to the wind collect large amounts of snow on the leeward slope.
The majority of sloped roofs have slopes less than 50". Observations show that, contrary to common be- lief, snow loads on sloped roofs are not significantly less than loads on flat roofs. Immediately after a snow storm sheltered flat roofs and sheltered sloped roofs have similar uniform depths of snow, while sloped roofs in exposed areas have more snow on their leeward slopes than any of the neighbouring flat roofs (Fig. 7). Fur- thermore, slide-off from non-metallic roof surfaces, seldom occurs on roof slopes of less than 40" to 50" (Fig. 8).
Fig. 4. Example of a triangular snow drift on a small residential canopy in Ottawa. Note deflection of outer stringer beam. Several of these canopies collapsed.
building, however, one may not be justified in allowing for the full re- duction in the snow load. An out- standing example of a high load on a sloped roof of an unheated building was observed at Glacier, B.C., where a load of 160 p.s.f. (equal to the ground load) occurred on a 38" roof (Fig. 9).
Curved Roofs
Curved roofs are not as common as sloped or flat roofs, yet they are often used in larger buildings such as arenas
usually accumulate to the top of the wall and cxtend about 10 to 1 5 ft. from the parapet wall. Beyond this distance, parapets do not provide much shelter from the wind. The drifts tend to occur not only on the windward side behind the parapet, but also along all the other parapets, producing a saucer-shaped snow sur- face.
Cl~irnne!j.s ancl Other Projections
The snow is usually swept away
Fig. 6. Example of deep uniform snow accumulation on a sloped roof in a sheltered location in Ottawa.
vious section have indicated that wind is the most important modifying factor in the consideration of snow load dis- tribution on roofs. The effect of wind therefore nceds to be examined more closely.
Snow storms are frequently accom- panied by wind which transports the snow horizontally as it falls. I n ad- dition, snow that has been deposited is often blown from one part of the roof to another either during o r after the snow storm. The way in which snow is deposited on (or eroded from) roofs will depend on the flow pattern of the wind around the building. "
To assist in predicting al.c:ls pro~lc to acci~mulatc snow, it m;ly be useful to consider in gencrtll tcrms thc dis- t i ~ ~ b n ~ l c c of tlle air flo\r~ c.reatet1 bv ;I
building. I n doing this it is pcrrnis- siblc to ;lssumc that thc \vind ; ~ p - p o ; ~ c h i n g :I building is n laminar,
strcumlinetl flow and t.onsitlcr as tur- bulence only thc ,~cldition;~l tlisturb- ~11cc c:~uscd by thc. building.
Fig. 11 sho\vs how the streamlines of air flow are dcflcctcd by u bllild-
' inz. Over the root and nrou11(1 thc
or aircraft hangars. Although rcla- tively few of these curved roofs havc been observed for snow loads, they deserve special attention, since snow loads accumulati~lg on these roots arc usually asymmetrical (Fig. 10). The deepest snow is likely to be observed on the lower area of the arch on one side only. Since these structurcs (bow- string trusses, arches) are oftcn quite susceptible to high stresses from un- balanced loads, this loading condition must be considered in the design cal-
from a narrow area around chimneys hind other small projections and drifts are piled up a few feet away. The sizc of these drifts depends 011 the
sizc of thc projection. Larger chimneys and l>cntllouses can cause significant accumulations.
sidves the streamlines are crowded to- gether, resulting in an increase in wind velocity to maintain t l ~ e same quantity rate of air flow past the building as in the undisturbed wind. The streamlines are approximately parallel to the windward roof slope, but n separation point occurs at the peak where the ill-erttia of flow makes
The Effect of Wind the streamlines "overshoot" the sharp The examples discussed in the pre- change in direction of the roof sur-
Fig. 7. Example of a heavy localized snow load on the west slope of a hip roof in an exposed location in Ottawa (photographed on the same day as Figure 6). cr~lations. Some of the
failures
thathave occiurcd in i,urvcd ~.oofs can, at least i l l part, be att~ibutcd to this ~ I I -
balanced 1o;ld.
