Lessons Learned From The Florida Hurricanes

May 15, 2005

TEven as we continue to view
images of hurricane devastation
remaining in the Gulf
Coast and eastern North
Carolina from 2005’s storm
season, roofing professionals
from RICOWI (Roofing
Industry Committee on
Weather Issues) and
SPRI (Single Ply Roofing
Industry) are in the
final phase of analyzing
data collected in Florida
just after the 2004 hurricanes.
It may be a dim
memory, given all that’s
happened since, but in
2004, Hurricanes Bonnie,
Charley, Frances,
Ivan, and Jeanne crisscrossed
various parts of
Florida, leaving virtually
no portion of the
state unscathed. Each
presented a set of unique
wind-event anomalies,
allowing teams
of RICOWI investigators
to fan out over the
storm-battered Sunshine
State to examine
a representative sampling
of more than 50
single-ply roofing systems
that either failed,
were damaged, or remained
“Many people claim
that as buildings become
more energy efficient, they also are
more susceptible to damage,” said André
Desjarlais, a member of the RICOWI team
from the Oak Ridge National Laboratory in
Tennessee. However, energy-efficient buildings
are also better designed structures.
That was just one preliminary finding by
RICOWI, but the emphasis of the mission
was to formally document the performance
and damage that occurred to roofing systems
during substantiated hurricane-level
wind speeds.
Pensacola School Building 1: RICOWI inspectors found pressurization and billowing at overhangs. There were
cleat deformations and disengagement of edge metal. The cleat gauge was less than currently recommended
by FM Global LPDS 1-49 and ANSI/SPRI ES-1. Corrosion reduced effective uplift resistance. Overhangs need to
be designed and constructed to resist higher loads. (Source: André Desjarlais’s Oak Ridge National Laboratory
report, “RICOWI and its Investigations of Hurricanes Charley and Ivan”)
26 • I N T E R FA C E J A N U A RY 2006
About the Study
The RICOWI field performance investigations
of roofing assemblies were specifically
conducted on structures that were in
the direct paths of Hurricanes Charley and
The study was jointly funded by the
roofing industry and the U.S. Department of
Energy (DOE). The government was interested
in determining if energy-efficient roofing
installations, advocated by the federal
agency, would exhibit favorable performance
characteristics in the presence of
hurricane-force winds.
Following its landfall on Friday, August
13, 2004, near Punta Gorda, seven teams of
investigators were deployed to conduct 90
inspections from August 18-21. After
Category 3 Hurricane Ivan, another five
teams were deployed to conduct 75 additional
inspections from September 22-25
near the Florida and Alabama borders.
David Roodvoets, RICOWI Wind Event
coordinator, who is also a technical director
for SPRI, headed the study.
All types of low-slope and steep-slope
roofs were included in the study, including
roofing systems with SPF, TPO, BUR,
EPDM, and vinyl membranes, Roodvoets
explained. Three attachment methods were
studied – mechanically-attached, adhered,
and ballasted – but all combinations of
membrane and attachment types were not
available in the field. The study was limited
to approximately 24 single-ply roofing systems,
restricted primarily by the available
time allotted for data collection.
“The study was somewhat opportunistic
for low-slope roofs. We tried to concentrate
on essential facilities such as shelters,
schools, hospitals, fire, municipal, and
postal facilities,” Roodvoets said.
For purposes of comparison, the
RICOWI teams of investigators selected geographic
areas where significant damage
occurred, as well as areas where damage
was minimal. All roofing systems investigated
sustained some damage, and patterns
developed after analysis of the inspection
Buildings closest to the ocean were not
inspected because their structural integrity
may have been compromised by water damage
experienced during the storm surge. In
many cases, the remains of coastline buildings
were only evident by piles of rubble,
and such remains were beyond the scope of
the RICOWI study. This was because the
focus was on roofing performance in a wind
event and not a water event.
Roof System Performance
“Failure of roofing systems was because
of system failure at the perimeter and punctures
and tears from debris,” Roodvoets
said. “The membrane attachment to the
deck cannot resist the loads created when
the perimeter securement fails, and this
leads to progressive loss of membrane coverage.”
An example of such a catastrophic roof
failure is a Port Charlotte office building
that had a poorly installed roof. Investigators
found that the roof system’s poor
performance was the result of inconsistent
edge metal clip installation. During the
wind event, an intensely high clip load was
demanded, and eventually a peel-off failure
occurred to the membrane.
Desjarlais noted, “It’s a fair statement to
say membranes themselves were not the
major cause of failure. There were a lot of
perimeter attachment failures. If you can’t
hold the edge down, no matter what kind of
roof you have, the whole thing is going to be
subject to loss.”
