Life Safety Issues in Roof Design

Life Safety Issues in Roof Design
STEPHEN L. PATTERSON, RRC, PE
Roof Technical Services, Inc.
1944 Handley Drive, Fort Worth, TX 76112
P: 8174964631
• Email:
spatterson@rooftechusa.com
MADAN MEHTA, PE, PhD
University of Texas at Arlington
Email:
mmehta@uta.edu
Proceedings of the RCI 25th International Convention Patterson and Mehta 191
ABSTRACT
The purpose of this paper is to increase awareness and emphasize the importance of designing
roofs to meet critical life safety standards. Failure to properly consider these life safety
issues can result in catastrophic failures, including the loss of life. The focus will be design
issues related to fire, wind, and drainage. This paper will include an historical overview of
design requirements and trace the changes of these design standards in past and current
codes and industry standards. Case studies and examples of major fire losses, wind losses, and
roof collapses will be provided.
SPEAKER
Mr. Patterson has been working as an engineer in the roofing industry for more than 35 years
and as a practicing roof consultant and engineer for more than 25 years. He has coauthored
Roof Design and Practice, a design textbook published by Prentice Hall. He coauthored Wind
Uplift Pressures on Low Sloped Roofs and Drainage Design, design manuals published by the
RCI Foundation. Mr. Patterson has evaluated thousands of roofing failures, including failures
involving serious life safety issues. He is a roof consultant and a longtime advocate for the
importance of life safety issues in roof design.
Patterson and Mehta 192
Proceedings of the RCI 25th International Convention
Life Safety Issues in Roof Design
INTRODUCTION
The roof is one of the most important
elements of a building. The basic
purpose of a building is to provide
shelter from the elements, and a
weathertight
roof is essential to the
function of a building. There is no
lack of literature about the problems
associated with leaking roofs or problems
in general with roofs and the
roofing industry. People forget that
roofing was a target industry for
OSHA. At one time or another, roofing
has shown up at the top of various
lists associated with construction,
including:
• Number 1 source of litigation in
construction
• Number 1 source of litigation for
architects
• Number 1 source of insurance
losses
• Number 1 building maintenance
cost
This dubious status for the roofing
industry has a long history. The first
building codes in the United States
dealt with roofing problems. City ordinances
banned thatch roofs in New
Amsterdam and New Boston in the
1600s due to issues with leaks and
fire hazards. The cyclical weather
cycles in the Americas were problematic
for thatch roofing. Roofing was
also in the forefront of the first modern
building codes. Like the first
building ordinances in the 1600s, fire
resistance was the focal point of the
first modern codes dealing with roofing.
During the latter part of the 19th
century, fires literally destroyed cities
across the United States.
The 1970s became a watershed
decade for the industry. Prior to 1970,
roofing was a mature and stable
industry. There were a limited number
of roofing manufacturers, most of
which had been around for decades.
More than 95% of the lowslope
roofs
installed were builtup
roofs, and
roofing technology had changed little
in 100 years. The building codes were
relatively simple. The codes required
the roofs to be either nailed or
mopped to the roof deck, and roofs
had to be either “fire retardant” or
“ordinary roofs.”
Prior to WWII, buildings were relatively
simple, but they changed dramatically
during and after WWII. Cost
became a critical factor in postWWII
construction, as the industry looked
for less expensive methods of construction.
Roofing was particularly
affected as industrial buildings
increased substantially in their footprint.
Consequently, roofing costs
became an increasingly important
consideration in such buildings. By
the 1970s, wood and concrete decks
had given way to lighter, more costeffective
roof structures like steel,
lightweight insulating concrete, gypsum,
cementfiber,
plywood, and precast
concrete decks. The roofing
industry was still mopping and nailing
roofs to these decks, yet each of
these roof decks posed a different
problem for builtup
roofs. Also, the
quality of builtup
roofs was reduced,
and the twoply,
builtup
roof introduced
in the 1960s began to fail at an
alarming rate in the 1970s.
Perhaps the biggest impact in the
1970s occurred as a result of the oil
embargo in 1973 when oil prices
increased exponentially. In 1973,
builtup
roofs in Texas cost $0.25/sq
ft with no insulation and $0.35/sq ft
with insulation. By 1976, these costs
had risen to more than $1.50/sq ft.
Increased costs and performance
issues opened the door to singleply
roofing and other alternatives to the
builtup
roof. The roofing industry
went from a very stable industry in
the early 1970s to a volatile and
chaotic industry by 1980. Singleply
roofing represented less than 1% of
the market at the beginning of the
1970s but represented more than
50% of the market by 1980.
The 1970s are also important
technically, as the roofing industry
began addressing these roofing problems
as they developed. The National
Roofing Contractors Association
(NRCA) published its first NRCA
Roofing Manual in 1970, and the
NRCA established itself as one of the
most important sources of technical
information in the roofing industry.
Also in 1970, the American Institute
of Architects (AIA) commissioned the
writing of the Manual of Builtup
Roofing Systems, which became an
important addition to roofing literature.
The increased problems in the
roofing industry in the 1970s also led
to the development of roof consulting
as a specialty in the roofing industry.
The tumultuous 1970s gave way
to the 1980s and the roofing industry
began to recognize life safety design
issues. The codes changed dramatically
in the 1980s and groups like the
Roof Consultants Institute (now RCI
Inc.), Factory Mutual (FM), AIA, and
NRCA helped bring roofing problems
and design issues to the forefront.
1. FIRE SAFETY ISSUES IN
ROOF DESIGN
Fire safety is the first of the three
roofingrelated
life safety issues introduced
in the codes. Thatch roofs
caught fire in the 1600s, leading to
their restriction by the first building
ordinances in America. Between 1860
and 1915, there were more than $1.2
billion in fire losses. The primary
Proceedings of the RCI 25th International Convention Patterson and Mehta 193
Figure 1 – One of the earliest test apparatus used by the Underwriters Laboratory for testing
of roof coverings.
cause in these disasters was the
spread of fire as the burning brands
blew from roof to roof. The 1871
Chicago fire alone destroyed more
than 17,450 buildings. In 1911, more
than 165,000 buildings burned to the
ground. The end of the nineteenth
century was the precursor to the
decade of the 1970s for roofing design
issues. Roof testing, begun in the
nineteenth century, became the basis
for the first national codes.
Underwriters Laboratories (UL)
began testing roofing materials and
systems for fire resistance at the
beginning of the twentieth century.
The tested roof assemblies were rated
based upon the relative fire resistance
of the roofs, and from these tests,
today’s UL 790 and ASTM E108 standards
evolved. Figure 1 shows one of
the earliest fire tests of roof coverings
by UL (The History of the National
Board of Fire Underwriters, Ref. 1).
As a result of the staggering fire
losses in the latter half of the nineteenth
century and continuing into
the twentieth century, the National
Board of Fire Underwriters developed
a model code in 1905 and distributed
it to all cities with 5,000 inhabitants
or more. Two additional editions were
published before the model code was
completely rewritten and published in
1915. This building code, recommended
by the National Board of Fire
Underwriters (NBFU), was the basis of
the first modern building codes, and
the roofing requirements for fire resistance
are remarkably similar to
today’s codes. Below are the requirements
for firerated
roofing assemblies
from the 1915 NBFU Building
Code.
Section 80. Roof Coverings
1. All buildings, except
as given below, shall have
roof coverings approved
standard quality, such as
brick, concrete, tile, or
slate; or highest grade of tin
roofing, or of asbestos shingles,
or of builtup
roofing
felt with gravel or slag surface,
or of builtup
asbestos
roofing; or other roofings of
like grade which would rank
as Class A under the test
specifications of the National
Board of Fire Under writers.
Patterson and Mehta 194
Proceedings of the RCI 25th International Convention
Exceptions were provided for
dwellings, frame buildings, and buildings
not exceeding 2,500 sq ft and not
used for factories, warehouses, or
mercantile purposes. The roofs for the
exceptions, however, were required to
meet a minimum Class C rating.
Further, the code stated the following:
8. This section shall not be
construed to prohibit the
repairing of a wooden shingle,
provided the building is
not increased in height, but
the renewal of such a roof is
forbidden. No existing
wooden shingle roof, if damaged
more than 10 per cent,
[sic] shall be repaired with
other than approved roofing.
Wood roofs had been identified as
one of the key problems with fire moving
from roof to roof, causing fire to
spread across neighborhoods. The
apparent intent of the 1915 Rec om mended
Building Code was to ultimately
eliminate the use of wood
shingles or other nonrated roofing
systems. This sentiment is consistent
throughout many of the early modern
codes, which were largely used in
areas where fire danger was highest.
These were in places where buildings
were close together and fires could
move rapidly from building to building,
such as in central cities as well as
industrial and manufacturing districts.
These areas were typically
identified as “fire districts.” The first
building code adopted by the city of
Fort Worth included requirements to
eliminate wood roofs in the “fire district.”
