In-Situ Measurement of Wind Performance of Roof Edge Systems

3 2 n d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma rc h 1 6 – 2 1 , 2 0 1 7 B a s k a r a n e t a l . • 8 9
In-Situ Measurement of Wind Performance
of Roof Edge Systems
Dr. A. Baskaran, PEng and Sudhakar Molleti
National Research Council Canada
1200 Montreal Road, Ottawa, ON K1A 0R6
Phone: 613-990-3616 • E-mail: bas.baskaran@nrc.ca
Tony Mallinger
Metal-Era, Inc.
1600 Airport Road, Waukesha, WI 53188
B. Martín-Pérez
Department of Civil Engineering, University of Ottawa
161 Louis-Pasteur, Ottawa, ON K1N 6N5
Abstract
In commercial roofing, metal edges are the first line of defense against the effects of wind.
As part of the Roofing Industry Committee on Weather Issues’ (RICOWI’s) Wind Investigation
Program (WIP), North American roofing professionals completed a major fact-finding investigation
immediately following the landfall of Hurricanes Charley, Ike, Ivan, and Katrina.
Field data clearly supported that the majority of the roof failures were due to the failure of
metal roof edges. These findings suggest that current building codes (i.e., NBCC and ASCE)
do not accurately specify wind design loads acting on roof edge metal systems. The speaker
will present the findings of measured wind-induced pressure acting on all surfaces of three
different edge configurations.
Speaker
Dr. Appupillai Baskaran, PEng – National Research Council of Canada, Ottawa, ON
Dr. Baskaran is a group leader with the NRC. He is a member of technical committees
with RCI, RICOWI, ASCE, SPRI, ICBEST, ASCE, and CIB and a research advisor to various
task groups of the National Building Code of Canada. Baskaran has authored and/or coauthored
over 200 research articles and received over 25 awards, including the Frank Lander
award from the Canadian Roofing Contractors Association and the Carl Cash Award from
ASTM. Dr. Baskaran was recognized by Queen Elizabeth II with a Diamond Jubilee medal
for his contribution to his fellow Canadians.
Nonpresenting Coauthors
Sudhakar Molleti – National Research Council Canada, Ottawa, ON
Tony Mallinger – Metal-Era, Inc., Waukesha, WI
B. Martín-Pérez – Department of Civil Engineering, University of Ottawa, Ottawa, ON
9 0 • B a s k a r a n e t a l . 3 2 n d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma rc h 1 6 – 2 1 , 2 0 1 7
ABSTRACT
In commercial roofing, metal edges are
the first line of defense against wind effects.
As part of the Roofing Industry Committee
on Weather Issues’ (RICOWI’s) Wind
Investigation Program (WIP), several North
American roofing professionals completed
a major fact-finding investigation for the
cause of wind failures. WIP collected factual
data immediately following the landfalls of
Hurricanes Charley, Ike, Ivan, and Katrina.
Field data clearly supported
that the majority of the roof
failures were due to the
failure of metal roof edges.
These findings suggest two
major contributing factors
causing failures are: 1) current
building codes and
standards in North America
(i.e., NBCC and IBC/ASCE)
do not accurately specify
wind design load criteria
for roof edge metal systems,
and 2) there is failure of
attachments (nails) due to
inadequate design or application
error. The objectives
of this paper are:
1. Give a brief summary
of the in-situ
measurement site
(the Canada Post
building located
in Vancouver,
Canada).
2. Compare the measured
wind-induced
pressures/suctions
acting on all surfaces
of the three
different edge configurations—
namely
the anchor clip
configuration (ACC),
continuous cleat
configuration (CCC),
and discontinuous
cleat configuration
(DCC) with the
current ASCE 7-10
wind load criteria for parapets.
3. Correlate the design pressure coefficients
of edge metal to that of the
roof design pressure coefficient, to
simplify the cladding and components
design process.
INTRODUCTION
Many commercial roofs with parapets
are often covered by metal components.