“The issue of fastener corrosion was
seen over and over again,” he continued,
noting that coastal specifications should
require fasteners with coatings that are
unaffected by the marine environment.
Wind-blown missile damage to roofs was
significant, according to Desjarlais. Membrane
damage was typically limited to punctures
from lightning rods and wire that
swung during the hurricane and whipped
the membrane with sharp metal edges.
Some poorly anchored HVAC units rolling
across the roofs also punctured membranes
in a few cases. But panels and other components
of the HVAC units were more commonly
blown off because the equipment was
not properly tied down either during installation
or following periodic servicing.
On some single-ply systems, the damage
was easier to locate, and this may be an
advantage to those types of systems,
Desjarlais revealed. Conversely, it was less
obvious to determine damage on systems
such as ballasted roofs, because the aggregate
overburden can conceal the indicators
of compromised performance.
Another factor affecting roof performance,
which was anticipated, was intensely
high-speed winds entering buildings
through openings such as shattered windows
and open doors. However tightly
sealed a building remained during the hurricanes
often determined the fate of the roof
and, in some cases, the entire structure.
Some roofing membranes were compro-
J A N U A RY 2006 I N T E R FA C E • 2 7
mised after walls or windows
were damaged,
providing uplift to the
roof in excess of what
the building was designed
to endure.
Buildings built after
Hurricane Andrew devastated
Florida in 1992
performed much better
than buildings with
roofs installed before
Andrew, according to
The overall findings
were that post-Andrew
roof designs and installations
performed well
unless there were
installation problems,
internal pressurization
from openings (typically
created by failed accessories),
perimeter detachment, or corrosion
The Punta Gorda area also contained a
significant amount of vinyl roofing membranes,
which performed very well.
Before Hurricane Charley, the last
recorded storm to make landfall in Punta
Gorda was in 1943, and many of the older
buildings there had been through previous
storms. Furthermore, one of the on-site
indications was that many of the roofs seen
there by the inspection teams had been
poorly maintained. “What we saw was a lot
of deferred maintenance that compromised
the roofing system,” Desjarlais said.
But not every roof on every building in
the path of Hurricane Charley failed. A
Punta Gorda Elementary School was well
designed, and its roofing system installation
and performance were both good. It
remained intact.
The teams of inspectors assigned to the
Florida panhandle in the wake of Hurricane
Ivan found similar results. The roof on a
Pensacola school building was damaged.
The key points that surfaced from the investigation
were that damage was initiated by
pressurization and billowing at overhangs.
This pressurization phenomenon caused
cleats to yield, compromising the attachment
of the edge metal. Progressive failure
then rapidly followed.
Other findings were that the cleat gauge
was less than currently recommended by
FM Global LPDS 1-49 and ANSI/SPRI ES-1.
In addition, corrosion of roof edge fasteners
28 • I N T E R FA C E J A N U A RY 2006
Pensacola School
Building 2: Cleat
deformed, edge metal
lifted, membrane peeled.
Cleat gauge less than
currently recommended
by FM Global LPDS 1-49
Corrosion of roof edge
fasteners contributed to
expansion of edge
damage into field.
(Source: André
Desjarlais’s Oak Ridge
National Laboratory
report, “RICOWI and its
Investigations of Hurricanes
Charley and Ivan.)
contributed to expansion of edge damage
into the field.
A second school building inspected in
Pensacola had damage and propagation.
The edge metal was bent upward, nailers
lifted, and the membrane base sheet torn
around fasteners and peeled back.
On closer inspection, the edge metal
gauge and nailer securement were also less
than currently recommended by FM Global
LPDS 1-49 and ANSI/SPRI ES-1, and the
lack of enhanced perimeter base sheet fastening
contributed to propagation of edge
damage into the field area.
Lessons Learned
According to another RICOWI inspection
team member, Peter Garrigus, vice
president of engineering for Trufast Corp. in
Bryan, Ohio, “Corrosion problems were due
to the wrong roofing products” for the hostile
atmosphere along the coast.
Addressing the consistent failure of edge
metal roofing clips and fasteners, Garrigus
explained, “It really boiled down to workmanship
and proper application of materials.”
Roofing systems performed acceptably
in cases where the proper fasteners and
roofing components were applied in accordance
with the manufacturers’ requirements
and industry standards established
by FM and others, Garrigus observed. In
nearly 95 percent of the cases studied where
roofing system failure was seen, the problems
were because of poor workmanship
and substandard materials, explained
Garrigus, who had made similar observations
during inspections conducted after
Hurricanes Hugo and Andrew in years past.
In some cases, uncoated nails from local
hardware stores were substituted for Jbolts
or other proper fasteners to secure
nailers. In addition, the nails were often
Pensacola School Building
3: Edge metal bent upward,
nailers lifted, membrane
base sheet torn around
fasteners and peeled back.
