119
Uplift on rooftop walkunays and cable trays
Abstract: A code review is presented,along with current state-of-knowledgeand in situ observations,in order to gain
a better understanding of the effect of wind-induced uplift on timber walkways and metal cable trays located on
rooftops.
Key words: wind uplift, pavers, small structural elements, timber walkways, metal cable trays, pressure equalization.
R6sum6 : Une 6tude de code est pr6sent6e,ainsi que l'6tat actuel de connaissances,et des observationsin-situ, afin
d'acqu6rir une meilleure compr6hensionde l'effet de souldvementprovoqu6 par le vent de passerellesen bois et de
plateaux de cdbles en m6tal situ6s au sommets des toitures.
Mots clds : souldvementpar Ie vent, paveurs, petits 6l6ments structuraux,passerellesen bois, plateaux de cdbles en
m6ta1,6galisation de presslon.
[Traduit par la R6daction]
Introduction
Cellular phone technology has been with us for decades,
but it was not until the late 1980s that its popularity spread
to the general public. As a result of the increaseddemand
for mobile telephone service, a proliferation of cell sites has
resulted.
In a mobile telephone network, signals are transmitted to
and from wireless phone units through antennas,located on
existing buildings or on freestanding towers. The antenna
signals are controlled from nearby switching equipment. In
residential and commercial areas, the cell sites are generally
located in medium to high-rise buildings. The antennasare
roof mounted,and the equipmentis housedin unusedspaces
within the basementor roof penthouse.If such spaceis unavailable, then the equipment is housed in a shelter (either
site fabricated or prefabricated) located in an appropriate
area on the roof itself.
Other elements that commonly form part of the wireless
telephone network infrastructure, for cell sites on existing
buildings, are walkways and cable trays. The walkways provide a level surface for pedestrian traffic. These walkways
are usually 1200 mm (4 ft) wide with continuous handrails,
and are made of pressure-treated timber, assembled with
standard nailing. The cable trays protect otherwise exposed
antennacables, and typically are fabricated from heavy gage
Received March 18, 1998.
Revised manuscript acceptedAugust 26,1998.
A. Mattacchione and L. Mattacchione. Prosum Engineering
Ltd., 56 Goddard Street, North York, ON M3H 5E2, Canada.
Written discussionof this note is welcomed and will be
received by the Editor until July 31, 1999 (addressinside
front cover).
Ca n. J . Civ . E n g . 2 6 : I l 9 - 1 2 2 ( 19 9 9 )
galvanizedmetal, connectedwith self tapping screws.They
are about 75 mm (3 in.) deep and 200-300 mm (8-12 in.)
wide, and rest on short timber spreaders located at about
1200 mm (4 ft) centres.
Timber walkways have a dead load of approximately
0.48 kPa (10 psf). Metal cable trays have a dead load of
0.34 kPa (7 psf) and O.62 kPa (13 psf) for the small and
large trays respectively.Becauseof the relatively light dead
loads of theseelements,and their exposureto rooftop winds,
their susceptibility to wind uplift has been questioned, especially when the noted dead loads are comparedwith the uplift forces generated at the top of tall, flat-roofed buildings,
using the simple procedure delineated in the National Building Code of Canada (NRCC 1995).
On some sites, the authors have observedthese elements
having been mechanically anchoredto the roof, or ballasted
to increasethe dead load. But, are such efforts (l) required
by code, (li) suppofiedby our cuffent state-of-knowledgeof
behaviour of elementson roofs exposed to wind loads, and
(iil) justified by observationsof in situ conditions?
Gode
For the simple procedure given in the National Building
Code of Canada(NBCC) for calculating wind forces, the design wind load would be
tll
p = qC"CrC,
where p is the specified external pressure; q is the reference
velocity pressure;C" is the exposurefactor; Cn is the gust effect factor; and Co is the external pressure cbefficient. For
most high-rise buildings in Ontario, these parametershave
the following values:
l 2l
q(l l l })
= 0.39 kP a
@ 1999 NRC Canada
C an.J. C i v.E ng.V ol.26, 1999
120
where q(1/10) is the referencevelocity pressurebased on a
probability of being exceeded in any one year of 1 in 10;
t3l
C" = 1 .3
for buildings from 10 to 15 storeYs;
l4l
Ce = 2.5
t5 l
c o = - 1. 0
(or higher along roof edges and at corners). Thus, eq. [1]
gives
llal
p = qC" Cr Cr= 0 .3 9 x 1 .3 x 2 .5 x (-1 .0 )
= -1.27 kPa (-26.5 psf)
which exceedsthe dead load of the walkways or cable trays.