The parapet can be constructed with various
types of substrates, including wood,
concrete, masonry, and steel. Normally,
the metal components cover the parapet
by means of an inner layer, namely a cleat
(nailed to the substrate), and an outer layer,
namely coping (typically, mechanically
engaged to the cleat on the exterior side of
the parapet and attached to the roof side of
the parapet with exposed fasteners). As part
of the roof’s perimeter, the roof edge acts
as an effective termination and transition
In-Situ Measurement of Wind Performance
of Roof Edge Systems
3 2 n d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma rc h 1 6 – 2 1 , 2 0 1 7 B a s k a r a n e t a l . • 9 1
Figure 1 – Typical roof edge failures during high wind events: A) coping disengagement from the
cleat failure, B) cleat fastener failure, C) partial failure, and D) complete failure.
A
C
B
D
between the roof and the wall. An adequate
edge termination is required to ensure the
integrity of the roof and the contents of the
building against the natural elements.
According to a study derived from FM
Global loss claims,6 59% of the losses
involving built-up roof systems occurred
due to failure of the roof edges. Reports
from RICOWI13-15 and observations made by
Baskaran et al.5 also point out that a significant
number of roof failures originated
from the poor performance of roof edges. In
1990, Smith16 documented the failure mode
of edge metal flashings after Hurricane
Hugo. Also, several FEMA publications provide
guidelines for edge metal design following
Hurricanes Charley (P-488),7 Ivan
(P-489),8 Katrina (P-549),9 and Ike (P-757)10.
ANSI/SPRI ES-11 is widely used by the
roofing industry. This standard has two
parts: 1) it provides wind load criteria for
the design of edge metals, and 2) it specifies
three different test methods for the resistance
evaluation of the edge metals. The
former is mainly based on ASCE 7-10.2 Thus,
there is a lack of design specifications for
wind loads acting on roof edge metal systems
in current building codes and standards.
In practice, there is a large variability
in manufacturing and installation of roof
edge metals. As wind separates from the
roof edge, it breaks down into vortices,
which create high-pressure differences at
the roof’s perimeter. The roof edge is particularly
susceptible to the effects of wind
because it can be fully immersed in the
separation bubble. These areas are characterized
by significant wind-induced pressure
fluctuations that can potentially damage
roof edges.
The three most common failure mechanisms
of roof edges are illustrated in Figure
1. The mode of failure caused by a billowing
membrane originates with the repeated
pulling force of a membrane that is partially
detached in the vicinity of the roof edge
(Figure 1A). It has been observed that billowing
membranes can actually pull out
fasteners or damage the coping-membrane
connection. A second type of failure arises
when the wind suction on the roof edges is
strong enough to pull out the frontal fasteners
that keep the flashings in place, as
depicted in Figure 1B. Figure 1C represents
the third mechanism of failure that occurs
when wind suction overcomes the resistance
provided by the connection between the
coping and its supporting element (known
as a cleat or clip). For instance, a failure in
the cleat can cause coping failure; billowing
of the membrane can induce failure of any
roof edge component. Improper nail attachment
can also cause failure.[17] To understand roof edge wind performance
and to incorporate wind load criteria
in the North America building codes (NBCC
and IBC/ASCE 7), a project, namely Roof
Edge Systems and Technologies (REST),
was formulated as explained in previous
papers by the authors.[3 and 4] The objective
9 2 • B a s k a r a n e t a l . 3 2 n d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma rc h 1 6 – 2 1 , 2 0 1 7
Figure 2 – A) Front view of the Canada Post building, showing B) the instrumented
Penthouse 6, and details of the configuration positioning. C) ACC, CCC, and DCC.
A
C
B
of this paper is to give a summary of the
in-situ measurement site, the Canada Post
building in Vancouver, and compare the
measured data of all three edge configurations.
In addition, this paper intends to
correlate design pressure coefficients of
edge metal to the measured roof pressure
coefficients.
The Canada Post mail distribution
facility located on the south side of the
Vancouver International Airport was identified
as a suitable field-monitoring site
(Figure 2). Positioned at an airport location,
the building can be categorized as having
an open terrain exposure (i.e., exposure C
from ASCE 7). The climatic wind data for
the airport, obtained from Environment
Canada, reveals that the predominant wind
direction ranges from southwest to northwest.
In order to be exposed to the critical
wind direction range, instrumentation was
placed on penthouse 6, located on the west
side of the building, with a parapet facing
the incoming west wind. This arrangement
exposes the instrumented roof edge to
perpendicular and oblique wind flow conditions.