Today’s building codes are very
similar to the earliest building codes
developed 100 years ago. The fire rating
standards are UL 790 and ASTM
E108. As shown in Table 1, there are
three fire rating classifications for
roofs (Class A, B, and C) in the 2009
International Building Code that are
very similar to the 1915 Recommended
Building Code. Until the
introduction of the 2000 IBC, the
Type of construction IA IB IIA IIB IIIA IIIB IV (HT)VA VB
Roof cover B B B C B C B C C
Table 1 – Minimum roof covering classification and the building’s
type of construction (adapted from 2009 IBC).
codes did not distinguish between
Class A and B roofs. Class A and B
roofs were considered to be “fire retardant”
and Class C roofs were considered
to be “ordinary.” The 2000 IBC
made a distinction between Class A
and B roofs, requiring Class A roofs to
be used on buildings such as hospitals
and jails. However, this distinction
was dropped in the 2003 IBC.
Basically, Class A and B roofs are
required on most commercial and
most large buildings. Class C roofs
are allowed on small commercial
buildings and most residential constructions.
Untreated wood shingles
can still be used on buildings two stories
or less in height and on buildings
with less than 6,000 sq ft when there
is a 10ft
separation between buildings.
It is important to understand that
there are distinctions within the Class
A, B, and C ratings. There are three
tests included in the test method. The
tests include a spreadofflame,
burning
brand, and an intermittentflame
test. The type of roof deck and the
slope of the roof are critical factors in
the rating of a roofing assembly. Roof
decks are categorized as combustible
or noncombustible. The only test a
roof has to pass for a noncombustible
deck is the spreadofflame
test. The
spreadofflame
test is limited to six ft
for Class A, nine ft for Class B, and 13
ft for Class C. In this test, slope is the
critical variable; the greater the slope,
the greater the spread of flame. Figure
2 shows the spread of flame test
apparatus, which is also used for the
intermittent flame test.
Combustible roof decks must also
meet the burning brand and intermittent
flame tests. These tests evaluate
the roof system’s ability to protect the
Figure 2 – Test apparatus used to determine the spread of flame
(ASTM E108 or UL 790). The same apparatus is used for intermittent
flame test and burning brand test.
Proceedings of the RCI 25th International Convention Patterson and Mehta 195
Figure 3 – Wood brands for testing Class A, B, and C roofs.
roof deck from spreading fire below
the roof. The intermittent flame test
measures the penetration of fire into
the deck by a pulsating flame source.
The flame temperature is maintained
at 1,400ºF with 15 on/off cycles for
Class A and eight on/off cycles for
Class B. The flame temperature is
maintained at 1,300ºF with three
on/off cycles for Class C. The burning
brand test measures the roof covering’s
ability to protect the roof deck
from external fire. Blocks of wood are
set on the roof and incinerated. The
burning brands are 12 in x 12 in x 2
11/23 in for a Class A rating; 6 in x 6
in x 2 11/23 in for a Class B rating,
and 1.5 in x 1.5 in by 25/32 in for
Class C. Figure 3 shows different sizes
of burning brands used in the test.
The fire rating for the roof system
is not the only life safety fire issue in
roof design. In 1953, a 1.5 million sf
General Motors plant in Livonia,
Michigan, burned to the ground as a
result of roofing issues. For many
years, this was the largest insurance
loss in the history of the United
States. The roof was a gravelcovered,
builtup
roof with a layer of rigid
board insulation installed over a 2ply
vapor barrier that had been solidly
mopped to the steel deck using hot
bitumen. The hot bitumen was
mopped over the entire roof deck and
bitumen accumulated in the ribs of
the deck. Sparks from a welder’s
torch started the fire within the plant,
hot bitumen in the ribs of the roof
deck contributed to the spread of fire
throughout the entire building, and
ultimately, the entire building collapsed.
Issues related to this fire were
evaluated by UL and Factory Mutual
(FM), and UL and FM began to
address the issue of how much the
roof and insulation would contribute
to a fire that started on the inside of
the building. FM developed fire standards
in its Class I category. UL also
developed fire ratings for different
assemblies to help address these
issues. Both UL and FM provide listings
for roof construction based upon
the criteria developed from their evaluations
of the Livonia fire.
More than 100 years have elapsed
since the fire ratings for roofs and
standards for fire resistance were
developed, yet there continue to be
issues related to fires and roofs. Still,
many roof designers are unaware of
the intricacies of fire ratings.
Manufacturers typically label roof
systems as Class A. Class A ratings
may only apply to noncombustible
decks or roofs with relatively low
slopes. The roof designer must ascertain
that the roof system is appropriately
rated for the slope and deck.
Even though the roof system is
labeled Class A, the roof system may
not meet the building codes on plywood
decks or on roofs with greater
slopes. Few lowslope
roofing systems
have Class A or B fire ratings on roof
slopes above 1:12. Drive around most
cities and look at the barrel or domed
roofs with modified bitumen, smoothsurfaced
builtup,
or EPDM roofs.
These domed or barrel roofs often
have sections of roofing with relatively
steep slopes. Chances are these
roofs do not meet the fire ratings
required by code, even though the
roof systems are labeled Class A.
2. LIFE SAFETY ISSUES IN
ROOF DRAINAGE DESIGN
Every year, roofs collapse as a
result of improper roof drainage
design and other drainagerelated
issues. Roof collapses are catastrophic
roof failures that cause millions of
dollars in property damage. More
Patterson and Mehta 196
Proceedings of the RCI 25th International Convention
Figure 4 – An example of a typical roof collapse caused by inadequate roof drainage.
importantly, roof collapses also result
in the loss of life. On May 5, 1995, a
series of severe thunderstorms
passed over the Dallas/Fort Worth
area, causing more than $2 billion in
property damage from large hail.
Softballsized
hail injured hundreds
of people attending an outdoor festival,
but no fatalities occurred as a
result of the hailstorm. Unfortunately,
there were fatalities from that
storm, occurring as the result of a
roof collapse due to heavy rains.
A correctly designed roofdrainage
system provides adequate slope and a
properly designed primary and secondary
drainage system to prevent an
excessive buildup
of water on the
roof. This buildup
of water is almost
always the key factor in roof collapses.
Figure 4 is a typical example of
such roof collapse.
Codes and Roof Drainage
The early model building codes
provided by the National Fire Un derwriters
included limited re quire ments
for roof drainage. How ever, roof
drainage requirements were included
in the first edition of the Uniform
Building Code (UBC) published in
1927. Issues of slope, drainage, and
overflow were addressed simply
(rather eloquently) in Section 3206 of
the code, which stated the following.
Roof Drainage, Section
3206. Roofs of all buildings
shall be sloped so that they
will drain to gutters and
downspouts, which shall be
connected with conductors
to carry the water down
from the roof underneath
the sidewalk to and through
the curb. Overflows shall be
installed at each low point
of the roof to which the
water drains.
From 1927, the evolution was
gradual. It was not until the 1964
Edition of the UBC that roof drains
were specifically mentioned in the
code. Section 3206 of the 1964 UBC
stated:
Roof drains shall be
installed when required at
each low point of the roof to
which the water drains, and
shall be adequate in size to
drain the roof. Overflow
drains shall be installed
with the inlet flow line located
two inches (2″) above the
low point of the roof, or
overflow scuppers may be
installed in parapet walls at
each low point of the roof
with the flow line not more
than two inches (2″) above
the adjacent roof.
There is no reference to a specific
plumbing code in the Uniform
Building Code until 1967. However, it
should be noted that the National
Bureau of Standards published the
first national plumbing code in 1928
(Publication BH 13). This publication
was commonly referred to as the
Hoover Code. The National Asso cia tion
of Master Plumbers published its
Standard Plumbing Code in 1933.
The American Standards Association
published a preliminary Plumbing
Code A40 in 1942. The Western
Plumbing Code drafted its first
Uniform Plumbing Code in 1938.
After World War II a joint committee,
the Uniform Plumbing Code Com mit tee,
was formed and extensive re search
was performed at the National
Bureau of Standards and several universities,
which ultimately culminated
in the American Standard National
Proceedings of the RCI 25th International Convention Patterson and Mehta 197
Plumbing Code, ASA A40.81955.
Today’s International Building
Code represents the merging of the
three national building codes – the
BOCA Building Code, the Uniform
Building Code, and the Standard
Building Code – into one unified
building code. The goal was to have a
single national building and plumbing
code. The International Building Code
has been universally accepted for the
most part, but there are still three
national plumbing codes, which
include the Uniform Plumbing Code,
the National Standard Plumbing
Code, and the International Plumbing
Code, each with slightly different
philosophies.
Key Drainage Design Issues in
the Codes
The key roof design issues
addressed in the codes are:
1. Roof slope.
2. Structural issues related to ponding,
3. The design of the primary roof
drainage system, and
4. The design of the overflow or secondary
drainage systems.
From the beginning, building
codes have stated the obvious:
1. Roofs should drain,
2. There needs to be a backup
drainage system in case something
happens to the primary drainage
system, and
3. Roof structures should be designed
to support all the anticipated loads
so they do not collapse.
These are pretty simple concepts,
but it is alarming how many roof
drainage problems persist and how
many roofs collapse every year.
Roof Slope
The need for sloping roofs is obvious.
