Roof Drainage Design, Roof Collapses, and the Codes

Roof Drainage Design, Roof Collapses,
and the Codes
Dr. Stephen L. Patterson, RRC, PE
Roof Technical Services, Inc.
1944 Handley Drive, Fort Worth, TX 76112
Phone: 800-256-6693 • E-mail: spatterson@rooftechusa.com
and
Dr. Madan Mehta, PE
University of Texas at Arlington
PO Box 19108, Arlington, TX 76019
Phone: 817-272-2801 • E-mail: mmehta@uta.edu
3 3 r d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma r c h 2 2 – 2 7 , 2 0 1 8 P a t t e r s o n a n d Me h t a • 1 2 1
Abstract
Every year, roofs in the United States collapse because of roof drainage-related design
issues. These collapses result in large financial losses and serious safety consequences,
including loss of life. This paper is the result of more than three decades of forensic investigations
of dozens of catastrophic roof collapses, and addresses recent changes in the codes
that have profound life-safety implications. The paper includes an in-depth discussion of
drainage design fundamentals, flaws in current and past code design standards, examples
of actual collapses, and the drainage design issues contributing to the collapses.
Speakers
Dr. Stephen L. Patterson, RRC, PE — Roof Technical Services, Inc., Fort Worth, TX
STEPHEN PATTERSON is a licensed engineer and Registered Roof
Consultant with more than 40 years of roofing industry experience,
including 34 years as a consulting engineer designing and evaluating
roofs. Patterson coauthored Roofing Design and Practice, Drainage
Design, and Wind Pressures on Low-Slope Roofs, as well as many other
technical papers and articles on roofing. He has evaluated more than
30 roof collapses.
Dr. Madan Mehta, PE — University of Texas at Arlington, Arlington, TX
MADAN MEHTA is a professor of architecture at the University of
Texas at Arlington and a licensed engineer. Dr. Mehta has authored several
books, monographs, and research papers, which include coauthoring roofrelated
books, monographs, and papers with Mr. Patterson. His 1,000-page
book on Building Construction, published by Pearson Inc., is one of the two
most widely used books by practicing architects and students of architecture
and construction engineering in North America.
1 2 2 • P a t t e r s o n a n d Me h t a 3 3 r d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma r c h 2 2 – 2 7 , 2 0 1 8
1. INTRODUCTION
Every year, there are several catastrophic
collapses of roofs in the United
States related to roof drainage, resulting in
prolonged legal proceedings that involve the
consideration of life-safety consequences
and monetary losses from property damage,
business interruptions, inventory loss, losses
to employees, and legal costs.1,2,3 Almost
all drainage-related roof collapses occur in
relatively flat (low-slope) roofs with parapet
walls that have inadequate provisions
for overflow. Roofs that drain water over
the edges of the roof into external gutters
and downspouts are not as subject to such
collapses and, hence, are not reviewed in
this paper. This paper addresses collapses
occurring in parapeted low-slope roofs with
internal roof drains and/or scuppers.
Most such collapses occur in one-story,
large-footprint, big box-type (warehouse
and retail-type) buildings, whose roofs consist
of long-span, lightweight steel framing
members, typically using open-web joists
and joist girders. Designed to the minimum
permissible code design criteria, they are
prone to collapses when the load from rainwater
accumulation on them exceeds the
design values. An example of such a collapse
is shown in Figure 1. Note that smallfootprint
buildings or reinforced concrete
frame buildings are less likely to collapse
from the accumulation of water.
There are several reasons for the collapses
just mentioned. The important ones,
discussed in greater detail in subsequent
sections, are as follows:
1. A large number of existing buildings
were built before the codes
addressed requirements related to
roof slope and overflow drains or
scuppers. Many, if not most, of these
buildings have inadequate overflow
and/or slope.
2. The steel structure is one of the most
expensive parts of a large box-type
building. Reducing steel tonnage by
increasing the spacing and spans of
framing members has a pronounced
effect on the overall cost of the building.
The reduction in steel lowers
the strength and stiffness of
components and increases the
likelihood of a collapse.
3. Low-slope roof drainage design,
though theoretically simple, is
complicated by the fact that
it involves the input of three
design professionals: the project
architect, the structural
engineer, and the plumbing
engineer (Figure 2). Educated
in disparate disciplines, few
of these professionals have a
comprehensive understanding
of drainage design and its
relationship with the building’s
structure. Therefore, although
the respective roles of each in
the design process are articulated
in practice regimes, putting
the entire design together
is not. Each design discipline
assumes that if they design their
specific part to meet the code, their
job is done and the building is safe.
In practice, therefore, there is a
general lack of communication and/
or coordination among the three
members. This can yield a faulty
design, which can be aggravated by
poor execution by the contracting
community and deterioration of the
building due to age, resulting in a
collapse under unfavorable weather
conditions.
4. It is generally forgotten that code
provisions are minimum requirements,
often arrived at through consensus
of only those stakeholders
who are present in code development
meetings. The provisions may
not be comprehensive, may ignore
important design considerations,
and often do not represent the best
of building science information.