In this shot, the edge metal
is bent upwards. The edge
metal gauge and nailer
securement are less than
that currently recommended
by FM Global LPDS 1-49
and ANSI/SPRU ES-1. Also,
a lack of enhanced perimeter
base sheet fastening
contributed to propagation
of edge damage into field
area. (Source: André Desjarlais’s
Oak Ridge National
Laboratory report, “RICOWI
and its Investigations of
Hurricanes Charley and
J A N U A RY 2006 I N T E R FA C E • 2 9
spaced improperly and at less frequency
along the roofs’ perimeter than what was
called for by the manufacturers of the edge
or roofing system. And in other cases, the
carbon steel nails were completely rusted
Garrigus believes roofs as poorly constructed
as those that failed during hurricane-
force winds would have failed in lower
than hurricane-speed winds. To counter
installation and material problems, he
encouraged the trend of manufacturers
offering contractor training programs be
accelerated. Also, inspections during application
– either by the system supplier or
consultant – would further decrease system
failures during high wind events.
The integrity of the entire building envelope
would benefit from improved wind
resistance in the design and installation of
windows and doors to minimize their failure
in high winds and reduce the deleterious
effects created from building pressurization.
The effect of compromised openings
can result in excessive damage on both the
roof and in the structural integrity of the
building. These losses can be catastrophic
in many cases; therefore, building teams
need to install impact-resistant doors and
windows to safeguard the interior from the
pressurization of the enclosure.
Many coastal property owners’ housing
docking and transportation operations simply
left loading doors open in their warehouses
and lost entire buildings. So, property
storm planning is also an issue.
RICOWI’s preliminary results were
recently presented by Roodvoets and
Desjarlais to SPRI. The earliest findings
contained information about six types of
roofing systems in the path of the storms. A
full report on every roof inspected was
released in October.
Steve Moskowitz is the current president of SPRI (Single Ply
Roofing Industry), and also vice president of technical and
warranty services at Stevens Roofing Systems, Holyoke,
Massachusetts. Moskowitz has been active in SPRI for 10
years. He has directed a number of key association committees
and task forces for SPRI, such as the technical committee
and the member services committee, and has served two
terms on the board of directors. Moskowitz also led SPRI’s fire
testing task force.
Steven Moskowitz
30 • I N T E R FA C E J A N U A RY 2006
to any roofing system, cedar shakes and
shingles included. For a cedar roof, the following
areas are key elements of a sound
wind resisting roofing system:
Correct Fasteners
Roofing contractors should use fasteners
that will not rust or break
away during high windstorms.
Proper Exposure
Ensure that the cedar shake or
shingle is exposed the correct number
of inches to the weather; overexposure
will dramatically reduce the
roof’s life and integrity.
Correct Keyways
Follow proper application directions
to ensure the shakes or shingles are
spaced and applied with side laps as
required by applicable building
codes; excessively wide or narrow
keyways should be avoided.
38 • I N T E R FA C E F E B R U A RY 2005
Photo 3: Overhangs increase uplift pressure. Special design is required.
Felt Interlayment
Place approved type of interlayment
at correct measurement on successive
courses to ensure protection
from UV degradation.
Metal Buildings
The metal building team investigated
the performance of metal roofing on building
systems, as well as on conventional construction.
Similar observations were made
at both of the hurricane areas investigated.
In general, post-Andrew metal roof designs
and installations performed well.
Exceptions were usually associated with
installation problems, insufficient attention
to details, or from internal pressurization
caused by a breech of the building envelope,
typically created by failed accessories.
Metal roofs are either through-fastened
or standing seam systems. It is common
knowledge that standing seam roofs require
better trained installers and are more likely
to suffer from poor installation, inspection,
or both. The spacing and attachment of
clips is essential to the performance, but
field seaming is crucial. Failures of standing
seam roofs were almost always due to the
separation of the seam from the clip.
F E B R U A RY 2005 I N T E R FA C E • 3 9
Photo 4: Typical metal edge damage.
Many metal roof failures initiated at the
eave, ridge, or rake because weaknesses were
introduced there due to details or flashing
attachments. For example, the attachment of
a gutter should not create a situation where
the failure of that component pulls up the
edge of the panel. Failures were commonly
observed with soffit panels and hip and ridge
flashing. These don’t appear to always be
designed for the expected wind pressures,
and though relatively minor, improvements
could be easily realized.
Wind damages ranged from minor to
extensive on low-sloped roof coverings. The
most common initiating conditions for wind
damage were lifting of edge metal and billowing
of membranes near windward edges.