Roofs that drain well last longer
than roofs that do not drain well, and
roofs that do not accumulate water
are not likely to collapse. Clearly,
however, there are still misunderstandings
regarding the issues related
to roof slope. In 1927, the UBC
stated, “Roofs of all buildings shall be
sloped so that they will drain.” How
could it be any clearer? The NRCA
published its first Manual of Roof
Practice and stated: “enough slope
should be builtin
so that water does
not collect in the low areas between
the roofframing
members…”
Unfortunately, the NRCA finished the
sentence with “… so that the roof is
completely dry 48 hours after it stops
raining.” Water standing for 48 hours
does not constitute good drainage,
but this 48hour
standard found its
way into the roofing literature and
became part of the first International
Building Code in 2000. The 2009
International Building Code defines
positive drainage as:
The drainage condition in
which consideration has
been made for all loading
deflections of the roof deck,
and the additional slope has
been provided to ensure
drainage of the roof with 48
hours of precipitation.
The Code further requires lowslope
roofs to slope a minimum of ¼
in/ft except for coaltar
pitch builtup
roofs, which can be sloped 1/8 in/ft It
is important, however, to understand
that there can be serious problems
with roofs with a slope of less than ¼
in/ft. These issues are addressed in
the structural requirements in the
codes for ponding instability.
Structural Issues Related to
Ponding
Roof slope is not only important
for roof performance, but roof slope is
extremely important in the structural
design of a roof. Roof structures with
slopes less than 1/4:12 must be evaluated
for ponding instability. The
concept of ponding instability is not
new. The requirement to specifically
design roofs for ponding water first
appeared in the Uniform Building
Code in 1967 in Section 2305(f).
2305(f) Water Accu mu la tion.
All roofs shall be
designed with sufficient
slope or camber to assure
adequate drainage after the
longtime
deflection from
dead load or shall be
designed to support maximum
loads including possible
ponding of water due to
deflection. See Section 2307
for deflection criteria.
Ponding instability is critical in
roof drainage design, because one of
the typical factors in many roof collapses
involves the failure of the
structure due to progressive deflection
under water loads. As water
builds up on a roof, the roof structure
deflects, and if the slope is inadequate
to permit free drainage, the accumulation
of water will result. Water continues
to build up deeper and deeper
as the structure deflects. This progressive
buildup
of water is one of
the key factors in many collapses.
Below is the section from the 2009
IBC dealing with ponding instability.
1622.2 Ponding instability.
For roofs with a slope
less than ¼ inch per foot
[1.19 degrees (0.0208 rad)],
the design calculations
shall include verification of
adequate stiffness to preclude
progressive deflection
in accordance with Section
8.4 of ASCE 7.
Water weighs 62.4 pounds per
cubic ft (pcf), and therefore, each inch
of water weighs approximately 5.2
pounds per sq ft (psf). It is precisely
this weight of water that provides the
load on most roofs that leads to roof
collapses. The codes have long provided
for live load reductions, which
commonly allow the reduction in the
live load on joists and other roofframing
members to as low a 12 psf
(depending on the tributary area of
the supporting member), which
equals slightly more than a 2in
uniform
water depth on the roof.
Standard deflection criteria for a
roof’s structural elements allows for a
deflection equal to (span/240), which
Patterson and Mehta 198
Proceedings of the RCI 25th International Convention
is two inches of deflection for a 40ft
span. Sloping a roof ¼ in/ft should be
considered the very minimum because
of variations in construction
tolerances. Figure 5 shows a roof that
slopes 1/8 in/ft over a span of 40 ft.
With a deflection of two inches, the
center of the roof’s span is only ½ in
above the lowest end (drainage point)
of the roof. A deviation of ½ inch in
construction tolerances is commonplace,
which implies that a part of the
roof – from midspan
to the eave – has
no slope (2½ inches in deflection and
2½ inches in fall).
Any additional deflection (resulting
from greater construction tolerances
and/or ponding instability) can
result in a serious problem – even a
collapse. Figure 5 also shows that a
roof with the 1¼ in/ft slope has a net
slope of 1/8 in/ft (after full deflection)
from midspan
to the eave, which
eliminates any potential ponding.
This is an important reason for having
a minimum slope of ¼ in/ft. The
International Building Code, which
allows roof slopes to be reduced to
1/8 in/ft for coaltar
pitch, ignores
the importance of positive drainage.
The Code, however, does require that
if the slope is less than ¼ in/ft, the
roof should be investigated for ponding
instability.
Figure 6 (adopted from Roof
Drainage, Ref. 2) shows the importance
of slope and the increase in
depth of water accumulation with
decreasing slope. Decreasing the
slope from ¼:12 to 1/8:12 doubles
the amount of water on the roof. In
other words, the lack of adequate
slope can result in catastrophic consequences
unless the structure is
designed for these loads.
The building codes have taken a
step backward in their requirements
relating to slope, particularly as it
relates to reroofing. The 1988
Uniform Building Code (UBC)
addressed the issues related to roof
slope and reroofing better than any
other code before or after. Presented
below is an excerpt from Chapter 32
in the Appendix to the 1988 UBC.
Figure 5 – A roof slope of 1/8 in/ft provides virtually
zero net slope after the roof deck assumes its
allowable deflection under rain load. A ¼ in/ft roof
slope, on the other hand, leaves a net slope of 1/8
in/ft under the same condition.
Figure 6 – For the same head of water over a drainage element
(scupper or roof drain), rain load on the roof increases with
decreasing roof slope.
Inspections
Sec. 3210. New roof
coverings shall not be
applied without first obtaining
an inspection by the
building official and written
approval from the building
official. A final inspection
and approval shall be
obtained from the building
official when the reroofing is
complete. The preroofing
inspection shall pay special
attention to evidence of
accumulation of water.
Where extensive ponding of
water is apparent, an analysis
of the roof structure for
compliance with Section
3207 shall be made and
corrective measures, such
Proceedings of the RCI 25th International Convention Patterson and Mehta 199
as relocation of roof drains
or scuppers, resloping of
the roof or structural
changes, shall be made.
An inspection covering the
abovelisted
topics prepared
by a special inspector may
be accepted in lieu of the
preinspection by the building
official.
Further, the 1988 UBC required
that the new roof meet the requirements
of Chapter 32 of the Code,
which required the roof to have a minimum
slope of ¼:12. Essentially, the
roof had to be inspected, and if there
were drainage problems, the roof
structure had to be analyzed and the
drainage corrected.
It is a mistake to assume that
because a building has been around
for a long time that the drainage system
is adequate. One of the first collapses
investigated by the author’s
firm involved a 50yearold
building
that never had a drainage problem—
that is until someone tossed a Sunday
newspaper up on the roof. That
Sunday paper was the perfect size to
block the scupper drains on the
building, and unfortunately for the
owner of the building, the roof was
not provided with any overflow
drainage system.
Slope is one of the three critical
elements of roof drainage design. It is
essential that the roof has adequate
slope to drain the roof without accumulating
excessive amount of water.
The following section deals with the
concept of overflow design in order to
prevent excessive water buildup.
Overflow Drainage System—
Controlling Water Depth on
Roofs
Probably the most important element
of roof drainage design is controlling
the depth of water that can
accumulate on the roof. Controlling
this depth is a function of the overflow
drainage system, also known as
the secondary drainage system.
Overflow scuppers are prominently
mentioned in the 1929 Standard
Practice in Sheet Metal Work, which is
the forerunner to the Architectural
Sheet Metal Manual published by the
Sheet Metal and Air Conditioning
Contractors National Association
(SMACNA). Also, overflow was one of
the requirements in the first Uniform
Building Code. Why, then, is it that so
many existing buildings have no overflow
drainage system?
Investigation of a number of roof
collapses has revealed that the collapses
were either due to the absence
or inadequacy of the overflow
drainage system. At some point, it is
likely that a roof drain will be blocked.
It may take 50 years before someone
throws a Sunday newspaper on the
roof or there is a severe hailstorm, or
blowing debris from a hurricane
occurs that is severe enough to block
the drains. At some point, the primary
drainage system will likely become
blocked. Water flowing to the drains
will carry debris and deposit it at the
low point of the roof, which is where
the primary drains are located. Some
drains are more likely to become
blocked than others. Drainage scuppers
are more susceptible to blockage
than roof drains. Figure 7 (adopted
from Roof Drainage, Ref. 2) shows
that a roof drain with a proper strainer
can still function, even with debris
accumulated around the perimeter of
the strainer, whereas the scupper is
virtually blocked by the same debris.
The American Society of Civil
Engineers recognized the importance
of blocked drains and provided the
following recommendations for structural
engineers in the first ASCE
Standard, entitled ANSI/ASCE 788,
Minimum Design Loads for Buildings
and Structures, which replaced ANSI
A58.11982.
8.2 Blocked Drains. Each
portion of a roof shall be
designed to sustain the load
of all rainwater that could
accumulate on it if the primary
drainage system for
that portion is blocked.
Ponding instability shall be
considered in this situation.
If the overflow drainage provisions
contain drain lines,
such lines shall be independent
of any primary drain
lines.
Overflow Drainage
Requirements
There have been revisions of ASCE
7, but the issues related to overflow
Figure 7 – The relative performance of roof drains and scuppers.
Patterson and Mehta 200
Proceedings of the RCI 25th International Convention
have remained virtually unchanged in
ASCE 7 since its inception in 1988,
and also in the International Building
Code, since its inception in 2000. The
structural engineer must assume that
the primary drains are blocked, calculate
the depth of water that could
accumulate on the roof, and design
the structure to support those loads.