Roof Drainage Design, Roof Collapses,
and the Codes
3 3 r d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma r c h 2 2 – 2 7 , 2 0 1 8 P a t t e r s o n a n d Me h t a • 1 2 3
Figure 1 – A typical drainage-related collapse of a low-slope roof. Photo by
Stephen Patterson.
Figure 2 – Professionals involved in the
design of drainage systems of low-slope
roofs. (This illustration applies only to
new construction. In reroofing, the design
professionals may not be involved.)
Lack of due diligence by the design
professionals to obtain appropriate
guidance from standards and other
publications can subsequently present
a serious problem.
For example, neither the
International Building Code (IBC)
nor the International Plumbing Code
(IPC) addresses the drainage flow
rates through roof drains as a function
of hydraulic head. The information
given in the IPC is the maximum
drainage capacity of roof drains of
various sizes with no reference to the
hydraulic head (Figure 3), erroneously
implying that the hydraulic head
is not a consideration in the drainage
design process.
The problem is more serious with
scuppers, as there is no information
related to scupper design in the IBC
or the IPC, which leaves it to the
designer to seek it. To the best of
the authors’ knowledge, the 2003
RCI Foundation (RCIF) monograph,
titled Roof Drainage4 is one of the
few publications that deals comprehensively
with scuppers.
5. The requirement for overflow drains
or scuppers did not appear in building
codes until the 1960s, and slope
was not addressed in them until the
1980s. Consequently, many existing
buildings have inadequate or no
overflow drains at all and are at risk
of their roofs collapsing.
6. Since the 1980s, there has been
a gradual weakening in several
requirements of the regulatory
apparatus (building code and
plumbing code provisions) related
to roof slope and overflow drainage.
For example, when roof system
replacement (referred to as “reroofing”)
takes place, it must conform
to the provisions of various codes
in force at the time of reroofing.
Because roof drainage is intimately
related to the roof system (functioning
as the carrier for rainwater), a
reroofing should automatically trigger
a scrutiny of the existing drainage
system of the building, so as to
bring it to par with the provisions of
the current building and plumbing
codes.
This, however, is not the case
today. The 2015 IBC has, for the first
time since its introduction, eliminated
the requirements for the building’s
drainage system to meet the
code’s drainage requirements when
reroofing, setting a dangerous precedent,
discussed in detail in Section
2(v).
WHAT THIS PAPER
ADDRESSES
This paper is the result of
the forensic work of its primary
author on several dozen roof collapses
over a span of 40 years,
with research collaboration provided
by the secondary author.
Its basic purpose is to highlight
the deceptive simplicity of
low-slope roof drainage design,
which can be quite complex in
practice because of its multidisciplinary
nature. The situation
has been aggravated by the regulatory
provisions not keeping
apace with the demands of the profession.
In fact, the reverse has happened as some
regulatory provisions have become increasingly
more permissive.
Therefore, the paper begins with a discussion
of drainage design provisions for
low-slope roofs. Because roof collapses more
frequently occur in buildings designed in
the past, a brief discussion of how the
drainage design provisions have evolved is
provided in the same section. A comprehensive
discussion of drainage design fundamentals
and the various parameters that
must be considered during the design development
stage are provided next, followed by
a few design examples selected from recent
roof collapses.
2. DRAINAGE DESIGN PROVISIONS
FOR LOW-SLOPE ROOFS
The basic elements of proper low-slope
roof drainage design are:
• Overflow drainage
• Roof slope
• Hydraulic head over the overflow
drains or scuppers
• Rain loads due to ponded water
• Design rainfall rates for primary
drainage and overflow drainage
• Verification that the roof structure
has been designed to carry the rain
load
• Investigation of the roof structure for
ponding instability
Overflow Drainage and Roof Slope
The earliest direct mention of overflow
drainage appeared in the 1964 Uniform
Building Code (UBC) when it required the
overflow drains or scuppers to be installed
2 in. above the low point of the roof. There
was no requirement for roof slope and no
1 2 4 • P a t t e r s o n a n d Me h t a 3 3 r d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma r c h 2 2 – 2 7 , 2 0 1 8
Figure 3 – Hydraulic head over a roof drain is an
important determinant of flow rate through it.
Figure 4 – Excerpt from Chapter 32: “Roof Construction and Coverings,” 1988
Uniform Building Code.
reference to plumbing codes or standards.
The requirement for providing roof slope
first appeared in the 1988 UBC (Section
3207), requiring a minimum ¼-in.-per-ft.
slope. The provision for overflow drains or
scuppers was also a part of the code, along
with reference to the plumbing code for sizing
the roof drains. The minimum required opening
height of scuppers was 4 in. (Figure 4).5
The provision of overflow drainage and
a minimum ¼-in.-per-ft. slope are now
universally accepted design requirements.
They are a part of the 2015 IBC for roofing
(except for reroofing, covered at the end
of this section). A ¼-in.-per-ft. slope helps
ensure rapid drainage and reduces the
probability of ponding instability.
In many ways, the 1988 UBC was a high
point for roofing and reroofing provisions
with respect to roof drainage, as it required
that all reroofing shall conform to the same
provisions of the code that are applicable to
(new) roofing, including the minimum slope
and overflow requirements. In other words,
no distinction was made between the provisions
for roofing and reroofing (Figure 5).6
The 1988 UBC also required roof inspection
before starting to reroof in addition to
a professional analysis of the roof structure
if extensive ponding of water was observed.