Purc~pct Walls
Parapet walls, and all otl~cr vertit.;ll projections on a roof, tend to incrcasc or changc thc accumulation of snow. ?'ria~lgular d ~ i f t s along parapet \r,;~lls arc the type that occurs most trc- cluel~tly. 011 esposcd roofs these drifts
Fig. 8. ExanlpIe of snow clinging to steep
face. The wake \\~hich lies between the streamlined flow m ~ d the leeward roof and wall surfaces is a region of turbulence where the direction of flow is random. For sharp-edged buildings separation points usually occur along the edges. Major turbulent regions are thus formed over leeward slopes of pitched roofs, over the lower levels of split-level roofs to the leeward or windward of the higher level, and in triangular areas adjacent to parapet walls. Curved buildings are more complicated since the position of the separation point on a curved surface such as an arched roof varies with the wind velocity and the roughness of the surface.
Snow tends to "drop out" of the wind flow and accumulate in the tur- bulent \vakes where there is no strong sustained flow in any particular direc- tion. Whcre the streamlines are ad- jacent to tlic roof, ho\\~cvcr, the sur- face is swept clcar ol sno\\l because any snow that may fall on it is es- posed to thc direct and continuous
action of wind at high velocity.
asphalt shingle roof (over 45") in Ottawa.
T h e encrgy available in the wind to cause di?iting increases with the square of the velocity. Snow normally begins drifting at wind speeds of 8 to 10 m.p.h., although this figure may vary with the consistency of the snow. Wind spccd increases with height above ground, so that higher roofs
experience wind effects in greaccr measure.
Obviously, the
prevailing
wind direction is an important factor in lxedicting drifting conditions. In some areas, particularly in Eastern Canada, the winds frequently come from the East during storms, but change to a westerly dilection after the storm. In most locations, howcvcr, all \vind directions must b e con-sidered.
A final consideration is the ratio of width to heipht of the buildinp. Wind
-
-
is more likely to pass over the roof rather than around the ends of a low, broad building, whereas for a tall, narrow structure, a greater part of the wind will pass around t h e sides. By taking account of these factors, i t may be possible for designers to pre- dict at least in a general way t h e most likely patterns for snow drifting for a given situation.Fig. 10. Removal of snow from the north side of the small remaining portion of the curved roof of Listowel Arena after collapse on 28 February 1959. The snow cover aloilg the north eave was 2% feet deep and weighed up to 80 p.s.f. while there was very little snow on the south side and the upper parts of the roof. The snow on the ground was 2 feet deep and weighed approximately 40 p.s.f. (courtesy M. Rice, Listowel).
Fig. 9. Exan~ple of deep snow accu~nulation (absence of slide-off) on steep wood shingle roof (38') in sheltered location in Glacier, B.C. (roof load 160 p.s.f.; ground
load 160 P . s . ~ . ) . Other Effects on Snow Accun~ulation
Heat Loss
The melting of snow on roofs due to heat loss will take place only if the roof and ceiling system is poorly in- sulated. And even if the heat loss is sufficient to cause melting of t h e snow on the roof, there must be, in order to reduce thc load, sufficient drainage to rcmove the melt water. Flat roofs usually do not provide adequate drainage to remove melt water from the \vet snow during mild winter days. 0 1 1 the othcr hand sloped roofs
ordinarily do provide sufficient drain-
age.
A special load condition call be caused however, on sloped roofs with large overhanging eaves. Melt water drainiilg from the warmer upper parts of the roof, may refreeze on t h e over- hanging eaves, which are colder. This lwoduces a build-up of ice at the caves, causing high loads. In additionCOMPRESSED STREAMLINES both the siiom 011 the ground and thc
snow on the roof are affected by solar radiation. Consequently no fur- ther adjustment seems justified, al- though some roofs that slope to the south, may experience some load re- TURBULENT AREA duction from solar radiation.
DBR Snow Load Survey
In view of the importance of the problems involved in the proper as- sessment of snow loads on roofs, the Associate Committee on the National Building Code recommended a de- tailed study b e undertaken. In 1956 the Division of Building Research be-
t gan a country-wide survey of snow
Fig. 11. Schematic presentation of air flow around a small building, showing areas of "compressed" streamlines (snow removal) and turbulence (snow deposi- tion).
leakage problems are encountered under such conditions on shingled roofs when the water backs up behind these ice dams.