Again, these are pretty simple, commonsense
concepts that have been
universally accepted in current design
standards.
However, there are variations in
overflow drainage design theory.
Historically, most primary and secondary
roof drainage systems were
designed based upon the maximum
onehour
duration, 100year
rainfall
rate. This is the maximum amount of
rain that is likely to occur in one
hour, once every 100 years. However,
in 1991, the Standard Building and
Plumbing Codes changed the requirements
for overflow design. The 1991
Standard Plumbing Code adopted the
15minute
rainfall rate for the design
of overflow drainage systems, which
is the maximum rainfall rate expected
in 15 minutes once every 100 years.
To put it in perspective, the 15minute,
100year
rainfall rate is
approximately twice the onehour,
100year
rainfall rate.
1507.3 Maximum Rainfall
Rate for Secondary
Drains. Secondary (emergency)
roof drain systems or
scuppers shall be sized
based on the flow rate
caused by the 100year,
15minute
precipitation as
indicated on Figure 1507.3.
The flow through the primary
system shall not be considered
when sizing the secondary
roof drain system.
The National Standard Plumbing
Code has also adopted this criterion.
The rationale for using the 15minute,
100year
rainfall rate is based on the
concept that roof drains may become
blocked during extreme storms like
hailstorms and hurricanes. The
Standard Plumbing Code was formerly
the Southern Building Code, which
was adopted primarily throughout the
Southeastern region of the United
States, including the hurricaneprone
Gulf and Atlantic Coast states.
Blowing debris during hurricanes
commonly blocks roof drains, and
hurricanes often produce extraordinary
rainfall. The combination of
blowing debris and extraordinary
rainfall can be catastrophic for roof
drainage systems. The first
International Plumbing Code (1997)
also adopted this 15minute,
100year
rainfall rate, but reverted to
using the onehour,
100year
rainfall
rate for overflow systems in 2000.
Also gone is the old Uniform
Building Code standard of using overflow
scuppers three times the area of
the roof drains. The Uniform Building
Code first provided the overflow scupper
option in the 1967 Edition, which
is shown below.
(c) Overflow Drains and
Scuppers. Where roof
drains are required, overflow
drains having the same
size as the roof drains shall
be installed with the inlet
flow line located two inches
(2″) above the low point of
the roof, or overflow scuppers
having three times the
size of the roof drains may
be installed in adjacent
parapet walls with the inlet
flow line located two inches
(2″) above the low point of
the adjacent roof and having
a minimum opening
height of four inches (4″).
Overflow drains shall be
connected to drain lines
independent from the roof
drains.
There is a major flaw in this
drainage design criterion. This criterion
does not take into account the
depth of water that develops at the
scupper based upon the geometry of
the scupper. The fundamental function
of the scupper is to control the
depth of water, so as to limit the load
on the roof structure and prevent a
collapse. The geometry of a scupper is
critical. For example, a 6in
roof drain
has an opening of approximately 28
sq in. Using the old Uniform Building
Code’s standard of providing an overflow
scupper three times the area of
the roof drain, the overflow scupper
size would be 84 sq in. The problem
with this standard was that a designer
could select an 8in
wide by 10.5in
high scupper and meet the code
requirements. However, the head of
water at the scupper would have to be
8.4 in to achieve the required flow
rate. The net result would be a 10.4in
water depth at the scupper (8.4 in
of head plus 2.0 in for the scupper
inlet elevation). This depth of water
could easily cause a collapse in most
circumstances. Based upon the old
UBC, it was possible to collapse a roof
using overflow scuppers three times
the area of the roof drains. Roof
drains are roughly three times more
efficient than scuppers in terms of
flow rate. The 8.0in
wide by 10.5in
high scupper has the flow capacity of
almost 790 gallons per minute (gpm).
The scupper has about three times
the area of a 6in
drain.
Unfortunately, the head of water
required to achieve the 790 gpm of
flow through the scupper is 8.4 in,
while the head of water required to
achieve approximately the same flow
is only about 4.0 in.
In 1991, the Standard Plumbing
Code introduced a method to determine
the depth of water that could
accumulate on the roof and required
the structural engineer to design the
structure based upon this depth of
water, assuming the blockage of primary
drains. This is the basis of
design requirement provided in the
1998 edition of ASCE 7 and the 2000
edition of the International Building
Code. This is the most logical
approach and simply reinforces the
concept that the structural engineer
should design the structure for the
rain loads expected to occur on a roof.
ASCE/SEI 705
provides a basis
for calculating the head of water over
roof drains and scuppers. There are
also methods for calculating the head
Proceedings of the RCI 25th International Convention Patterson and Mehta 201
of water over scuppers and drains (in
Roof Drainage Design, Ref. 2). Table 2
(adapted from Roof Drainage, Ref. 2,
and ASCE/SEI 705)
shows how the
drainage capacity of a roof drain
changes with the head of water.
Table 3 (also from Roof Drainage)
shows the head of water that must
exist over a roof drain to achieve its
(maximum) discharge capacity. For
instance, the discharge capacity of a
5in
drain is 261 gpm, which can only
be achieved if the minimum head of
water above the inlet level of the drain
is 2.5 in.
The overflow drainage system controls
the depth of water that can
accumulate on a roof, and it is the
most critical part of the drainage system.
The contemporary codes, therefore,
require that the roof designer
Discharge rate (gpm)
Roof drain
diameter
Depth of water above drain, hydraulic head (in)
1.0 2.0 2.5 3.0 3.5 4.0
4 in
6 in
8 in
80 170 180
100 190 270 380 540
125 230 340 560 850 1100
Table 2 – Drainage capacities of roof drains as a function of the
head of water (Roof Drainage, Ref. 2).
Roof drain Discharge capacity Corresponding head
diameter (gpm) from Table 3.5 of water (in)
4 in 144 1.75
5 in 261 2.50
6 in 424 3.25
8 in 913 3.75
calculate the depth of water that can
accumulate on the roof, assuming
that the primary drainage system is
blocked, and design the structure to
support those loads. The structure
must be evaluated for ponding instability
for roof slopes less than ¼:12.
Note, however, that these basic concepts
have been articulated in codes
in one form or the other for several
years.
Location of Overflow Drains
and Scuppers
Overflow drains and scuppers
should be located above the low point
of the roof to help prevent them from
becoming blocked. Debris is carried
by the flow of water on the roof, which
is to the low point. The primary drains
are located at the low point, which is
what makes them susceptible to
becoming blocked by debris.
Placing the overflow drain or scupper
approximately two inches above
the low point of the roof reduces the
likelihood that debris will flow into the
overflow system and become blocked.
Additionally, overflow drain outlets
should be located in a prominent
location so that maintenance personnel
can readily observe water flowing
out of the overflow. This is an indication
that the primary drains may be
blocked and maintenance needs to be
Table 3 – Minimum head of water over a roof drain to achieve its
drainage capacity.
performed.
Sometimes it is difficult to place
overflow scuppers two inches above
the low point of the roof due to the
location of the drains. Today, there is
no specific requirement to locate the
overflow drain or scupper two inches
above the low point of the roof. The
requirement is to find out how much
water can accumulate above the overflow
system and design the structure
accordingly. Locating the overflow
drain or scupper approximately two
inches above the low point helps prevent
debris from blocking the overflow,
but the scupper can be located
at a higher elevation, provided the
structure is adequate to support the
load.
Primary Drainage System’s
Design
The primary drainage system is an
important element in drainage
designs, but in many ways, it is the
least important of the other three elements
of drainage design: the slope of
the roof, ponding instability, and the
overflow drainage system. However, it
is important that roofs drain freely,
and the primary drainage system is
designed to remove water efficiently.
The primary drainage system generally
consists of either roof drains or
scupper drains. Most roof drains
today are manufactured by companies
like Josam, Zurn, and J.R.
Smith, and have standard flow rates
that are reflected in various plumbing
codes. These drains are designed with
sumps and strainers that conform to
the codes. Strainers are important, as
they block the debris from getting into
the drain lines. Also, strainers can
actually improve the flow into drains
by breaking up the vortex of the water
flowing into the drain.
Scupper drains, on the other
hand, are generally shopor
fieldfabricated,
and normally, flow rates must
be calculated. There are no standard
strainers designed to promote water
flow and function with debris.
Scuppers also generally require a
greater depth of water to achieve the
designed flow, so the depth of water at
scuppers can be significant. Scupper
drains and overflow drains should be
separate and should not be connected.
Patterson and Mehta 202
Proceedings of the RCI 25th International Convention
Table 3.5 – Size of vertical conductors as per the National Standard Plumbing Code and the Uniform Plumbing Code
Pipe diameter Design flow Maximum drainable (projected) roof area (sq ft)
(in) (gpm)
Rainfall intensity (in/h)
1 2 3 4 5 6
2 23 2,176 1,088 725 544 435 363
3 67 6,440 3,220 2,147 1,610 1,288 1,073
4 144 13,840 6,920 4,613 3,460 2,768 2,307
5 261 25,120 12,560 8,373 6,280 5,024 4,187
6 424 40,800 20,400 13,600 10,200 8,160 6,800
7 913 88,000 44,000 29,333 22,000 17,600 14,667
Adapted from the National Standard Plumbing Code, 2003, and Uniform Plumbing Code, 2003, with permission. This table is
the same as Table 2.1(a).