Inspection of the roof after reroofing was
also required. Sadly, these requirements
were deleted from the subsequent versions
of the UBC and never included in various
editions of the UBC or the International
Building Code (IBC) that followed.
The 1988 UBC did not provide any rational
procedure for determining the scupper
size except to state that the scupper opening
area must be at least three times the
roof drain area (Figure 4). It was a flawed
provision because the scupper’s opening
size is a function of the head of water at the
scupper—necessary to provide the required
flow rate. Significant and unsafe buildup
of water can easily occur on a roof using
1988 UBC criterion (see Example 2 under
the section on Determining the Depth of
Ponded Water on a Roof).
Hydraulic Head and Rain Load
The requirement for determining the
rain load—load of water accumulating
on the roof (with all primary drains
blocked)—was first introduced in the 1988
publication of the ANSI/ASCE7-88 standard.
The consideration of rain load on
low-slope roofs from ponded water is now
a standard requirement for the design
of all low-slope roofs with raised edges.
However, none of the code publications (in
their various editions) provide any design
aid or guidance for determining the depth
of ponded water.
The first such design aid appeared
in the 1994 Standard Plumbing Code
(SPC)7 and subsequently in 1995 edition of
ASCE/SEI 7-95 standard8 and remained
unchanged up to ASCE/SEI 7-10 standard,
9 but was updated in ASCE/SEI 7-16
standard (see Section 4, Table 2).
Primary Drainage, Overflow Drainage,
and Design Rainfall Rate
That the primary and overflow drainage
systems should be completely independent
of each other has been mandated by the
codes since 1964. Each system was to be
designed using the maximum of one-hour
rainfall with a mean return period (MRP)
of 100 years. However, the 1991 SPC made
a significant change by requiring that the
overflow drainage system be designed for
15-minute rainfalls with a 100-year MRP.
The 15-minute, 100-year MRP rainfall is
approximately twice the one-hour, 100-
year MRP rainfall, providing the necessary
safety provision against roof collapses (see
Section 7).
The first International Plumbing Code
(IPC), published in 1995,10 required that the
drainage capacities of roof drains given in
IPC tables be divided by a factor of two for
the design of overflow systems. This provision
effectively doubled the design rainfall
rate for overflow drainage, making it virtually
identical to the 1991 SPC provision.
Unfortunately, the IPC, which became the
governing plumbing code after the merger of
all three legacy codes into the International
Code Council (ICC) in 2000, eliminated the
effective doubling of design rainfall rate for
overflow drainage design. The current (2015)
IPC requires the overflow drainage system
to be designed for the same rainfall as the
primary drainage system (one-hour rainfall
with a 100-year MRP).
Based on the analysis of several roof
collapses and the study of hydrological
cycles, the authors had recommended the
use of 15-minute, 100-year MRP rainfall for
overflow drainage design in the monograph
on Roof Drainage, published by the RCI
Foundation (RCIF) in 2003.11
3 3 r d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma r c h 2 2 – 2 7 , 2 0 1 8 P a t t e r s o n a n d Me h t a • 1 2 5
Figure 5 – Excerpt from Appendix Chapter 32: “Reroofing,” 1988 Uniform Building
Code.
Figure 6 – Excerpt from ASCE/SEI 7-10, Minimum Design Loads for Buildings and
Other Structures, Chapter 8, Rain Loads.
Ponding Instability
Ponding instability is defined as the
progressive increase in the accumulation
(ponding) of water on the roof due to the
lack of sufficient stiffness in roof framing.
As the ponded water exerts load on the roof,
the roof deflects, leading to greater accumulation
of water, which further increases the
roof’s deflection. As the deflection increases,
more water accumulates on the roof,
increasing the deflection further, and so
on—leading to the roof’s ultimate collapse.
Note that ponding instability may also
occur from the accumulation of snow or the
combined effects of snow and rain.
The consideration of ponding instability
has been a part of the codes and standards
for a long time. As previously indicated, IBC
and ASCE/SEI 7 standards do not require
the investigation of ponding instability for
roofs with a slope greater than or equal
to ¼ in. per ft. (Figure 6).12 As shown in
Section 6, this carte blanche assumption is
incorrect. Therefore, roofs designed for slope
equaling or exceeding ¼ in. per ft. may need
to be checked for ponding instability.
Code Provisions for Reroofing
As shown in Figure 5, at one time, the
drainage-related code provisions for reroofing
were the same as for roofing (including
those for overflow drainage and roof slope).
However, gradually, the reroofing provisions
have been watered down. Several years ago,
the requirement for a minimum roof slope
(¼ in. per ft.) was eliminated and replaced
by the requirement that the existing roof
should provide positive drainage.
Positive drainage is defined in 2015
IBC (Section 202) as “the drainage condition
in which consideration has been made
for all loading deflections of the roof deck,
and additional slope has been provided to
ensure drainage of roof within 48 hours of
precipitation.” Water standing on a roof for
two days is the definition of “poor drainage,”
not “positive drainage.”
Eliminating the requirement for ¼-in.-
per-ft. slope has been a retrograde step
because:
1. The criterion is imprecise and has
no relationship to good drainage.