The roof elevation was 18 m (58 ft.).
RESULTS and DISCUSSION
The period of October 2013 to September
2014 was selected for the present analysis.
Only one-hour data segments, for which
the wind direction statistical mode was in
the interval SW–NW and the hourly peak
wind speed was greater than 48 kph (30
mph), were considered. The above procedure
yielded 16 wind speed and pressure
data records during the one-year monitoring
period. Taking November 15, 2013, as a
typical windy day, Figure 3 shows the wind
pressure/suction acting on the three faces
(front, top, and back) of the ACC metal edge.
The presented data represent the maximum
measured pressure/suction for each
minute. The westerly peak winds of 97 kph
(60 mph) attack the instrumented edge configuration
with perpendicular wind direction.
The X axis shows the time in minutes,
whereas the measured wind and pressure/
suctions are displayed in the Y axis. At
the front face, wind induces both pressure
(positive) and suction (negative), whereas
the other two faces are subjected to only
suction. The design value for the day can be
labelled as peak. The corresponding peak
suction values are -335, -646, and -416 Pa
(-7, -13.5, and -8.7 psf) for the front, top,
and back faces, respectively. Even though
there is a difference in the peak magnitudes,
all three peaks occur at the middle of
the hour at about 30 minutes. The range of
the suction fluctuations is another observation
of interest. In the case of the front and
back faces, during the represented hour,
the suction fluctuates in the range of -192
to -431 Pa (-4 to -9 psf). This is significantly
different on the top surface, which experiences
higher fluctuations ranging from -192
to -670 Pa (-4 to -14 psf). This clearly shows
that in designing roof edges, the outward
negative pull force is higher at the top surface.
This is due to wind flow separation
from the leading edge. To understand the
performance of CCC, ACC, and DCC, the
measured data are plotted in Figure 4.[4] The X axis shows the roof height wind
speed squared (VR
2). Measured data from
the top and front surfaces of the edge metal
are shown in the Y axis. When observing
the effects of wind speed on the CCC, it can
be seen that both pressure and suction on
each face increase as wind speed increases.
And the rate of suction increase is greater
on the top face of the CCC compared to the
front face. As the top face is fully immersed
in the separation bubble, it is not experiencing
any pressures. Even though the
front face is subjected to pressure and suctions,
the magnitude of pressure is only
approximately half of the induced suction.
Moreover, when the edge metals are considered
as components and cladding for wind
design, the design for suction forces is more
critical than that of pressure. Hence, the
suction that is acting away from the parapet
surface tends to pull the cladding metal
3 2 n d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma rc h 1 6 – 2 1 , 2 0 1 7 B a s k a r a n e t a l . • 9 3
Figure 3 – Typical pressure distribution on the front face of ACC and the corresponding wind speed and wind direction
elements away from its substrates and the
attachment locations. This can lead to coping
disengagement from the cleat and/or
fastener pullout from the cleat or the cleat
pulling over the fastener. Examples of such
failures are shown in Figure 2.
Due to lack of wind load criteria for proper
roof edge design process in the Canadian
building code, it has been decided to perform
further data
analysis such that
pressure coefficients
that are unique to
the metal edges
can be specified.
Inspired by available
engineering approaches in developing pressure
coefficients, the authors developed a
procedure for pressure coefficient calculations
as explained in a companion paper.
[3] The design pressure coefficients based
on the Gumbel distribution are -3.1 and
-4.6 for the front and top legs, respectively.
The code-comparable pressure coefficients
with a 50-year return period (obtained from
Peterka & Shahid, 1998[12]) are -5.1 and -7.6
for the front and top legs, respectively, as a
design pressure coefficient. Note that these
coefficients are limited to low-rise buildings,
H <18 m (60 ft.), and are not suitable for
use with ASCE 7. Additional data processing
is needed to develop ASCE 7 pressure
coefficients. These pressure coefficients are
higher than ES-1 coefficients.
Discussions with industrial partners
revealed that without sacrificing the technical
merit of the acquired field data, simplifications
are needed in specifying the design
pressure coefficient. This is mainly to
avoid a separate set of calculations for the
front, top, and back faces of the edge metal.
9 4 • B a s k a r a n e t a l . 3 2 n d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma rc h 1 6 – 2 1 , 2 0 1 7
Figure 4 – Effect of wind speed on CCC, DCC, and ACC.