Table 3.6 – Size of vertical conductors as per the International Plumbing Code
Pipe diameter Maximum drainable (projected) roof area (sq ft)
(in)
Rainfall intensity (in/h)
1 2 3 4 5 6 7 8 9
2 2,880 1,440 960 725 575 480 410 360 320
3 8,800 4,400 2,930 2,147 1,760 1,470 1,260 1,100 980
4 18,400 9,200 6,130 4,613 3,680 3,070 2,630 2,300 2,045
5 34,600 17,300 11,530 8,373 6,920 5,765 4,945 4,325 3,845
6 54,000 27,000 17,995 13,600 10,800 9,000 7,715 6,750 6,000
7 116,000 58,000 38,660 29,333 23,200 19,315 16,570 14,500 12,890
Copyright 2003, International Plumbing Code, Falls Church, Virginia: International Code Council, Inc., reproduced with permission.
All rights reserved. This table is the same as Table 2.1(b).
Table 4 – Drainage capacities of roof drains as provided in the three plumbing codes.
Drainage Capacities for Roof
Drains
The various plumbing codes generally
agree on the design criteria in
principle. However, there are variations
in the drainage capacities of roof
drains between the International
Plumbing Code and the other two
plumbing codes—the National
Standard Plumbing Code and the
Uniform Plumbing Code. Table 4
(adopted from Roof Drainage, Ref. 2)
shows these variations.
The National Standard and
Uniform Plumbing Codes are more
conservative. All these flow rates are
based upon Manning’s equations, but
there are slightly different assumptions
regarding the amount of open
area in the pipe, which results in the
differences in the charts. Roof drain
manufacturers also publish drainage
design literature, and most of these
manufacturers use the same drainage
design assumptions as the International
Plumbing Code’s standards.
Scuppers as Drainage Elements
and Their Drainage Rates
There are no design provisions for
scuppers in the International Build ing
Code or the International Plumb ing
Code, so scupper drains fall into
the category of “Alternate Methods of
Construction.” Section 105 of the
2009 International Plumbing Code
provides the requirements for
“Alternate Methods of Construction,”
and these requirements specifically
state (Sections 105.4.1 through
105.4.6):
1. The alternate design shall
conform to the intent of
the provisions of the code.
2. The registered design professional
must indicate on
the plumbing permit
application that the system
is an alternative engineered
system.
3. The registered design professional
shall submit
technical data to substantiate
the proposed alternative
design.
4. The registered design professional
shall submit two
sets of signed and sealed
construction documents
for the alternative design.
Proceedings of the RCI 25th International Convention Patterson and Mehta 203
Head of Discharge capacity of scupper, Q (gpm)
water, H, in
scupper (in) Length of scupper, L (in)
6 8 10 12 18 24 30 36 42
0.5* 6.3 8.4 10.5 12.6 19.0 25.3 31.7 38.2 44.5
1.0 17.4 23.4 29.4 35.4 53.4 71.4 89.4 107.4 125.4
1.5 31.4 42.4 53.5 64.5 97.6 130.6 163.7 196.8 229.8
2.0 47.5 64.5 81.5 98.4 149.3 200.3 251.2 302.0 353.0
2.5 65.2 88.9 112.7 136.4 207.5 278.7 349.8 421.0 492.1
3.0 84.2 115.4 146.5 177.7 271.2 364.8 458.3 551.8 645.4
3.5 102.7 141.4 180.2 218.9 335.2 451.4 575.6 683.9 800.1
4.0 124.8 172.8 220.8 2688 412.8 556.8 700.8 844.8 988.8
Table 2.3 – Drainage Capacity of a Rectangular Scupper (gpm)
Note: the accuracy of discharge capacity values for 0.5 inch head Note: Scupper length, L, below this (stepped) line is less than four times
of water may not be accurate due to the effect of surface tension. the head, 4H. These scupper dimensions should be avoided.
Table 5 – Drainage capacities of rectangular scuppers relative to the head of water.
Of course, there are design data
for scuppers, but there are also variations
in the formulas used for calculating
the flow through a scupper. It is
important to understand that water
has to build up to a relatively significant
depth on the roof to achieve the
design flow rate though the scupper.
The depth of water that develops is
primarily an issue of the width of the
scupper. The wider the scupper, the
lower the head of water that will
develop at the scupper, which is
desirable even if the structure is
designed to support the loads from a
large head of water.
The flow rate of water through
scuppers is generally determined by
the derivation of an equation known
as the Francis Formula, which is Q =
3.33LH1.5, where Q is the flow rate, L
is the length of the weir (scupper),
and H is the head of water. Because
experiments have shown there is a
contraction in the water flowing
through the weir, the equation has
been modified to adjust for this
reduction. The modified form is Q =
3.33(L0.2H)
H1.5. Table 5 (from Roof
Drainage, Ref. 2) has been derived
from this equation.
The design of the primary drainage
system is relatively straightforward.
There are variations in flow rates of
drains and scuppers, and further
research into these variables would
be helpful in establishing consistent
design guidelines.
Drainage Design Requirements
for Roof Replacements
The 1967 Uniform Building Code
added Chapter 32 to the Appendix of
the code, which was titled ReRoofing,
and the first section (3209) in that
chapter stated that all rroofing
had
to comply with Section 32 in the
Building Code. This was a significant
change in the code. However, the
most significant change came in 1988
with the addition of the statement
that roof systems shall be sloped a
minimum ¼ in/ft for drainage. This
requirement to provide a minimum
slope in Chapter 32 of the 1988 UBC,
in combination with the changes
made to Chapter 32 in the Appendix
for Reroofing, had enormous implications
for reroofing design. Chapter 32
in the Appendix required a reroof to
conform to Chapter 32 of the Code,
which required the roof to be sloped a
minimum ¼ in/ft.
There was no reference to draining
within 48 hours or allowing 1/8 in/ft
for coaltar
pitch. From a fundamental
design perspective, this was the
most appropriate code dealing with
roof drainage and roof slope. A minimum
slope of ¼ in/ft has long been
recognized as the most appropriate
minimum slope for good drainage. A
minimum ¼ in/ft is also important
from a structural design perspective,
as any roof with less than this slope
has to be designed for ponding instability.
Deflections in structural elements
with less than ¼ in/ft slope
can result in a progressive collapse
due to deflection. In other words, the
roof deflects, allowing more water to
accumulate until the roof collapses.
Clearly, the authors of the 1988 UBC
were addressing the issues that cause
roof collapses. These changes in the
1988 UBC were met with less than an
enthusiastic response from elements
of the roofing community. A great
number of existing buildings did not
have a minimum ¼ in/ft slope. In
some cases, it was not only impractical,
but it was virtually impossible to
provide the minimum ¼ in/ft slope.
Then there were the issues from the
coaltar
pitch industry, where a ¼
in/ft slope could be too much slope
for the system. The result of these
issues and others was a lessening of
the requirements.
Current Drainage Design
Standards
Today’s International Building
Code is relatively ambiguous regarding
positive drainage. As defined, positive
drainage is based upon ensuring
drainage within 48 hours of precipitation.
As previously stated, water
Patterson and Mehta204
Proceedings of the RCI 25th International Convention
standing for 48 hours does not constitute
good drainage. What does 48
hours from precipitation mean? Does
it mean 48 hours in summertime conditions
or wintertime conditions? A
properly sloped roof should drain
freely. Other than anomalies in the
roof created around penetrations or
crickets and valleys, there should be
no water ponding after a rain.
The code requires reroofs to have
positive drainage, but since the definition
is ambiguous and enforcement
is difficult. Some building officials
have ruled that the design professional
is responsible for making the determination
of what constitutes positive
drainage, but often there is no design
professional in the case of reroofing.
Certainly, positive drainage is a benefit
in terms of roofing longevity and
performance, and providing ¼ in/ft
eliminates many structural concerns.
In those cases where achieving ¼
in/ft is not practical, care should be
taken to help limit the amount of
water that can accumulate on the
roof, and involving a structural engineer
should be considered.
It is important to understand that
simply resloping a roof with tapered
insulation may not be adequate. It is
imperative that the drainage system
function properly after the tapered
insulation is installed. Often, increases
in insulation thickness can restrict
drains and overflow systems.
One of the most important re quire
ments for reroofing is to make
sure there is a proper overflow system.
This is a code requirement that
is often overlooked by many in the
roofing industry, but the overflow system
is critical in terms of limiting the
amount of water that can accumulate
on a roof and in preventing roof collapses.
As a rule of thumb, some try
to limit the depth of water to a maximum
of four inches, which prevents
the load from water buildup
from
exceeding the minimum 20 psf live
load used throughout much of the
southern regions of the country. It is
also important to understand that
even though there is a minimum 20
psf liveload
requirement in the code,
there are liveload
reductions that are
allowed by code that can reduce live
loads on certain structural elements
to as low as 12 psf.