2. Water should drain freely and quickly
and not stand on a roof for an
extended period of time, let alone
two days.
3. Hardly any roofing contractor will
test the roof (before reroofing) by
flooding it,
then waiting
for 48 hours
to observe
the areas
where the
water is still
present.
4. There is no
mandate in
the code for
a third-party
inspection of
the process
and how the
faulty situation
is to be
corrected.
5. A roof could be dead flat, surrounded
by parapet walls without
any overflow drainage, and could
still meet the requirement of positive
drainage, while remaining highly
prone to ponding instability and collapse.
A more serious degrading of reroofing
provision occurred with the 2015 IBC, which
deleted the requirement for the overflow
drainage if the existing roof was not previously
provided with one. Additionally, if the
overflow drainage exists and is below the
current code, an upgrade is not required.
Reroofing provisions, as they exist in the
2015 IBC, are illustrated in Figure 7.13
3. THE BASIC PROBLEM
The problem is that many, if not most,
existing buildings were built either with no
overflow drainage or an inadequate overflow.
It is a serious design and construction
defect that has the potential for catastrophic
consequences. The most logical time to
correct the situation in such cases is when
the roof is replaced.
The argument made against making
the correction is that if the building has
performed well during all its previous years
and even decades, it will perform well in
the future as well. The argument is made
not only by laypersons but also by some
architects, engineers, and even by the code
officials, who are supposedly the guardians
of ensuring health, safety, and welfare in
buildings.
One of the first collapses the primary
author investigated was a 30-year-old grocery
store with scuppers as the primary
drainage and no overflow drainage. In 30
years, there was no problem until someone
threw a Fort Worth Star-Telegram Sunday
newspaper on the roof. The newspaper
floated into the scupper, forming a perfect
plug, resulting in a catastrophic collapse.
Fortunately, no one was hurt.
There is a chance that had it not been
for the newspaper, the buildings would
have never collapsed, but it happened. As
consultants and designers, we cannot (and
are not permitted to) rely on chance. In
another grocery store collapse investigated
by the primary author, blocked scuppers
caused a collapse that claimed two lives.
(Unfortunately, the conditions that led to
these failures are now permitted in the
Reroofing section of the 2015 IBC.)
The primary author recently inspected
the collapse of a large warehouse facility
in the Dallas area that was constructed in
the early 1980s. The roof had 1/8-in.-perft.
slope, which met the code requirements
at the time it was constructed. There was
no significant ponding of water. The roof
drained freely and exceeded the requirements
for “positive drainage.” The roof
drains were slightly oversized per the design
requirements of the UBC in force at the time
of its construction; i.e., it was overdesigned
per the requirements of 2015 IBC.
The overflow drainage was provided
through scuppers, which were sized based
on the then-applicable UBC provision,
requiring that scupper opening area should
be at least three times the area of the roof
drains (see Section 2, Overflow Drainage
and Roof Slope.) The evidence suggested that
the roof drains were blocked with debris.
Our calculation indicated that, assuming
the drains were blocked, the rain load
1 2 6 • P a t t e r s o n a n d Me h t a 3 3 r d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma r c h 2 2 – 2 7 , 2 0 1 8
Figure 7 – Comparison of the requirements for overflow
drainage and roof slope for roofing and reroofing in 2015
International Building Code [Ref. 13].
on joists would be in excess of two times the
typical design live load for the joists. These
calculations did not take into account the
additional load that could possibly occur
from positive wind load on the roof. In other
words, a roof drainage design can meet the
code in force at the time when the building
was built, but the roof can still collapse.
4. DETERMINING THE DEPTH OF
PONDED WATER ON A ROOF
The determination of rain load on a roof
requires calculating the depth of ponded
water. This must be preceded by the design
of both the primary and overflow drainage
systems. The drainage system design is
based on one-hour, 100-year MRP rainfall
for the location and the use of IPC table for
the drainage capacities of drains of various
sizes, shown in Table 1.14
The process just described will be illustrated
using two examples. In Example 1,
both primary and overflow drainage systems
consist of roof drains. In Example 2, the primary
system consists of roof drains and the
secondary system consists of scuppers.
Example 1
In consultation with the architect, the
project’s plumbing engineer has prepared
the layout of roof drains for a 300-ft. x
450-ft. distribution center (Figure 8). The
roof slopes ¼ in. per ft. on either side of a
central ridge, and the drains are located 90
ft. on center along the 450-ft.-long parapets
(five drains next to each parapet)—a
total of ten drains on the roof. It has been
decided to use a side-by-side combination
of primary and overflow drains, with inlet
of the overflow drain elevated 2 in. above
that of the primary drain, using an overflow
collar dam.
The architect has asked the plumbing
engineer to provide 1) primary and overflow
drain sizes conforming with 2015 IPC and
2) the depth of ponded water on the roof
when the primary drains are blocked. The
one-hour, 100-year MRP rainfall for the
location is 4 in.
Plumbing Engineer’s Solution
Total roof area =
300 x 450 ft. = 135,000 sq. ft.
Total rainfall on roof in 60 minutes =
135,000 sq. ft. (4 in.) = 45,000 cft
Total rainfall on roof in 1 minute =
(45,000/60) = 750 cft =
750 x 7.48 = 5610 gallons
(Note: 1 cft = 7.48 gallons.)