T h e reduction of snow loads by heat loss is insignificant at very low temperatures. Furthermore it is a slow process, and although it is sometimes effective in reducing accumulations of snow over the winter, it does not affect the loads caused by a single storm. Frequently, single storms cause the large drifts near projectipns and on lower levels, and consequently such drift loads are important even on roofs with large heat losses. Solar Radiation
Solar radiation is not generally an important factor in modifying snow- loads on roofs. Since only about 10% of the energy available from solar radiation is absorbed by clean snow, this energy is sufficient to melt the snow only at the higher temperatures of the late winter. Observations in the Ottawa and Edmonton areas show that for a six-week ~ e r i o d without new snowfalls the ro& loads did not decrease appreciably. Furthermore, ground snow loads are used as the basis for determining roof loads and
roads on roofs to obtain more factual information of these loads and to pro- duce further refinements of the snow load specifications in the National Building Code. This survey will be continued for a few more winters to lend the necessary statistical strength to the findings.
Observations are made at various stations which can b e grouped into three categories, A, B and C Stations. The major stations are the 1 8 A Sta- tions which range from Vancouver to Gander and from Inuvik to Toronto. At these stations snow depths and densities are measured on three or more roofs each week and after every snow storm (Fig. 12). The roofs gener- ally include one sloped roof, one flat roof of residential size, and one other larger flat roof. Snow depth on the roof is measured at several predeter- mined positions and on the surround- ing ground. The density of the snow is also measured both on the roof and on the ground. At the eight C Stations (RCAF Stations) similar ab- servations are made by Air Force personnel on large flat and curved roofs.
At B Stations occasional observa- tions only are made of unusual ac- cumulations. These B-Station measure- ments (depth only) are usually made by building inspectors and at present more than 500 building inspectors have been asked to assist in the sur- vey. I t is hoped that measurements
Fig. 12. Observer taking snow density samples on roof for DBR snow load survey,
from B Stations will yield useful ill- formation in providing case records of large snow loads.
On the basis of the prelimiilary re- sults of this survey a uumber of im- provements were made in the treat- ment of suow loads in the 1960 edition of the National Building Code. Further refinements will no doubt be possible when the survey is completed and all the results havc been analysed.
Treatment of Snow Loads in Building Codes
Canada
In the National Building Codc of Canada snow loads are dealt with in two separate places: (a) in Section 4.1, "Structural Loads and Proce- dures" where the roof load applicable to one given locality and its modifica- tions for different roof shapes are given, and (b) in Supplement No. 1 (Climate) to the Code which contains snow load information for all of Canada. This "geographic7' snow load information is given in two forms, as a map of ground snow loads for Canada (Chart 7 of Supplement No. 1, see Fig. 13) and as a list of ground loads and roof loads for about 200 selected municipalities across Canada (Table "Design Data for Selected Municipalities", p. 33 to 36 of Supple- ment No. 1).
Grotlnd Loads-Grouncl loads are based on observations of snow depths
011 the ground taken a t more than
200 stations across Canada for periods ranging from 10 to 1 8 years. These records were made available by the Meteorological Branch of the Depart- ment of Transport. From these records the probable 30-year snow depths (depths that will be equalled or ex- ceeded on the average once in 30 years) were calculated."ecause the records have all been collected in populated areas, the values for certain regions (for example, B.C.), apply to the valleys only and may be too low for higher mountain areas. I n such cases, local weather observers may be able to give advice on conditions in the mountai~ls.
Ground snow depths were con- verted to ground loads by assuming a specific gravity for snow of allout 0.2 gm/cc corresponding to a load of 12 p.s.f. for each foot of snow. An addition was made for rain being nb- sorbecl by the snow and thus iucrcas- ing the load.
Roof Loncls-Tllc Insic roof load was set at 0.8 of t11c gralmcl load by the Associate Comrnittec on the National Building Code, based 011 the
results of the snow load survey to take into account the fact that the
majority of roofs experience lower average snow loads than the ground. At the same time some new modifica- tions to allow for exposure and shapc of roof were made as follows (see ar- ticles 4.1.2.8 to 4.1.2.11 of Section 4.1 of the Code).