Yet the NBCC makes a distinction between the forces on
a building and those on small elements' When discussing
wind on a building, arlicle 4.1.8 addressesthe effect of wind
loads on all or part of a surface of a building. The NBCC
user's guide (NRCC 1996) not only provides more in-depth
commentary but, in the Wind Loads section, also includes
data for pressure or combined pressure and gust coefficients
for (i) low- to high-rise buildings and (ii) a variety of other
structures and elements, including one for structural members, single and assembledsections.
The data reveal that for rectangular sections subjected to a
horizontal wind normal to the axis of the element, no uplift
occurs (for sectionsin free flow).
Finally, for protected membrane roofs, the NBCC user's
guide indicates that unbonded insulation in an inverled roof
system is not subjectedto the same uplift pressureas is applied through the depth of the entire roof assembly, because
of air leakage and partial pressure equalization between the
top and bottom of the insulation boards.
Therefore from the stand point of a code, wind forces on
buildings and those on small elements are not necessarily
the same. Thus, uplift for which the roof of a building must
be designed may not apply to smaller members located on
the roof.
Gurrent state-of-knowledge
The present knowledge of uplift forces on small elements
located on rooftops is derived mostly from studies done on
loose-laid insulation ballasted either by gravel or by concrete pavers. Notwithstanding the scope of the studies, they
provide information that may be applicable to other small
rooftop elements.
In one of the two works referenced by the NBCC Structural Commentaries in its discussion of pressure equalization, Kind and Wardlaw (1976) studied a variety of factors
influencing the susceptibility of gravel blowoff from wind.
One of the factors considered was the effect that the use of
paving slabs had on raising critical wind speed for blowoff.
The paving slabs replace the roofing gravel where the aerodynamic forces on stoneswould otherwise be most intense.
In the second work referenced by the NBCC Structural
Commentaries, Kind and Wardlaw (1919) found that static
pressurepatterns under paver arrays or under roof-insulation
iystems were similar to those on the exterior surface of the
rooftop, as a result of air entering and leaving the area beneath the arrays via the joints between the elements.
In the follow-up study, Kind et al. (1984) confirmed that
static pressures underneath boards or pavers tended to become equal to those on exterior surfaces because of air flow
through the joints between boards or pavers' and that gust
speed at rooftop level, rather than mean speed,was the pertinent value for use in assessingthe resistanceof the roofing
systemsto wind damage.
In another study to correlate the wind speed at which
loose-laid insulation and concrete paving slabs dislodged
(Kind et al. 1988), a prediction schemewas developed,but
the authors cautioned that the correlation was based on a
limited body of experimental data and thus applied only to
relatively simple building configurations.
More recently, McDonald et al. (1995) carried out fullscale wind load measurementson concrete paving slabs, to
serve as a benchmark for calibrating wind tunnel numerical
results. They reported that wind loading on a paver depends
on the pressure distribution on both top and bottom of the
paver; and if there was sufficient permeability, the response
of pressureunderneath a paver to pressure change above the
paver was almost instantaneous.As a result, they found that
the net wind load on a paver may be significantly lower than
that on an impermeable roof surface, because of the existence of a pressure field underneath the pavers which instantly responded to the pressure change on the top of the
paver. They pointed out the importance of leaving proper
gaps between and underneath pavers to allow pressure
equalization between top and bottom of pavers.
Based on their thorough literature review, as well as their
own full-scale tests, McDonald et al. concludedthat the net
pressure (difference between pressure on top and underneath
the paver) was not necessarilyupward, and was generally
downward on pavers at the corner of a roof. Pavers located
away from corners may not have net uplift pressures, depending on paver location and direction of the wind (largest
net uplift occurs along the edge of the wall)' There was an
insignificant time lag between pressure fluctuations on tops
and bottomsqf pavers.
Observations of in situ conditions
Over the past decade, the authors have been involved in
numerous mobile telephone sector projects, in areasthroughout Ontario. Of these, many required rooftop cable trays and
timber walkways.
As indicated earlier, the uplift force on the top of tall, flar
roofed buildings, basedon the simple procedure,exceedsthe
inherent dead load of the timber walkways or of the cable
tray and cables. Therefore it would appear to present a problem. However, no problems associated with movement of
the cable trays or timber walkways have been reported, in
spite of the cell sites being visited weekly by a technician to
monitor the functioning of the switching equipment and to
report any out of the ordinary items observed. This regular
and ongoing monitor of the behaviour of the assemblieshas
yet to reveal any problem associatedwith uplift.