Figure 5 – Codification of wind load pressure coefficient of
low-slope membrane roof edge metals.
Windward Parapet
Load Case A
1. Windward parapet pressure (p1) is determined using
the positive wall pressure (p5) zones ⓔ or ⓦ from
the applicable figure.
2. Leeward parapet pressure (p2) is determined using
the negative roof pressure (p7) zones ⓒ or ⓢ from
the applicable figure.
Leeward Parapet
Load Case B
1. Windward parapet pressure (p3) is determined using
the positive wall pressure (p5) zones ⓔ or ⓦ from
the applicable figure.
2. Leeward parapet pressure (p4) is determined using
the negative wall pressure (p6) zones ⓔ or ⓦ from
the applicable figure.
Component Covering Parapet
Pressure (pe) is determined using the negative roof
pressure (p7 x 1.4) zones ⓒ or ⓢ from the applicable
figure.
Based on industry input, and considering
the variability and complexity of the
wind dynamics at the separation bubble, it
has been decided to propose a representative
design pressure coefficient for the edge
metals, by taking the worst-case scenario
between top and front legs. This resulted
in a design pressure coefficient of -7.6, corresponding
to the top leg.
Also, to account for practicality in the
design process, it has been decided to correlate
the proposed coefficient with the existing
roof cladding and component pressure
coefficient. A correlation factor of 1.4 can be
obtained by dividing the above-calculated
pressure coefficient of -7.6 for edge metals
by -5.4, where -5.4 is the maximum external
CpCg value corresponding to low-slope
roof components and cladding design specification
in accordance with the NBCC 2015
(refer to the Figure 4.1.7.6 of NBCC 2015).[11] Based on the above discussions and roofing
industry consultations, the proposed CpCg
for edge metals are shown in Figure 5.
For completeness of the design, windward
and leeward loads that are directly
taken from ASCE 7 are also shown. This
present proposal only enhances the existing
parapet wind load specification of ASCE/
SEI 7-10.[2] Edge metal pressure (Pe) is
determined using 1.4 times the negative
roof pressure (P7) corresponding to zone
C or S from the appropriate figure of
the NBCC. This explicit specification will
facilitate proper design and the attachment
mechanism to transfer the edge metal
load to the parapet structure. Doing so not
only simplifies the design process, but also
provides an explicit wind load specification
for the component that offers the first line
of defense against commercial roof failures
due to wind effects. A similar attempt has
also been made to develop a pressure coefficient
suitable for the ASCE 7 by considering
necessary transformations, as the ASCE 7
pressure coefficient (GCp) values refer to a
velocity pressure based on a three-second
gust wind speed. However, only recently
major changes were balloted for the components
and cladding sections to the ASCE
7-2016. Hence, for the time being, it has
been decided to limit the proposal to the
NBCC 2020.
CONCLUSIONS
This paper developed appropriate design
pressure coefficients for possible incorporation
into NBCC. During this process, the
following observations were made:
• Comparison of the in-situ measured
pressures with the existing ANSI/
ES-1[1] wind standards suggested
that the ES-1 wind load criteria are
inadequate.
• Based on consultation with the roofing
industry, it is proposed that the
edge metal pressure (Pe) be determined
using a factor of 1.4 applied
to the negative roof pressure (P7)
corresponding to zone C or S from
the appropriate figure of the NBCC.
This simplifies the components and
cladding design process.
This explicit specification will facilitate
proper design, and implicitly, facilitate the
design of attachment mechanism to transfer
the metal load to the parapet structure.
This load path transfer depends on the
capacity of the metal and pullout resistance
of the fasteners with the substrate. This
will simplify the design process, and clearly
state a wind load specification for the component
that offers the first line of defense
against commercial roof failures due to
wind effects.