The basic design roof load is to bc applied uniformly over the entire roof. Non-uniform loading, consisting of half the design load over some parts and full design load on the other parts of the roof, must also be considered, however, if this would cause greater stress in the roof members. Provision is made for unbalanced loading on pitched and curved roofs by requiring the roof to be designed for zero load on one side and 1.25 times the basic roof load on the other as well as for uniform load. Reductions of snow load for sloped roofs begin at slopes of 30" and proceed in increasing steps to slopes of 70°, after which the snow load is zero. Increased loads 1.5 times the basic roof load are specified for lower levels of multi-level roofs (in- cluding canopies, marquees, and porch roofs) to be applied over an area equal in width to three times the dif- ference of elevation between the two levels but not more than 15 ft.
Codes in Other Countries
Many other countries have also had
to deal with snow loads in their build- ing codes. Some parts of these codes are similar to the National Building Code of Canada in their treatment of snow loads, but others contain pro- visions or procedures that deserve examination with a view to possible future adaptation to Canadian con- ditions.
The American Standards Associa- tion map "f snow loads for the United States is based partly on ground snow depth measurements and partly on snowfall measurements. Whereas in Canada, the loads are based on a probability of one in 30 for any winter, those of the ASA are based on a probability of one in 10. A minimum roof load of 20 p.s.f. for flat roofs is specified in the U.S. in areas of light snow (or rain loads) to cover incidental loadings during con- struction or repairs, or possible ice loads resulting from poor drainage of melt water.
The mountainous regions of Europe present a special problem. In the Alps, the main consideration for depth of snowfall is the altitude of the loca- tion. The snow load codes of France 6 ,
and Switzerland 7 therefore relate the
design snow load to the altitude. In Switzerland, for example, the mini- mum design load of 18 p.s.f. (- 90 kg/m2) is increased in proportion to
the square of the altitude so that at 2600 feet (800 m) the snow load has reached 50 p.s.f. (-280 kg/m2). This procedure might be followed for some regions in Canada such as the mountainous regions of Alberta and British Columbia.
France has three basic snow load zones for those areas below 200 m in altitude. For higher elevations there is a linear increase in load with altitude. The French code also recommends an additional load (20 p.s.f. 100 kg/m2) for flat roofs where poor drainage of melt water and possible refreezing might cause increased load.
Snow load zones in Japan 8 are re-
stricted mainly to the mountain re- gions, where the loads may reach values as high as 240 p.s.f. A unique feature of the Japanese code is that reduced long-term loads (equal to 70% of the short-term loads) are taken into consideration, Considerable re- search on snow loads on buildings has been carried on in Japan with the re- sult that a number of provisions for reductions and modification in roof loads, depending on such factors as wind, shape, and heat loss, are quite detailed. For example, the snow load reductions for a windy location de- pend on the wind velocity, and sepa- rate values apply for the two sides of a pitched roof.
L O A D I N G
I
2
2
L O A D I N G
2
FOR LOADING 2:
but not greater than
4 for trusses and beams in roofs of less t h u ~ 30 Ib/ft2 weight.
c, = 1 . 5 [ 1
+
0.6((1 h , ) ] 2.5 for trusses and beams in roofs of more than 30 Ib/ft' weight.c, = 1.5[1
+
0.4((1;1?,)] 2.5 for roof slabs of spans greater than 20 ft. ant1 rafters independent of span.2.0 for roof slabs of spans less than 20 ft.
Fig. 14. Example of "shape factor" diagram for snow loads from t k Russian "Instruction for the Determination of Snow Loads on Roofs of Buildings, SN69-59, published by the State Committee of the Soviet Ministries of the U.S.S.R. for Construction, Moscow, 1960.
Some of thc most detailed snow Conclusion
loatl specifications of any country Snow loads arc the most important arc those adopted by the U.S.S.R. dcsign loads for roofs in Canatla. 1960. several tables show one or D c s i p values for snow loads are spe- design loadings for various tyl,cs of cified in many building codes, inclucl- ing thc National Building Code of roofs. Extensive use is made of cocf-
Canada. It sllould be remembered, ficients for designating the intensities however, that the design loads givcll of loadings based mainly on the in the National Building Code are shapes of roofs. A11 example is given minimum loads and must not be con- in Fig. 14. sidered by the designer as the greatest
load that could ever occur. Rathcr, ~t is the load that, on the average, will be equalled or exceeded once in thirty ycars. During any winter it is possible that a structure may experience a greatcr load, say the 100-year masi- mum. Furthermore, and quite apart from the statistical probability of ex- periencing greater loads, the actual loads on various parts of a roof de- end on m a w factors and are often
I J
significantly increased over t h e mini- mum loads due to exposure of thc structure to the wind and t h e shape of the roof. While a casual observer may never have noticed a snow load on a roof that reached the design value, the rarity of extreme loads does not provide any proof that t h e design tigures are unjustified and that thc design loads could not have occurred or been exceeded.