FINAL COMMENTS ON
DRAINAGE DESIGN
The basic concepts of proper roof
drainage design have been around for
many years, and there are extensive
data and design guides available. The
Roof Drainage monograph (Ref. 2)
published by the RCI Foundation,
provides a much more complete discussion
of roof drainage, and every
roof consultant should have a copy in
his or her library. There are still
issues that need clarification and
additional research is needed, particularly
in the area of water accumulation
on roofs and the appropriate flow
rates of drains and scuppers.
The roof consultant can play an
important role in preventing roof collapses.
It is essential that there is a
properly functioning overflow system
on a building. Checking roofs for an
appropriate overflow system and recommending
corrective action to add
or enlarge overflow drains and/or
scuppers should be a standard part of
the roof investigation process.
Further, improving the drainage and
the design and installation of overflow
systems should be part of a reroofing
project where these systems are inadequate.
LIFE SAFETY ISSUES IN WIND
UPLIFT DESIGN OF A ROOF
Few catastrophes are as dramatic
as a major roof collapse or 17,000
buildings burning to the ground.
However, wind damage to roofs and
buildings can be just as dramatic
when tornados or hurricanes make a
direct strike on a major city. An F5
tornado made a direct hit on the city
of Lubbock, Texas, on May 11, 1970,
and once again, the 1970s proved to
be a pivotal time for roofing design.
The Great Plains Life Building, a 20story
office building in downtown
Lubbock, was in the direct path of the
F5
tornado. The building sustained
major damage (Ref. Wikipedia). The
faculty members in the College of
Engineering in Texas Tech University,
Lubbock evaluated the damaged
building and found it restorable. The
building remained vacant in damaged
condition until 1975, when it was
restored and renamed the Metro
Tower.
Most of the building’s structure
was sound, despite major damage
caused to some areas that were
stressed beyond their elastic limit and
were permanently deformed. The permanent
deformations in the structure
provided an opportunity for Texas
Tech engineering faculty to determine
the loads required to cause them,
allowing them to indirectly calculate
the wind loads imposed by the tornado.
The 1970 Lubbock tornado provided
one of the first opportunities to
study the pressures associated with a
major tornado. Normally, F5
tornadoes
completely destroy what they
hit, so there is little left to evaluate.
This time, the tornado hit a highrise
building. This tornado and the subsequent
evaluations of the damages
associated with this tornado were significant
to the understanding of wind
pressures, moving Texas Tech
University to the forefront of research
related to wind pressures on buildings.
Wind uplift issues on roofs were
not something that was dealt with in
the early codes. Most roofs were
mopped or nailed to wood or concrete
structures, and wind was not much of
an issue on that type of construction.
The methods of attaching roofs to
buildings became a major issue after
the 1953 General Motors fire in
Livonia, Michigan. The method of
adhering the insulation to the steel
deck on the Livonia GM plant had
contributed to a massive fire loss.
Studies by both UL and FM showed
that the bitumen used to attach the
roof could contribute significantly to a
fire that developed on the inside of a
building. Construction standards for
adhering insulation to steel decks
changed. The new standards required
that the adhesives used to adhere
insulation to a steel deck had to be
tested and approved and that the
application of the adhesive must be
limited to the top rib of the steel deck.
Proceedings of the RCI 25th International Convention Patterson and Mehta 205
Insulation was still adhered to steel
decks until the 1980s.
In 1983, FM reported (FM 12883)
that just over 30% of the wind losses
between 1971 and 1980 involved losses
to steel decks, and in losses where
the insulation lifted off the deck, there
were inadequate adhesives, particularly
at the edge and perimeter.
Further, FM’s investigations indicated
that mechanically attached insulation
fared far better than adhered insulation.
FM stopped approving the
installation of insulation to steel
decks using adhesives and only
approved mechanically attached in sulation.
This standard was adopted
industrywide
by the mid to late
1980s. FM also recognized that wind
losses were particularly problematic
at the perimeter and in the corners. In
1983, FM required that the fasteners
be increased by 50% in the corners of
FM 160
systems and increased by
50% in the both the corners and the
perimeter of the FM 190
systems.
Factory Mutual provided the first
sophisticated roof design guidelines
for wind uplift pressures. These
design guidelines included charts to
determine the appropriate wind uplift
pressures on roofs for various exposures
and heights. In addition, FM
provided lists of roofing systems that
had been tested to meet those wind
uplift pressures. FM has long been at
the forefront of roof design issues and
by the 1980s had become the de facto
design standard
for wind
uplift on roofs.
Factory Mu tual
tested more
lowslope
roofing
systems for
wind uplift
than any other
agency and
provided so phisticated
de sign
standards
for wind uplift
on roofs. The
codes ad dressed
wind
uplift, but there
was little guidance
for roof design. Most roofs were
de signed using FM standards.
The national codes and FM used
the fastest mile wind speeds until
2000, when the first IBC was published.
However, seven years earlier,
the American Society of Engineers
(ASCE) changed from using fastest
mile wind speed to using threesecond
gust wind speeds (ASCE 793
Minimum Design Loads for Buildings
and Other Structures). The 2000
International Building Code adopted
the ASCE standards. In 2000, there
was a divergence between the code
requirements and Factory Mutual.
ASCE 793
was the most current and
accurate design standard of the day,
but the use of the ASCE 7 design
standards was slow to take hold in
the roofing design community. How ever,
by the year 2000, ASCE 7 was
clearly the design standard for wind
uplift. Even so, many (perhaps most)
in the roofing design community continued
to use the FM design standards.
Fundamentals of Wind Uplift
Design
The standard wind speeds used
for wind uplift design are based upon
the threesecond
gust wind speed
with a 50year
recurrence interval.
These wind speeds are based upon
measurements taken at 33 ft above
the ground in an area with relatively
open terrain and are provided as wind
speed contours superimposed on the
map of the United States. The pressure
related to wind speed is referred
to as the velocity pressure (q) and is
obtained from Bernoulli’s Equation.
The velocity pressure for a wind speed
of 100 mph would be 25.6 psf. It is
important to realize that the velocity
pressure is a function of the square of
the wind velocity, so a 50mph
wind
would only have a velocity pressure of
6.4 psf, or one fourth of the pressure
at 100 mph. With respect to wind
pressure, Bernoulli’s Equation is generally
stated as:
q = 0.00256 V2,
where q = velocity pressure in lbs per
sq ft (psf) and V is the wind speed in
miles per hour (mph).
Bernoulli’s Equation is the basis
for determining wind uplift pressure
on roofs, but it is only the first step.
There are a variety of other factors
that must be taken into consideration
when calculating the wind uplift pressure
on roofs. It is helpful to understand
how wind pressures develop on
roofs. The wind uplift pressure occurs
as a result of differences in pressure.
The atmospheric pressure is lowered
when wind blows across the roof. This
is the Bernoulli Effect, and this effect
allows airplanes to fly. As an airplane
picks up velocity, the air pressure on
the top of the wing is lowered, resulting
in lift. The same thing happens on
a roof. Wind lowers the pressure on
Figure 8 – Uplift pressure on a roof is equal to the difference between the interior
air pressure (generally the atmospheric pressure of 2,100 psf) and the exterior air
pressure.
Patterson and Mehta 206
Proceedings of the RCI 25th International Convention
Figure 9 – Wind uplift pressure on a mechanically fastened singleply
roof on a steel deck.
top of the roof, and the difference
between the interior pressure and
exterior pressure results in wind
uplift pressure. Figure 8 shows a
building with no wind, where the air
pressure inside and outside are equal
(to atmospheric pressure of 2,100 psf)
and thus there is no uplift pressure
on the roof. This figure also shows a
building subjected to wind, which
lowers the pressure on top of the roof
(to say 2,050 psf), while the interior
pressure remains at the atmospheric
pressure (of 2,100 psf), resulting in
uplift pressure of 50 psf on the roof.
It is also helpful to understand
how this difference in wind uplift
pressure results in wind uplift on the
roof. Wind uplift pressure develops as
a result of the difference in pressures
between the inside and the outside of
the building. The uplift pressure acts
on the element that prevents the wind
pressure from equalizing. This is generally
the roof, which provides an airtight
or semiairtight
barrier between
the inside and outside of the building
that we call the critical layer. Figure 9
shows how the wind uplift pressure
develops on a mechanically attached
roof system.
Bernoulli’s Equation gives us wind
pressure and Bernoulli’s Effect
explains how wind pressure develops.
The next step is to modify the equation
to develop a velocity pressure for
a specific building. As stated earlier,
wind speed maps are based on threesecond
wind gusts measured 33 ft
above the ground in relatively open
terrain. In order to determine the
wind uplift pressure on a roof, the
basic velocity pressure for the given
wind speed must be modified by certain
factors to adjust the pressures
based upon the building parameters.
These factors include an allowance for
ground roughness, because the winds
are different, depending upon the
roughness of the terrain. The more
open the terrain, the higher the wind
speed; and conversely, the rougher
the terrain, the lower the wind speed.
These variations in ground roughness
are categorized into Surface
Roughness Categories B, C, and D.
There was a Surface Roughness A for
downtown urban terrain, but this category
was eliminated. Also, up to a
point, the wind speeds increase with
an increase in elevation, so the winds
are higher on top of a tall building
than on top of a short building. The
pressure formula is modified based
on the ground roughness and height
of the building in order to adjust the
wind pressure determined using the
standard wind speed maps to the
actual conditions of the building. This
modifier is called the Velocity
Pressure Exposure Coefficient, Kz.