Because the roof contains ten drains,
the minimum required flow rate of each
drain = (5610/10) = 561 gpm.
From Table 2,15 the primary drainage
system will comprise 6-in.-diameter
drains. The flow rate of each drain = 563
gpm > 561 gpm (minimum required flow
rate).
The 2015 IPC requires that the overflow
drains have the same flow rate as
the primary drains. Therefore, the overflow
roof drains will also be 6 inches in
diameter.
While the 2015 IPC provides the flow
rates of roof drains, it does not provide
the head of water that must exist over the
drain to produce
that flow rate. To
determine the head
of water corresponding
to the flow rate,
we refer to ASCE/
SEI 7-16 Standard
data, given in Table
1. From this table,
the head of water for
the flow rate of 561
gpm is approximately
5.5 in.
Thus, the total
depth of water on the
roof when the primary
system is blocked
= s tatic h ead +
hydraulic head = 2.0
+ 5.5 = 7.5 in. (Figure
9). This information
is sent to the project
architect for onward
transmission to the
structural engineer
for determining the
rain load and the
design of the roof
assembly. The weight
of water corresponding
to a depth of 7.5
in. = (5.2)(7.5) = 39.0
psf.
(Note that this
example does not
provide the entire
drainage solution, as
the remaining part
is not relevant to this paper; i.e., the design
of below-deck drainage elements, such as
tail pipes, horizontal pipes, and the conductors.)
3 3 r d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma r c h 2 2 – 2 7 , 2 0 1 8 P a t t e r s o n a n d Me h t a • 1 2 7
Table 1 – Maximum flow rate (drainage
capacities) of roof drains in gallons per
minute.
Table 2 – Flow rate through a 6-in.-diameter roof drain as a
function of hydraulic head.
Figure 8 – Roof plan of the building in Example 1. RD is the
acronym for “roof drains”—in this case, a set of primary and
overflow drains.
1 2 8 • P a t t e r s o n a n d Me h t a 3 3 r d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma r c h 2 2 – 2 7 , 2 0 1 8
Example 2
The project architect of the building of
Example 1 has asked the plumbing engineer
to provide an alternative drainage
solution in which the overflow drainage will
be provided by scuppers located close to
each primary drain.
Plumbing Engineer’s Solution
Minimum required flow rate of each scupper
= 561 gpm
The flow rate of a scupper is given by the
following equation:
Q = 2.9 (L) H1.5
Equation 1
Where Q = flow rate through scupper
(gpm), L = length of scupper opening (in.), and
H = head of water (in.)
Setting Q = 561 and L = 18 inches in
Eq. (1), we obtain H = 4.87 in., (say 5.0 in.).
Because the inlet level of the scupper is
raised 2 in. above the primary drain, the
head of water at its lowest point = 2.0 + 5.0 =
7.0 in. This information is sent to the architect.
The weight of water (at the lowest point
on roof) corresponding to a depth of 7.0 in. =
(7.0)(5.2) = 36.4 psf.
The scupper opening size = 18 in. x 6 in.
(Note: A minimum 1-in. clearance is required
above the head of water.) The total scupper
opening area = 108 sq. in. Note that the scupper
opening height of 6 in. is greater than the
4-in. minimum required by 2015 IPC.
Authors’ Observations
It is the authors’ experience that the
design process illustrated in the given examples
seldom occurs in practice. In theory,
the plumbing engineer should calculate
the head of water
above the overflow
drains or overflow
scuppers and submit
this information
to the architect
who, after reviewing
it, would send
it to the structural
engineer. However,
the code places
the responsibility
on the structural
engineer (who is
typically unfamiliar
with plumbing
design and the
issues discussed in this paper) to verify that
the structure will support the load from rainwater
accumulation.
In both examples, the
rain load at the lowest point
on the roof (39.0 psf in
Example 1 and 36.4 psf in
Example 2) are greater than
the roof’s design live load
of 20 psf. They are much
greater if the live load reduction
has been assumed by
the project’s structural
engineer.
It is worth pointing out
that it is difficult to locate
scuppers 2 in. above the low
point of a roof because of crickets and other
variations in rooftop elevation (Figure 10).
Consequently, the scuppers are typically located
4 in. above the roof’s low point, which further
increases the depth of water. In the case
of Example 2, this will give a ponding depth of
9.0 inches in place of 7.0 inches, increasing the
weight of water at the lowest point of the roof to
46.8 psf in place of 36.4 psf.
Returning to the 1988 UBC’s (arbitrary)
provision that the overflow scupper opening
area be three times the primary roof drain
area, we see that in Example 2, the scupper
opening area = 108 sq. in. The area of each
6-in.-diameter roof drain = 28.27 sq. in.
Hence, three times the area of roof drains
= 84.81 sq. in., which is well below that
obtained from rational analysis (108 sq. in.)
in Example 2.
Figure 9 – Hydraulic head and static head over the overflow
drains of the building of Example 1 (not to scale).
Figure 10 – Because of crickets and tapered insulation
near a parapet wall, the typical difference between the
inlet levels of scupper and the primary roof drain is 4
in. or greater.