A proper understanding of the clauses in thc Code by which the basic roof snow load is modified to account for exposure and shape is csscntial tor a designer. Since each design problcm brings together a tresh combinatio~l of the various fac- tors, thc detailed assessment must be made by thc designer and cannot bc made by thc Code. Witll a ge~leral ~ i n d ~ r s t a n d i n g of the phenomena, he should consider the conditions and make the best possible prediction of thc probable magnitude and distribu- tion of the snow loatl on the roof to l)c designed.
This paper is a contribution from thc Division of Building Research, National Research Council and is l)ublishcd with thc approval of thc Ilirc~rto~ of thc Division.
. ..
63i2j-
5. American standard building code re-
quirements for minimum design loads in buildings and other structures.
American Standards Association, A58. 1-1955, New York, NY.. 1955.
6. R-les dbfinissant les effets de la neige ei du vent sur les consfruciions. Minis- tgre de la Reconstruction e t de l'Urba- &m&-paris. 1947.
7. N o h e n fiir die Belastungsannahem, die
Inbetriebnahme und die Uberwachung der Bauten. (Standaids f o r Load As- sumtions, Acceptance a,nd Inspection of Structures.) Schweizerischer Ingenieur und Architekten Verein (Swiss Associa- tion of Eneineers and Architects). No.
Japanese).
9. Instructions for the determination of s l o w loads on roofs of buildings. SN69-59, published b y the State Com- mittee of t h e Soviet ,Ministries of the U.S.S.R. for Construction. Moscow. 1960.
DISCUSSION
Discussion by A. G . Davenport,
University of Western Ontario. There are a great many reasons f o r supposing that some of the greatest strides in structural engineering that will be made in the next few years will be in the realm of the improved matching of design loads to the actual loads structures have to withstand. At prescnt our knowledge of the real loading conditions lags far behind our ability to assess a structure's response to loads. I n this respect the authors have outlined work which is likely to prove of grcat benefit to structural engineers working in snowbou~ld climates.
All the studies so far have bee11 conducted in the field. The writer wonders whether consideration has been given to the possibility of model :studies in artificial snow storms. Al- though a difficult problem to tackle due to the large number of variables,
model studies provide a quick way to gain much experience and feel for this subject at small cost.
Author's Reply
Professor Davenport has raised a very good ~ o i n t . Model studies are often
u
resorted to, first, because full-scale observations are too difficult or pro- hibitively expensive; this is, however, not particularly true of snowload ob- servations. Secondly model research permits control of conditions such that individual factors may be considered separately. As Prof. Davenport has indicated, there are a great many variables in snow loads on roofs, and model studies might well be employ- ed in a more detailed investigation of their separate influences.
The authors are aware that model studies in wind tunnels or water flumes have been employed elsewhere to i n v e s t i g a t e c e r t a i n s n o w d r i f t i n g patterns around buildings. Professor Strom of New York University used a large wind tunnel to assist in the planning and layout of buildings in
the Arctic. Professor Theakston of the Ontario Agricultural College has ob- tained interesting results by which he has been able to give advice on effec- tive location of farm buildings, wind breaks and fences to avoid unwanted drifting. It is quite possible that model research might also prove helpful in the prediction of location and depth of drifts on roofs, yielding "shape factors" similar to those now in use for wind loads.
Field information is, of course, in- dispensable when dealing with snow loads on roofs, particularly with regard to the statistical approach that must be taken to loads, especially "climate loads". This is also the case because of the variety of snow types (e.p. wet or dry) which lead to different forms of drifts. Furthermore, since field ob- servations take a number of winters to allow reliable conclusions, it was im- portant to start such observations as soon as the problem was recognized. The authors consider that field observ- ations have served to give a general "feel" for the subject as well as quanti- tative results in an economical fashion.