ACKNOWLEDGEMENT
This research and development work
was carried out under the auspices of the
Natural Sciences and Engineering Research
Council of Canada’s (NSERC’s) Collaborative
Research Grant (CDR – 395869). Industrial
partners were Firestone Building Products,
JRS Engineering Group, Menzies Metal
Products, Metal-Era, Inc., the Roofing
Contractors Association of BC, and Soprema
Inc., whom we thank for the contributions,
including coping configurations. The
authors also acknowledge the assistance
of NRC technical officers Steven Ko, Amor
Duric, David Van Reenen, Aarti Singla, and
Maha Dabas. Material requirements were
coordinated by Carlisle Syntech, and sensor
installation and repairs were completed
by Larry Lemke of Marine Roofing. Access
to the building was graciously given by the
Canada Post, and we appreciate the help of
Michael Bryson, Alan Shopland, and Reyes
Ronald of JLL for coordinating the access.
REFERENCES
1 ANSI/SPRI. 2011. Wind Design
Standard for Edge Systems Used
With Low Slope Roofing Systems,
ANSI/SPRI/FM 4435/ES-1.
2 ASCE. 2010. Minimum Design Loads
of Buildings and Other Structures,
ASCE/SEI 7-10.
3 A. Baskaran, J. Bysice, B. Martin-
Perez. “In-Situ Measurement of Wind
Performance of Roof Edge Systems –
Effect of Edge Metal Configurations.”
Journal of Testing and Evaluation,
ASTM International (JAI), under
review.
4 A. Baskaran, S. Molleti, N. Martins,
B. Martin-Perez. 2016. “Development
of Wind Load Specifications for
Commercial Roof Edge Metals.”
Journal of Architectural Engineering,
(ASCE), under review.
5 A. Baskaran, S. Molleti, and D.
Roodvoets. 2007. “Understanding
Low-sloped Roofs Under Hurricane
Charley From Field to Practice,”
Journal of ASTM International, Vol.
4, No. 10.
6 Factory Mutual Research Corporation,
2016. Loss Prevention Data
Sheet 1-49, Factory Mutual Research
Corporation, Norwood, MA, 2016.
7 FEMA 488. 2005. “Hurricane Charley
in Florida: Mitigation Assessment
Team Report, Observations, Recommendations,
and Technical Guidance,
Hurricane Recovery Advisories.”
http://www.fema.gov/library/
viewRecord.do?id=1444.
8 FEMA 489. 2005. “Hurricane Ivan
in Alabama and Florida: Mitigation
Assessment Team Report,
Observations, Recommendations,
and Technical Guidance.” http://
www.fema.gov/library/viewRecord.
do?id=1569.
9 FEMA 549. 2006. “Hurricane
Katrina in the Gulf Coast: Mitigation
Assessment Team Report, Building
Performance Observations, Recommendations,
and Technical
Guid-ance.” http://www.fema.gov/
library/viewRecord.do?id=1857.
10 FEMA 757. 2009. “Hurricane Ike
in Texas and Louisiana: Mitigation
Assessment Team Report,
Building Performance Observations,
Recommendations, and Technical
Guidance.” https://www.
f e m a . g o v / m e d i a – l i b r a r y –
data/20130726-1648-20490-9826/
fema757.pdf.
11 NBCC. 2015. National Building Code
of Canada, Institute for Research
3 2 n d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma rc h 1 6 – 2 1 , 2 0 1 7 B a s k a r a n e t a l . • 9 5
in Construction. National Research
Council of Canada (NRCC).
12 J.A. Peterka and S. Shahid, 1998.
“Design Gust Wind Speeds in the
United States.” Journal of Structural
Engineering. 124(2), 207–214.
13 RICOWI, Inc. 2006. “Hurricanes
Charley and Ivan Wind Investigation
Report.” Roofing Industry Committee
on Weather Issues Inc., McDonough,
Georgia.
14 RICOWI, Inc. 2007. “Hurricane
Katrina Wind Investigation Report.”
Roofing Industry Committee on
Weather Issues Inc., Powder Springs,
Georgia.
15 RICOWI, Inc., 2007. “Hurricane Ike
Wind Investigation Report.” Roofing
Industry Committee on Weather
Issues Inc., Powder Springs, Georgia.
16 T. Smith, 1990. “Hurricane Hugo’s
Effects on Metal Edge Flashings.”
International Journal of Roofing
Technology.
17 T. Smith, 2015. “Nailer Attachment
is One Key to Achieving Wind-Uplift
Performance.” Professional Roofing.
NRCA, page 65-70.
9 6 • B a s k a r a n e t a l . 3 2 n d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma rc h 1 6 – 2 1 , 2 0 1 7