In addition, over the years, the authors have observed the
followins:
@ 1999 NRC Canada
121
Note
since 1985'
Tabte 1..Uplift forces correspondingto highest maximum wind gust speed in each month
Date
Gust wind
velocity (kn/h)
Wind
direction
Jan. 1996
Feb. 199'7
Mar. r996
Apr. r996
May 1990
June 1990$
July t99l
Aug. 1990
Sept. r995
Oct. 1989
Nov. r995
Dec. 1996n
109
97
97
93
93
107
98
83
89
104
106
109
SW
w
NE
w
w
wNw
ENE
NM
w
NW
w
SSW
Mean hourly wind velocitY
(km/h)
(m/s)
68.9
61.3
6r.3
58.8
58.8
67.7
62.0
52.5
56.3
65.8
67.0
68.9
19.1
r7.0
r7.0
IO.J
16.3
18.8
t7.2
14.6
15.6
18.3
18.6
19.1
Pressure q
(kPa)r
UPlift P
(kPa)+
0.24
0.19
0.19
0.17
0.17
0.23
0.19
0.14
0.16
0.22
0.22
0.24
4.7',7
-0.61
-0.61
-0.56
-0.56
-0.74
4.62
-0.45
-0.51
-0.70
4."/3
-0.7'7
Notes: Wind velocities were recorded at Toronto Pearson International Airpoft.
'Mean hourly wind velocity = gust velocity / r/2.5.
tPressure,q'(kPa) = 0.0006464-5x (mean irourly wind velocity (m/s))'?'
,Upliit.
p tt<Pat= qC.C.Co.
SHighestrecorded in June in the past 58 years'
Highest recorded in December in the past 59 years'
(a) unballasted/unanchored timber walkways on rooftops,
which have begun to rot and warp after 20+ years of
service.but which have not moved:
(b) unballasted/unanchored cable trays, which have not
moved after two decades;
(c) 2 ft x 2 ft x 2 in. concrete pavers used on rooftops,
which remain in place (and which are used along edges
and corners where the NBCC indicates that uplift forces
are higher still); and
(d) loose-iaid rigid insulation on protected membrane roofs,
ballasted with 50-75 mm (2-3 in.) of stone, which remain in place.
Interestingly, none of these items possessesa dead load equal
to the uplift calculated earlier, let alone a dead load equal to
or greaterthan twice the uplift force, as required by the code'
Over the past decade,high winds have been recorded at
the Toronto Pearson International Airport by Environment
Canada(Table 1), some of which have been among the highest recorded for their respective months, with winds gusts
reaching a velocity of up to 109 km/h. This is noteworthy
because
(a) a gust velocity of 109 km./h translates to a mean hourly
wind velocity of 69 km/h (gust velocity = mean hourly
velocity x r/2.5);
(b) a mean hourly wind velocity of 69 km/h (19.2 m/s) exerts a pressureof 0.24 kPa; and
(c) the uplift force from this pressure is
I Lb|
p = qC" Cr Cp= O .2 4 x 1 .3 x 2 .5 x (-1 ' 0 )
= -0.78 kPa (-16.2 Psf)
It is worth recognizing that wind speedsin urban environments are affected by factors such as upstream terrain conditions, surrounding buildings, and the geometry of the
specific building under question. As such, winds from
weather stations at airports (which are typically flat and
open areas)may not be indicative of corresponding winds at
oi around a given building located away from the weather
station. Nevertheless, considering the velocity of the winds
which have been recorded at the weather station, it is likely
that during the life of these installations (some of which are
on 3O-year-old buildings of up to 20 storeys in height)' the
walkways and cable trays have undergone and sustained
high winds, with no ill effects.
There has been one recorded incident of uplift on a rooftop timber track (Toronto Star 1982). It was located on the
exierior podium roof level (3rd floor) at Toronto City Hall.
On 28 December 1982, a portion of the timber jogging track
was lifted by an episode of high winds. This incident has often been cited as reason for justification of the use of the
NBCC simple procedure for calculating wind forces on
small unanchored rooftop elements. But research into the incident indicates that this may not be warranted.
The Pearson International Airport weather station recorded wind gusts of 104 km/h that day, which was the
highest on record for the month of December (not having
been exceededuntil December I,1996, by 109 km/h gusts)'
An investigation inio the incident was ordered (Globe and
Mail 1982), and council authorized funds for a wind study
(City of Toronto Executive Committee 1984), whose findings showed that neither the shape nor the orientation of the
City Hall building lead to any unusual or undesirable aerodynamic effects in public areas. While wind conditions on
the podium roof were more severethan those experienced on
the bivic Square,this was not consideredunusual as winds
generally increase with height above ground (City Services
Committee 1986).
But the details of the running track itself may be more impofiant than the severity of the winds that day. The newspap". report relates the then property commissioner's
description of the construction of the 5-ft wide (1500 mm)
3/+
track is two-by-four (39 x 89 mm) framing overlaid with
in. (20 mm) plywood.