Figure 11 – Effect of roof slope on rain load on a roof. The maximum rain load in
all three illustrations is 39.0 psf. The rain load distribution shown in illustration
(a) relates to the building of Example 1.
3 3 r d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma r c h 2 2 – 2 7 , 2 0 1 8 P a t t e r s o n a n d Me h t a • 1 2 9
5. IMPORTANCE OF ROOF SLOPE
To know the depth of water at the low
point of the roof is the first step in determining
the rain load on the roof. The next
step is to account for the roof’s slope, which
affects the total rain load on the roof and
its distribution. Figure 11(a) illustrates the
distribution of load on the roof of Example
1. Assuming that the joist span is 45 ft., the
rain load extends to a length of 30.0 ft. over
the first joist.
Note that the total load on the first joist
of Figure 11(a) is [0.5(39.0)30] = 585 pounds,
which is equivalent to a uniform load of 19.5
psf over a 30-ft. length of joist. The high
concentration of load near the parapet may
cause deflection-related distress in the deck
and local failure of the joist, but is not likely
to cause ponding instability because the
deflection of the joist should normally not
exceed the allowable deflection.
If roof slope is 1/8 in. per ft., the submerged
area of roof is 60.0 ft. long (Figure
11(b)). In this case, the rain load on the roof
extends over two joists and is twice that of
Figure 11(a). The average load on the first
joist = 0.5[39 + 39(15/60)] = 24.4 psf, which
exceeds the design load of 20 psf, indicating
a fair probability of ponding instability in a
framing system designed to the minimum
structural design provisions of the code.
If the roof were dead flat, the entire roof
would be submerged in water (Figure 11(c)).
In this case, the entire roof is under a load
of 39.0 psf. This is 95% greater than the
live load of 20 psf and 144% greater than
16 psf (if live load reduction was assumed
in the design of the joists). This roof is the
most likely candidate for ponding instability
failure, unless its structural framing
has been designed with adequate stiffness
to prevent it.
The illustrations in Figure 11 highlight
the importance of roof slope in the structural
design of the building for ponding considerations.
They also explain why the building
code historically required the roof structure
to be analyzed for ponding instability if the
roof slope was less than ¼ in. per ft.
6. ORIENTATION OF STRUCTURAL
FRAMING AND DIRECTION OF
ROOF SLOPE
Figure 11(a) shows the rain load on the
roof of Example 1, where the secondary
framing members of the roof (members
that provide direct support to the deck, i.e.,
the joists) are oriented in the direction of
roof slope. Figure 12(a) shows the rain load
distribution of the same building, but the
secondary framing members (joists) are oriented
perpendicular to roof slope.
Note that although the rain load distribution
and the total rain load on the roofs
of Figure 11(a) and Figure 12(a) are identical,
there is great difference in their structural
implications. In Figure 12(a), the average
rain load on the first joist is 32.5 psf along
the entire span of the joist—much higher
than the roof’s design load. This situation
is similar to the joist of the dead-flat roof
of Figure 11(c), and hence prone to ponding
instability. (Note: In Figure 12(a), we have
assumed that the joists are spaced 5 ft. on
center—typical for roofs with steel deck and
steel joists.)
Figure 12(a) shows that the building
code provision stating that a roof with a
slope ≥¼ in. per ft. is not required to be
investigated for ponding instability is not
always correct. This observation is further
endorsed by Figure 12(b), where the joists
from opposite directions are supported by
a joist girder forming a valley, creating a
possibility of substantial overload on the
joist girder.
This situation is particularly serious
because the tributary area of a typical joist
girder is so large that it qualifies for the
roof live load reduction of up to 40% by the
building codes—from 20 to 12 psf. This can
result in the design of joist girders that are
highly deficient in stiffness and strength
to support the weight of ponded water that
may exceed the load for which they have
been designed by 100% to 200%.
7. DESIGN RAINFALL RATE
The discussion and the examples provided
thus far are based on the design rainfall
rate as the maximum one-hour rainfall
with an MRP of 100 years for both primary
and overflow drainage. The design rainfall
rate assumes that it is uniform within the
entire one-hour duration. In other words,
the assumption is that if the 100-year MRP
rainfall at a location is 4 in. per hour, that
location will receive 1 in. rainfall every 15
minutes, or 0.5 in. every 7.5 minutes, or
(4/60) in. = 0.067 in. per minute.
The actual rainfall seldom occurs at a
uniform rate, particularly during thunderstorms,
tropical storms, and hurricanes.
In such situations, bursts of rainfall may
occur in short durations, but the one-hour
rainfall may be the same as the design rainfall.
Therefore, there is a strong rationale
for using a higher design rainfall rate for
overflow drainage.
Note that the primary drainage system
is designed to drain water off the roof
within a reasonable time. There are no life
safety issues related to the primary drainage
design. Therefore, designing the primary
drainage system assuming a uniform
rainfall rate is fine. The overflow drainage
system, on the other hand, is the safety
valve—to prevent unsafe accumulation of
water on the roof. A strong rationale therefore
exists that the overflow drainage design
should account for the bursts of rainfall
within short durations.
Figure 12 – Effect of the orientation of roof framing members on rain load.