Figure 10 illustrates how the height of
the building and the ground roughness
affect the wind speed and gives a
description of the Ground Roughness
Categories.
The final equation includes the
Importance Factor (I), the Direc tion ality
Factor (Kd), and Topo graphical
Factor (Kzt), which further refine the
equation. The Importance Factor is
like a safety factor that increases or
decreases the load, depending upon
the importance of the building. The
directionality factor reduces the load
based on the low probability that the
maximum wind speed will occur from
the direction that produces the maximum
wind pressure. The Topographical
Factor increases loads to
account for buildings located on elevated
topography. The final equation
for velocity pressure is:
qz = 0.00256(Kz)(Kzt)(Kd)V2)(I)
This equation gives us the velocity
pressure for a specific building. The
velocity pressure has now been
adjusted for the height of the building,
the terrain, the type of building,
and the topography. There is one
more key issue in developing wind
Proceedings of the RCI 25th International Convention Patterson and Mehta 207
Figure 10 – The effect of ground roughness categories on wind speed, and hence on wind uplift
pressure on the roof.
uplift pressure, and that involves ballooning,
which occurs as a result of
an increase in internal pressure
caused by an opening on the windward
side. Ballooning typically occurs
when a door or window on the windward
side of the building blows in,
becomes trapped, and increases the
interior pressure. The design for wind
uplift loads on the roof must be
increased in these cases. Figure 11 is
an illustration of the ballooning effect
where there is an opening on the
windward side and no (or extremely
small) openings on the other sides to
allow the additional pressure to dissipate.
Such buildings are referred to
as “partially enclosed” buildings by
wind standards and codes.
The next step is to convert the
velocity pressure to uplift pressure on
the roof. This step involves the pressure
zones on the roof and the internal
and external pressure coefficients.
Wind uplift is higher along the
perimeter and highest at the corners.
Typically, there are three zones:
1. The field of the roof,
2. The perimeter of the roof, and
3. The corner of the roof.
Patterson and Mehta 208
Each zone has a different wind
uplift pressure, and there is a different
External Pressure Coefficient
(GCp) for each zone. Table 6 (designated
as Table 8 in Wind Pressures on
Low Slope Roofs, Ref. 3) shows typical
zones on roofs and the different coefficients
used in the equation to develop
wind uplift pressure.
As previously discussed, the uplift
pressure on the roof is the difference
between the internal and external
pressures. ASCE gives us the Ex ter nal
Pressures Coefficient (CGp) based
upon the zone on the roof (field,
perimeter, and corner) and the height
of the building (above and below 60
Figure 11 – Balooning effect in a partially enclosed building.
Proceedings of the RCI 25th International Convention
Table 6 – External pressure coefficients and wind zones on a
lowslope
roof.
ft). ASCE also provides the Internal
Pressure Coefficient (GCpi), as shown
in Table 7 (designated as Table 9 in
Wind Pressures on Low Slope Roofs,
Ref. 3).
The formula for calculating wind
uplift pressure on roofs is shown
below with pressure (p) equal to the
velocity pressure (qh) times the difference
between the Internal Pressure
Coefficient (CGp) and External
Pressure Coefficient (CGpi).
p = qh [(GCp) – (GCpi)] For ease of use, these equations
have been converted into tabular format
that provides winduplift
pressure
for various wind speeds, height
of the roof, and various wind zones (in
Wind Pressures on Low Slope Roofs,
Ref. 3). Table 8 shows one such table.
Factory Mutual Design
Standards
Factory Mutual (FM) has long been
a leader in roof design and for many
years its ratings have provided the de
facto standard for designing roofs to
resist wind uplift pressures. The FM
design standards for wind uplift are
based upon the same concepts
described above. In addition, however,
FM provided design guidelines that
allowed the designer to take the wind
uplift pressures and select a roof
assembly that would resist those
uplift pressures. One major advantage
for FM was the comprehensive
testing performed at FM’s laboratories.
Roofing manufacturers had to
test their roofing assemblies in FM’s
laboratories in order to sell their roofing
systems to FMinsured
clients. As
a result, FM developed extensive,
unparalleled data on the wind uplift
resistance of almost every roofing
assembly available.
In addition to the extensive testing,
FM developed charts for wind
uplift pressure for different building
heights, different wind speeds, and
different ground roughness coefficients.
The roof designer selected the
ground roughness coefficient, the
wind speed from the
Table 9 – Values of Internal Pressure Coefficients
(GCpi) for a Building of Any Mean Roof Height
Enclosure Classification (GCpi)
Open buildings 0.00
Partially enclosed buildings 0.55
Enclosed buildings 0.18
map, and the wind
uplift pressure from
the chart. The
designer’s next step
was to select a roof
system designed to
resist those pressures
from FM’s
database.
Historically, there
Table 7 – Internal pressure coefficients as a were three pressure
function of the enclosure classification of zones and two roof
the story below the roof.
Table 8 – Wind uplift pressures on various wind zones of a lowslope
roof as a function of roof heights and site exposure.
Proceedings of the RCI 25th International Convention Patterson and Mehta 209
classifications. Zone 1 was for pressures
up to 30 psf, and roofs with FM
Class 160
were acceptable. Zone 2
was for pressures between 31 and 45
psf, and roofs with FM Class 190
were acceptable. Zone 3 was for pressures
above 45 psf, and only monolithic
decks like concrete were
approved. Roofs tested to meet FM
Class 160
were tested for pressures
up to 60 psf, and these roofs were
acceptable for design wind uplift pressures
up to 30 psf. Roofs tested to FM
Class 190
were tested to pressures
up to 90 psf and were acceptable for
wind uplift pressures up to 45 psf.
This established the safety factor of
two for wind uplift on roof assemblies.
There were additional requirements
for the roof systems. FM recognized
that the wind uplift pressures
were greater along the perimeter and
in the corners of the roof, so FM
required enhancements in the attachment
at the corners and perimeter. In
1983, FM 128
required that fasteners
be increased by 50% in the corners
for Class 160
roofs and required
that fasteners be increased by 50%
both at the perimeter and in the corners
for Class 190
roofs. However,
there appeared to be a disconnect
between the actual pressures occurring
at the perimeter and corners and
the attachment enhancements recommended
in FM design standards.
The 1983 FM 128
Loss Prevention
Data Sheet stated that the uplift pressures
for the perimeter could be calculated
by multiplying the field pressures
by two, and the corner pressures
by multiplying the field pressures
by three. However, FM only
required increasing the fasteners by
50% in the corners for Class 160
roof
assemblies and by 50% at the perimeter
and in the corners for Class 190
roof assemblies.
This disconnect was somewhat
rectified in 1991. FM required that
the fasteners be increased by 50%
along the perimeter and 100% in the
corners for Class 190
roof assemblies,
but made no change in the
requirements for FM Class 160.
FM
also recognized an increase in pressure
on roofs from ballooning and
required the pressures be increased
by 25% for partially enclosed buildings.
Otherwise, there were no other
significant changes. Roofs were classified
as either Class 160
for design
uplift pressures up to 30 psf or Class
190
for design wind uplift pressures
up to 45 psf. Steel decks were not recommended
for pressures over 45 psf.
The requirement to increase fasteners
by 50% at the perimeter and 100% in
the corners for the Class 190
systems
more closely represented the
increased uplift pressures occurring
at the perimeter and in the corners.
FM made substantial changes in
1996. The highest wind uplift rating
for roofs was FM Class 190
until
1996, at which time FM went from
three zones to five zones and from a
maximum of FM Class 190
to a maximum
FM Class 1180.
FM also
increased pressures resulting from
ballooning to 50% instead of 25%.
Curiously, however, FM lowered the
enhancement requirement in the corners
from an increase in fasteners of
100% to an increase in fasteners of
75%. This lowering of the enhancement
requirement was especially
interesting considering FM required
basesheet
fasteners to be increased
by 75% at the perimeter and by 160%
in the corners, which was more in line
with the actual increases in pressures.
Again, there appeared to be a
disconnection between the enhancement
requirements and the actual
uplift loads at the perimeter and corners.
ASCE adopted the threesecond
gust wind speed as its standard in
1993, but FM maintained the fastest
mile wind speed standard until 2002.
There were major changes in the
codes and design standards in 2000.
The three national building codes
were consolidated into the
International Building Code (IBC),
which was published in 2000. The
2000 IBC adopted ASCE 7 as the
standard for wind design, which
included using the threesecond
gust
wind speed. This change also effectively
raised the building code wind
uplift standards above FM’s standards.
Roofing Design and Practice
(Ref. 4) was also published in 2000 by
Prentice Hall and recognized the discrepancies
between FM’s design standards
and the new IBC requirements.
FM’s wind uplift pressures were below
the IBC requirements and the FM
enhancements at the perimeter and
corners did not compensate for the
increase in wind uplift pressures in
these areas. Roofing Design and
Practice presented a design standard
for wind that included the requirement
to design the roof system for the
actual wind uplift pressures in the
field, perimeter, and corners. This
was a significant departure from FM’s
design standards and was necessary
in order to ensure the wind uplift
design met the requirements of the
IBC. This departure was also recognized
in the 2000 AIA Master
Specification in Section 1.4.C Roof
System Design, which stated the following:
1.4.C Roof System Design:
Provide a roofing system
that is identical to systems
that have been successfully
tested by a qualified testing
and inspecting agency to
resist uplift pressures calculated
according to ASCE
7.