At most locations, deterioration of roof
attachment systems (e.g., rusted fasteners)
and installation variations from typical
industry recommendations (e.g., thin gauge
cleats) reduced available roof wind uplift
resistance and contributed to the wind
Suggestions to enhance wind resistance
of installed roof coverings include:
1. Use the currently available high
wind roof design guidelines by those
Photo 6: Old BUR attachment. Improper nailing of base sheet did not provide adequate attachment for
BUR or new SPF roof system. Side lap nailing is typically 9″ O.C.
Photo 5: Corrosion is common
in hurricane prone areas –
especially near the ocean.
40 • I N T E R FA C E F E B R U A RY 2005
Photo 7: Edge flashing
failure – SPF roof system
depended upon attachment
to rotten edge nailer.
who design and specify roof coverings.
2. Understand common wind damage
initiation conditions (e.g., roof edge
metal, peel, pressurization of membranes
above roof overhangs) and
ways to enhance resistance against
these conditions.
3. Use calculations from ASCE -7 to
determine rooftop wind loads. Use
edge metal products that comply
with the building code required
4. Use highly durable materials as part
of roof attachment systems (e.g.,
stainless steel or corrosion protected
fasteners, preservative treated
5. In hurricane-prone zones, make
sure additional perimeter and corner
securement as required by
Factory Mutual or the membrane
supplier is in place. It is advisable to
add a secondary securement strap
that significantly exceeds the potential
uplift load within three feet of
the perimeter on membrane roofs to
avoid damage.
Spray Polyurethane Foam Roofs
Damage observed on SPF roofs again
confirms three very significant points:
1. SPF has tenacious adhesion to any
properly prepared surface.
2. Most built-up roofs require additional
edge and field fastening prior to
SPF application.
3. Proper fastening of edge flashing is
The practice of reusing existing BUR
edge flashing routinely resulted in failure.
Photo 8: SPF edge repair – Adjacent BUR blew off, while the SPF roof only suffered
perimeter damage due to inadequate nailing of metal edge flashing. www.chemlinkinc.com
Cheap wood
keep pipes,
and equipment
high and dry.
• Install quickly
(with M-1 Structural Sealant)
• No fasteners
• Weather proof
• Long Term Solution
damage your
work and

F E B R U A RY 2005 I N T E R FA C E • 4 1
SPF roofs over BURs were found with rotten
edge nailers and metal flashing fastened at
12″ on center and greater. Blown off existing
BURs with SPF roofs were seen with base
sheet fastening substantially less than the
industry recommended patterns. Minor
missile damage to the coating and SPF was
observed, but this damage was not the
source of leaks and could have been
repaired with a tube of sealant.
Suggestions for improved performance
of SPF roofing include:
1. Remove all edge flashing, inspect
nailers and replace as necessary,
and install new edge metal fastened
in accordance with industry standards.
2. Cut the existing BUR to determine
its attachment, and add fastening as
3. Without cutting the existing BUR,
consider adding screws and discs to
the 10-foot perimeter at the rate of
one for every 2 sq.ft.
4. Prime new edge metal as recommended
by the SPF manufacturer.
5. Ensure that rooftop mechanical
equipment is properly
In summary, a hurricane does
not discriminate. It finds the details
that did not work and takes the system
apart from there. RICOWI has
trained over 80 people to understand
wind loading. This training was put
into practice in these investigations,
and the areas understood to be
potential for damage were easily
found after the hurricanes. If you
design for roofs that will be subjected to
high wind, you will want to attend the
RICOWI seminar in March to learn more.
The findings noted in this article are preliminary.
Much more detail will be provided
in the final report.
42 • I N T E R FA C E F E B R U A RY 2005
Photo 9: Edge flashing attachment –
metal edge improperly fastened at
up to 22″ O.C.
Lynne Christensen – Cedar Shake & Shingle Bureau
Phil Dregger – Technical Roof Services/RCI representative
Dave Roodvoets – DLR Consultants/SPRI
Lee Shoemaker – Metal Building Manufacturers Assn.
Robb Smith – Amtech Roofing Consultants, Inc.
Patty Wood-Shields – RICOWI, Inc.
The Authors
The U.S. Army Corps of Engineers
(USACE) issued a report to Congress in
August stating that reverse auctions
should not be used to procure construction
services and that reverse auctions fail
to realize any additional financial savings
over the sealed bid process. In its final
determinations, USACE found that the
acquisition of construction services cannot
and should not be equated with commodities
for the following reasons:
• Within the operational parameters
of Department of Defense contracting
regulations, the dynamics are
much too diverse between construction
services and commodities;
• Virtually all of the USACE construction
services…are one-of-akind
projects under one-of-a-kind
conditions with numerous and
consistent variables for cost and
no-cost factors;
For construction, the USACE “could
find no factual, significant, or marginal
savings in the use of reverse auctioning
methodology over the standard sealed bid