1 3 0 • P a t t e r s o n a n d Me h t a 3 3 r d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma r c h 2 2 – 2 7 , 2 0 1 8
Hydrological studies16 have shown that
a location can get half its one-hour rainfall
in 15 minutes. Thus, if a location receives
a 100-year MRP rainfall of 4 inches in one
hour, it can receive up to 2 inches of rainfall
in 15 minutes—a rainfall rate of 8 in. per
hour. Figure 13 illustrates this narrative.
In addition to the high rainfall rate over
a short duration that can overload the roof,
hailstorms are another problematic event
for a drainage system. Small hail is particularly
problematic as it can easily block
or impede the flow. Because hail sometimes
occurs with severe thunderstorms, one can
expect a large amount of rainfall coupled
with hail, increasing the probability of
drain blockages.
The accumulation of debris on roofs is
another problem related to roof drainage.
Serious and frequent blockages of drains
and scuppers from the accumulation of
debris has been reported by investigators.17
The types of debris found on roofs includes
dirt, leaves, plastic bags, paper, soda cans,
bottles, and so on. Although regular roof
observation and maintenance can prevent
this problem, it is neither enforceable nor
practical.
Designing the overflow drainage with
15-minute-per-hour rainfall rate is an
insurance, not only against nonuniform
rainfall rate, but also against blockage
caused by hail, as well as debris accumulation.
As stated in Section 2 under “Ponding
Instability,” the 1991 SPC and 1995 IPC
required the overflow drainage design to be
based on 15-minute, 100-year MRP rainfall
rates. There is a need to revert to that
provision in future editions of the plumbing
codes.
8. AUTHORS’ INVESTIGATIONS
AND ASCE/SEI 7-16 STANDARD
In Section 6, examples of how lowslope
roofs of large-footprint, big-box-type
structures can be substantially overloaded
by ponded water and the resultant ponding
instability, are given. Historically, the
evaluation for ponding instability has been
required on roofs with slopes ≤¼ in. per
foot. For example, ASCE/SEI 7-05 Standard
required: “Roofs with a slope of ¼ in./
ft. (1.19°) shall be investigated for structural
analysis to assure that they possess
adequate stiffness to preclude progressive
deflection (i.e., instability) as rain falls on
them or meltwater is created from snow on
them.”
The above provision was simplified in
ASCE/SEI 7-10 Standard (see Figure 6)
by requiring that “Susceptible bays shall
be investigated by structural analysis to
assure that they possess adequate stiffness
to preclude progressive deflection (i.e.,
instability) as rain falls on them or meltwater
is created from snow on them. …Roof
surfaces with a slope of at least ¼ in. per ft.
(1.19°) towards points of free drainage need
not be considered as susceptible bays.”
The 2015 IBC refers to ASCE/SEI 7-10,
published in 2010. ASCE/SEI 7-10 has now
been replaced by ASCE/SEI 7-16, published
in July 2017, and will be referenced in the
2018 IBC.
It is important to mention that the
provisions of ASCE/SEI 7-16 for ponding
instability analysis are far more stringent
than those given in the standard’s previous
editions and agree with the authors’ investigations,
summarized in Section 6. This
is a positive vindication of the many years
of the authors’ work on various collapses.
ASCE/SEI 7-16 requires ponding instability
analysis for the following conditions:
1. Bays with a roof slope less
than ¼ in. per foot (1.19°)
when the secondary members
are perpendicular to the
free draining edge,
2. Bays with a roof slope less
than 1 in. per foot (4.76°)
when the secondary members
are parallel to the free
draining edge,
3. Bays with a roof slope less
than 1 in. per foot (4.76°)
and a span-to-spacing ratio
for the secondary members
greater than 16 when the secondary
members are parallel
to the free drainage edge, or
4. Bays on which water accumulates
(in whole or in part)
when the primary drain system
is blocked but the overflow
drain system is functional.
The larger of the snow load
or the rain load equal to the
design condition for a blocked
primary drain system shall
be used in this analysis.
ASCE/SEI 7-16 has also recognized
the importance of a higher design rainfall
rate for overflow drainage, as it now
requires the 15-minute, 100-year rainfall
rate. Section 8.2 of the standard states:
“The design flow rate of the secondary
(overflow) drains (including roof drains and
downstream piping) or scuppers and their
resulting hydraulic head (dh) shall be based
on a rainfall intensity equal to or greater
than the 15-min duration/100-year return
period (frequency) storm. Primary drainage
systems shall be designed for a rainfall
intensity equal to or greater than the
60-min duration/100-year return period
(frequency) storm.”
As the current IBC and IPC requirement
for overflow drainage design are still
based on 60-minute, 100-year MRP rainfall,
the authors hope that in the 2018 editions
of IBC and IPC, it will be revised to
that required by ASCE/SEI 7-16. This will
substantially reduce the potential for roof
collapses. Not doing so will be tantamount
to ignoring the expertise of the two major
organizations—the American Society of
Civil Engineers (ASCE) and the Structural
Engineering Institute (SEI)—that develop
ASCE/SEI 7 with the help of several hundred
engineering experts.
9. CONCLUSION
Roof drainage design is one of the most
important roof design elements, and the
overflow drainage design is its most critical
part. The function of the overflow drainage
is to prevent the roof from collapsing—an
important life safety issue in roofs.