Corner Uplift Pressure:
<Insert Number> psf
Perimeter Uplift Pressure:
<Insert Number> psf
FieldofRoof
Uplift Pres sure:
<Insert Num ber> psf
FM again revised FM 128
in 2001.
FM retained the fastest mile wind
speed, which maintained FM’s uplift
pressures well below the ASCE 7
standard and IBC. However, FM eliminated
zones and introduced a design
standard similar to that included in
Roofing Design and Practice and the
AIA Master Spec. The new FM design
standard was based upon taking the
actual wind uplift pressure, multiplying
the uplift pressure by two (for a
safety factor), and then selecting a
roof system that was tested to meet
the design winduplift
pressure. FM
Patterson and Mehta 210
Proceedings of the RCI 25th International Convention
provided a chart showing the system
ratings that were required for the
field, perimeter, and corner, which
was in line with the actual design
requirements based on actual load. If
the roof had an uplift pressure of 30
psf, an FM 160
system would be
used in the field, an FM 190
system
would be used at the perimeter, and
an FM 1135
system would be used in
the corner. This method provided a
roof system that was tested for the
uplift pressures in the field, perimeter,
and corners. Below is an example
of FM’s 2001 chart for selecting an
appropriate roof.
Field Perimeter Corner
160
190
1135
190
1105
1165
1105
1150
1225
The new method still used the
uplift pressures with the lower wind
speeds from the fastest mile wind
maps, so the design pressures were
below the ASCE and IBC standards.
This new method of selecting a roof
system based on the actual uplift
pressures resolved the design problems
associated with perimeters and
corners. It should have been obvious
that increasing the fastening pattern
of an FM Class 160
system by 75%
would in no way increase the rating of
the roof system to an FM Class 1135.
In fact, there were very few roof systems
over steel decks that would meet
a 1135
system on a steel deck in
2001. Again, FM retained the option
to increase the fasteners by 50% at
the perimeter and 75% in the corners,
even though it appeared to be clear
that the enhancement standards were
inadequate. In 2002, FM finally
adopted the ASCE 798
standard
based upon the threesecond
gust
wind speed, which brought FM design
standards in line with ASCE and IBC
standards. The differences in uplift
pressure using the different wind
speeds were significant. Below is an
example of a typical distribution center
in Houston, Texas.
FM finally amended its standards
in 2007 and brought the FM design
standards in line with the ASCE 7
standard and IBC. In 2007, FM only
2001 FM Standard Using Fastest Mile 2002 FM Standard Using 3Secong
Gusts
Wind Speed: 90 mph Wind Speed: 120 mph
Pressure: 34.5 psf Pressure: 54 psf
System: FM 175
System: FM 1120
permitted enhancements in the
attachment under very restricted circumstances.
Enhancements can only
be used for buildings in nonhurricaneprone
regions (with wind speeds
of 90 mph or less) with roof heights
below 75 ft where the partially
enclosed rating does not exceed 175.
The roof height is further limited to 30
ft for buildings located in areas with
ground exposure D. The enhancement
options were also significantly
increased. The perimeter enhancement
was maintained at 50% but with
the restriction of having at least one
fastener per two sq ft, which results
in 16 fasteners for a 4x
8ft
insulation
board. The corner enhancement
was changed to one fastener per sq ft
or 32 fasteners for a 4x
8ft
insulation
board. These were substantial
increases in the previous enhancements
of 50% at the perimeter and
75% in the corners.
While Factory Mutual’s data
sheets became de facto design standards
in roof design, the FM standards
technically only applied to FMinsured
projects and were not a substitute
for designing roofs to meet the
building codes or to meet the most
current design standards from ASCE
7. Until 2007, the design of roofs
using FM design standards likely
resulted in roofs that were underdesigned
when wind speeds
approached the upper limits provided
for in the codes. The most vulnerable
area of the roof was the corner, where
FM inexplicably reduced its requirements
for enhancements in 1996 and
failed to amend the enhancement
requirement until 2007. Certainly,
there have been many wind uplift roof
failures since 1993, but it is important
to note that the vast majority of
the roofs designed using FM standards
have performed well and have
not blown off.
The deficiencies in the FM wind
uplift design criteria are mitigated by
several factors. First of all, it is relatively
unusual for the wind speeds to
reach the maximum design wind
speeds. The wind uplift pressure is a
function of the square of the velocity,
so the uplift pressure associated with
an 80mph
wind is 64% of the wind
pressure from a 100mph
wind.
Therefore, roofs are rarely exposed to
the full uplift pressure. Another mitigating
factor is the safety factor of
two. Doubling the uplift pressure is a
substantial safety factor, especially
when the loads are high. The safety
factor is 30 psf in an FM Class 160,
but the safety factor is 120 psf in an
FM Class 1120.
Another mitigating
factor is the gaps in the FM design
zones. In 1996, a roof with an FM
design winduplift
pressure of 31 psf
would fall into an FM Zone 2, requiring
an FM Class 190,
tested for a
winduplift
pressure of 90 psf. This
effectively increased the safety factor
to almost three. Also, roof designers
often simply specified an FM Class 190
roof assembly, even when an FM
Class 160
was required.
COMBINED IMPACT OF RAIN
LOADS AND DOWNWARD WIND
PRESSURES
A commonly ignored consideration
with respect to the structural design
of lowslope
roofs is the downward
wind pressure on them. Although
wind uplift is a major concern with
lowslope
roofs, the downward pressure
cannot be ignored in many situations.
For lowheight
roofs in lowwind
regions, the downward wind pressure
is lower than the roof’s live load.
Ignoring it is, therefore, of little or no
consequence. However, in several situations
(such as in highwind
regions
and/or tall buildings, the downward
wind pressure may be substantially
larger than the roof live load, and can
Proceedings of the RCI 25th International Convention Patterson and Mehta 211
cause serious problems when it
occurs in combination with high rain
load on the roof due to inadequate
drainage.
Additionally, during severe
storms, the rainfall intensity over a
short duration (5,
10,
and 15minute
durations) is far greater than
the standard design intensity of one
hour. These intense rainfall intensities
also typically occur during conditions
of high wind speeds. The combination
of downward wind pressure
(resulting from high wind speeds) and
the additional load occurring as a
result of higherthandesigned
rainfall
can become catastrophic for
buildings that are designed to the
very minimum standards allowed by
the codes. The codes and standards
must require architects and engineers
to evaluate such a possibility.
CONCLUDING REMARKS
The design of roofs has evolved
over the years with changes in our
understanding of the roofrelated
life
safety issues. The evolution in our
understanding has resulted in
changes in codes and design standards.
The National Board of Fire
Underwriters (Ref. 4) made major contributions
at the beginning of the
twentieth century and literally provided
the forerunner of the modern
building code. UL and FM have made
major contributions to roof design for
more than 100 years. Codes and
standards continued to evolve over
the years with the development of the
modern building and plumbing codes.
In the 1970s, NRCA published the
first NRCA Roofing Manual and the
AIA commissioned the Manual of
Builtup
Roofing, both of which contributed
significantly to the body of
literature on roofing and roof design.
Soon after, the International Roofing
and Waterproofing Consultants and
the Roof Consultants Institute were
established, promulgating ethical
standards and laying the foundation
for the field of professional roof
design. The American Society of Civil
Engineers and the Structural
Engineering Institute also contributed
significantly toward both wind and
drainage design, particularly in the
1990s.
Major changes occurred in the
decade beginning in 2000. The first
International Building Code and
International Plumbing Code were
published in 2000 and were the consolidation
of the three national codes.
Roofing Design and Practice (Ref. 4)
was published in 2000 and was the
first roofing book that specifically
addressed roofrelated
life safety
issues. The RCI Foundation commissioned
and published Wind Pressures
on Low Slope Roofs (Ref. 3) and
Drainage Design (Ref. 2), which were
comprehensive design manuals related
to wind and drainage design
respectively.
There is still significant work to be
done in the field of roof design for life
safety issues. There is limited data
and research related to the buildup
of water over drains, which is a critical
component related to the determination
of water loads on roof. The
roofing industry has had a turbulent
history but has come a long way. The
challenge is to take this information
to the rest of the design community,
lead the way in research to improve
the design standards, and improve
our understanding of roof design for
future generations.
REFERENCES
1. Harry C. Brearly, The History
of the National Board of Fire
Underwriters, Fredrick A.
Stokes Company, New York,
1916.
2. Stephen Patterson, Madan
Mehta, and Richard Wagner,
Roof Drainage, Roof Consultants
Institute Foundation
Publication No. 02.03.
3. Stephen Patterson and
Madan Mehta, Wind Pres sures
on LowSlope
Roofs,
Roof Consultants Institute
Foundation Publication No.
01.01 (Revised 2005).
4. Stephen Patterson and
Madan Mehta, Roofing Design
and Practice, Prentice Hall,
2001.
5. 1915 Building Code recommended
by the National
Board of Fire Underwriters.
Patterson and Mehta 212
Proceedings of the RCI 25th International Convention