Discussion
The NBCC makes a distinction between forces on buildings -and those on small structural elements. It also recog@ 1999 NRC Canada
Can.J, Civ.Eng.Vol.26, 1999
tal
nizes the phenomenon of pressure equalization, at least with
respectto wind-induced uplift forces on loose-laidinsulation
on protected membrane roofing.
Current state-of-knowledge with respect to wind-induced
forces on concrete pavers tells us that the responseof pressure underneath a paver to a pressurechange above is almost
instantaneous,with the effect being that the net uplift force
on a paver is significantly less than that on an impermeable
surface.Also, gaps and holes, which would minimize resistance to air flow, would increase the effect of pressure
equalization.
Observations of in situ conditions reveal that unanchored/unballasted timber walkways and cable trays have
not displayed a susceptibility to uplift over a period of 20+
years, in spite of their low dead loads and the presenceof
numerous recent high winds in the past decade. The only
mitigating factor the authors can identify is that the elements
are constructed in a fashion to allow the free flow of air
fully around them, thus allowing for the phenomenon of
pressure equalization which would act to decrease (if not
eliminate) the wind-induced uplift on these elements.
Finally, the only case of an unanchored timber track being
subjected to uplift occurred where continuous plywood was
used as the surfacing material, where pressure equalization
would not have been possible.
Gonclusions
Research has shown that the uplift forces that the NBCC
describes for the tops of flat-roofed buildings do not translate into net uplift forces on smaller elements such as looselaid insulation or concrete pavers. These elements allow for
sufficient air flow around and under them, to enable pressure
equalization to occur, thus either reducing or eliminating net
uplift pressure.
Likewise, timber walkways and cable trays are constructed in a manner that allows free air flow around and under them. The phenomenon of pressure equalization, which
has been observed and documented in a number of wind tunnel and one full-scale study, may well be occuruing on these
elements. This appears to be the only reason that rooftop
timber walkways and metal cable trays have displayed the
stability observed of in situ conditions over the past two decades.
Consequently,for elements that are located on the roofs of
flat-topped medium to high-rise buildings and are built such
that pressure equalization may be acting, a review of the applicability of the uplift forces given by the NBCC simple
procedure (NRCC 1995) appearswarranted.
Acknowledgments
Assistance in locating and providing background information and reference material was provided by Don Taylor and
Mike Savage of the National ResearchCouncil of Canada.
Alexandra Radecki of Environment Canada supplied historical wind data.
References
City of Toronto Executive Committee. 1984. Report No. 12, Appendix A, Item 44, April 16, Toronto, Ont.
City Services Committee. 1986. Report No. 11, Appendix A, Item
5, July 14, Toronto, Ont.
Dyrbye, C., and Hansen, S.O. 1996. Wind loads on structures.
John Wiley & Sons Ltd., Chichester,England.
Globe and Mail. 1982. Track put on ice while city probes wind
accident. December 30, Toronto, Ont., p.1.
Kind, R.J., and Warlaw, R.L. 1976. Design of rooftops against
gravel blow-off. National Aeronautical Establishment, National
Research Council of Canada, Ottawa, Ont., September, NRC
15544.
Kind, R.J., and Warlaw, R.L. 1979. Model studies of the wind resistance of two loose-laid roof-insulation systems. National
Aeronautical Establishment. National Research Council of
Canada, Ottawa, Ont., May, Laboratory Technical Report LTRLA-235.
Kind, R.J., Savage,M.G., and Warlaw, R.L. 1984. Furlher model
studies of the wind resistanceof two loose-laid roof-insulation
systems (high rise buildings). National Aeronautical Establishment, National Research Council of Canada, Ottawa, Ont.,
April, Laboratory Technical Report LTR-LA-269.
Kind, R.J., Savage, M.G., and Warlaw, R.L. 1988. Prediction of
wind-induced failure of loose laid roof cladding systems. Journal of Wind Engineering and Industrial Aerodynamics, 29: 29uc"oonald, J.R., Wand, W., and Smith, D.A. 1995. Field experiments for wind loads on pavers. Proceedings of the 9th International Conference on Wind Engineering, New Delhi, India, pp.
488499.
NRCC. 1995. National building code of Canada. Canadian Commission on Building and Fire Codes,National ResearchCouncil
of Canada. Ottawa. Ont.
NRCC. 1996. User's guide - NBCC 1995 structural commentaries (part 4). CanadianCommission on Building and Fire Codes,
National Research Council of Canada, Ottawa, Ont.
Toronto Star. 1982. High winds blow family of 4 18 feet off deck
at City Hall. December 29, Toronto, Ont., pp. A1 and ,A4.
@ 1999 NRC Canada