ASCE/SEI 7-16 has recognized the deficiencies
in drainage design with respect to
ponding instability and has made major
revisions from its previous edition, which,
when implemented, will dramatically
reduce the potential for roof collapses. This
Figure 13 – Maximum 15-minute, 100-
year MRP rainfall is approximately
half of one-hour, 100-year MRP rainfall,
implying that the rainfall rate in 15
minutes is twice the design rainfall
rate.
3 3 r d RC I I n t e r n a t i o n a l C o n v e n t i o n a n d T r a d e S h ow • Ma r c h 2 2 – 2 7 , 2 0 1 8 P a t t e r s o n a n d Me h t a • 1 3 1
is a major step forward at a time when the
IBC has been moving in the opposite direction.
The standard has also recognized
the importance of the 15-minute duration
rainfall rate for overflow drainage design.
The ICC should re-evaluate the drainage
design requirements in the IBC and IPC
and provide appropriate provisions that are
in compliance with ASCE 7-16 to ensure
public safety.
This is equally important for reroofing
because there is no reason why the IBC’s
requirements for reroofing should not be
the same as for roofing in new construction.
The costs involved to add overflow drainage
at the time of reroofing (if it does not
already exist) or to modify it to comply with
the current code provisions for roofing are
relatively insignificant compared with the
monetary losses (not counting the injuries
and fatalities) that may occur as a result
of the collapses. Many existing buildings
needlessly fall in the impending-collapse
category, which can be easily prevented
at the time of reroofing through overflow
drainage. Figure 14 summarizes the cost
data case studies from two of the several
collapses investigated by the primary
author in support of this statement.
Additionally, there is a real problem
with the definition and use of the term
“positive drainage” in the IBC with respect
to reroofing. Positive drainage as defined in
the IBC is a poor and unworkable definition.
By deleting the relationship between drainage
efficiency and minimum ¼ in. per ft.
slope, the IBC has placed the roofs of many
existing buildings at more serious risk.
Fundamentally, any roof that has drainage
issues—including, but not limited to
the lack of appropriate slope or the lack of
adequate overflow—should be evaluated by
a design professional when a building is
reroofed, in the same way as required for
roofing.
ACKNOWLEDGMENTS
The authors gratefully acknowledge
the following experts for their meticulous
review of this paper: Thomas Smith, AIA, TL
Smith Consulting, Inc.; Scott Hinesley, PE,
REI Engineers, Inc.; and William Waterston,
AIA, WJE Associates, Inc.
REFERENCES
1. Stephen Patterson. Forensic consulting
related to roof collapses and
private communications with several
leading forensic consultants
engaged in the same activity over a
period of 40 years (unpublished).
2. J. Vamberskey. “Roof Failures
Due to Ponding—A Symptom of
Underestimated Development.”
HERON, Vol. 51, No. 2/3. Delft
University of Technology and
Corsmit Consulting Engineers.
2006. pp. 83-96.
3. John Lawson. “Roof Drainage—Not
My Problem…May Be,” Proceedings
of Structural Engineers Association
of California (SEAOC) Convention.
2012. pp. 136-151.
4. Stephen Patterson and Madan Mehta.
Roof Drainage. Roof Consultants
Institute Foundation (RCIF)
Publication No. 02-03. 2003. p. 34.
5. International Council of Building
Officials. “Roof Construction and
Roof Coverings.” Uniform Building
Code. Chapter 32. 1988. p. 617.
6. Vamberskey. Op. cit. Chapter 32,
Reroofing. p. 855.
7. Standard Plumbing Code. Plumbing-
Heating-Cooling Contractors (PHCC)
Association. Section 1109.3. 1994. p.
141.
8. American Society of Civil Engineers/
Structural Engineering Institute.
ASCE/SEI 7-95 Standard, Minimum
Design Loads for Buildings and
Other Structures. 1995. p. 187.
9. American Society of Civil Engineers/
Structural Engineering Institute.
ASCE/SEI 7-10 Standard, Minimum
Design Loads for Buildings and
Other Structures. 2010. p. 452.
10. International Code Council.
International Plumbing Code. Section
1108.3. 1995. p. 358.
11. Patterson and Mehta. p. 73.
12. ASCE 7-10. p. 43.
13. American Society of Civil Engineers/
Structural Engineering Institute.
ASCE/SEI 7-05 Standard, Minimum
Design Loads for Buildings and
Other Structures. 2005. p. 95.
14. International Code Council.
International Plumbing Code. 2015.
p. 94.
15. American Society of Civil Engineers/
Structural Engineering Institute.
ASCE/SEI 7-16 Standard, Minimum
Design Loads and Associated
Criteria for Buildings and Other
Structures. July 2017. p. 509.
16. Patterson and Mehta. p 73.
17. Jim Koontz. “The Effects of Debris
on the Flow Rates of Roof Drains
and Scuppers.” Proceedings of the
RCI 25th International Convention.
RCI, Inc. 2010. pp. 149-156.
Figure 14 – Cost of correcting/modifying roof drainage system as a function of
monetary loss from roof collapses. Note the data does not include life safety
consequences from collapses.